1 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3 * machmode.h (decimal_float_mode_for_size): New function.
4 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14 (get_builtin_sync_mode): Likewise.
15 (expand_ifn_atomic_compare_exchange): Likewise.
16 (expand_builtin_atomic_clear): Likewise.
17 (expand_builtin_atomic_test_and_set): Likewise.
18 (fold_builtin_atomic_always_lock_free): Likewise.
19 * calls.c (compute_argument_addresses): Likewise.
20 (emit_library_call_value_1): Likewise.
21 (store_one_arg): Likewise.
22 * combine.c (combine_instructions): Likewise.
23 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
24 * config/arm/arm.c (arm_function_value): Likewise.
25 (aapcs_allocate_return_reg): Likewise.
26 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
27 * config/i386/i386.c (construct_container): Likewise.
28 (ix86_gimplify_va_arg): Likewise.
29 (ix86_expand_sse_cmp): Likewise.
30 (emit_memmov): Likewise.
31 (emit_memset): Likewise.
32 (expand_small_movmem_or_setmem): Likewise.
33 (ix86_expand_pextr): Likewise.
34 (ix86_expand_pinsr): Likewise.
35 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
36 * config/microblaze/microblaze.c (microblaze_block_move_straight):
38 * config/mips/mips.c (mips_function_value_1) Likewise.
39 (mips_block_move_straight): Likewise.
40 (mips_expand_ins_as_unaligned_store): Likewise.
41 * config/powerpcspe/powerpcspe.c
42 (rs6000_darwin64_record_arg_advance_flush): Likewise.
43 (rs6000_darwin64_record_arg_flush): Likewise.
44 * config/rs6000/rs6000.c
45 (rs6000_darwin64_record_arg_advance_flush): Likewise.
46 (rs6000_darwin64_record_arg_flush): Likewise.
47 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
48 (sparc_function_value_1): Likewise.
49 * config/spu/spu.c (adjust_operand): Likewise.
50 (spu_emit_branch_or_set): Likewise.
51 (arith_immediate_p): Likewise.
52 * emit-rtl.c (gen_lowpart_common): Likewise.
53 * expr.c (expand_expr_real_1): Likewise.
54 * function.c (assign_parm_setup_block): Likewise.
55 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
56 * reload1.c (alter_reg): Likewise.
57 * stor-layout.c (mode_for_vector): Likewise.
58 (layout_type): Likewise.
60 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
62 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
63 (spu_convert_move): Likewise.
64 * lower-subreg.c (resolve_simple_move): Likewise.
66 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
69 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
70 define_insn to a define_expand.
71 (altivec_vsum2sws_direct): New define_insn.
72 (altivec_vsumsws): Convert from a define_insn to a define_expand.
74 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
76 * config/arm/arm.c (arm_option_params_internal): Improve setting of
79 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
84 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
85 realign stack if stack alignment needed is less than incoming
88 2017-09-05 Marek Polacek <polacek@redhat.com>
91 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
94 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
96 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
98 2017-09-05 Richard Biener <rguenther@suse.de>
100 PR tree-optimization/82084
101 * fold-const.c (can_native_encode_string_p): Handle wide characters.
103 2017-09-05 Richard Biener <rguenther@suse.de>
105 PR tree-optimization/82102
106 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
108 2017-09-05 Martin Liska <mliska@suse.cz>
110 PR tree-optimization/82032
111 * tree-cfg.c (generate_range_test): New function.
112 * tree-cfg.h (generate_range_test): Declared here.
113 * tree-cfgcleanup.c (convert_single_case_switch): New function.
114 (cleanup_control_expr_graph): Use it.
115 * tree-switch-conversion.c (try_switch_expansion): Remove
117 (emit_case_nodes): Use generate_range_test.
119 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
122 * config/i386/i386.md (*<btsc><mode>_mask): Add
123 TARGET_USE_BT to insn constraint.
124 (*btr<mode>_mask): Ditto.
126 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
128 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
129 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
131 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
134 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
135 TARGET_NEON and TARGET_IWMMXT.
136 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
137 TARGET_NEON and TARGET_IWMMXT.
138 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
140 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
142 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
143 (ix86_rewrite_tls_address): Ditto.
144 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
145 (ix86_rewrite_tls_address_1): Ditto.
146 (ix86_rewrite_tls_address): Ditto.
147 * config/i386/predicates.md (tls_address_pattern): New predicate.
148 * config/i386/i386.md (TLS address splitter): New splitter.
150 2017-09-04 Richard Biener <rguenther@suse.de>
152 PR tree-optimization/82084
153 * fold-const.h (can_native_encode_string_p): Declare.
154 * fold-const.c (can_native_encode_string_p): Factor out from ...
155 (native_encode_string): ... here.
156 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
157 vectorizing stores from constants we later cannot handle.
159 2017-09-04 Marek Polacek <polacek@redhat.com>
162 * doc/invoke.texi: Update -Wtautological-compare documentation.
164 2017-09-04 Jeff Law <law@redhat.com>
166 PR tree-optimization/64910
167 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
168 swap the first and last operand if the last is a constant.
170 2017-09-04 Marek Polacek <polacek@redhat.com>
173 * convert.c (do_narrow): When sanitizing signed integer overflows,
174 bail out for signed types.
175 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
177 2017-09-04 Richard Biener <rguenther@suse.de>
179 PR tree-optimization/82060
180 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
181 Move devirtualization after stmt folding and before EH/AB/noreturn
182 cleanup to get the stmt refs canonicalized. Use a bool instead
183 of gimple_modified_p since that doesn't work for NOPs. Schedule
184 NOPs generated by folding for removal.
186 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
187 Alan Hayward <alan.hayward@arm.com>
188 David Sherwood <david.sherwood@arm.com>
190 * coretypes.h (pad_direction): New enum.
191 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
192 (FUNCTION_ARG_PADDING): Likewise.
193 * target.def (function_arg_padding): New hook.
194 * targhooks.h (default_function_arg_padding): Declare.
195 * targhooks.c (default_function_arg_padding): New function.
196 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
197 (TARGET_FUNCTION_ARG_PADDING): ...this.
198 * doc/tm.texi: Regenerate.
199 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
200 instead of direction.
201 (compute_argument_addresses): Likewise.
202 (load_register_parameters): Likewise.
203 (emit_library_call_value_1): Likewise.
204 (store_one_arg): Use targetm.calls.function_arg_padding instead
205 of FUNCTION_ARG_PADDING.
206 (must_pass_in_stack_var_size_or_pad): Likewise.
207 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
208 (emit_group_store): Likewise.
209 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
210 instead of FUNCTION_ARG_PADDING.
211 (emit_push_insn): Likewise, and propagate enum change throughout
213 * function.h (direction): Delete.
214 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
216 * function.c (assign_parm_find_stack_rtl): Likewise.
217 (assign_parm_setup_block_p): Likewise.
218 (assign_parm_setup_block): Likewise.
219 (gimplify_parameters): Likewise.
220 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
221 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
223 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
224 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
225 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
226 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
227 (aarch64_function_arg_padding): ...this new function.
228 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
229 (TARGET_FUNCTION_ARG_PADDING): Redefine.
230 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
231 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
232 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
233 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
234 (arm_pad_arg_upward): Replace with...
235 (arm_function_arg_padding): ...this new function.
236 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
238 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
239 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
240 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
241 (ia64_hpux_function_arg_padding): Replace with...
242 (ia64_function_arg_padding): ...this new function. Use pad_direction
243 instead of direction. Check for TARGET_HPUX.
244 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
245 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
246 (iq2000_function_arg_padding): New function.
247 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
248 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
249 (mips_function_arg_padding): ...this new function.
250 (mips_pad_reg_upward): Update accordingly.
251 (TARGET_FUNCTION_ARG_PADDING): Redefine.
252 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
253 targetm.calls.function_arg_padding.
254 (FUNCTION_ARG_PADDING): Delete.
255 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
256 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
257 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
258 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
259 (nios2_block_reg_padding): Return pad_direction instead of direction.
260 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
261 instead of direction.
262 (nios2_function_arg_padding): Likewise. Make static.
263 (TARGET_FUNCTION_ARG_PADDING): Redefine.
264 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
265 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
266 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
267 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
268 (pa_function_arg_padding): Make static. Return pad_direction instead
270 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
271 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
272 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
273 instead of direction. Use targetm.calls.function_arg_padding.
274 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
275 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
276 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
277 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
278 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
280 (function_arg_padding): Rename to...
281 (rs6000_function_arg_padding): ...this. Make static. Return
282 pad_direction instead of direction.
283 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
284 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
285 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
286 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
287 instead of direction. Use targetm.calls.function_arg_padding.
288 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
289 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
290 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
291 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
292 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
293 (function_arg_padding): Rename to...
294 (rs6000_function_arg_padding): ...this. Make static. Return
295 pad_direction instead of direction.
296 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
297 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
298 * config/s390/s390.c (s390_function_arg_padding): New function.
299 (TARGET_FUNCTION_ARG_PADDING): Redefine.
300 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
301 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
302 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
303 (function_arg_padding): Rename to...
304 (sparc_function_arg_padding): ...this. Make static. Return
305 pad_direction instead of direction.
306 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
307 * config/spu/spu.c (spu_function_arg_padding): New function.
308 (TARGET_FUNCTION_ARG_PADDING): Redefine.
309 * system.h (FUNCTION_ARG_PADDING): Poison.
311 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * target.def (modes_tieable_p): New hook.
316 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
317 (TARGET_MODES_TIEABLE_P): ...this.
318 * doc/tm.texi.in: Regenerate.
319 * hooks.h (hook_bool_mode_mode_true): Declare.
320 * hooks.c (hook_bool_mode_mode_true): New function.
321 * combine.c (subst): Use targetm.modes_tieable_p instead of
323 * dse.c (find_shift_sequence): Likewise.
324 * expmed.c (extract_low_bits): Likewise.
325 * lower-subreg.c: Include target.h.
326 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
328 * rtlanal.c (rtx_cost): Likewise.
329 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
330 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
331 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
332 (TARGET_MODES_TIEABLE_P): Redefine.
333 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
334 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
335 (TARGET_MODES_TIEABLE_P): Redefine.
336 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
337 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
338 (arc_modes_tieable_p): New function.
339 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
340 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
341 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
342 (arm_modes_tieable_p): Make static.
343 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
344 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
345 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
346 (TARGET_MODES_TIEABLE_P): Redefine.
347 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
348 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
349 (TARGET_MODES_TIEABLE_P): Redefine.
350 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
351 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
352 (cr16_modes_tieable_p): New function.
353 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
354 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
355 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
356 (TRULY_NOOP_TRUNCATION): Update comment.
357 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
358 (TRULY_NOOP_TRUNCATION): Update comment.
359 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
360 (frv_modes_tieable_p): New function.
361 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
362 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
363 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
364 (TARGET_MODES_TIEABLE_P): Redefine.
365 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
366 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
367 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
368 (TARGET_MODES_TIEABLE_P): Redefine.
369 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
370 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
371 (ia64_modes_tieable_p): New function.
372 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
373 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
374 (iq2000_modes_tieable_p): New function.
375 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
376 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
377 (lm32_modes_tieable_p): New function.
378 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
379 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
380 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
381 (TARGET_MODES_TIEABLE_P): Redefine.
382 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
383 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
384 (m32r_modes_tieable_p): New function.
385 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
386 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
387 (m68k_modes_tieable_p): New function.
388 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
389 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
390 (mcore_modes_tieable_p): New function.
391 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
392 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
394 (TARGET_MODES_TIEABLE_P): Redefine.
395 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
396 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
397 * config/mips/mips.c (mips_modes_tieable_p): Make static.
398 (TARGET_MODES_TIEABLE_P): Redefine.
399 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
400 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
401 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
402 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
403 (mn10300_modes_tieable_p): ...this and make static.
404 (TARGET_MODES_TIEABLE_P): Redefine.
405 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
406 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
407 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
408 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
409 (msp430_modes_tieable_p): Make static.
410 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
411 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
412 (TARGET_MODES_TIEABLE_P): Redefine.
413 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
414 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
415 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
416 (TARGET_MODES_TIEABLE_P): Redefine.
417 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
418 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
419 * config/pa/pa.c (pa_modes_tieable_p): Make static.
420 (TARGET_MODES_TIEABLE_P): Redefine.
421 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
422 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
423 (pdp11_modes_tieable_p): New function.
424 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
425 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
426 (rs6000_modes_tieable_p): New function.
427 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
428 * config/powerpcspe/powerpcspe.md: Update comment.
429 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
430 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
431 (TARGET_MODES_TIEABLE_P): Redefine.
432 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
433 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
434 (rl78_modes_tieable_p): New function.
435 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
436 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
437 (rs6000_modes_tieable_p): New function.
438 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
439 * config/rs6000/rs6000.md: Update comment.
440 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
441 * config/rx/rx.c (rx_modes_tieable_p): New function.
442 (TARGET_MODES_TIEABLE_P): Redefine.
443 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
444 * config/s390/s390.c (s390_modes_tieable_p): New function.
445 (TARGET_MODES_TIEABLE_P): Redefine.
446 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
447 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
448 (sh_modes_tieable_p): New function.
449 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
450 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
451 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
452 (sparc_modes_tieable_p): Make static.
453 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
454 * config/spu/spu.c (spu_modes_tieable_p): New function.
455 (TARGET_MODES_TIEABLE_P): Redefine.
456 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
457 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
458 (TARGET_MODES_TIEABLE_P): Redefine.
459 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
460 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
461 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
462 * config/v850/v850.c (v850_modes_tieable_p): New function.
463 (TARGET_MODES_TIEABLE_P): Redefine.
464 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
465 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
466 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
467 (visium_modes_tieable_p): New function.
468 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
469 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
470 (xtensa_modes_tieable_p): New function.
471 * system.h (MODES_TIEABLE_P): Poison.
473 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
474 Alan Hayward <alan.hayward@arm.com>
475 David Sherwood <david.sherwood@arm.com>
477 * target.def (hard_regno_mode_ok): New hook.
478 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
479 (TARGET_HARD_REGNO_MODE_OK): ...this.
480 * doc/tm.texi.in: Regenerate.
481 * hooks.h (hook_bool_uint_mode_true): Declare.
482 * hooks.c (hook_bool_uint_mode_true): New function.
483 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
485 * genpreds.c (write_insn_preds_c): Add an include of target.h.
486 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
487 instead of HARD_REGNO_MODE_OK.
488 * caller-save.c: Include target.h.
489 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
491 * combine.c (can_combine_p): Likewise.
492 (combinable_i3pat): Likewise.
493 (can_change_dest_mode): Likewise.
494 * expr.c (init_expr_target): Likewise.
495 (convert_move): Likewise.
496 (convert_modes): Likewise.
497 * ira.c (setup_prohibited_class_mode_regs): Likewise.
498 (setup_prohibited_mode_move_regs): Likewise.
499 * ira.h (target_ira): Likewise.
500 * lra-assigns.c (find_hard_regno_for_1): Likewise.
501 * lra-constraints.c (process_alt_operands): Likewise.
502 (split_reg): Likewise.
503 * recog.c (peep2_find_free_register): Likewise.
504 * ree.c (combine_reaching_defs): Likewise.
505 * regcprop.c (maybe_mode_change): Likewise.
506 * reginfo.c (init_reg_sets_1): Likewise.
507 (choose_hard_reg_mode): Likewise.
508 (simplifiable_subregs): Likewise.
509 * regrename.c (check_new_reg_p): Likewise.
510 * reload.c (find_valid_class): Likewise.
511 (find_valid_class_1): Likewise.
512 (reload_inner_reg_of_subreg): Likewise.
513 (push_reload): Likewise.
514 (combine_reloads): Likewise.
515 (find_dummy_reload): Likewise.
516 (find_reloads): Likewise.
517 * reload1.c (find_reg): Likewise.
518 (set_reload_reg): Likewise.
519 (allocate_reload_reg): Likewise.
520 (choose_reload_regs): Likewise.
521 (reload_adjust_reg_for_temp): Likewise.
522 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
523 (simplify_subreg_regno): Likewise.
524 * sel-sched.c (init_regs_for_mode): Likewise.
525 * varasm.c (make_decl_rtl): Likewise.
526 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
527 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
529 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
530 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
532 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
533 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
535 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
536 (arc_mode_class): Delete.
537 (HARD_REGNO_MODE_OK): Delete.
538 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
539 (arc_hard_regno_mode_ok): Rename old array to...
540 (arc_hard_regno_mode_ok_modes): ...this.
541 (arc_conditional_register_usage): Update accordingly.
542 (arc_mode_class): Make static.
543 (arc_hard_regno_mode_ok): New function.
544 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
545 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
546 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
547 (arm_hard_regno_mode_ok): Make static.
548 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
550 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
551 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
552 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
555 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
556 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
557 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
558 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
560 * config/bfin/predicates.md (valid_reg_operand): Use
561 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
562 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
563 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
565 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
566 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
567 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
568 (cr16_hard_regno_mode_ok): Make static and return a bool.
569 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
570 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
571 (cris_hard_regno_mode_ok): New function.
572 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
573 (epiphany_mode_class): Delete.
574 (HARD_REGNO_MODE_OK): Delete.
575 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
576 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
577 (hard_regno_mode_ok): Rename to...
578 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
579 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
580 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
582 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
583 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
584 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
585 (frv_hard_regno_mode_ok): Make static and return a bool.
586 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
588 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
589 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
590 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
591 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
594 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
595 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
596 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
599 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
600 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
601 (ia64_hard_regno_mode_ok): New function.
602 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
603 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
604 (iq2000_hard_regno_mode_ok): New function.
605 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
606 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
607 (lm32_hard_regno_mode_ok): New function.
608 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
609 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
610 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
611 instead of HARD_REGNO_MODE_OK.
612 (m32c_hard_regno_ok): Rename to...
613 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
614 (m32c_cannot_change_mode_class): Update accordingly.
615 (TARGET_HARD_REGNO_MODE_OK): Redefine.
616 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
617 (m32r_mode_class): Delete.
618 (HARD_REGNO_MODE_OK): Delete.
619 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
620 (m32r_hard_regno_mode_ok): Rename to...
621 (m32r_hard_regno_modes): ...this.
622 (m32r_mode_class): Make static.
623 (m32r_hard_regno_mode_ok): New function.
624 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
625 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
626 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
627 (m68k_hard_regno_mode_ok): Make static.
628 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
629 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
630 (mcore_hard_regno_mode_ok): New function.
631 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
632 (HARD_REGNO_MODE_OK): Delete.
633 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
635 (microblaze_hard_regno_mode_ok_p): ...this and make static.
636 (microblaze_hard_regno_mode_ok): New function.
637 (TARGET_HARD_REGNO_MODE_OK): Redefine.
638 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
639 (mips_hard_regno_mode_ok): Delete.
640 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
641 (mips_hard_regno_mode_ok_p): ...this and make static.
642 (mips_hard_regno_mode_ok_p): Rename to...
643 (mips_hard_regno_mode_ok_uncached): ...this.
644 (mips_hard_regno_mode_ok): New function.
645 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
646 of HARD_REGNO_MODE_OK.
647 (mips_option_override): Update after above name changes.
648 (TARGET_HARD_REGNO_MODE_OK): Redefine.
649 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
650 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
651 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
652 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
653 (TARGET_HARD_REGNO_MODE_OK): Redefine.
654 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
655 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
656 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
657 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
658 (msp430_hard_regno_mode_ok): Make static and return a bool.
659 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
660 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
661 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
664 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
665 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
666 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
667 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
668 (PA_HARD_REGNO_MODE_OK): ...this
669 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
670 (PA_HARD_REGNO_MODE_OK): ...this.
671 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
672 (pa_hard_regno_mode_ok): New function.
673 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
674 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
675 (pdp11_hard_regno_mode_ok): New function.
676 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
677 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
679 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
681 (TARGET_HARD_REGNO_MODE_OK): Redefine.
682 (rs6000_hard_regno_mode_ok): Rename to...
683 (rs6000_hard_regno_mode_ok_uncached): ...this.
684 (rs6000_init_hard_regno_mode_ok): Update accordingly.
685 (rs6000_hard_regno_mode_ok): New function.
686 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
687 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
688 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
689 (riscv_hard_regno_mode_ok): ...this and make static.
690 (TARGET_HARD_REGNO_MODE_OK): Redefine.
691 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
692 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
693 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
694 (rl78_hard_regno_mode_ok): Make static and return bool.
695 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
696 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
698 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
700 (rs6000_hard_regno_mode_ok): Rename to...
701 (rs6000_hard_regno_mode_ok_uncached): ...this.
702 (rs6000_init_hard_regno_mode_ok): Update accordingly.
703 (rs6000_hard_regno_mode_ok): New function.
704 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
705 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
706 (TARGET_HARD_REGNO_MODE_OK): Redefine.
707 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
708 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
709 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
711 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
712 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
713 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
714 (sh_hard_regno_mode_ok): Make static.
715 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
716 instead of HARD_REGNO_MODE_OK.
717 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
718 (sparc_mode_class): Delete.
719 (HARD_REGNO_MODE_OK): Delete.
720 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
721 (hard_regno_mode_classes): Make static.
722 (sparc_mode_class): Likewise.
723 (sparc_hard_regno_mode_ok): New function.
724 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
725 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
726 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
729 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
730 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
731 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
732 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
734 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
735 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
736 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
737 (visium_hard_regno_mode_ok): New function.
738 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
739 instead of HARD_REGNO_MODE_OK.
740 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
741 (HARD_REGNO_MODE_OK): Delete.
742 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
743 (xtensa_hard_regno_mode_ok_p): ...this and make static.
744 (xtensa_option_override): Update accordingly.
745 (TARGET_HARD_REGNO_MODE_OK): Redefine.
746 (xtensa_hard_regno_mode_ok): New function.
747 * system.h (HARD_REGNO_MODE_OK): Poison.
749 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
750 Alan Hayward <alan.hayward@arm.com>
751 David Sherwood <david.sherwood@arm.com>
753 * target.def (hard_regno_call_part_clobbered): New hook.
754 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
755 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
756 * doc/tm.texi: Regenerate.
757 * hooks.h (hook_bool_uint_mode_false): Declare.
758 * hooks.c (hook_bool_uint_mode_false): New function.
759 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
760 * cselib.c (cselib_process_insn): Use
761 targetm.hard_regno_call_part_clobbered instead of
762 HARD_REGNO_CALL_PART_CLOBBERED.
763 * ira-conflicts.c (ira_build_conflicts): Likewise.
764 * ira-costs.c (ira_tune_allocno_costs): Likewise.
765 * lra-constraints.c (need_for_call_save_p): Likewise.
766 * lra-lives.c: Include target.h.
767 (check_pseudos_live_through_calls): Use
768 targetm.hard_regno_call_part_clobbered instead of
769 HARD_REGNO_CALL_PART_CLOBBERED.
770 * regcprop.c: Include target.h.
771 (copyprop_hardreg_forward_1): Use
772 targetm.hard_regno_call_part_clobbered instead of
773 HARD_REGNO_CALL_PART_CLOBBERED.
774 * reginfo.c (choose_hard_reg_mode): Likewise.
775 * regrename.c (check_new_reg_p): Likewise.
776 * reload.c (find_equiv_reg): Likewise.
777 * reload1.c (emit_reload_insns): Likewise.
778 * sched-deps.c (deps_analyze_insn): Likewise.
779 * sel-sched.c (init_regs_for_mode): Likewise.
780 (mark_unavailable_hard_regs): Likewise.
781 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
782 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
783 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
785 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
786 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
787 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
789 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
791 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
792 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
793 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
795 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
796 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
797 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
799 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
800 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
802 * config/powerpcspe/powerpcspe.c
803 (rs6000_hard_regno_call_part_clobbered): New function.
804 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
805 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
806 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
808 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
809 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
810 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
812 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
813 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
814 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
816 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
817 Alan Hayward <alan.hayward@arm.com>
818 David Sherwood <david.sherwood@arm.com>
820 * rtl.h (subreg_memory_offset): Declare.
821 * emit-rtl.c (subreg_memory_offset): New function.
822 * expmed.c (store_bit_field_1): Use it.
823 * expr.c (undefined_operand_subword_p): Likewise.
824 * simplify-rtx.c (simplify_subreg): Likewise.
826 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
828 PR rtl-optimization/57448
831 * optabs.c (expand_atomic_load): Place compiler memory barriers if
832 using atomic_load pattern.
833 (expand_atomic_store): Likewise.
835 2017-09-04 Jakub Jelinek <jakub@redhat.com>
838 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
839 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
840 handling. Use replace_call_with_value with NULL instead of
841 gsi_replace, unlink_stmt_vdef and release_defs.
843 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
846 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
848 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
851 * rtl.h (emit_library_call_value_1): Declare.
852 (emit_library_call): Replace declaration with a series of overloads.
853 Remove the parameter count argument.
854 (emit_library_call_value): Likewise.
855 * calls.c (emit_library_call_value_1): Make global. Replace varargs
856 with an "rtx_mode_t *".
857 (emit_library_call_value): Delete.
858 (emit_library_call): Likewise.
859 * asan.c (asan_emit_stack_protection): Update calls accordingly.
860 (asan_emit_allocas_unpoison): Likewise.
861 * builtins.c (expand_builtin_powi): Likewise.
862 (expand_asan_emit_allocas_unpoison): Likewise.
863 * cfgexpand.c (expand_main_function): Likewise.
864 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
865 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
866 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
867 * config/arm/arm.c (arm_trampoline_init): Likewise.
868 (arm_call_tls_get_addr): Likewise.
869 (arm_expand_divmod_libfunc): Likewise.
870 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
871 (smulsi3_highpart): Likewise.
872 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
873 (c6x_expand_compare): Likewise.
874 (c6x_expand_movmem): Likewise.
875 * config/frv/frv.c (frv_trampoline_init): Likewise.
876 * config/i386/i386.c (ix86_trampoline_init): Likewise.
877 (ix86_expand_divmod_libfunc): Likewise.
878 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
879 (ia64_expand_compare): Likewise.
880 (ia64_profile_hook): Likewise.
881 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
882 (nonlocal_goto): Likewise.
883 (restore_stack_nonlocal): Likewise.
884 * config/m32r/m32r.c (block_move_call): Likewise.
885 (m32r_trampoline_init): Likewise.
886 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
887 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
888 (m68k_call_m68k_read_tp): Likewise.
889 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
890 (microblaze_expand_divide): Likewise.
891 * config/mips/mips.h (mips_args): Likewise.
892 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
893 (MIPS_ICACHE_SYNC): Likewise.
894 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
895 (nios2_trampoline_init): Likewise.
896 * config/pa/pa.c (hppa_tls_call): Likewise.
897 (pa_trampoline_init): Likewise.
898 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
899 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
900 (expand_strn_compare): Likewise.
901 (rs6000_generate_compare): Likewise.
902 (rs6000_expand_float128_convert): Likewise.
903 (output_profile_hook): Likewise.
904 (rs6000_trampoline_init): Likewise.
905 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
906 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
907 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
908 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
909 (rs6000_generate_compare): Likewise.
910 (rs6000_expand_float128_convert): Likewise.
911 (output_profile_hook): Likewise.
912 (rs6000_trampoline_init): Likewise.
913 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
914 * config/sh/sh.c (sh_trampoline_init): Likewise.
915 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
916 (sparc_emit_float_lib_cmp): Likewise.
917 (sparc32_initialize_trampoline): Likewise.
918 (sparc64_initialize_trampoline): Likewise.
919 (sparc_profile_hook): Likewise.
920 * config/spu/spu.c (ea_load_store): Likewise.
921 * config/spu/spu.md (floatunssidf2): Likewise.
922 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
923 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
924 * config/visium/visium.c (expand_block_move_4): Likewise.
925 (expand_block_move_2): Likewise.
926 (expand_block_move_1): Likewise.
927 (expand_block_set_4): Likewise.
928 (expand_block_set_2): Likewise.
929 (expand_block_set_1): Likewise.
930 (visium_trampoline_init): Likewise.
931 (visium_profile_hook): Likewise.
932 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
933 (xtensa_setup_frame_addresses): Likewise.
934 (xtensa_trampoline_init): Likewise.
935 * except.c (sjlj_emit_function_enter): Likewise.
936 (sjlj_emit_function_exit): Likewise.
937 * explow.c (allocate_dynamic_stack_space): Likewise.
938 (probe_stack_range): Likewise.
939 * expr.c (convert_mode_scalar): Likewise.
940 * optabs.c (expand_binop): Likewise.
941 (expand_twoval_binop_libfunc): Likewise.
942 (expand_unop): Likewise.
943 (prepare_cmp_insn): Likewise.
944 (prepare_float_lib_cmp): Likewise.
945 (expand_float): Likewise.
946 (expand_fix): Likewise.
947 (expand_fixed_convert): Likewise.
948 (maybe_emit_sync_lock_test_and_set): Likewise.
949 (expand_atomic_compare_and_swap): Likewise.
950 (expand_mem_thread_fence): Likewise.
951 (expand_atomic_fetch_op): Likewise.
953 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
955 * doc/generic.texi (OpenACC): Adjust URL.
956 * doc/invoke.texi (C Dialect Options): Ditto.
958 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
960 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
961 predicate for operand 1. Add (m,<S>) constraint.
962 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
963 Prevent memory operand 1 with register operand 2.
965 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
967 PR rtl-optimization/82024
968 * combine.c (try_combine): If the combination result is a PARALLEL,
969 and we only need to retain the SET in there that would be placed
970 at I2, check that we can place that at I3 instead, before doing so.
972 2017-09-01 Jakub Jelinek <jakub@redhat.com>
975 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
977 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
978 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
981 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
982 Jeff Law <law@redhat.com>
984 * varasm.c (bss_initializer_p): Do not put constants into .bss
985 (categorize_decl_for_section): Handle bss_initializer_p returning
986 false when DECL_INITIAL is NULL.
988 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
991 * config/s390/s390.c (s390_can_inline_p): New function.
993 2017-09-01 Jeff Law <law@redhat.com>
995 PR tree-optimization/82052
996 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
997 Always initialize the returned slot after a hash table miss
1000 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1002 * config/s390/s390.md (mem_signal_fence): Remove.
1003 * doc/md.texi (mem_signal_fence): Remove.
1004 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1006 * target-insns.def (mem_signal_fence): Remove.
1008 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1011 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1014 * asan.c (create_odr_indicator): Strip name encoding from assembler
1015 name before appending it after __odr_asan_.
1017 2017-09-01 Martin Liska <mliska@suse.cz>
1019 PR tree-optimization/82059
1020 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1021 frequency only when an edge is redirected.
1023 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1025 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1026 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1027 (arc_conditional_register_usage): Remove ARC600 lp_count
1029 (arc_file_start): Emit Tag_ARC_CPU_variation.
1030 (arc_can_use_doloop_p): New conditions to use ZOLs.
1031 (hwloop_fail): New function.
1032 (hwloop_optimize): Likewise.
1033 (hwloop_pattern_reg): Likewise.
1034 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1035 (arc_reorg_loops): New function, calls reorg_loops.
1036 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1037 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1039 (arc_loop_hazard): Remove function, functionality moved into
1041 (arc_hazard): Remove arc_loop_hazard call.
1042 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1043 into hwloop_optimize.
1044 (arc_label_align): Remove ZOL handling.
1045 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1046 * config/arc/arc.md (doloop_begin): Remove pattern.
1047 (doloop_begin_i): Likewise.
1048 (doloop_end_i): Likewise.
1049 (doloop_fallback): Likewise.
1050 (doloop_fallback_m): Likewise.
1051 (doloop_end): Reimplement expand.
1052 (arc_lp): New pattern for LP instruction.
1053 (loop_end): New pattern.
1054 (loop_fail): Likewise.
1055 (decrement_and_branch_until_zero): Likewise.
1056 * config/arc/arc.opt (mlpc-width): New option.
1057 * doc/invoke.texi (mlpc-width): Document option.
1059 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1061 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1062 (arc_ccfsm_advance): Fix checking for delay slots.
1063 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1065 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1067 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1069 (movhi_insn): Update store instruction constraint which are saving
1070 6-bit short immediates.
1071 (movsi_insn): Consider also short scaled load operations.
1072 (zero_extendhisi2_i): Use Usd constraint instead of T.
1073 (extendhisi2_i): Add q constraint.
1074 (arc_clzsi2): Add type and length attributes.
1075 (arc_ctzsi2): Likewise.
1076 * config/arc/constraints.md (Usc): Update constraint, the
1077 assembler can parse two relocations for a single instruction.
1079 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1081 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1082 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1084 2017-08-31 Olivier Hainque <hainque@adacore.com>
1086 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1087 match as powerpc-wrs-vxworks*.
1089 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1091 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1092 register constraint for by-element operand.
1093 (aarch64_mls_elt_merge<mode>): Likewise.
1095 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1097 * config/arc/arc.c (arc_can_follow_jump): Check for short
1100 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1102 * config.gcc: Use g.opt for arc.
1103 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1104 functionality moved to ...
1105 (legitimate_scaled_address_p): New function, ...here.
1106 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1107 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1108 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1110 (arc_override_options): Handle G option.
1111 (arc_output_pic_addr_const): Correct function definition.
1112 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1113 (arc_decl_anon_ns_mem_p): Delete.
1114 (arc_in_small_data_p): Overhaul this function to take into
1115 consideration the value given via G option.
1116 (arc_rewrite_small_data_1): Renamed and corrected old
1117 arc_rewrite_small_data function.
1118 (arc_rewrite_small_data): New function.
1119 (small_data_pattern): Don't use pic_offset_table_rtx.
1120 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1121 * config/arc/simdext.md (movmisalignv2hi): Use
1122 prepare_move_operands function.
1124 (movmisalign*): Likewise.
1125 * doc/invoke.texi (ARC options): Document -G option.
1127 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1129 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1131 * config/arc/arc.c (arc_print_operand): Output scalled address for
1132 sdata whenever is possible.
1133 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1134 load/stores are available.
1135 (compact_sda_memory_operand): Check for the alignment required by
1136 code density instructions.
1137 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1139 * config/arc/constraints.md (Usd): Update constraint.
1140 (Us0): New constraint.
1141 (Usc): Update constraint.
1143 2017-08-31 Richard Biener <rguenther@suse.de>
1146 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1149 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1151 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1152 Resize type_signature.
1154 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1155 Alan Hayward <alan.hayward@arm.com>
1156 David Sherwood <david.sherwood@arm.com>
1158 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1159 subregs whose inner modes can be stored in GPRs.
1160 (aarch64_classify_index): Likewise.
1162 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1163 Alan Hayward <alan.hayward@arm.com>
1164 David Sherwood <david.sherwood@arm.com>
1166 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1167 (V_INT_EQUIV): ...this.
1168 (v_cmp_result): Rename to...
1169 (v_int_equiv): ...this.
1170 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1171 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1172 (copysign<mode>3): Likewise.
1173 (aarch64_simd_bsl<mode>_internal): Likewise.
1174 (aarch64_simd_bsl<mode>): Likewise.
1175 (vec_cmp<mode><mode>): Likewise.
1176 (vcond<mode><mode>): Likewise.
1177 (vcond<v_cmp_mixed><mode>): Likewise.
1178 (vcondu<mode><v_cmp_mixed>): Likewise.
1179 (aarch64_cm<optab><mode>): Likewise.
1180 (aarch64_cmtst<mode>): Likewise.
1181 (aarch64_fac<optab><mode>): Likewise.
1182 (vec_perm_const<mode>): Likewise.
1183 (vcond_mask_<mode><v_cmp_result>): Rename to...
1184 (vcond_mask_<mode><v_int_equiv>): ...this.
1185 (vec_cmp<mode><v_cmp_result>): Rename to...
1186 (vec_cmp<mode><v_int_equiv>): ...this.
1188 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1189 Alan Hayward <alan.hayward@arm.com>
1190 David Sherwood <david.sherwood@arm.com>
1192 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1194 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1195 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1196 (UNSPEC_LD4_DREG): New unspecs.
1197 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1198 (aarch64_ld2<mode>_dreg_be): Replace with...
1199 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1201 (aarch64_ld3<mode>_dreg_le)
1202 (aarch64_ld3<mode>_dreg_be): Replace with...
1203 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1205 (aarch64_ld4<mode>_dreg_le)
1206 (aarch64_ld4<mode>_dreg_be): Replace with...
1207 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1210 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1212 PR tree-optimization/81987
1213 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1214 insert an initializer in a location not dominated by the stride
1217 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1219 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1220 on the entire header of the finally block in the fallthru case.
1222 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1224 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1226 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1228 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1229 rs6000_emit_move_from_cr and call renamed function.
1230 (rs6000_emit_prologue): Call renamed functions.
1231 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1232 movesi_from_cr, remove volatile CRs.
1234 2017-08-30 Jon Beniston <jon@beniston.com>
1235 Richard Biener <rguenther@suse.de>
1237 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1238 of VECTOR_MODE_P check.
1239 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1240 element vector types.
1242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1244 * df.h (df_read_modify_subreg_p): Remove in favor of...
1245 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1246 const_rtx instead of an rtx.
1247 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1248 * df-problems.c (df_word_lr_mark_ref): Likewise.
1249 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1250 (mark_pseudo_reg_dead): Likewise.
1251 (mark_ref_dead): Likewise.
1252 * reginfo.c (init_subregs_of_mode): Likewise.
1253 * sched-deps.c (sched_analyze_1): Likewise.
1254 * df-scan.c (df_def_record_1): Likewise.
1255 (df_uses_record): Likewise.
1256 (df_read_modify_subreg_p): Remove in favor of...
1257 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1258 const_rtx instead of an rtx.
1260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1261 Alan Hayward <alan.hayward@arm.com>
1262 David Sherwood <david.sherwood@arm.com>
1264 * rtl.h (partial_subreg_p): New function.
1265 * caller-save.c (save_call_clobbered_regs): Use it.
1266 * calls.c (expand_call): Likewise.
1267 * combine.c (combinable_i3pat): Likewise.
1268 (simplify_set): Likewise.
1269 (make_extraction): Likewise.
1270 (make_compound_operation_int): Likewise.
1271 (gen_lowpart_or_truncate): Likewise.
1272 (force_to_mode): Likewise.
1273 (make_field_assignment): Likewise.
1274 (reg_truncated_to_mode): Likewise.
1275 (record_truncated_value): Likewise.
1276 (move_deaths): Likewise.
1277 * cse.c (record_jump_cond): Likewise.
1278 (cse_insn): Likewise.
1279 * cselib.c (cselib_lookup_1): Likewise.
1280 * expmed.c (extract_bit_field_using_extv): Likewise.
1281 * function.c (assign_parm_setup_reg): Likewise.
1282 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1283 * ira-build.c (create_insn_allocnos): Likewise.
1284 * lra-coalesce.c (merge_pseudos): Likewise.
1285 * lra-constraints.c (match_reload): Likewise.
1286 (simplify_operand_subreg): Likewise.
1287 (curr_insn_transform): Likewise.
1288 * lra-lives.c (process_bb_lives): Likewise.
1289 * lra.c (new_insn_reg): Likewise.
1290 (lra_substitute_pseudo): Likewise.
1291 * regcprop.c (mode_change_ok): Likewise.
1292 (maybe_mode_change): Likewise.
1293 (copyprop_hardreg_forward_1): Likewise.
1294 * reload.c (push_reload): Likewise.
1295 (find_reloads): Likewise.
1296 (find_reloads_subreg_address): Likewise.
1297 * reload1.c (alter_reg): Likewise.
1298 (eliminate_regs_1): Likewise.
1299 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1301 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1303 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1304 back to if statements, including unpack.
1306 2017-08-30 Martin Liska <mliska@suse.cz>
1309 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1311 (func_checker::compare_asm_inputs_outputs): ... this function.
1312 (func_checker::compare_gimple_asm): Use the function to compare
1313 also ASM constrains.
1314 * ipa-icf-gimple.h: Rename the function.
1316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1317 Alan Hayward <alan.hayward@arm.com>
1318 David Sherwood <david.sherwood@arm.com>
1320 * coretypes.h (complex_mode): New type.
1321 * gdbhooks.py (build_pretty_printer): Handle it.
1322 * machmode.h (complex_mode): New class.
1323 (complex_mode::includes_p): New function.
1324 (is_complex_int_mode): Likewise.
1325 (is_complex_float_mode): Likewise.
1326 * genmodes.c (get_mode_class): Handle complex mode classes.
1327 * function.c (expand_function_end): Use is_complex_int_mode.
1329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * coretypes.h (scalar_mode_pod): New typedef.
1334 * gdbhooks.py (build_pretty_printer): Handle it.
1335 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1336 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1337 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1338 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1342 Alan Hayward <alan.hayward@arm.com>
1343 David Sherwood <david.sherwood@arm.com>
1345 * machmode.h (mode_for_vector): Take a scalar_mode instead
1347 * stor-layout.c (mode_for_vector): Likewise.
1348 * explow.c (promote_mode): Use as_a <scalar_mode>.
1349 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
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 * target.def (preferred_simd_mode): Take a scalar_mode
1356 instead of a machine_mode.
1357 * targhooks.h (default_preferred_simd_mode): Likewise.
1358 * targhooks.c (default_preferred_simd_mode): Likewise.
1359 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1360 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1361 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1362 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1363 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1364 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1365 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1366 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1368 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1369 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1370 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1371 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1372 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1373 * doc/tm.texi: Regenerate.
1374 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1378 Alan Hayward <alan.hayward@arm.com>
1379 David Sherwood <david.sherwood@arm.com>
1381 * target.def (scalar_mode_supported_p): Take a scalar_mode
1382 instead of a machine_mode.
1383 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1384 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1385 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1386 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1387 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1388 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1389 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1390 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1391 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1392 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1393 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1394 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1395 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1396 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1398 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1399 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1400 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1401 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1402 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1404 * doc/tm.texi: Regenerate.
1406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1407 Alan Hayward <alan.hayward@arm.com>
1408 David Sherwood <david.sherwood@arm.com>
1410 * coretypes.h (opt_scalar_mode): New typedef.
1411 * gdbhooks.py (build_pretty_printers): Handle it.
1412 * machmode.h (mode_iterator::get_2xwider): Add overload for
1414 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1416 * expr.c (convert_mode_scalar): Likewise.
1417 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1418 * optabs.c (expand_float): Likewise.
1419 (expand_fix): Likewise.
1420 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1423 Alan Hayward <alan.hayward@arm.com>
1424 David Sherwood <david.sherwood@arm.com>
1426 * optabs.c (expand_float): Explicitly check for scalars before
1427 using a branching expansion.
1428 (expand_fix): Likewise.
1430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1431 Alan Hayward <alan.hayward@arm.com>
1432 David Sherwood <david.sherwood@arm.com>
1434 * expr.c (convert_mode): Split scalar handling out into...
1435 (convert_mode_scalar): ...this new function. Treat the modes
1438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1439 Alan Hayward <alan.hayward@arm.com>
1440 David Sherwood <david.sherwood@arm.com>
1442 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1444 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1447 Alan Hayward <alan.hayward@arm.com>
1448 David Sherwood <david.sherwood@arm.com>
1450 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1451 rather than a machine_mode.
1452 (fixed_from_string): Likewise.
1453 (fixed_convert): Likewise.
1454 (fixed_convert_from_int): Likewise.
1455 (fixed_convert_from_real): Likewise.
1456 (real_convert_from_fixed): Likewise.
1457 * fixed-value.c (fixed_from_double_int): Likewise.
1458 (fixed_from_string): Likewise.
1459 (fixed_convert): Likewise.
1460 (fixed_convert_from_int): Likewise.
1461 (fixed_convert_from_real): Likewise.
1462 (real_convert_from_fixed): Likewise.
1463 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1466 Alan Hayward <alan.hayward@arm.com>
1467 David Sherwood <david.sherwood@arm.com>
1469 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1470 of separate mode class checks. Do not allow vector modes here.
1471 (immed_wide_int_const): Use as_a <scalar_mode>.
1472 * explow.c (trunc_int_for_mode): Likewise.
1473 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1474 (wi::shwi): Likewise.
1475 (wi::min_value): Likewise.
1476 (wi::max_value): Likewise.
1477 * dwarf2out.c (loc_descriptor): Likewise.
1478 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1482 Alan Hayward <alan.hayward@arm.com>
1483 David Sherwood <david.sherwood@arm.com>
1485 * tree.h (SCALAR_TYPE_MODE): New macro.
1486 * expr.c (expand_expr_addr_expr_1): Use it.
1487 (expand_expr_real_2): Likewise.
1488 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1489 (fold_convert_const_fixed_from_int): Likewise.
1490 (fold_convert_const_fixed_from_real): Likewise.
1491 (native_encode_fixed): Likewise
1492 (native_encode_complex): Likewise
1493 (native_encode_vector): Likewise.
1494 (native_interpret_fixed): Likewise.
1495 (native_interpret_real): Likewise.
1496 (native_interpret_complex): Likewise.
1497 (native_interpret_vector): Likewise.
1498 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1499 (simd_clone_adjust_argument_types): Likewise.
1500 (simd_clone_init_simd_arrays): Likewise.
1501 (simd_clone_adjust): Likewise.
1502 * stor-layout.c (layout_type): Likewise.
1503 * tree.c (build_minus_one_cst): Likewise.
1504 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1505 * tree-inline.c (estimate_move_cost): Likewise.
1506 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1507 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1508 (vectorizable_reduction): Likewise.
1509 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1510 (vect_recog_mixed_size_cond_pattern): Likewise.
1511 (check_bool_pattern): Likewise.
1512 (adjust_bool_pattern): Likewise.
1513 (search_type_for_mask_1): Likewise.
1514 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1515 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1516 (vectorizable_load): Likewise.
1517 (vectorizable_store): Likewise.
1518 * ubsan.c (ubsan_encode_value): Likewise.
1519 * varasm.c (output_constant): Likewise.
1521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1522 Alan Hayward <alan.hayward@arm.com>
1523 David Sherwood <david.sherwood@arm.com>
1525 * coretypes.h (scalar_mode): New class.
1526 * machmode.h (scalar_mode): Likewise.
1527 (scalar_mode::includes_p): New function.
1528 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1529 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1530 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1531 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1532 * expmed.c (store_bit_field_1): Likewise.
1533 (extract_bit_field_1): Likewise.
1534 * expr.c (write_complex_part): Likewise.
1535 (read_complex_part): Likewise.
1536 (emit_move_complex_push): Likewise.
1537 (expand_expr_real_2): Likewise.
1538 * function.c (assign_parm_setup_reg): Likewise.
1539 (assign_parms_unsplit_complex): Likewise.
1540 * optabs.c (expand_binop): Likewise.
1541 * rtlanal.c (subreg_get_info): Likewise.
1542 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1543 * varasm.c (output_constant_pool_2): Likewise.
1545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1551 (expmed_mult_highpart_optab): Likewise.
1552 (expmed_mult_highpart): Likewise.
1554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1555 Alan Hayward <alan.hayward@arm.com>
1556 David Sherwood <david.sherwood@arm.com>
1558 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1559 instead of a machine_mode.
1560 (builtin_memset_read_str): Likewise.
1561 * builtins.c (c_readstr): Likewise.
1562 (builtin_memcpy_read_str): Likewise.
1563 (builtin_strncpy_read_str): Likewise.
1564 (builtin_memset_read_str): Likewise.
1565 (builtin_memset_gen_str): Likewise.
1566 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1567 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1568 instead of a machine_mode.
1569 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1571 (make_extraction): Likewise.
1572 (try_widen_shift_mode): Take and return scalar_int_modes instead
1574 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1575 a scalar_int_mode instead of a machine_mode.
1576 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1577 (avr_addr_space_pointer_mode): Likewise.
1578 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1579 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1580 (msp430_unwind_word_mode): Likewise.
1581 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1582 (spu_addr_space_pointer_mode): Likewise.
1583 (spu_addr_space_address_mode): Likewise.
1584 (spu_libgcc_cmp_return_mode): Likewise.
1585 (spu_libgcc_shift_count_mode): Likewise.
1586 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1587 (rl78_addr_space_pointer_mode): Likewise.
1588 (fl78_unwind_word_mode): Likewise.
1589 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1591 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1592 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1593 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1594 (mips_valid_pointer_mode): Likewise.
1595 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1596 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1597 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1599 (ft32_addr_space_address_mode): Likewise.
1600 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1601 scalar_int_mode instead of a machine_mode.
1602 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1604 (m32c_addr_space_address_mode): Likewise.
1605 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1606 (rs6000_eh_return_filter_mode): Likewise.
1607 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1608 (rs6000_eh_return_filter_mode): Likewise.
1609 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1610 (s390_libgcc_shift_count_mode): Likewise.
1611 (s390_unwind_word_mode): Likewise.
1612 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1614 * target.def (mode_rep_extended): Likewise.
1615 (valid_pointer_mode): Likewise.
1616 (addr_space.valid_pointer_mode): Likewise.
1617 (eh_return_filter_mode): Return a scalar_int_mode rather than
1619 (libgcc_cmp_return_mode): Likewise.
1620 (libgcc_shift_count_mode): Likewise.
1621 (unwind_word_mode): Likewise.
1622 (addr_space.pointer_mode): Likewise.
1623 (addr_space.address_mode): Likewise.
1624 * doc/tm.texi: Regenerate.
1625 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1627 (do_jump): Use scalar_int_mode for local variables.
1628 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1629 rather than a machine_mode.
1630 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1631 (scompare_loc_descriptor_wide): Likewise.
1632 (scompare_loc_descriptor_narrow): Likewise.
1633 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1635 * except.c (sjlj_emit_dispatch_table): Likewise.
1636 (expand_builtin_eh_copy_values): Likewise.
1637 * explow.c (convert_memory_address_addr_space_1): Likewise.
1638 Take a scalar_int_mode rather than a machine_mode.
1639 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1640 than a machine_mode.
1641 (memory_address_addr_space): Use scalar_int_mode for local variables.
1642 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1643 rather than a machine_mode.
1644 * expmed.c (mask_rtx): Likewise.
1645 (init_expmed_one_conv): Likewise.
1646 (expand_mult_highpart_adjust): Likewise.
1647 (extract_high_half): Likewise.
1648 (expmed_mult_highpart_optab): Likewise.
1649 (expmed_mult_highpart): Likewise.
1650 (expand_smod_pow2): Likewise.
1651 (expand_sdiv_pow2): Likewise.
1652 (emit_store_flag_int): Likewise.
1653 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1655 (extract_low_bits): Likewise.
1656 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1658 * expr.c (pieces_addr::adjust): Likewise.
1659 (can_store_by_pieces): Likewise.
1660 (store_by_pieces): Likewise.
1661 (clear_by_pieces_1): Likewise.
1662 (expand_expr_addr_expr_1): Likewise.
1663 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1664 (expand_expr_real_1): Likewise.
1665 (try_casesi): Likewise.
1666 * final.c (shorten_branches): Likewise.
1667 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1668 type of "mode" to machine_mode.
1669 * internal-fn.c (expand_arith_overflow_result_store): Take a
1670 scalar_int_mode rather than a machine_mode.
1671 (expand_mul_overflow): Use scalar_int_mode for local variables.
1672 * loop-doloop.c (doloop_modify): Likewise.
1673 (doloop_optimize): Likewise.
1674 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1675 than a machine_mode.
1676 (expand_doubleword_shift_condmove): Likewise.
1677 (expand_doubleword_shift): Likewise.
1678 (expand_doubleword_clz): Likewise.
1679 (expand_doubleword_popcount): Likewise.
1680 (expand_doubleword_parity): Likewise.
1681 (expand_absneg_bit): Use scalar_int_mode for local variables.
1682 (prepare_float_lib_cmp): Likewise.
1683 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1684 rather than a machine_mode.
1685 (convert_memory_address_addr_space): Likewise.
1686 (get_mode_bounds): Likewise.
1687 (get_address_mode): Return a scalar_int_mode rather than a
1689 * rtlanal.c (get_address_mode): Likewise.
1690 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1691 than a machine_mode.
1692 * targhooks.c (default_mode_rep_extended): Likewise.
1693 (default_valid_pointer_mode): Likewise.
1694 (default_addr_space_valid_pointer_mode): Likewise.
1695 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1696 than a machine_mode.
1697 (default_libgcc_cmp_return_mode): Likewise.
1698 (default_libgcc_shift_count_mode): Likewise.
1699 (default_unwind_word_mode): Likewise.
1700 (default_addr_space_pointer_mode): Likewise.
1701 (default_addr_space_address_mode): Likewise.
1702 * targhooks.h (default_eh_return_filter_mode): Likewise.
1703 (default_libgcc_cmp_return_mode): Likewise.
1704 (default_libgcc_shift_count_mode): Likewise.
1705 (default_unwind_word_mode): Likewise.
1706 (default_addr_space_pointer_mode): Likewise.
1707 (default_addr_space_address_mode): Likewise.
1708 (default_mode_rep_extended): Take a scalar_int_mode rather than
1710 (default_valid_pointer_mode): Likewise.
1711 (default_addr_space_valid_pointer_mode): Likewise.
1712 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1714 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
1715 rather than a machine_mode.
1716 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
1717 for local variables.
1718 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1719 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
1720 than a machine_mode.
1722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1723 Alan Hayward <alan.hayward@arm.com>
1724 David Sherwood <david.sherwood@arm.com>
1726 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
1727 the mode argument to scalar_int_mode.
1728 (do_jump_by_parts_zero_rtx): Likewise.
1729 (do_jump_by_parts_equality_rtx): Likewise.
1730 (do_jump_by_parts_greater): Take a mode argument.
1731 (do_jump_by_parts_equality): Likewise.
1732 (do_jump_1): Update calls accordingly.
1734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1735 Alan Hayward <alan.hayward@arm.com>
1736 David Sherwood <david.sherwood@arm.com>
1738 * is-a.h (safe_dyn_cast): New function.
1739 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
1740 (jump_table_for_label): Likewise.
1741 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
1742 instead of an rtx_insn *.
1743 (shorten_branches): Use dyn_cast instead of LABEL_P and
1744 JUMP_TABLE_DATA_P. Use jump_table_for_label and
1745 rtx_jump_table_data::get_data_mode.
1746 (final_scan_insn): Likewise.
1748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1749 Alan Hayward <alan.hayward@arm.com>
1750 David Sherwood <david.sherwood@arm.com>
1752 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1753 trying to combine a full-register integer set with a subreg
1756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1757 Alan Hayward <alan.hayward@arm.com>
1758 David Sherwood <david.sherwood@arm.com>
1760 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1761 that is always either address_mode or pointer_mode.
1763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1764 Alan Hayward <alan.hayward@arm.com>
1765 David Sherwood <david.sherwood@arm.com>
1767 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1768 when the two are known to be equal.
1770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1771 Alan Hayward <alan.hayward@arm.com>
1772 David Sherwood <david.sherwood@arm.com>
1774 * simplify-rtx.c (simplify_const_unary_operation): Use
1775 is_a <scalar_int_mode> instead of checking for a nonzero
1776 precision. Forcibly convert op_mode to a scalar_int_mode
1777 in that case. More clearly differentiate the operand and
1778 result modes and use the former when deciding what the value
1779 of a count-bits operation should be. Use is_int_mode instead
1780 of checking for a MODE_INT. Remove redundant check for whether
1781 this mode has a zero precision.
1783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1784 Alan Hayward <alan.hayward@arm.com>
1785 David Sherwood <david.sherwood@arm.com>
1787 * optabs.c (widen_leading): Change the type of the mode argument
1788 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1789 (widen_bswap): Likewise.
1790 (expand_parity): Likewise.
1791 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1792 (expand_ffs): Likewise.
1793 (epand_unop): Check for scalar integer modes before calling the
1796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1797 Alan Hayward <alan.hayward@arm.com>
1798 David Sherwood <david.sherwood@arm.com>
1800 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1802 (expand_expr_real_1): Update call accordingly.
1804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1805 Alan Hayward <alan.hayward@arm.com>
1806 David Sherwood <david.sherwood@arm.com>
1808 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1809 value_mode arguments. Use scalar_int_mode internally.
1810 (store_bit_field_1): Rename the new integer mode from imode
1811 to op0_mode and use it instead of GET_MODE (op0). Update calls
1812 to store_split_bit_field, store_bit_field_using_insv and
1813 store_fixed_bit_field.
1814 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1815 Use scalar_int_mode internally. Use a bit count rather than a mode
1816 when calculating the largest bit size for get_best_mode.
1817 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1818 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1819 Remove assertion that OP0 has a scalar integer mode.
1820 (store_split_bit_field): Add op0_mode and value_mode arguments.
1821 Update calls to extract_fixed_bit_field.
1822 (extract_bit_field_using_extv): Add an op0_mode argument.
1823 Use scalar_int_mode internally.
1824 (extract_bit_field_1): Rename the new integer mode from imode to
1825 op0_mode and use it instead of GET_MODE (op0). Update calls to
1826 extract_split_bit_field, extract_bit_field_using_extv and
1827 extract_fixed_bit_field.
1828 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1829 to extract_split_bit_field and extract_fixed_bit_field_1.
1830 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1831 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1833 (extract_split_bit_field): Add an op0_mode argument. Update call
1834 to extract_fixed_bit_field.
1836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1837 Alan Hayward <alan.hayward@arm.com>
1838 David Sherwood <david.sherwood@arm.com>
1840 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1841 * explow.c (hard_function_value): Likewise.
1842 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1843 convert_to_mode call outside the loop.
1844 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1845 for the mode iterator. Require the mode specified by max_pieces
1847 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1849 (copy_blkmode_to_reg): Likewise.
1850 (set_storage_via_setmem): Likewise.
1851 * optabs.c (prepare_cmp_insn): Likewise.
1852 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1853 * stor-layout.c (finish_bitfield_representative): Likewise.
1855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1856 Alan Hayward <alan.hayward@arm.com>
1857 David Sherwood <david.sherwood@arm.com>
1859 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1860 * expr.c (convert_move): Use them.
1861 (convert_modes): Likewise.
1862 (store_expr_with_bounds): Likewise.
1864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1865 Alan Hayward <alan.hayward@arm.com>
1866 David Sherwood <david.sherwood@arm.com>
1868 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1869 parameter for the mode of "x". Remove the "known_x", "known_mode"
1870 and "known_ret" arguments. Change the type of the mode argument
1872 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1873 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1874 (reg_num_sign_bit_copies_for_combine): Likewise.
1875 * rtlanal.c (nonzero_bits1): Likewise.
1876 (num_sign_bit_copies1): Likewise.
1877 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1878 (reg_num_sign_bit_copies_general): Likewise.
1879 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1880 (reg_nonzero_bits_general): Likewise.
1882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1887 than in subroutines. Return 1 for non-integer modes.
1888 (cached_num_sign_bit_copies): Change the type of the mode parameter
1890 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1891 classes. Handle CONST_INT_P first and then check whether X also
1892 has a scalar integer mode. Check the same thing for inner registers
1893 of a SUBREG and for values that are being extended or truncated.
1895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1896 Alan Hayward <alan.hayward@arm.com>
1897 David Sherwood <david.sherwood@arm.com>
1899 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1900 in subroutines. Return the mode mask for non-integer modes.
1901 (cached_nonzero_bits): Change the type of the mode parameter
1903 (nonzero_bits1): Likewise. Remove early exit for other mode
1904 classes. Handle CONST_INT_P first and then check whether X
1905 also has a scalar integer mode.
1907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1908 Alan Hayward <alan.hayward@arm.com>
1909 David Sherwood <david.sherwood@arm.com>
1911 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1912 Return a scalar_int_mode and assert that the size is greater than
1914 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1915 (op_by_pieces_d::op_by_pieces_d): Likewise.
1916 (op_by_pieces_d::run): Likewise.
1917 (can_store_by_pieces): Likewise.
1919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1920 Alan Hayward <alan.hayward@arm.com>
1921 David Sherwood <david.sherwood@arm.com>
1923 * combine.c (extract_left_shift): Add a mode argument and update
1925 (make_compound_operation_int): Change the type of the mode parameter
1926 to scalar_int_mode and update the call to extract_left_shift.
1928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1929 Alan Hayward <alan.hayward@arm.com>
1930 David Sherwood <david.sherwood@arm.com>
1932 * combine.c (simplify_and_const_int): Change the type of the mode
1933 parameter to scalar_int_mode.
1934 (simplify_and_const_int_1): Likewise. Update recursive call.
1936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1937 Alan Hayward <alan.hayward@arm.com>
1938 David Sherwood <david.sherwood@arm.com>
1940 * combine.c (simplify_compare_const): Check that the mode is a
1941 scalar_int_mode (rather than VOIDmode) before testing its
1943 (simplify_comparison): Move COMPARISON_P handling out of the
1944 loop and restrict the latter part of the loop to scalar_int_modes.
1945 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
1946 and when considering SUBREG_REGs. Use is_int_mode instead of
1947 checking GET_MODE_CLASS against MODE_INT.
1949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1950 Alan Hayward <alan.hayward@arm.com>
1951 David Sherwood <david.sherwood@arm.com>
1953 * combine.c (try_widen_shift_mode): Move check for equal modes to...
1954 (simplify_shift_const_1): ...here. Use scalar_int_mode for
1955 shift_unit_mode and for modes involved in scalar shifts.
1957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1958 Alan Hayward <alan.hayward@arm.com>
1959 David Sherwood <david.sherwood@arm.com>
1961 * combine.c (force_int_to_mode): New function, split out from...
1962 (force_to_mode): ...here. Keep xmode up-to-date and use it
1963 instead of GET_MODE (x).
1965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1966 Alan Hayward <alan.hayward@arm.com>
1967 David Sherwood <david.sherwood@arm.com>
1969 * optabs-query.h (extraction_insn::struct_mode): Change type to
1970 opt_scalar_int_mode and update comment.
1971 (extraction_insn::field_mode): Change type to scalar_int_mode.
1972 (extraction_insn::pos_mode): Likewise.
1973 * combine.c (make_extraction): Update accordingly.
1974 * optabs-query.c (get_traditional_extraction_insn): Likewise.
1975 (get_optab_extraction_insn): Likewise.
1976 * recog.c (simplify_while_replacing): Likewise.
1977 * expmed.c (narrow_bit_field_mem): Change the type of the mode
1978 parameter to opt_scalar_int_mode.
1980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1981 Alan Hayward <alan.hayward@arm.com>
1982 David Sherwood <david.sherwood@arm.com>
1984 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
1985 to a scalar_int_mode instead of a machine_mode.
1986 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
1987 (get_best_mode): Return a boolean and use a pointer argument to store
1988 the selected mode. Replace the limit mode parameter with a bit limit.
1989 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
1990 for the values returned by bit_field_mode_iterator::next_mode.
1991 (store_bit_field): Update call to get_best_mode.
1992 (store_fixed_bit_field): Likewise.
1993 (extract_fixed_bit_field): Likewise.
1994 * expr.c (optimize_bitfield_assignment_op): Likewise.
1995 * fold-const.c (optimize_bit_field_compare): Likewise.
1996 (fold_truth_andor_1): Likewise.
1997 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
1998 Update for new type of m_mode.
1999 (get_best_mode): As above.
2001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2002 Alan Hayward <alan.hayward@arm.com>
2003 David Sherwood <david.sherwood@arm.com>
2005 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2006 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2007 (store_bit_field): Check is_a <scalar_int_mode> before calling
2008 strict_volatile_bitfield_p.
2009 (extract_bit_field): Likewise.
2011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2012 Alan Hayward <alan.hayward@arm.com>
2013 David Sherwood <david.sherwood@arm.com>
2015 * target.def (cstore_mode): Return a scalar_int_mode.
2016 * doc/tm.texi: Regenerate.
2017 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2018 * targhooks.h (default_cstore_mode): Likewise.
2019 * targhooks.c (default_cstore_mode): Likewise, using a forced
2021 * expmed.c (emit_cstore): Expect the target of the cstore to be
2024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2025 Alan Hayward <alan.hayward@arm.com>
2026 David Sherwood <david.sherwood@arm.com>
2028 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2030 (niter_desc): Likewise mode.
2031 (iv_analyze): Add a mode parameter.
2033 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2034 and change its type to scalar_int_mode.
2035 * loop-iv.c: Update commentary at head of file.
2036 (iv_constant): Pass the mode paraeter before the rtx it describes
2037 and change its type to scalar_int_mode. Remove VOIDmode handling.
2038 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2039 (iv_extend): Likewise.
2040 (shorten_into_mode): Likewise.
2041 (iv_add): Use scalar_int_mode.
2042 (iv_mult): Likewise.
2043 (iv_shift): Likewise.
2044 (canonicalize_iv_subregs): Likewise.
2045 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2046 it describes and change its mode to scalar_int_mode. Also change
2047 the type of the returned inner_mode to scalar_int_mode.
2048 (get_biv_step): Likewise, turning outer_mode from a pointer
2049 into a direct parameter. Update call to get_biv_step_1.
2050 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2051 iv_constant and get_biv_step.
2052 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2053 and change its type to scalar_int_mode. Don't initialise iv->mode
2054 to VOIDmode and remove later checks for its still being VOIDmode.
2055 Update calls to iv_analyze_op and iv_analyze_expr. Check
2056 is_a <scalar_int_mode> when changing the mode under consideration.
2057 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2058 Update call to iv_analyze_expr.
2059 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2060 inner register is not also a scalar_int_mode. Update call to
2062 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2063 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2064 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2065 separate mode class checks. Update calls to iv_analyze. Remove
2066 fix-up of VOIDmodes after iv_analyze_biv.
2067 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2068 don't have a scalar_int_mode. Update call to biv_p.
2070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2071 Alan Hayward <alan.hayward@arm.com>
2072 David Sherwood <david.sherwood@arm.com>
2074 * cfgexpand.c (convert_debug_memory_address): Use
2075 as_a <scalar_int_mode>.
2076 * combine.c (expand_compound_operation): Likewise.
2077 (make_extraction): Likewise.
2078 (change_zero_ext): Likewise.
2079 (simplify_comparison): Likewise.
2080 * cse.c (cse_insn): Likewise.
2081 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2082 (mem_loc_descriptor): Likewise.
2083 (loc_descriptor): Likewise.
2084 * expmed.c (init_expmed_one_mode): Likewise.
2085 (synth_mult): Likewise.
2086 (emit_store_flag_1): Likewise.
2087 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2088 of a comparison with size.
2089 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2090 (reduce_to_bit_field_precision): Likewise.
2091 * function.c (expand_function_end): Likewise.
2092 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2093 * loop-doloop.c (doloop_modify): Likewise.
2094 * optabs.c (expand_binop): Likewise.
2095 (expand_unop): Likewise.
2096 (expand_copysign_absneg): Likewise.
2097 (prepare_cmp_insn): Likewise.
2098 (maybe_legitimize_operand): Likewise.
2099 * recog.c (const_scalar_int_operand): Likewise.
2100 * rtlanal.c (get_address_mode): Likewise.
2101 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2102 (simplify_cond_clz_ctz): Likewise.
2103 * tree-nested.c (get_nl_goto_field): Likewise.
2104 * tree.c (build_vector_type_for_mode): Likewise.
2105 * var-tracking.c (use_narrower_mode): Likewise.
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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2112 * builtins.c (expand_builtin_signbit): Use it.
2113 * cfgexpand.c (expand_debug_expr): Likewise.
2114 * dojump.c (do_jump): Likewise.
2115 (do_compare_and_jump): Likewise.
2116 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2117 * expmed.c (make_tree): Likewise.
2118 * expr.c (expand_expr_real_2): Likewise.
2119 (expand_expr_real_1): Likewise.
2120 (try_casesi): Likewise.
2121 * fold-const-call.c (fold_const_call_ss): Likewise.
2122 * fold-const.c (unextend): Likewise.
2123 (extract_muldiv_1): Likewise.
2124 (fold_single_bit_test): Likewise.
2125 (native_encode_int): Likewise.
2126 (native_encode_string): Likewise.
2127 (native_interpret_int): Likewise.
2128 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2129 * internal-fn.c (expand_addsub_overflow): Likewise.
2130 (expand_neg_overflow): Likewise.
2131 (expand_mul_overflow): Likewise.
2132 (expand_arith_overflow): Likewise.
2133 * match.pd: Likewise.
2134 * stor-layout.c (layout_type): Likewise.
2135 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2136 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2137 * tree-ssanames.c (get_range_info): Likewise.
2138 * tree-switch-conversion.c (array_value_type) Likewise.
2139 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2140 (vect_recog_divmod_pattern): Likewise.
2141 (vect_recog_mixed_size_cond_pattern): Likewise.
2142 * tree-vrp.c (extract_range_basic): Likewise.
2143 (simplify_float_conversion_using_ranges): Likewise.
2144 * tree.c (int_fits_type_p): Likewise.
2145 * ubsan.c (instrument_bool_enum_load): Likewise.
2146 * varasm.c (mergeable_string_section): Likewise.
2147 (narrowing_initializer_constant_valid_p): Likewise.
2148 (output_constant): Likewise.
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 * machmode.h (NARROWEST_INT_MODE): New macro.
2155 * expr.c (alignment_for_piecewise_move): Use it instead of
2156 GET_CLASS_NARROWEST_MODE (MODE_INT).
2157 (push_block): Likewise.
2158 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2160 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2163 Alan Hayward <alan.hayward@arm.com>
2164 David Sherwood <david.sherwood@arm.com>
2166 * postreload.c (move2add_valid_value_p): Change the type of the
2167 mode parameter to scalar_int_mode.
2168 (move2add_use_add2_insn): Add a mode parameter and use it instead
2170 (move2add_use_add3_insn): Likewise.
2171 (reload_cse_move2add): Update accordingly.
2173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2174 Alan Hayward <alan.hayward@arm.com>
2175 David Sherwood <david.sherwood@arm.com>
2177 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2179 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2180 * optabs.c (expand_unop): Likewise.
2182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2183 Alan Hayward <alan.hayward@arm.com>
2184 David Sherwood <david.sherwood@arm.com>
2186 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2187 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2188 (popcount_loc_descriptor): Likewise.
2189 (bswap_loc_descriptor): Likewise.
2190 (rotate_loc_descriptor): Likewise.
2191 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2192 calling the functions above.
2194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2195 Alan Hayward <alan.hayward@arm.com>
2196 David Sherwood <david.sherwood@arm.com>
2198 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2200 (try_combine): Likewise.
2201 (simplify_if_then_else): Likewise.
2202 * cse.c (cse_insn): Likewise.
2203 * dwarf2out.c (mem_loc_descriptor): Likewise.
2204 * emit-rtl.c (gen_lowpart_common): Likewise.
2205 * simplify-rtx.c (simplify_truncation): Likewise.
2206 (simplify_binary_operation_1): Likewise.
2207 (simplify_const_relational_operation): Likewise.
2208 (simplify_ternary_operation): Likewise.
2209 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2212 Alan Hayward <alan.hayward@arm.com>
2213 David Sherwood <david.sherwood@arm.com>
2215 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2216 * reload.c (push_reload): Likewise.
2217 (find_reloads): Likewise.
2219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2220 Alan Hayward <alan.hayward@arm.com>
2221 David Sherwood <david.sherwood@arm.com>
2223 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2224 (make_compound_operation_int): Likewise.
2225 (change_zero_ext): Likewise.
2226 * expr.c (convert_move): Likewise.
2227 (convert_modes): Likewise.
2228 * fwprop.c (forward_propagate_subreg): Likewise.
2229 * loop-iv.c (get_biv_step_1): Likewise.
2230 * optabs.c (widen_operand): Likewise.
2231 * postreload.c (move2add_valid_value_p): Likewise.
2232 * recog.c (simplify_while_replacing): Likewise.
2233 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2234 (simplify_binary_operation_1): Likewise. Remove redundant
2235 mode equality check.
2237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2238 Alan Hayward <alan.hayward@arm.com>
2239 David Sherwood <david.sherwood@arm.com>
2241 * combine.c (combine_simplify_rtx): Add checks for
2242 is_a <scalar_int_mode>.
2243 (simplify_if_then_else): Likewise.
2244 (make_field_assignment): Likewise.
2245 (simplify_comparison): Likewise.
2246 * ifcvt.c (noce_try_bitop): Likewise.
2247 * loop-invariant.c (canonicalize_address_mult): Likewise.
2248 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2251 Alan Hayward <alan.hayward@arm.com>
2252 David Sherwood <david.sherwood@arm.com>
2254 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2255 is_a <scalar_int_mode> instead of != BLKmode.
2257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2258 Alan Hayward <alan.hayward@arm.com>
2259 David Sherwood <david.sherwood@arm.com>
2261 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2262 instead of != VOIDmode.
2263 * combine.c (if_then_else_cond): Likewise.
2264 (change_zero_ext): Likewise.
2265 * dwarf2out.c (mem_loc_descriptor): Likewise.
2266 (loc_descriptor): Likewise.
2267 * rtlanal.c (canonicalize_condition): Likewise.
2268 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2271 Alan Hayward <alan.hayward@arm.com>
2272 David Sherwood <david.sherwood@arm.com>
2274 * simplify-rtx.c (simplify_binary_operation_1): Use
2275 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2278 Alan Hayward <alan.hayward@arm.com>
2279 David Sherwood <david.sherwood@arm.com>
2281 * wide-int.h (int_traits<unsigned char>) New class.
2282 (int_traits<unsigned short>) Likewise.
2283 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2284 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2286 * combine.c (set_nonzero_bits_and_sign_copies): Use
2287 is_a <scalar_int_mode>.
2288 (find_split_point): Likewise.
2289 (combine_simplify_rtx): Likewise.
2290 (simplify_logical): Likewise.
2291 (expand_compound_operation): Likewise.
2292 (expand_field_assignment): Likewise.
2293 (make_compound_operation): Likewise.
2294 (extended_count): Likewise.
2295 (change_zero_ext): Likewise.
2296 (simplify_comparison): Likewise.
2297 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2298 (ucompare_loc_descriptor): Likewise.
2299 (minmax_loc_descriptor): Likewise.
2300 (mem_loc_descriptor): Likewise.
2301 (loc_descriptor): Likewise.
2302 * expmed.c (init_expmed_one_mode): Likewise.
2303 * lra-constraints.c (lra_constraint_offset): Likewise.
2304 * optabs.c (prepare_libcall_arg): Likewise.
2305 * postreload.c (move2add_note_store): Likewise.
2306 * reload.c (operands_match_p): Likewise.
2307 * rtl.h (load_extend_op): Likewise.
2308 * rtlhooks.c (gen_lowpart_general): Likewise.
2309 * simplify-rtx.c (simplify_truncation): Likewise.
2310 (simplify_unary_operation_1): Likewise.
2311 (simplify_binary_operation_1): Likewise.
2312 (simplify_const_binary_operation): Likewise.
2313 (simplify_const_relational_operation): Likewise.
2314 (simplify_subreg): Likewise.
2315 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2316 * var-tracking.c (adjust_mems): Likewise.
2317 (prepare_call_arguments): Likewise.
2319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2320 Alan Hayward <alan.hayward@arm.com>
2321 David Sherwood <david.sherwood@arm.com>
2323 * machmode.h (is_int_mode): New fuction.
2324 * combine.c (find_split_point): Use it.
2325 (combine_simplify_rtx): Likewise.
2326 (simplify_if_then_else): Likewise.
2327 (simplify_set): Likewise.
2328 (simplify_shift_const_1): Likewise.
2329 (simplify_comparison): Likewise.
2330 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2331 * cse.c (notreg_cost): Likewise.
2332 (cse_insn): Likewise.
2333 * cselib.c (cselib_lookup_1): Likewise.
2334 * dojump.c (do_jump_1): Likewise.
2335 (do_compare_rtx_and_jump): Likewise.
2336 * dse.c (get_call_args): Likewise.
2337 * dwarf2out.c (rtl_for_decl_init): Likewise.
2338 (native_encode_initializer): Likewise.
2339 * expmed.c (emit_store_flag_1): Likewise.
2340 (emit_store_flag): Likewise.
2341 * expr.c (convert_modes): Likewise.
2342 (store_field): Likewise.
2343 (expand_expr_real_1): Likewise.
2344 * fold-const.c (fold_read_from_constant_string): Likewise.
2345 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2346 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2347 * optabs.c (expand_binop): Likewise.
2348 (expand_unop): Likewise.
2349 (expand_abs_nojump): Likewise.
2350 (expand_one_cmpl_abs_nojump): Likewise.
2351 * simplify-rtx.c (mode_signbit_p): Likewise.
2352 (val_signbit_p): Likewise.
2353 (val_signbit_known_set_p): Likewise.
2354 (val_signbit_known_clear_p): Likewise.
2355 (simplify_relational_operation_1): Likewise.
2356 * tree.c (vector_type_mode): Likewise.
2358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2359 Alan Hayward <alan.hayward@arm.com>
2360 David Sherwood <david.sherwood@arm.com>
2362 * machmode.h (smallest_mode_for_size): Fix formatting.
2363 (smallest_int_mode_for_size): New function.
2364 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2365 instead of smallest_mode_for_size.
2366 * combine.c (make_extraction): Likewise.
2367 * config/arc/arc.c (arc_expand_movmem): Likewise.
2368 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2369 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2370 * config/s390/s390.c (s390_expand_insv): Likewise.
2371 * config/sparc/sparc.c (assign_int_registers): Likewise.
2372 * config/spu/spu.c (spu_function_value): Likewise.
2373 (spu_function_arg): Likewise.
2374 * coverage.c (get_gcov_type): Likewise.
2375 (get_gcov_unsigned_t): Likewise.
2376 * dse.c (find_shift_sequence): Likewise.
2377 * expmed.c (store_bit_field_1): Likewise.
2378 * expr.c (convert_move): Likewise.
2379 (store_field): Likewise.
2380 * internal-fn.c (expand_arith_overflow): Likewise.
2381 * optabs-query.c (get_best_extraction_insn): Likewise.
2382 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2383 * stor-layout.c (layout_type): Likewise.
2384 (initialize_sizetypes): Likewise.
2385 * targhooks.c (default_get_mask_mode): Likewise.
2386 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2389 Alan Hayward <alan.hayward@arm.com>
2390 David Sherwood <david.sherwood@arm.com>
2392 * machmode.h (opt_mode::else_blk): New function.
2393 (int_mode_for_mode): Declare.
2394 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2395 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2397 * cfgexpand.c (expand_debug_expr): Likewise.
2398 * combine.c (gen_lowpart_or_truncate): Likewise.
2399 (gen_lowpart_for_combine): Likewise.
2400 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2401 * config/avr/avr.c (avr_to_int_mode): Likewise.
2402 (avr_out_plus_1): Likewise.
2403 (avr_out_plus): Likewise.
2404 (avr_out_round): Likewise.
2405 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2406 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2407 (s390_expand_vcond): Likewise.
2408 * config/spu/spu.c (spu_split_immediate): Likewise.
2409 (spu_expand_mov): Likewise.
2410 * dse.c (get_stored_val): Likewise.
2411 * expmed.c (store_bit_field_1): Likewise.
2412 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2414 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2415 (extract_low_bits): Likewise.
2416 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2417 handling rather than repeating the check.
2418 (emit_group_store): Likewise.
2419 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2420 * optabs.c (expand_absneg_bit): Likewise.
2421 (expand_copysign_absneg): Likewise.
2422 (expand_copysign_bit): Likewise.
2423 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2424 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2425 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2426 * var-tracking.c (prepare_call_arguments): Likewise.
2427 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2428 int_mode_for_mode instead of mode_for_size.
2429 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2432 Alan Hayward <alan.hayward@arm.com>
2433 David Sherwood <david.sherwood@arm.com>
2435 * machmode.h (int_mode_for_size): New function.
2436 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2437 instead of mode_for_size.
2438 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2440 * combine.c (expand_field_assignment): Use int_mode_for_size
2441 instead of mode_for_size.
2442 (make_extraction): Likewise.
2443 (simplify_shift_const_1): Likewise.
2444 (simplify_comparison): Likewise.
2445 * dojump.c (do_jump): Likewise.
2446 * dwarf2out.c (mem_loc_descriptor): Likewise.
2447 * emit-rtl.c (init_derived_machine_modes): Likewise.
2448 * expmed.c (flip_storage_order): Likewise.
2449 (convert_extracted_bit_field): Likewise.
2450 * expr.c (copy_blkmode_from_reg): Likewise.
2451 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2452 * internal-fn.c (expand_mul_overflow): Likewise.
2453 * lower-subreg.c (simple_move): Likewise.
2454 * optabs-libfuncs.c (init_optabs): Likewise.
2455 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2456 * tree.c (vector_type_mode): Likewise.
2457 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2458 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2459 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2460 * tree-vect-stmts.c (vectorizable_load): Likewise.
2461 (vectorizable_store): Likewise.
2463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2464 Alan Hayward <alan.hayward@arm.com>
2465 David Sherwood <david.sherwood@arm.com>
2467 * coretypes.h (pod_mode): New type.
2468 (scalar_int_mode_pod): New typedef.
2469 * machmode.h (pod_mode): New class.
2470 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2471 * genmodes.c (emit_mode_int_n): Update accordingly.
2472 * lower-subreg.h (target_lower_subreg): Change type to
2473 scalar_int_mode_pod.
2474 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2475 scalar_int_mode_pod.
2477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2478 Alan Hayward <alan.hayward@arm.com>
2479 David Sherwood <david.sherwood@arm.com>
2481 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2482 machine_mode to scalar_int_mode.
2483 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2484 (rs6000_option_override_internal): Remove cast to int.
2485 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2486 machine_mode to scalar_int_mode.
2487 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2488 (rs6000_option_override_internal): Remove cast to int.
2489 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2490 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2492 * config/s390/s390.c (s390_expand_builtin): Likewise.
2493 * coretypes.h (scalar_int_mode): New type.
2494 (opt_scalar_int_mode): New typedef.
2495 * machmode.h (scalar_int_mode): New class.
2496 (scalar_int_mode::includes_p): New function.
2497 (byte_mode): Change type to scalar_int_mode.
2498 (word_mode): Likewise.
2499 (ptr_mode): Likewise.
2500 * emit-rtl.c (byte_mode): Likewise.
2501 (word_mode): Likewise.
2502 (ptr_mode): Likewise.
2503 (init_derived_machine_modes): Update accordingly.
2504 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2505 and MODE_PARTIAL_INT.
2506 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2507 opt_scalar_int_mode.
2509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2510 Alan Hayward <alan.hayward@arm.com>
2511 David Sherwood <david.sherwood@arm.com>
2513 * target.def (libgcc_floating_mode_supported_p): Take a
2515 * doc/tm.texi: Regenerate.
2516 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2518 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2519 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2523 Alan Hayward <alan.hayward@arm.com>
2524 David Sherwood <david.sherwood@arm.com>
2526 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2527 * doc/tm.texi: Regenerate.
2528 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2529 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2530 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2531 * targhooks.h (default_floatn_mode): Likewise.
2532 * targhooks.c (default_floatn_mode): Likewise.
2533 * tree.c (build_common_tree_nodes): Update accordingly.
2535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2536 Alan Hayward <alan.hayward@arm.com>
2537 David Sherwood <david.sherwood@arm.com>
2539 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2540 (mode_iterator::iterate_p): Likewise.
2541 (mode_iterator::get_wider): Likewise.
2542 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2545 Alan Hayward <alan.hayward@arm.com>
2546 David Sherwood <david.sherwood@arm.com>
2548 * coretypes.h (opt_scalar_float_mode): New typedef.
2549 * machmode.h (float_mode_for_size): New function.
2550 * emit-rtl.c (double_mode): Delete.
2551 (init_emit_once): Use float_mode_for_size.
2552 * stor-layout.c (layout_type): Likewise.
2553 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2556 Alan Hayward <alan.hayward@arm.com>
2557 David Sherwood <david.sherwood@arm.com>
2559 * output.h (assemble_real): Take a scalar_float_mode.
2560 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2561 * config/arm/arm.md (consttable_4): Likewise.
2562 (consttable_8): Likewise.
2563 (consttable_16): Likewise.
2564 * config/mips/mips.md (consttable_float): Likewise.
2565 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2566 * varasm.c (assemble_real): Take a scalar_float_mode.
2567 (output_constant_pool_2): Update accordingly.
2568 (output_constant): Likewise.
2570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2571 Alan Hayward <alan.hayward@arm.com>
2572 David Sherwood <david.sherwood@arm.com>
2574 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2575 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2576 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2577 (native_encode_real): Likewise.
2578 (native_interpret_real): Likewise.
2579 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2580 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2583 Alan Hayward <alan.hayward@arm.com>
2584 David Sherwood <david.sherwood@arm.com>
2586 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2587 <scalar_float_mode>. Simplify.
2588 (gen_extend_conv_libfunc): Likewise.
2590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2591 Alan Hayward <alan.hayward@arm.com>
2592 David Sherwood <david.sherwood@arm.com>
2594 * coretypes.h (scalar_float_mode): New type.
2595 * machmode.h (mode_traits::from_int): Use machine_mode if
2596 USE_ENUM_MODES is defined.
2597 (is_a): New function.
2599 (dyn_cast): Likewise.
2600 (scalar_float_mode): New class.
2601 (scalar_float_mode::includes_p): New function.
2602 (is_float_mode): Likewise.
2603 * gdbhooks.py (MachineModePrinter): New class.
2604 (build_pretty_printer): Use it for scalar_float_mode.
2605 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2606 (format_helper::format_helper): Turn into a template.
2607 * genmodes.c (get_mode_class): New function.
2608 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2609 or machine_mode if none.
2610 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2611 as_a <scalar_float_mode>.
2612 * dwarf2out.c (mem_loc_descriptor): Likewise.
2613 (insert_float): Likewise.
2614 (add_const_value_attribute): Likewise.
2615 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2616 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2617 (expand_unop): Update accordingly.
2618 (expand_abs_nojump): Likewise.
2619 (expand_copysign_absneg): Take a scalar_float_mode.
2620 (expand_copysign_bit): Likewise.
2621 (expand_copysign): Update accordingly.
2623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2624 Alan Hayward <alan.hayward@arm.com>
2625 David Sherwood <david.sherwood@arm.com>
2627 * coretypes.h (opt_mode): New class.
2628 * machmode.h (opt_mode): Likewise.
2629 (opt_mode::else_void): New function.
2630 (opt_mode::require): Likewise.
2631 (opt_mode::exists): Likewise.
2632 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2633 (GET_MODE_2XWIDER_MODE): Likewise.
2634 (mode_iterator::get_wider): Update accordingly.
2635 (mode_iterator::get_2xwider): Likewise.
2636 (mode_iterator::get_known_wider): Likewise, turning into a template.
2637 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2638 forcing a wider mode to exist.
2639 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2640 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2641 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2642 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2643 * lower-subreg.c (init_lower_subreg): Likewise.
2644 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2645 on the final iteration.
2646 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2647 a wider mode exists before asking for a move pattern.
2648 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2649 forcing a wider mode to exist.
2650 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2651 returning false if no such mode exists.
2652 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2653 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2654 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2655 Avoid checking for a MODE_INT if we already know the mode is not a
2657 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2658 forcing a wider mode to exist.
2659 (expmed_mult_highpart_optab): Likewise.
2660 (expmed_mult_highpart): Likewise.
2661 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2663 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2664 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2665 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2666 * internal-fn.c (expand_mul_overflow): Update use of
2667 GET_MODE_2XWIDER_MODE.
2668 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2669 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2670 GET_MODE_WIDER_MODE.
2671 (convert_plusminus_to_widen): Likewise.
2672 * tree-switch-conversion.c (array_value_type): Likewise.
2673 * var-tracking.c (emit_note_insn_var_location): Likewise.
2674 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2675 Return false inside rather than outside the loop if no wider mode
2677 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2678 and GET_MODE_2XWIDER_MODE
2679 (can_compare_p): Use else_void.
2680 * gdbhooks.py (OptMachineModePrinter): New class.
2681 (build_pretty_printer): Use it for opt_mode.
2683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2684 Alan Hayward <alan.hayward@arm.com>
2685 David Sherwood <david.sherwood@arm.com>
2687 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2688 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2691 Alan Hayward <alan.hayward@arm.com>
2692 David Sherwood <david.sherwood@arm.com>
2694 * machmode.h (mode_traits): New structure.
2695 (get_narrowest_mode): New function.
2696 (mode_iterator::start): Likewise.
2697 (mode_iterator::iterate_p): Likewise.
2698 (mode_iterator::get_wider): Likewise.
2699 (mode_iterator::get_known_wider): Likewise.
2700 (mode_iterator::get_2xwider): Likewise.
2701 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2702 (FOR_EACH_MODE): Likewise.
2703 (FOR_EACH_MODE_FROM): Likewise.
2704 (FOR_EACH_MODE_UNTIL): Likewise.
2705 (FOR_EACH_WIDER_MODE): Likewise.
2706 (FOR_EACH_2XWIDER_MODE): Likewise.
2707 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2708 * combine.c (simplify_comparison): Likewise
2709 * config/i386/i386.c (type_natural_mode): Likewise.
2710 * cse.c (cse_insn): Likewise.
2711 * dse.c (find_shift_sequence): Likewise.
2712 * emit-rtl.c (init_derived_machine_modes): Likewise.
2713 (init_emit_once): Likewise.
2714 * explow.c (hard_function_value): Likewise.
2715 * expmed.c (extract_fixed_bit_field_1): Likewise.
2716 (extract_bit_field_1): Likewise.
2717 (expand_divmod): Likewise.
2718 (emit_store_flag_1): Likewise.
2719 * expr.c (init_expr_target): Likewise.
2720 (convert_move): Likewise.
2721 (alignment_for_piecewise_move): Likewise.
2722 (widest_int_mode_for_size): Likewise.
2723 (emit_block_move_via_movmem): Likewise.
2724 (copy_blkmode_to_reg): Likewise.
2725 (set_storage_via_setmem): Likewise.
2726 (compress_float_constant): Likewise.
2727 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2728 * optabs-query.c (get_best_extraction_insn): Likewise.
2729 * optabs.c (expand_binop): Likewise.
2730 (expand_twoval_unop): Likewise.
2731 (expand_twoval_binop): Likewise.
2732 (widen_leading): Likewise.
2733 (widen_bswap): Likewise.
2734 (expand_parity): Likewise.
2735 (expand_unop): Likewise.
2736 (prepare_cmp_insn): Likewise.
2737 (prepare_float_lib_cmp): Likewise.
2738 (expand_float): Likewise.
2739 (expand_fix): Likewise.
2740 (expand_sfix_optab): Likewise.
2741 * postreload.c (move2add_use_add2_insn): Likewise.
2742 * reg-stack.c (reg_to_stack): Likewise.
2743 * reginfo.c (choose_hard_reg_mode): Likewise.
2744 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2745 * stor-layout.c (mode_for_size): Likewise.
2746 (smallest_mode_for_size): Likewise.
2747 (mode_for_vector): Likewise.
2748 (finish_bitfield_representative): Likewise.
2749 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2750 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2751 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2752 * var-tracking.c (prepare_call_arguments): Likewise.
2754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2755 Alan Hayward <alan.hayward@arm.com>
2756 David Sherwood <david.sherwood@arm.com>
2758 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2759 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2760 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2761 * machmode.h (mode_size): Move earlier in file.
2762 (mode_precision): Likewise.
2763 (mode_inner): Likewise.
2764 (mode_nunits): Likewise.
2765 (mode_unit_size): Likewise.
2766 (unit_unit_precision): Likewise.
2767 (mode_wider): Likewise.
2768 (mode_2xwider): Likewise.
2769 (machine_mode): New class.
2770 (mode_to_bytes): New function.
2771 (mode_to_bits): Likewise.
2772 (mode_to_precision): Likewise.
2773 (mode_to_inner): Likewise.
2774 (mode_to_unit_size): Likewise.
2775 (mode_to_unit_precision): Likewise.
2776 (mode_to_nunits): Likewise.
2777 (GET_MODE_SIZE): Use mode_to_bytes.
2778 (GET_MODE_BITSIZE): Use mode_to_bits.
2779 (GET_MODE_PRECISION): Use mode_to_precision.
2780 (GET_MODE_INNER): Use mode_to_inner.
2781 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2782 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2783 (GET_MODE_NUNITS): Use mode_to_nunits.
2784 * system.h (ALWAYS_INLINE): New macro.
2785 * config/powerpcspe/powerpcspe-c.c
2786 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2787 int for arg1_mode and arg2_mode.
2789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2790 Alan Hayward <alan.hayward@arm.com>
2791 David Sherwood <david.sherwood@arm.com>
2793 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2794 Prefix mode names with E_ in case statements.
2795 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2796 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2797 (aarch64_split_simd_move): Likewise.
2798 (aarch64_gen_storewb_pair): Likewise.
2799 (aarch64_gen_loadwb_pair): Likewise.
2800 (aarch64_gen_store_pair): Likewise.
2801 (aarch64_gen_load_pair): Likewise.
2802 (aarch64_get_condition_code_1): Likewise.
2803 (aarch64_constant_pool_reload_icode): Likewise.
2804 (get_rsqrte_type): Likewise.
2805 (get_rsqrts_type): Likewise.
2806 (get_recpe_type): Likewise.
2807 (get_recps_type): Likewise.
2808 (aarch64_gimplify_va_arg_expr): Likewise.
2809 (aarch64_simd_container_mode): Likewise.
2810 (aarch64_emit_load_exclusive): Likewise.
2811 (aarch64_emit_store_exclusive): Likewise.
2812 (aarch64_expand_compare_and_swap): Likewise.
2813 (aarch64_gen_atomic_cas): Likewise.
2814 (aarch64_emit_bic): Likewise.
2815 (aarch64_emit_atomic_swap): Likewise.
2816 (aarch64_emit_atomic_load_op): Likewise.
2817 (aarch64_evpc_trn): Likewise.
2818 (aarch64_evpc_uzp): Likewise.
2819 (aarch64_evpc_zip): Likewise.
2820 (aarch64_evpc_ext): Likewise.
2821 (aarch64_evpc_rev): Likewise.
2822 (aarch64_evpc_dup): Likewise.
2823 (aarch64_gen_ccmp_first): Likewise.
2824 (aarch64_gen_ccmp_next): Likewise.
2825 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2826 (alpha_emit_xfloating_libcall): Likewise.
2827 (emit_insxl): Likewise.
2828 (alpha_arg_type): Likewise.
2829 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2830 (arc_preferred_simd_mode): Likewise.
2831 (arc_secondary_reload): Likewise.
2832 (get_arc_condition_code): Likewise.
2833 (arc_print_operand): Likewise.
2834 (arc_legitimate_constant_p): Likewise.
2835 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2836 * config/arc/arc.md (casesi_load): Likewise.
2837 (casesi_compact_jump): Likewise.
2838 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2839 (cc_use_register): Likewise.
2840 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2841 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2842 (arm_init_iwmmxt_builtins): Likewise.
2843 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2844 (neon_expand_vector_init): Likewise.
2845 (arm_attr_length_move_neon): Likewise.
2846 (maybe_get_arm_condition_code): Likewise.
2847 (arm_emit_vector_const): Likewise.
2848 (arm_preferred_simd_mode): Likewise.
2849 (arm_output_iwmmxt_tinsr): Likewise.
2850 (thumb1_output_casesi): Likewise.
2851 (thumb2_output_casesi): Likewise.
2852 (arm_emit_load_exclusive): Likewise.
2853 (arm_emit_store_exclusive): Likewise.
2854 (arm_expand_compare_and_swap): Likewise.
2855 (arm_evpc_neon_vuzp): Likewise.
2856 (arm_evpc_neon_vzip): Likewise.
2857 (arm_evpc_neon_vrev): Likewise.
2858 (arm_evpc_neon_vtrn): Likewise.
2859 (arm_evpc_neon_vext): Likewise.
2860 (arm_validize_comparison): Likewise.
2861 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2862 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2863 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2864 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2865 (c6x_preferred_simd_mode): Likewise.
2866 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2867 (epiphany_rtx_costs): Likewise.
2868 * config/epiphany/predicates.md (proper_comparison_operator):
2870 * config/frv/frv.c (condexec_memory_operand): Likewise.
2871 (frv_emit_move): Likewise.
2872 (output_move_single): Likewise.
2873 (output_condmove_single): Likewise.
2874 (frv_hard_regno_mode_ok): Likewise.
2875 (frv_matching_accg_mode): Likewise.
2876 * config/h8300/h8300.c (split_adds_subs): Likewise.
2877 (h8300_rtx_costs): Likewise.
2878 (h8300_print_operand): Likewise.
2879 (compute_mov_length): Likewise.
2880 (output_logical_op): Likewise.
2881 (compute_logical_op_length): Likewise.
2882 (compute_logical_op_cc): Likewise.
2883 (h8300_shift_needs_scratch_p): Likewise.
2884 (output_a_shift): Likewise.
2885 (compute_a_shift_length): Likewise.
2886 (compute_a_shift_cc): Likewise.
2887 (expand_a_rotate): Likewise.
2888 (output_a_rotate): Likewise.
2889 * config/i386/i386.c (classify_argument): Likewise.
2890 (function_arg_advance_32): Likewise.
2891 (function_arg_32): Likewise.
2892 (function_arg_64): Likewise.
2893 (function_value_64): Likewise.
2894 (ix86_gimplify_va_arg): Likewise.
2895 (ix86_legitimate_constant_p): Likewise.
2896 (put_condition_code): Likewise.
2897 (split_double_mode): Likewise.
2898 (ix86_avx256_split_vector_move_misalign): Likewise.
2899 (ix86_expand_vector_logical_operator): Likewise.
2900 (ix86_split_idivmod): Likewise.
2901 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2902 (ix86_build_const_vector): Likewise.
2903 (ix86_build_signbit_mask): Likewise.
2904 (ix86_match_ccmode): Likewise.
2905 (ix86_cc_modes_compatible): Likewise.
2906 (ix86_expand_branch): Likewise.
2907 (ix86_expand_sse_cmp): Likewise.
2908 (ix86_expand_sse_movcc): Likewise.
2909 (ix86_expand_int_sse_cmp): Likewise.
2910 (ix86_expand_vec_perm_vpermi2): Likewise.
2911 (ix86_expand_vec_perm): Likewise.
2912 (ix86_expand_sse_unpack): Likewise.
2913 (ix86_expand_int_addcc): Likewise.
2914 (ix86_split_to_parts): Likewise.
2915 (ix86_vectorize_builtin_gather): Likewise.
2916 (ix86_vectorize_builtin_scatter): Likewise.
2917 (avx_vpermilp_parallel): Likewise.
2918 (inline_memory_move_cost): Likewise.
2919 (ix86_tieable_integer_mode_p): Likewise.
2920 (x86_maybe_negate_const_int): Likewise.
2921 (ix86_expand_vector_init_duplicate): Likewise.
2922 (ix86_expand_vector_init_one_nonzero): Likewise.
2923 (ix86_expand_vector_init_one_var): Likewise.
2924 (ix86_expand_vector_init_concat): Likewise.
2925 (ix86_expand_vector_init_interleave): Likewise.
2926 (ix86_expand_vector_init_general): Likewise.
2927 (ix86_expand_vector_set): Likewise.
2928 (ix86_expand_vector_extract): Likewise.
2929 (emit_reduc_half): Likewise.
2930 (ix86_emit_i387_round): Likewise.
2931 (ix86_mangle_type): Likewise.
2932 (ix86_expand_round_sse4): Likewise.
2933 (expand_vec_perm_blend): Likewise.
2934 (canonicalize_vector_int_perm): Likewise.
2935 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
2936 (expand_vec_perm_1): Likewise.
2937 (expand_vec_perm_interleave3): Likewise.
2938 (expand_vec_perm_even_odd_pack): Likewise.
2939 (expand_vec_perm_even_odd_1): Likewise.
2940 (expand_vec_perm_broadcast_1): Likewise.
2941 (ix86_vectorize_vec_perm_const_ok): Likewise.
2942 (ix86_expand_vecop_qihi): Likewise.
2943 (ix86_expand_mul_widen_hilo): Likewise.
2944 (ix86_expand_sse2_abs): Likewise.
2945 (ix86_expand_pextr): Likewise.
2946 (ix86_expand_pinsr): Likewise.
2947 (ix86_preferred_simd_mode): Likewise.
2948 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
2949 * config/i386/sse.md (*andnot<mode>3): Likewise.
2950 (<mask_codefor><code><mode>3<mask_name>): Likewise.
2951 (*<code><mode>3): Likewise.
2952 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
2953 (ia64_expand_atomic_op): Likewise.
2954 (ia64_arg_type): Likewise.
2955 (ia64_mode_to_int): Likewise.
2956 (ia64_scalar_mode_supported_p): Likewise.
2957 (ia64_vector_mode_supported_p): Likewise.
2958 (expand_vec_perm_broadcast): Likewise.
2959 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
2960 (iq2000_function_arg_advance): Likewise.
2961 (iq2000_function_arg): Likewise.
2962 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
2963 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
2964 (m68k_libcall_value): Likewise.
2965 (m68k_function_value): Likewise.
2966 (sched_attr_op_type): Likewise.
2967 * config/mcore/mcore.c (mcore_output_move): Likewise.
2968 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
2970 (microblaze_function_arg): Likewise.
2971 * config/mips/mips.c (mips16_build_call_stub): Likewise.
2972 (mips_print_operand): Likewise.
2973 (mips_mode_ok_for_mov_fmt_p): Likewise.
2974 (mips_vector_mode_supported_p): Likewise.
2975 (mips_preferred_simd_mode): Likewise.
2976 (mips_expand_vpc_loongson_even_odd): Likewise.
2977 (mips_expand_vec_unpack): Likewise.
2978 (mips_expand_vi_broadcast): Likewise.
2979 (mips_expand_vector_init): Likewise.
2980 (mips_expand_vec_reduc): Likewise.
2981 (mips_expand_msa_cmp): Likewise.
2982 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
2983 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
2984 (cc_flags_for_mode): Likewise.
2985 * config/msp430/msp430.c (msp430_print_operand): Likewise.
2986 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
2987 (nds32_output_casesi_pc_relative): Likewise.
2988 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2989 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
2990 (nvptx_gen_unpack): Likewise.
2991 (nvptx_gen_pack): Likewise.
2992 (nvptx_gen_shuffle): Likewise.
2993 (nvptx_gen_wcast): Likewise.
2994 (nvptx_preferred_simd_mode): Likewise.
2995 * config/pa/pa.c (pa_secondary_reload): Likewise.
2996 * config/pa/predicates.md (base14_operand): Likewise.
2997 * config/powerpcspe/powerpcspe-c.c
2998 (altivec_resolve_overloaded_builtin): Likewise.
2999 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3001 (rs6000_preferred_simd_mode): Likewise.
3002 (output_vec_const_move): Likewise.
3003 (rs6000_expand_vector_extract): Likewise.
3004 (rs6000_split_vec_extract_var): Likewise.
3005 (reg_offset_addressing_ok_p): Likewise.
3006 (rs6000_legitimate_offset_address_p): Likewise.
3007 (rs6000_legitimize_address): Likewise.
3008 (rs6000_emit_set_const): Likewise.
3009 (rs6000_const_vec): Likewise.
3010 (rs6000_emit_move): Likewise.
3011 (spe_build_register_parallel): Likewise.
3012 (rs6000_darwin64_record_arg_recurse): Likewise.
3013 (swap_selector_for_mode): Likewise.
3014 (spe_init_builtins): Likewise.
3015 (paired_init_builtins): Likewise.
3016 (altivec_init_builtins): Likewise.
3017 (do_load_for_compare): Likewise.
3018 (rs6000_generate_compare): Likewise.
3019 (rs6000_expand_float128_convert): Likewise.
3020 (emit_load_locked): Likewise.
3021 (emit_store_conditional): Likewise.
3022 (rs6000_output_function_epilogue): Likewise.
3023 (rs6000_handle_altivec_attribute): Likewise.
3024 (rs6000_function_value): Likewise.
3025 (emit_fusion_gpr_load): Likewise.
3026 (emit_fusion_p9_load): Likewise.
3027 (emit_fusion_p9_store): Likewise.
3028 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3029 (fusion_gpr_mem_load): Likewise.
3030 (fusion_addis_mem_combo_load): Likewise.
3031 (fusion_addis_mem_combo_store): Likewise.
3032 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3033 (fusion_gpr_mem_load): Likewise.
3034 (fusion_addis_mem_combo_load): Likewise.
3035 (fusion_addis_mem_combo_store): Likewise.
3036 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3038 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3039 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3040 (rs6000_preferred_simd_mode): Likewise.
3041 (output_vec_const_move): Likewise.
3042 (rs6000_expand_vector_extract): Likewise.
3043 (rs6000_split_vec_extract_var): Likewise.
3044 (reg_offset_addressing_ok_p): Likewise.
3045 (rs6000_legitimate_offset_address_p): Likewise.
3046 (rs6000_legitimize_address): Likewise.
3047 (rs6000_emit_set_const): Likewise.
3048 (rs6000_const_vec): Likewise.
3049 (rs6000_emit_move): Likewise.
3050 (rs6000_darwin64_record_arg_recurse): Likewise.
3051 (swap_selector_for_mode): Likewise.
3052 (paired_init_builtins): Likewise.
3053 (altivec_init_builtins): Likewise.
3054 (rs6000_expand_float128_convert): Likewise.
3055 (emit_load_locked): Likewise.
3056 (emit_store_conditional): Likewise.
3057 (rs6000_output_function_epilogue): Likewise.
3058 (rs6000_handle_altivec_attribute): Likewise.
3059 (rs6000_function_value): Likewise.
3060 (emit_fusion_gpr_load): Likewise.
3061 (emit_fusion_p9_load): Likewise.
3062 (emit_fusion_p9_store): Likewise.
3063 * config/rx/rx.c (rx_gen_move_template): Likewise.
3064 (flags_from_mode): Likewise.
3065 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3066 (s390_slb_comparison): Likewise.
3067 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3068 (s390_vector_mode_supported_p): Likewise.
3069 (s390_cc_modes_compatible): Likewise.
3070 (s390_match_ccmode_set): Likewise.
3071 (s390_canonicalize_comparison): Likewise.
3072 (s390_emit_compare_and_swap): Likewise.
3073 (s390_branch_condition_mask): Likewise.
3074 (s390_rtx_costs): Likewise.
3075 (s390_secondary_reload): Likewise.
3076 (__SECONDARY_RELOAD_CASE): Likewise.
3077 (s390_expand_cs): Likewise.
3078 (s390_preferred_simd_mode): Likewise.
3079 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3080 * config/sh/sh.c (sh_print_operand): Likewise.
3081 (dump_table): Likewise.
3082 (sh_secondary_reload): Likewise.
3083 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3084 * config/sh/sh.md (casesi_worker_1): Likewise.
3085 (casesi_worker_2): Likewise.
3086 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3087 (fcc_comparison_operator): Likewise.
3088 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3089 (emit_soft_tfmode_cvt): Likewise.
3090 (sparc_preferred_simd_mode): Likewise.
3091 (output_cbranch): Likewise.
3092 (sparc_print_operand): Likewise.
3093 (sparc_expand_vec_perm_bmask): Likewise.
3094 (vector_init_bshuffle): Likewise.
3095 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3096 (spu_vector_mode_supported_p): Likewise.
3097 (spu_expand_insv): Likewise.
3098 (spu_emit_branch_or_set): Likewise.
3099 (spu_handle_vector_attribute): Likewise.
3100 (spu_builtin_splats): Likewise.
3101 (spu_builtin_extract): Likewise.
3102 (spu_builtin_promote): Likewise.
3103 (spu_expand_sign_extend): Likewise.
3104 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3105 (tilegx_simd_int): Likewise.
3106 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3107 (tilepro_simd_int): Likewise.
3108 * config/v850/v850.c (const_double_split): Likewise.
3109 (v850_print_operand): Likewise.
3110 (ep_memory_offset): Likewise.
3111 * config/vax/vax.c (vax_rtx_costs): Likewise.
3112 (vax_output_int_move): Likewise.
3113 (vax_output_int_add): Likewise.
3114 (vax_output_int_subtract): Likewise.
3115 * config/visium/predicates.md (visium_branch_operator): Likewise.
3116 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3117 (visium_print_operand_address): Likewise.
3118 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3119 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3120 (xtensa_expand_conditional_branch): Likewise.
3121 (xtensa_copy_incoming_a7): Likewise.
3122 (xtensa_output_literal): Likewise.
3123 * dfp.c (decimal_real_maxval): Likewise.
3124 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3127 Alan Hayward <alan.hayward@arm.com>
3128 David Sherwood <david.sherwood@arm.com>
3130 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3131 (mode_nunits_inline): Likewise.
3132 (mode_inner_inline): Likewise.
3133 (mode_unit_size_inline): Likewise.
3134 (mode_unit_precision_inline): Likewise.
3135 (emit_insn_modes_h): Likewise. Also emit a #define of the
3137 (emit_mode_wider): Add an E_ prefix to mode names.
3138 (emit_mode_complex): Likewise.
3139 (emit_mode_inner): Likewise.
3140 (emit_mode_adjustments): Likewise.
3141 (emit_mode_int_n): Likewise.
3142 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3143 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3144 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3145 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3146 (CRC32_BUILTIN, ENTRY): Likewise.
3147 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3148 (aarch64_pop_regs): Likewise.
3149 (aarch64_process_components): Likewise.
3150 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3151 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3152 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3153 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3154 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3155 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3157 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3158 (pcmpistr): Likewise.
3159 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3160 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3161 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3163 * config/rl78/rl78.c (mduc_regs): Likewise.
3164 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3165 (htm_expand_builtin): Likewise.
3166 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3167 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3168 * config/xtensa/xtensa.c (print_operand): Likewise.
3169 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3170 (NUM_MODE_VECTOR_INT): Likewise.
3171 * genoutput.c (null_operand): Likewise.
3172 (output_operand_data): Likewise.
3173 * genrecog.c (print_parameter_value): Likewise.
3174 * lra.c (debug_operand_data): Likewise.
3176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3178 * dwarf2out.c (scompare_loc_descriptor_wide)
3179 (scompare_loc_descriptor_narrow): New functions, split out from...
3180 (scompare_loc_descriptor): ...here.
3181 * expmed.c (emit_store_flag_int): New function, split out from...
3182 (emit_store_flag): ...here.
3184 2017-08-30 Richard Biener <rguenther@suse.de>
3186 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3187 (dwarf2out_early_finish): Move setting of AT_pubnames from
3188 early debug output to early finish.
3190 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3192 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3193 and -mdata-region to the assembler.
3195 2017-08-30 Richard Biener <rguenther@suse.de>
3197 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3199 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3200 (dwarf2out_early_global_decl): Only generate a DIE for the
3201 abstract origin if it doesn't already exist or is a declaration DIE.
3202 (resolve_addr): Do not add the linkage name twice when
3203 generating a stub DIE for the DW_TAG_GNU_call_site target.
3205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3207 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3208 Use machine_mode rather than int for arg1_mode.
3210 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3213 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3214 that the second argument of the built-in functions to unpack
3215 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3216 switch statement instead a lot of if statements.
3217 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3218 Allow 64-bit values to be in Altivec registers as well as
3219 traditional floating point registers.
3220 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3222 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3224 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3225 MAX_REGS_PER_ADDRESS == 1.
3227 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3229 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3230 * config/i386/i386.c (ix86_option_override_internal): Simplify
3231 setting of opts->x_flag_entry.
3233 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3234 Jakub Jelinek <jakub@redhat.com>
3235 Richard Biener <rguenther@suse.de>
3237 PR tree-optimization/81503
3238 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3239 folded constant fits in the target type; reorder tests for clarity.
3241 2017-08-29 Martin Liska <mliska@suse.cz>
3243 * passes.def: Include pass_lower_switch.
3244 * stmt.c (dump_case_nodes): Remove and move to
3245 tree-switch-conversion.
3246 (case_values_threshold): Likewise.
3247 (expand_switch_as_decision_tree_p): Likewise.
3248 (emit_case_decision_tree): Likewise.
3249 (expand_case): Likewise.
3250 (balance_case_nodes): Likewise.
3251 (node_has_low_bound): Likewise.
3252 (node_has_high_bound): Likewise.
3253 (node_is_bounded): Likewise.
3254 (emit_case_nodes): Likewise.
3255 (struct simple_case_node): New struct.
3256 (add_case_node): Remove.
3257 (emit_case_dispatch_table): Use vector instead of case_list.
3258 (reset_out_edges_aux): Remove.
3259 (compute_cases_per_edge): Likewise.
3260 (expand_case): Build list of simple_case_node.
3261 (expand_sjlj_dispatch_table): Use it.
3262 * tree-switch-conversion.c (struct case_node): Moved from
3263 stmt.c and adjusted.
3264 (emit_case_nodes): Likewise.
3265 (node_has_low_bound): Likewise.
3266 (node_has_high_bound): Likewise.
3267 (node_is_bounded): Likewise.
3268 (case_values_threshold): Likewise.
3269 (reset_out_edges_aux): Likewise.
3270 (compute_cases_per_edge): Likewise.
3271 (add_case_node): Likewise.
3272 (dump_case_nodes): Likewise.
3273 (balance_case_nodes): Likewise.
3274 (expand_switch_as_decision_tree_p): Likewise.
3275 (emit_jump): Likewise.
3276 (emit_case_decision_tree): Likewise.
3277 (try_switch_expansion): Likewise.
3278 (do_jump_if_equal): Likewise.
3279 (emit_cmp_and_jump_insns): Likewise.
3280 (fix_phi_operands_for_edge): New function.
3281 (record_phi_operand_mapping): Likewise.
3282 (class pass_lower_switch): New pass.
3283 (pass_lower_switch::execute): New function.
3284 (make_pass_lower_switch): Likewise.
3285 (conditional_probability):
3286 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3287 * tree-pass.h: Add make_pass_lower_switch.
3289 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3292 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3295 2017-08-29 Richard Biener <rguenther@suse.de>
3297 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3298 we do not add a DW_AT_inline attribute twice.
3299 (gen_subprogram_die): Remove code setting DW_AT_inline on
3300 DECL_ABSTRACT_P nodes.
3302 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3304 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3305 calls to internal functions.
3306 (gimplify_modify_expr): Likewise.
3307 * tree-call-cdce.c (use_internal_fn): Likewise.
3308 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3309 (convert_to_divmod): Set the nothrow flag.
3310 * tree-if-conv.c (predicate_mem_writes): Likewise.
3311 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3312 (vectorizable_call): Likewise.
3313 (vectorizable_store): Likewise.
3314 (vectorizable_load): Likewise.
3315 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3316 (vect_recog_mask_conversion_pattern): Likewise.
3318 2017-08-29 Martin Liska <mliska@suse.cz>
3321 * gcc.c (driver_handle_option): Add new argument.
3322 * opts-common.c (handle_option): Pass
3323 target_option_override_hook.
3324 * opts-global.c (lang_handle_option): Add new option.
3325 (set_default_handlers): Add new argument.
3326 (decode_options): Likewise.
3327 * opts.c (target_handle_option): Likewise.
3328 (common_handle_option): Call target_option_override_hook.
3329 * opts.h (struct cl_option_handler_func): Add hook for
3330 target option override.
3331 (struct cl_option_handlers): Likewise.
3332 (set_default_handlers): Add new argument.
3333 (decode_options): Likewise.
3334 (common_handle_option): Likewise.
3335 (target_handle_option): Likewise.
3336 * toplev.c (toplev::main): Pass targetm.target_option.override
3339 2017-08-29 Richard Biener <rguenther@suse.de>
3340 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3342 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3343 life to the active subtree.
3345 2017-08-28 Jeff Law <law@redhat.com>
3347 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3348 derive_equivalences.
3349 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3351 (edge_info::derive_equivalences): New private member function
3353 * tree-ssa-dom.c (class edge_info): Changed from a struct
3354 to a class. Add ctor/dtor, methods and data members.
3355 (edge_info::edge_info): Renamed from allocate_edge_info.
3356 Initialize additional members.
3357 (edge_info::~edge_info): New.
3358 (free_dom_edge_info): Delete the edge info.
3359 (record_edge_info): Use new class & associated member functions.
3360 Tighten forms for testing for edge equivalences.
3361 (record_temporary_equivalences): Iterate over the simple
3362 equivalences rather than assuming there's only one per edge.
3363 (cprop_into_successor_phis): Iterate over the simple
3364 equivalences rather than assuming there's only one per edge.
3365 (optimize_stmt): Use operand_equal_p rather than pointer
3366 equality for mini-DSE code.
3368 2017-08-28 Nathan Sidwell <nathan@acm.org>
3370 * gcc.c (execute): Fold SIGPIPE handling into switch
3371 statement. Adjust internal error message.
3373 2017-08-28 Richard Biener <rguenther@suse.de>
3376 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3377 Do nothing for removed DIEs.
3379 2017-08-28 Richard Biener <rguenther@suse.de>
3381 PR tree-optimization/81977
3382 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3385 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3388 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3389 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3390 using targetm.gen_mem_thread_fence.
3392 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3395 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3396 predicate to register_operand. Reorder operands.
3397 (*btr<mode>): Ditto.
3398 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3399 (*btr<mode>_mask): Ditto.
3401 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3403 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3404 * config/rs6000/xmmintrin.h: New file.
3405 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3407 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3410 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3411 parameter and_insn and return it.
3412 (recombine_lvx_pattern): Insert a copy to ensure availability of
3413 the base register of the copied masking operation at the point of
3414 the instruction replacement.
3415 (recombine_stvx_pattern): Likewise.
3417 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3419 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3420 undocumented switches.
3421 (-mpower9-dform-vector): Likewise.
3422 (-mpower9-dform): Likewise.
3423 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3424 comments to delete references to -mpower9-dform* switches.
3425 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3426 Delete reference to -mpower9-dform* switches, test for
3427 -mpower9-vector instead.
3428 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3429 (OTHER_P9_VECTOR_MASKS): Likewise.
3430 (POWERPC_MASKS): Likewise.
3431 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3432 tests against -mpower9-dform* to -mpower9-vector. Delete code
3433 that checked for -mpower9-dform* consistancy with other options.
3434 Add test for -mpower9-misc to enable other power9 switches.
3435 (rs6000_init_hard_regno_mode_ok): Likewise.
3436 (rs6000_option_override_internal): Likewise.
3437 (rs6000_emit_prologue): Likewise.
3438 (rs6000_emit_epilogue): Likewise.
3439 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3440 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3441 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3443 (emit_fusion_p9_store): Likewise.
3444 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3445 resetting these macros if the assembler does not support ISA 3.0
3447 (TARGET_P9_DFORM_VECTOR): Likewise.
3448 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3449 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3451 2017-08-25 Alan Modra <amodra@gmail.com>
3453 PR rtl-optimization/81747
3454 * cse.c (cse_extended_basic_block): Don't attempt to record
3455 equivalences for degenerate conditional branches that branch
3456 to their fall-through.
3458 2017-08-24 Martin Sebor <msebor@redhat.com>
3461 * gimple-fold.c (size_must_be_zero_p): New function.
3462 (gimple_fold_builtin_memory_op): Call it.
3464 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3466 * config/rs6000/mm_malloc.h: New file.
3468 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3470 PR tree-optimization/81913
3471 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3472 analysis when either IVs in condition can wrap.
3474 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3476 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3477 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3479 2017-08-24 Richard Biener <rguenther@suse.de>
3482 * targhooks.c (default_target_can_inline_p): Properly
3483 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3484 is present and always compare.
3485 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3486 infer -mfpmath=sse from TARGET_SSE_P.
3487 (ix86_can_inline_p): Properly use target_option_default_node when
3488 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3490 2017-08-24 Richard Biener <rguenther@suse.de>
3493 * dwarf2out.c (output_die): Handle flag_generate_offload like
3495 (output_comp_unit): Likewise.
3496 (gen_array_type_die): Likewise.
3497 (dwarf2out_early_finish): Likewise.
3498 (note_variable_value_in_expr): Likewise.
3499 (dwarf2out_finish): Likewise. Adjust assert.
3500 * cgraphunit.c (symbol_table::compile): Move setting of
3501 flag_generate_offload earlier ...
3502 (symbol_table::finalize_compilation_unit): ... here, before
3503 early debug finalization.
3505 2017-08-24 Richard Biener <rguenther@suse.de>
3507 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3508 and ipa-fnsummary.h.
3509 (ix86_can_inline_p): When ix86_fpmath flags do not match
3510 check whether the callee uses FP math at all.
3512 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3515 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3516 instead of TYPE_PRECISION.
3518 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3519 Alan Hayward <alan.hayward@arm.com>
3520 David Sherwood <david.sherwood@arm.com>
3522 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3524 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3526 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3528 (ix86_frame::stack_realign_allocate): New field.
3529 (struct machine_frame_state): Modify comments.
3530 (machine_frame_state::sp_realigned_fp_end): New field.
3531 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3533 (sp_valid_at): Add assertion to assure no attempt to access invalid
3534 offset of a realigned stack.
3535 (fp_valid_at): Likewise.
3536 (choose_baseaddr): Modify comments.
3537 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3538 ix86_expand_prologue.
3539 (ix86_expand_prologue): Modify stack realignment and allocation.
3540 (ix86_expand_epilogue): Modify comments.
3541 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3542 avx2_runtime, avx512f, and avx512f_runtime.
3544 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3546 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3547 (mstackrealign): Do not init to -1.
3548 * config/i386/i386.c (ix86_option_override_internal):
3549 Check opts_set, not opts when setting default value of
3550 opts->x_ix86_force_align_arg_pointer.
3552 2017-08-23 Richard Biener <rguenther@suse.de>
3554 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3555 lang_hooks.decl_printable_name.
3556 * print-rtl-function.c (print_rtx_function): Likewise.
3557 * tree-pretty-print.c (dump_function_header): Likewise.
3559 2017-08-23 Richard Biener <rguenther@suse.de>
3562 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3563 -g0 at compile-time.
3565 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3568 * doc/sourcebuild.texi (Other hardware attributes):
3571 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3574 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3575 Fix single-use check.
3577 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3579 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3581 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3583 * doc/install.texi: Modify to add more details on running selected
3586 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3588 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3589 is combined with -mabi=ms.
3590 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3593 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3595 PR tree-optimization/81488
3596 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3597 and cached_basis fields.
3598 (MAX_SPREAD): New constant.
3599 (alloc_cand_and_find_basis): Initialize new fields.
3600 (clear_visited): New function.
3601 (create_phi_basis_1): Rename from create_phi_basis, set visited
3602 and cached_basis fields.
3603 (create_phi_basis): New wrapper function.
3604 (phi_add_costs_1): Rename from phi_add_costs, add spread
3605 parameter, set visited field, short-circuit when limits reached.
3606 (phi_add_costs): New wrapper function.
3607 (record_phi_increments_1): Rename from record_phi_increments, set
3609 (record_phi_increments): New wrapper function.
3610 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3611 (phi_incr_cost): New wrapper function.
3612 (all_phi_incrs_profitable_1): Rename from
3613 all_phi_incrs_profitable, set visited field.
3614 (all_phi_incrs_profitable): New wrapper function.
3616 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3617 Alan Hayward <alan.hayward@arm.com>
3618 David Sherwood <david.sherwood@arm.com>
3620 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3621 that takes the outer and inner modes.
3622 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3623 comparison as the canonical test for a paradoxical subreg.
3624 * combine.c (simplify_set): Use paradoxical_subreg_p.
3625 (make_extraction): Likewise.
3626 (force_to_mode): Likewise.
3627 (rtx_equal_for_field_assignment_p): Likewise.
3628 (gen_lowpart_for_combine): Likewise.
3629 (simplify_comparison): Likewise.
3630 * cse.c (equiv_constant): Likewise.
3631 * expmed.c (store_bit_field_1): Likewise.
3632 * final.c (alter_subreg): Likewise.
3633 * fwprop.c (propagate_rtx): Likewise.
3634 (forward_propagate_subreg): Likewise.
3635 * ira-conflicts.c (ira_build_conflicts): Likewise.
3636 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3637 * lra-constraints.c (curr_insn_transform): Likewise.
3638 (split_reg): Likewise.
3639 * lra-eliminations.c (move_plus_up): Likewise.
3640 (lra_eliminate_regs_1): Likewise.
3641 * recog.c (general_operand): Likewise.
3642 * ree.c (combine_reaching_defs): Likewise.
3643 * reload.c (push_reload): Likewise.
3644 (find_reloads): Likewise.
3645 * reload1.c (elimination_effects): Likewise.
3646 (compute_reload_subreg_offset): Likewise.
3647 (choose_reload_regs): Likewise.
3648 * rtlanal.c (subreg_lsb_1): Likewise.
3649 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3650 (simplify_subreg): Likewise.
3651 * var-tracking.c (track_loc_p): Likewise.
3652 * emit-rtl.c (byte_lowpart_offset): Likewise.
3653 (paradoxical_subreg_p): Delete out-of-line definition.
3655 2017-08-22 Jeff Law <law@redhat.com>
3657 PR tree-optimization/81741
3658 PR tree-optimization/71947
3659 * tree-ssa-dom.c: Include tree-inline.h.
3660 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3661 equivalences if one is more expensive to compute than the other.
3662 * tree-ssa-scopedtables.h (class const_or_copies): Make
3663 record_const_or_copy_raw method private.
3664 (class avail_exprs_stack): New method simplify_binary_operation.
3665 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3666 avail_exprs_stack::simplify_binary_operation as needed.
3667 (avail_exprs_stack::simplify_binary_operation): New function.
3669 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3671 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3672 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3673 (DOT_SYMBOLS): Likewise.
3674 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3675 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3676 (RS6000_ABI_NAME): Likewise.
3677 (TARGET_CMODEL): Likewise.
3678 (TOC_SECTION_ASM_OP): Likewise.
3679 (SET_CMODEL): New macro.
3680 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3682 2017-08-22 Richard Biener <rguenther@suse.de>
3684 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3685 to allow for free-lang-data replacements similar to verify_type_variant.
3687 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3688 Alan Hayward <alan.hayward@arm.com>
3689 David Sherwood <david.sherwood@arm.com>
3691 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3692 VOIDmode for the LABEL_REF.
3694 2017-08-22 Richard Biener <rguenther@suse.de>
3696 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3698 2017-08-22 Richard Biener <rguenther@suse.de>
3700 * common.opt (feliminate-dwarf2-dups): Ignore.
3701 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3702 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3703 same_die_p_wrap, compute_section_prefix,
3704 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3705 (comdat_symbol_id, comdat_symbol_number): Likewise.
3706 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3708 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3709 (output_die): Mark unreachable path unreachable.
3710 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3711 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3712 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3713 (dwarf2out_early_finish): Likewise.
3715 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
3717 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
3719 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
3722 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
3723 not VAR_P. Filter attribute warnings with OPT_Wattributes.
3724 (avr_attribute_table) <io, io_low, address>: Initialize
3725 .decl_required with true.
3727 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
3729 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
3730 undocumented debugging options.
3731 (-mvsx-scalar-double): Likewise.
3732 (-mallow-df-permute): Likewise.
3733 (-mvectorize-builtins): Likewise.
3734 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3735 (rs6000_builtin_vectorized_function): Likewise.
3736 (rs6000_builtin_md_vectorized_function): Likewise.
3737 (rs6000_opt_vars): Likewise.
3739 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
3742 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
3743 (*btrq_imm): Rename from *btrq.
3744 (*btcq_imm): Rename from *btcq.
3745 (btsc): New code attribute.
3746 (*<btsc><mode>): New insn pattern.
3747 (*btr<mode>): Ditto.
3748 (*<btsc><mode>_mask): New insn_and_split pattern.
3749 (*btr<mode>_mask): Ditto.
3751 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3752 Alan Hayward <alan.hayward@arm.com>
3753 David Sherwood <david.sherwood@arm.com>
3755 * function.c (pad_below): Simplify padding calculation.
3757 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3758 Alan Hayward <alan.hayward@arm.com>
3759 David Sherwood <david.sherwood@arm.com>
3761 * target.def (function_prologue): Remove frame size argument.
3762 (function_epilogue): Likewise.
3763 * doc/tm.texi: Regenerate.
3764 * final.c (final_start_function): Update call to function_prologue.
3765 (final_end_function): Update call to function_epilogue.
3766 (default_function_pro_epilogue): Remove frame size argument.
3767 * output.h (default_function_pro_epilogue): Likewise.
3768 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3769 (arm_output_function_prologue): Likewise.
3770 * config/frv/frv.c (frv_function_prologue): Likewise.
3771 (frv_function_epilogue): Likewise.
3772 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3773 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3774 (ia64_output_function_epilogue): Likewise.
3775 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3776 (m32r_output_function_epilogue): Likewise.
3777 * config/microblaze/microblaze.c (microblaze_function_prologue)
3778 (microblaze_function_epilogue): Likewise.
3779 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3780 (mips_output_function_epilogue): Likewise.
3781 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3782 (mmix_target_asm_function_epilogue): Likewise.
3783 * config/msp430/msp430.c (msp430_start_function): Likewise.
3784 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3785 (nds32_asm_function_epilogue): Likewise.
3786 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3787 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3788 (pa_output_function_epilogue): Likewise.
3789 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3790 (rs6000_output_function_epilogue): Likewise.
3791 * config/rl78/rl78.c (rl78_start_function): Likewise.
3792 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3793 (rs6000_output_function_epilogue): Likewise.
3794 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3795 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3796 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3797 (sparc_asm_function_epilogue): Likewise.
3799 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3801 * tree.h (type_has_mode_precision_p): New function.
3802 * convert.c (convert_to_integer_1): Use it.
3803 * expr.c (expand_expr_real_2): Likewise.
3804 (expand_expr_real_1): Likewise.
3805 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3806 * match.pd: Likewise.
3807 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3808 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3809 * tree-tailcall.c (process_assignment): Likewise.
3810 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3811 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3812 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3813 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3814 (vectorizable_assignment): Likewise.
3815 (vectorizable_shift): Likewise.
3816 (vectorizable_operation): Likewise.
3817 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3819 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3821 * match.pd: Add pow (C, x) simplification.
3823 2017-08-21 Richard Biener <rguenther@suse.de>
3825 PR tree-optimization/81900
3826 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3827 for blocks with abnormal predecessors.
3828 (compute_antic): Do not set visited flag prematurely.
3830 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3833 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3835 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3837 * stor-layout.h (vector_type_mode): Move to...
3838 * tree.h (vector_type_mode): ...here.
3839 * stor-layout.c (vector_type_mode): Move to...
3840 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3842 2017-08-21 Richard Biener <rguenther@suse.de>
3844 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3845 register_external_die hooks.
3846 (debug_false_tree_charstarstar_uhwistar): Declare.
3847 (debug_nothing_tree_charstar_uhwi): Likewise.
3848 * debug.c (do_nothing_debug_hooks): Adjust.
3849 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3850 (debug_nothing_tree_charstar_uhwi): Likewise.
3851 * dbxout.c (dbx_debug_hooks): Adjust.
3852 (xcoff_debug_hooks): Likewise.
3853 * sdbout.c (sdb_debug_hooks): Likewise.
3854 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3855 * dwarf2out.c (macinfo_label_base): New global.
3856 (dwarf2out_register_external_die): New function for the
3857 register_external_die hook.
3858 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3859 (dwarf2_debug_hooks): Use them.
3860 (dwarf2_lineno_debug_hooks): Adjust.
3861 (struct die_struct): Add with_offset flag.
3862 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3863 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3864 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3865 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3866 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3867 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3868 defining section names for the early LTO debug variants.
3869 (reset_indirect_string): New helper.
3870 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3871 (print_dw_val): Add support for offsetted symbol references.
3872 (get_ultimate_context): Split out from is_cxx.
3873 (is_cxx): Use get_ultimate_context.
3874 (is_fortran): Add decl overload.
3875 (compute_comp_unit_symbol): Split out worker from
3876 compute_section_prefix.
3877 (compute_section_prefix): Call compute_comp_unit_symbol and
3878 set comdat_type_p here.
3879 (output_die): Skip DIE symbol output for the LTO added one.
3880 Handle DIE symbol references with offset.
3881 (output_comp_unit): Guard section name mangling properly.
3882 For LTO debug sections emit a symbol at the section beginning
3883 which we use to refer to its DIEs.
3884 (add_abstract_origin_attribute): For DIEs registered via
3885 dwarf2out_register_external_die directly refer to the early
3886 DIE rather than indirectly through the shadow one we created.
3887 Remove obsolete call to dwarf2out_abstract_function for
3888 non-function/block origins.
3889 (gen_array_type_die): When generating early LTO debug do
3890 not emit DW_AT_string_length.
3891 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3892 late when in LTO. As suggested place a gcc_unreachable for
3893 the DECL_ABSTRACT_P case.
3894 (gen_subprogram_die): Avoid another specification DIE
3895 for early built declarations/definitions for the late LTO case.
3896 (gen_variable_die): Add type references for late duplicated VLA dies
3898 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3899 we have the abstract instance already.
3900 (process_scope_var): Adjust decl DIE contexts in LTO which
3901 first puts them in limbo.
3902 (gen_decl_die): Do not generate type DIEs late apart from
3903 types for VLAs or for decls we do not yet have a DIE. Do not
3904 call dwarf2out_abstract_function late.
3905 (dwarf2out_early_global_decl): Make sure to create DIEs
3906 for abstract instances of a decl first.
3907 (dwarf2out_late_global_decl): Adjust comment.
3908 (output_macinfo_op): With multiple macro sections use
3909 macinfo_label_base to distinguish labels.
3910 (output_macinfo): Likewise. Update macinfo_label_base.
3911 Pass in the line info label.
3912 (note_variable_value_in_expr): When generating LTO resolve
3913 all variable values here by generating DIEs as needed.
3914 (init_sections_and_labels): Add early LTO debug flag parameter
3915 and generate different sections and names if set. Add generation
3916 counter for the labels so we can have multiple of them.
3917 (reset_dies): Helper to allow DIEs to be output multiple times.
3918 (dwarf2out_finish): When outputting DIEs to the fat part of an
3919 LTO object first reset DIEs.
3920 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3921 (modified_type_die): Check for decl_ultimate_origin being self
3923 (gen_type_die_with_usage): Likewise.
3924 (gen_typedef_die): Allow decl_ultimate_origin being self.
3925 (set_decl_abstract_flags): Remove.
3926 (set_block_abstract_flags): Likewise.
3927 (dwarf2out_abstract_function): Treat the early generated DIEs
3928 as the abstract copy and only add DW_AT_inline and
3929 DW_AT_artificial here and call set_decl_origin_self.
3930 If the DIE has an abstract origin don't do anything.
3931 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
3932 if we have none yet (Go fails to build one, PR78628).
3933 (variably_modified_type_p): Prevent endless recursion for Ada
3934 cyclic pointer types.
3935 * lto-streamer-in.c: Include debug.h.
3936 (dref_queue): New global.
3937 (lto_read_tree_1): Stream in DIE references.
3938 (lto_input_tree): Register DIE references.
3939 (input_function): Stream DECL_DEBUG_ARGS.
3940 * lto-streamer-out.c: Include debug.h.
3941 (lto_write_tree_1): Output DIE references.
3942 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
3943 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
3944 (output_function): Stream DECL_DEBUG_ARGS.
3945 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
3946 Stream DECL_ABSTRACT_ORIGIN.
3947 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
3948 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
3949 DECL_CONTEXT for file-scope decls.
3950 * lto-streamer.h (struct dref_entry): Declare.
3951 (dref_queue): Likewise.
3952 * cfgexpand.c (pass_expand::execute): Do not call the
3953 outlining_inline_function hook here.
3954 * lto-wrapper.c (debug_obj): New global.
3955 (tool_cleanup): Unlink it if required.
3956 (debug_objcopy): New function.
3957 (run_gcc): Handle early debug sections in the IL files by
3958 extracting them to separate files, partially linkin them and
3959 feeding the result back as result to the linker.
3960 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
3961 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
3962 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
3963 sections into a separate segment.
3964 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
3966 (darwin_asm_dwarf_section): Likewise.
3967 (darwin_asm_output_dwarf_offset): Likewise.
3968 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
3970 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3971 Alan Hayward <alan.hayward@arm.com>
3972 David Sherwood <david.sherwood@arm.com>
3974 * read-md.h (md_reader::record_potential_iterator_use): Replace
3975 pointer argument with an rtx and an index.
3976 * read-rtl.c (iterator_group::apply_iterator): Likewise.
3977 (apply_mode_iterator): Likewise.
3978 (apply_code_iterator): Likewise.
3979 (apply_int_iterator): Likewise.
3980 (apply_subst_iterator): Likewise.
3981 (record_iterator_use): Likewise.
3982 (record_attribute_use): Likewise.
3983 (md_reader::record_potential_iterator_use): Likewise. Update calls
3984 to record_iterator_use and apply_iterator.
3985 (iterator_use): Replace ptr with x and index.
3986 (attribute_use): Likewise.
3987 (apply_attribute_uses): Update calls to apply_iterator.
3988 (apply_iterators): Likewise. Update initialization of iterator_use.
3989 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
3990 and record_potential_iterator_use.
3991 (rtx_reader::read_rtx_operand): Likewise.
3993 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3994 Alan Hayward <alan.hayward@arm.com>
3995 David Sherwood <david.sherwood@arm.com>
3997 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4000 2017-08-21 Richard Biener <rguenther@suse.de>
4003 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4004 at struct end conservatively when comparing common bases.
4006 2017-08-21 Richard Biener <rguenther@suse.de>
4008 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4009 (mem_ref_in_stmt): Remove.
4010 (determine_max_movement): Use ref index to get at the reference.
4011 (invariantness_dom_walker::before_dom_children): Deal with
4012 lim data already initialized.
4013 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4015 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4017 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4018 (TARGET_ISA_ROUND): Ditto.
4019 (TARGET_ROUND): Ditto.
4020 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4021 * config/i386/i386.md: Ditto.
4022 * config/i386/sse.md: Ditto.
4023 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4024 with OPTION_MASK_ISA_SSE4_1.
4026 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4029 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4030 __builtin_ia32_lzcnt_u16.
4032 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4035 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4036 (rs6000_set_current_function): Rewrite function to use it.
4038 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4041 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4042 and TYPE_WARN_IF_NOT_ALIGN.
4043 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4044 (handle_warn_if_not_align): New.
4045 (place_union_field): Call handle_warn_if_not_align.
4046 (place_field): Call handle_warn_if_not_align.
4047 Copy TYPE_WARN_IF_NOT_ALIGN.
4048 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4049 (layout_type): Likewise.
4050 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4052 (tree_decl_common): Add warn_if_not_align.
4053 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4054 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4055 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4056 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4057 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4058 * doc/extend.texi: Document warn_if_not_aligned attribute.
4059 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4061 2017-08-17 Martin Liska <mliska@suse.cz>
4064 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4065 (get_data_dependence): Use it as pointer type.
4066 (distribute_loop): Likewise.
4068 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4070 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4071 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4072 (p8_vmrgow_<mode>_direct): New define_insn.
4073 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4074 handle endianness for vmrgew and vmrgow permute patterns.
4076 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4078 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4079 * config/rs6000/rs6000-cpus.def: Remove comment.
4080 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4081 (POWERPC_MASKS): Likewise.
4082 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4083 use of TARGET_VSX_TIMODE.
4084 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4085 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4086 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4087 (rs6000_option_override_internal): Remove dead code.
4088 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4089 (rs6000_legitimize_reload_address): Likewise.
4090 (rs6000_legitimate_address_p): Likewise.
4091 (rs6000_opt_masks): Delete "vsx-timode".
4092 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4093 from function comment.
4094 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4095 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4096 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4098 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4099 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4100 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4103 (define_peephole2): Likewise.
4105 2017-08-17 Martin Sebor <msebor@redhat.com>
4108 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4109 past the end of an array.
4110 (test_pp_format): Add test cases.
4112 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4114 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4115 missing ECF_NOTHROW flags.
4117 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4120 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4121 operands residing in integer registers.
4122 (*vsx_le_perm_load_<mode>): Likewise.
4123 (*vsx_le_perm_store_<mode>): Likewise.
4124 (define_peephole2): Add peepholes to optimize the above.
4126 2017-08-17 Marek Polacek <polacek@redhat.com>
4129 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4130 to mimic what shorten_compare did. Change the return type to bool.
4131 (fold_cond_expr_with_comparison): Update call to
4132 operand_equal_for_comparison_p.
4133 (fold_ternary_loc): Likewise.
4135 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4137 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4139 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4141 2017-08-17 Richard Biener <rguenther@suse.de>
4143 * tree-ssa-structalias.c (solve_graph): When propagating
4144 to successors update the graphs succ edges and avoid duplicate work.
4146 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4149 * config/i386/i386.c (ix86_option_override_internal): Save target
4150 specific options after ix86_stack_protector_guard_reg was changed.
4152 2017-08-17 Richard Biener <rguenther@suse.de>
4154 PR tree-optimization/81827
4155 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4156 (new_var_info): Initialize it conservatively.
4157 (get_call_vi): Mark register vars.
4158 (new_scalar_tmp_constraint_exp): Likewise.
4159 (handle_rhs_call): Likewise.
4160 (handle_const_call): Likewise.
4161 (create_function_info_for): Likewise.
4162 (solve_constraints): Sort varinfos to separate register from
4163 non-register vars to pack points-to solution bitmaps during
4166 2017-08-17 Marek Polacek <polacek@redhat.com>
4168 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4170 2017-08-17 Richard Biener <rguenther@suse.de>
4172 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4173 to true when overflow is undefined and we saturated the result.
4175 2017-08-17 Alan Modra <amodra@gmail.com>
4178 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4179 Don't use store multiple if only one reg needs saving.
4180 (interesting_frame_related_regno): New function.
4181 (rs6000_frame_related): Don't emit frame info for regs that
4183 (rs6000_emit_epilogue): Likewise.
4185 2017-08-16 Nathan Sidwell <nathan@acm.org>
4187 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4188 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4189 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4190 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4191 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4192 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4193 (verify_type): Adjust for TYPE_BINFO move.
4194 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4195 process TYPE_BINFO directly.
4196 (hash_tree): Likewise.
4197 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4199 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4202 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4204 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4206 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4209 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4210 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4211 Add ix86_binary_operator_ok to insn constraint.
4212 (*iordi_1_bts): Ditto.
4213 (*xordi_1_btc): Ditto.
4214 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4215 Update corresponding peephole2 pattern.
4219 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4221 PR tree-optimization/81832
4222 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4223 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4225 2017-08-16 Marek Polacek <polacek@redhat.com>
4228 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4229 regarding size_zero_node.
4231 2017-08-16 Martin Liska <mliska@suse.cz>
4234 * config.gcc: Respect previously set extra_objs in case
4237 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4239 PR tree-optimization/81835
4240 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4241 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4242 not depend on the phi.
4244 2017-08-16 Alan Modra <amodra@gmail.com>
4246 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4249 2017-08-16 Alan Modra <amodra@gmail.com>
4251 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4252 (save_reg_p): ..into this. Update all callers.
4253 (first_reg_to_save): Simplify.
4255 2017-08-16 Alan Modra <amodra@gmail.com>
4257 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4260 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4264 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4265 constant count if that count is less than 32.
4267 2017-08-15 Nathan Sidwell <nathan@acm.org>
4269 * gcc.c (execute): Emit friendlier message if inferior is killed
4270 by an external cause.
4272 2017-08-15 Richard Biener <rguenther@suse.de>
4274 PR tree-optimization/81790
4275 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4276 CONSTRUCTORs from simplifying and VN.
4278 2017-08-14 Martin Sebor <msebor@redhat.com>
4280 * builtin-attrs.def: Add comments.
4282 2017-08-14 Martin Sebor <msebor@redhat.com>
4285 * doc/extend.texi (attribute nonstring): Document new attribute.
4287 2017-08-14 Martin Sebor <msebor@redhat.com>
4290 * tree-diagnostic.c (default_tree_printer): Handle %G.
4291 * gimple-pretty-print.h (percent_G_format): Declare new function.
4292 * gimple-pretty-print.c (percent_G_format): Define.
4293 * tree-pretty-print.c (percent_K_format): Add argument.
4295 2017-08-14 Martin Sebor <msebor@redhat.com>
4297 PR translation/79998
4298 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4299 Remove a stray space.
4301 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4304 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4305 (*iordi_1_bts): Ditto.
4306 (*xordi_1_btc): Ditto.
4308 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4311 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4312 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4314 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4315 quoted strings, and make more translator-friendly.
4316 (darwin_rs6000_override_options): Likewise.
4317 (rs6000_option_override_internal): Likewise.
4318 (rs6000_return_in_memory): Fix overlong line.
4319 (init_cmulative_args): Use quoted strings, and make more
4320 translator-friendly.
4321 (rs6000_pass_by_reference): Fix overlong line.
4322 (def_builtin): Use quoted strings.
4323 (altivec_expand_predicate_builtin): Use quoted strings, and make
4324 more translator-friendly.
4325 (htm_expand_builtin): Use quoted strings.
4326 (cpu_expand_builtin): Use quoted strings, and make more
4327 translator-friendly.
4328 (altivec_expand_builtin): Likewise.
4329 (paired_expand_predicate_builtin): Likewise.
4330 (rs6000_invalid_builtin): Likewise.
4331 (builtin_function_type): Use quoted strings.
4332 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4333 more translator-friendly.
4334 (rs6000_trampoline_init): Likewise.
4335 (rs6000_handle_altivec_attribute): Likewise.
4336 (rs6000_inner_target_options): Use quoted strings.
4337 (rs6000_disable_incompatible_switches): Likewise.
4338 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4339 strings, and make more translator-friendly.
4340 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4342 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4344 PR tree-optimization/81799
4345 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4346 cond_expr to simple gimple operand.
4348 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4351 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4353 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4357 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4358 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4359 TARGET_GASISR_PROLOGUES.
4360 * config/avr/avr.c (avr_option_override): Same.
4361 (avr_pass_pre_proep::execute): Same.
4363 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4366 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4367 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4369 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4371 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4372 define_insn_and_split. Split to a memory load from 0 in
4373 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4374 using PTR mode iterator.
4375 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4376 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4377 (*add_tp_<mode>): Redefine as define_insn_and_split.
4378 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4379 address space. Merge with *add_tp_x32 using PTR mode iterator.
4380 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4381 Split to an add with a memory load from 0 in
4382 DEFAULT_TLS_SEG_REG address space.
4384 2017-08-12 Andrew Pinski <apinski@cavium.com>
4386 * config/aarch64/aarch64-option-extensions.def (rdma):
4387 Fix feature string to what Linux prints out in /proc/cpuinfo.
4389 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4392 * dwarf2out.c (modified_type_die): For C typedef types that have
4393 an ultimate origin, process the ultimate origin instead of the
4395 (gen_typedef_die): Assert that input DECLs have no ultimate
4397 (gen_type_die_with_usage): For typedef variants that have an
4398 ultimate origin, just call gen_decl_die on the original DECL.
4399 (process_scope_var): Avoid creating DIEs for local typedefs and
4400 concrete static variables.
4402 2017-08-12 Alan Modra <amodra@gmail.com>
4406 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4407 match gnu-user.h startfile.
4408 (ENDFILE_LINUX_SPEC): Similarly.
4410 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4413 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4415 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4417 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4418 * config/aarch64/aarch64.md (mov<mode>): Change.
4419 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4420 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4421 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4423 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4425 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4426 for storage order barriers.
4428 2017-08-11 Martin Liska <mliska@suse.cz>
4431 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4432 variables of void type.
4434 2017-08-11 Martin Liska <mliska@suse.cz>
4436 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4437 TARGET_SUPPORTS_ALIASES.
4438 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4439 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4440 (optimize_weakref): Likewise.
4441 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4442 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4443 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4445 2017-08-11 Martin Liska <mliska@suse.cz>
4448 * config/i386/i386.c (make_resolver_func): Do complete
4449 refactoring of the function.
4451 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4454 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4455 * config/i386/i386.c (ix86_stack_protect_guard): Use
4456 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4457 * doc/invoke.texi (x86 Options): Document
4458 -mstack-protector-guard-symbol= option.
4460 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4462 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4463 * config/i386/i386.c (ix86_split_stack_guard): New function.
4464 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4465 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4466 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4467 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4468 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4469 (split_stack_space_check): Call ix86_split_stack_guard.
4471 2017-08-10 Martin Sebor <msebor@redhat.com>
4473 * print-tree.c (print_node): Print location using the established
4475 Replace spaces with colons.
4476 (debug_raw, debug): Ditto.
4478 2017-08-10 Martin Sebor <msebor@redhat.com>
4481 * pretty-print.c (pp_format): Correct the handling of %s precision.
4483 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4486 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4488 (ix86_finalize_stack_frame_flags): This. Also clear
4489 frame_pointer_needed if -fno-omit-frame-pointer is used without
4491 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4492 with ix86_finalize_stack_frame_flags.
4493 (ix86_expand_epilogue): Likewise.
4494 (ix86_expand_split_stack_prologue): Likewise.
4495 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4497 2017-08-10 Martin Liska <mliska@suse.cz>
4500 * c-attribs.c (handle_target_attribute):
4501 Report warning for an empty string argument of target attribute.
4503 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4506 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4508 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4509 or DECL_NONLOCAL labels.
4510 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4511 or DECL_NONLOCAL labels here.
4513 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4516 to indicate when early gimple folding has been disabled.
4517 (rs6000_gimple_fold_builtin): Add debug content.
4518 (rs6000_invalid_builtin): Fix whitespace.
4519 (rs6000_expand_builtin): Fix whitespace.
4520 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4522 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4525 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4526 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4528 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4530 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4531 (qdf24xx): Likewise.
4532 * config/aarch64/aarch64.md: Include falkor.md.
4533 * config/aarch64/falkor.md: New.
4535 2017-08-09 Marek Polacek <polacek@redhat.com>
4538 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4539 * diagnostic.c (emit_diagnostic): Add a comment.
4540 (emit_diagnostic_valist): New function.
4542 2017-08-09 Marek Polacek <polacek@redhat.com>
4545 * input.c (make_location): New overload.
4546 * input.h (make_location): Declare.
4548 2017-08-08 Alan Modra <amodra@gmail.com>
4549 H.J. Lu <hongjiu.lu@intel.com>
4552 * gcc.c (NO_PIE_SPEC): Delete.
4553 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4555 (LINK_PIE_SPEC): ..to here.
4556 (LINK_COMMAND_SPEC): Support -no-pie.
4557 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4558 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4560 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4561 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4562 (ENDFILE_CRTEND_SPEC): Similarly.
4564 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4567 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4568 (mstack-protector-guard-offset=): Ditto.
4569 * config/i386/i386.c (ix86_option_override): Handle
4570 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4572 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4573 ix86_stack_protect_guard_offset variables.
4574 (TARGET_STACK_PROTECT_GUARD): Always define.
4575 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4576 and -mstack-protector-guard-offset= options.
4578 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4580 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4581 boundary case for the last candidate.
4583 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4585 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4586 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4588 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4591 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4592 * config/aarch64/aarch64-builtins.c
4593 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4594 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4595 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4597 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4598 Andrew Pinski <pinskia@gmail.com>
4601 * internal-fn.def (XORSIGN): New.
4602 * optabs.def (xorsign_optab): New.
4603 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4604 (convert_expand_mult_copysign): New.
4605 (pass_optimize_widening_mul::execute): Call
4606 convert_expand_mult_copysign.
4608 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4610 PR tree-optimization/81354
4611 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4612 Insert on edges rather than explicitly creating landing pads.
4613 (analyze_candidates_and_replace): Commit edge inserts.
4615 2017-08-08 Richard Biener <rguenther@suse.de>
4618 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4619 (expand_simple_operations): Also look through ADDR_EXPRs with
4620 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4622 2017-08-08 Richard Biener <rguenther@suse.de>
4624 PR tree-optimization/81723
4625 * tree-vect-slp.c (struct bst_traits): New hash traits.
4626 (bst_fail): New global.
4627 (vect_build_slp_tree_2): New worker, split out from ...
4628 (vect_build_slp_tree): ... this now wrapping it with using
4629 bst_fail set to cache SLP tree build fails. Properly handle
4631 (vect_analyze_slp_instance): Allocate and free bst_fail.
4633 2017-08-08 Martin Liska <mliska@suse.cz>
4636 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4637 LABEL_DECLs that can be from a different function.
4639 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4641 PR tree-optimization/81744
4642 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4643 loop's number of iterations.
4645 2017-08-08 Martin Liska <mliska@suse.cz>
4647 * asan.c: Include header files.
4648 * attribs.c (build_decl_attribute_variant): New function moved
4650 (build_type_attribute_qual_variant): Likewise.
4651 (cmp_attrib_identifiers): Likewise.
4652 (simple_cst_list_equal): Likewise.
4653 (omp_declare_simd_clauses_equal): Likewise.
4654 (attribute_value_equal): Likewise.
4655 (comp_type_attributes): Likewise.
4656 (build_type_attribute_variant): Likewise.
4657 (lookup_ident_attribute): Likewise.
4658 (remove_attribute): Likewise.
4659 (merge_attributes): Likewise.
4660 (merge_type_attributes): Likewise.
4661 (merge_decl_attributes): Likewise.
4662 (merge_dllimport_decl_attributes): Likewise.
4663 (handle_dll_attribute): Likewise.
4664 (attribute_list_equal): Likewise.
4665 (attribute_list_contained): Likewise.
4666 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4667 (lookup_attribute_by_prefix): Likewise.
4668 * bb-reorder.c: Include header files.
4669 * builtins.c: Likewise.
4670 * calls.c: Likewise.
4671 * cfgexpand.c: Likewise.
4672 * cgraph.c: Likewise.
4673 * cgraphunit.c: Likewise.
4674 * convert.c: Likewise.
4675 * dwarf2out.c: Likewise.
4676 * final.c: Likewise.
4677 * fold-const.c: Likewise.
4678 * function.c: Likewise.
4679 * gimple-expr.c: Likewise.
4680 * gimple-fold.c: Likewise.
4681 * gimple-pretty-print.c: Likewise.
4682 * gimple.c: Likewise.
4683 * gimplify.c: Likewise.
4684 * hsa-common.c: Likewise.
4685 * hsa-gen.c: Likewise.
4686 * internal-fn.c: Likewise.
4687 * ipa-chkp.c: Likewise.
4688 * ipa-cp.c: Likewise.
4689 * ipa-devirt.c: Likewise.
4690 * ipa-fnsummary.c: Likewise.
4691 * ipa-inline.c: Likewise.
4692 * ipa-visibility.c: Likewise.
4694 * lto-cgraph.c: Likewise.
4695 * omp-expand.c: Likewise.
4696 * omp-general.c: Likewise.
4697 * omp-low.c: Likewise.
4698 * omp-offload.c: Likewise.
4699 * omp-simd-clone.c: Likewise.
4700 * opts-global.c: Likewise.
4701 * passes.c: Likewise.
4702 * predict.c: Likewise.
4703 * sancov.c: Likewise.
4704 * sanopt.c: Likewise.
4705 * symtab.c: Likewise.
4706 * toplev.c: Likewise.
4707 * trans-mem.c: Likewise.
4708 * tree-chkp.c: Likewise.
4709 * tree-eh.c: Likewise.
4710 * tree-into-ssa.c: Likewise.
4711 * tree-object-size.c: Likewise.
4712 * tree-parloops.c: Likewise.
4713 * tree-profile.c: Likewise.
4714 * tree-ssa-ccp.c: Likewise.
4715 * tree-ssa-live.c: Likewise.
4716 * tree-ssa-loop.c: Likewise.
4717 * tree-ssa-sccvn.c: Likewise.
4718 * tree-ssa-structalias.c: Likewise.
4719 * tree-ssa.c: Likewise.
4720 * tree-streamer-in.c: Likewise.
4721 * tree-vectorizer.c: Likewise.
4722 * tree-vrp.c: Likewise.
4724 * ubsan.c: Likewise.
4725 * varasm.c: Likewise.
4726 * varpool.c: Likewise.
4727 * tree.c: Remove functions moved to attribs.[ch].
4729 * config/aarch64/aarch64.c: Add attrs.h header file.
4730 * config/alpha/alpha.c: Likewise.
4731 * config/arc/arc.c: Likewise.
4732 * config/arm/arm.c: Likewise.
4733 * config/avr/avr.c: Likewise.
4734 * config/bfin/bfin.c: Likewise.
4735 * config/c6x/c6x.c: Likewise.
4736 * config/cr16/cr16.c: Likewise.
4737 * config/cris/cris.c: Likewise.
4738 * config/darwin.c: Likewise.
4739 * config/epiphany/epiphany.c: Likewise.
4740 * config/fr30/fr30.c: Likewise.
4741 * config/frv/frv.c: Likewise.
4742 * config/ft32/ft32.c: Likewise.
4743 * config/h8300/h8300.c: Likewise.
4744 * config/i386/winnt.c: Likewise.
4745 * config/ia64/ia64.c: Likewise.
4746 * config/iq2000/iq2000.c: Likewise.
4747 * config/lm32/lm32.c: Likewise.
4748 * config/m32c/m32c.c: Likewise.
4749 * config/m32r/m32r.c: Likewise.
4750 * config/m68k/m68k.c: Likewise.
4751 * config/mcore/mcore.c: Likewise.
4752 * config/microblaze/microblaze.c: Likewise.
4753 * config/mips/mips.c: Likewise.
4754 * config/mmix/mmix.c: Likewise.
4755 * config/mn10300/mn10300.c: Likewise.
4756 * config/moxie/moxie.c: Likewise.
4757 * config/msp430/msp430.c: Likewise.
4758 * config/nds32/nds32-isr.c: Likewise.
4759 * config/nds32/nds32.c: Likewise.
4760 * config/nios2/nios2.c: Likewise.
4761 * config/nvptx/nvptx.c: Likewise.
4762 * config/pa/pa.c: Likewise.
4763 * config/pdp11/pdp11.c: Likewise.
4764 * config/powerpcspe/powerpcspe.c: Likewise.
4765 * config/riscv/riscv.c: Likewise.
4766 * config/rl78/rl78.c: Likewise.
4767 * config/rx/rx.c: Likewise.
4768 * config/s390/s390.c: Likewise.
4769 * config/sh/sh.c: Likewise.
4770 * config/sol2.c: Likewise.
4771 * config/sparc/sparc.c: Likewise.
4772 * config/spu/spu.c: Likewise.
4773 * config/stormy16/stormy16.c: Likewise.
4774 * config/tilegx/tilegx.c: Likewise.
4775 * config/tilepro/tilepro.c: Likewise.
4776 * config/v850/v850.c: Likewise.
4777 * config/vax/vax.c: Likewise.
4778 * config/visium/visium.c: Likewise.
4779 * config/xtensa/xtensa.c: Likewise.
4781 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4784 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4785 constraints since the -mupper-regs-* switches have been
4787 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4788 into a vector from a double word element that was extracted from
4789 another vector, and eliminate extra XXPERMDI instructions.
4790 (vsx_concat_<mode>_2): Likewise.
4791 (vsx_concat_<mode>_3): Likewise.
4792 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4793 concat to allow optimizing inserts from previous extracts.
4795 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4797 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4798 memory reference to a SSP offset in TLS address space.
4799 (ix86_print_operand) <case '@'>: Remove.
4800 (ix86_print_operand_punct_valid_p): Remove '@' code.
4801 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4803 (stack_tls_protect_set_<mode>): Remove.
4804 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4805 (stack_tls_protect_test_<mode>): Remove.
4806 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4808 2017-08-07 Olivier Hainque <hainque@adacore.com>
4811 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4813 2017-08-07 Douglas Rupp <rupp@adacore.com>
4815 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4816 variable was referenced as multidir in command.
4818 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4821 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4824 2017-08-07 Martin Liska <mliska@suse.cz>
4826 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4827 * config/rl78/rl78.c: Add include of attribs.h.
4828 * config/sh/sh.c: Likewise.
4829 * config/v850/v850.c: Likewise.
4831 2017-08-07 Tom de Vries <tom@codesourcery.com>
4834 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4836 2017-08-07 Martin Liska <mliska@suse.cz>
4838 * config/mips/mips.c: Include attribs.h.
4840 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4843 * doc/invoke.texi: Document change in behvaior for -Ofast for
4846 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4848 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4850 (aarch64_pop_regs): Likewise.
4851 (aarch64_gen_load_pair): Likewise.
4852 (aarch64_save_callee_saves): Likewise.
4853 (aarch64_restore_callee_saves): Likewise.
4855 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4857 * config/i386/i386.c: Revert the last change.
4859 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4862 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4864 (ix86_finalize_stack_frame_flags): This. Also clear
4865 frame_pointer_needed if -fno-omit-frame-pointer is used without
4867 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4868 with ix86_finalize_stack_frame_flags.
4869 (ix86_expand_epilogue): Likewise.
4870 (ix86_expand_split_stack_prologue): Likewise.
4872 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4875 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4876 to P_AES for Westmere.
4878 2017-08-07 Jonathan Yong <10walls@gmail.com>
4880 * config/i386/mingw.opt (fset-stack-executable): Removed.
4881 * config/i386/cygming.opt (fset-stack-executable): Moved
4883 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4885 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4887 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4889 2017-08-07 Marek Polacek <polacek@redhat.com>
4892 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4894 2017-08-07 Martin Liska <mliska@suse.cz>
4896 * attribs.h (canonicalize_attr_name): New function.
4897 (cmp_attribs): Move from c-format.c and adjusted.
4898 (is_attribute_p): Moved from tree.h.
4899 * tree-inline.c: Add new includes.
4900 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4901 (private_is_attribute_p): Remove.
4902 (private_lookup_attribute): Likewise.
4903 (private_lookup_attribute_by_prefix): Simplify.
4904 (remove_attribute): Use is_attribute_p.
4905 * tree.h: Remove removed declarations.
4907 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4910 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4911 instead of computing it in the function. Formatting fix.
4912 (expand_case): Don't rely on default_edge being the first edge,
4913 clear it if removing it, pass default_edge to
4914 emit_case_dispatch_table.
4915 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4918 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4920 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4921 insn in the function, emit NOP after the insn.
4923 2017-08-06 Tom de Vries <tom@codesourcery.com>
4925 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
4928 2017-08-06 Tom de Vries <tom@codesourcery.com>
4930 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
4933 2017-08-04 Yury Gribov <tetra2005@gmail.com>
4935 PR tree-optimization/57371
4936 * match.pd: New pattern.
4938 2017-08-04 Marek Polacek <polacek@redhat.com>
4941 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
4942 perform the computation in offset_int.
4944 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4946 PR tree-optimization/81136
4947 * tree-vectorizer.h: Include tree-hash-traits.h.
4948 (vec_base_alignments): New typedef.
4949 (vec_info): Add a base_alignments field.
4950 (vect_record_base_alignments): Declare.
4951 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
4953 (DR_IS_CONDITIONAL_IN_STMT): New macro.
4954 (create_data_ref): Add an is_conditional_in_stmt argument.
4955 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
4956 the is_conditional_in_stmt field.
4957 (data_ref_loc): Add an is_conditional_in_stmt field.
4958 (get_references_in_stmt): Set the is_conditional_in_stmt field.
4959 (find_data_references_in_stmt): Update call to create_data_ref.
4960 (graphite_find_data_references_in_stmt): Likewise.
4961 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
4962 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
4963 (vect_record_base_alignment): New function.
4964 (vect_record_base_alignments): Likewise.
4965 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
4966 for nested statements even if we fail to compute a misalignment.
4967 Use pooled base alignments for unconditional references.
4968 (vect_find_same_alignment_drs): Compare base addresses instead
4970 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
4971 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4973 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4975 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
4976 Add an explicit name for the enum. Use auto_vec for slp_instances
4978 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
4980 (_bb_vec_info): Add a constructor and destructor.
4981 (vinfo_for_stmt): Return NULL for uids of -1 as well.
4982 (destroy_loop_vec_info): Delete.
4983 (vect_destroy_datarefs): Likewise.
4984 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
4985 (vec_info::vec_info): New function.
4986 (vec_info::~vec_info): Likewise.
4987 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
4988 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
4989 destroy_loop_vec_info.
4990 * tree-vect-loop.c (new_loop_vec_info): Replace with...
4991 (_loop_vec_info::_loop_vec_info): ...this.
4992 (destroy_loop_vec_info): Replace with...
4993 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
4994 the stmt_vec_infos. Leave handling of vec_info information to its
4995 destructor. Remove explicit vector releases.
4996 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
4997 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
4998 * tree-vect-slp.c (new_bb_vec_info): Replace with...
4999 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5000 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5001 (destroy_bb_vec_info): Replace with...
5002 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5003 information to its destructor.
5004 (vect_slp_analyze_bb_1): Use new and delete instead of
5005 new_bb_vec_info and destroy_bb_vec_info.
5006 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5009 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5011 * tree-data-ref.h (subscript): Add access_fn field.
5012 (data_dependence_relation): Add could_be_independent_p.
5013 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5014 (same_access_functions): Move to tree-data-ref.c.
5015 * tree-data-ref.c (ref_contains_union_access_p): New function.
5016 (access_fn_component_p): Likewise.
5017 (access_fn_components_comparable_p): Likewise.
5018 (dr_analyze_indices): Add a reference to access_fn_component_p.
5019 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5021 (constant_access_functions): Likewise.
5022 (add_other_self_distances): Likewise.
5023 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5024 (initialize_data_dependence_relation): Use XCNEW and remove
5025 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5026 of access functions that have the same type. Allow the
5027 subsequence to end with different bases in some circumstances.
5028 Record the chosen access functions in SUB_ACCESS_FN.
5029 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5030 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5031 (subscript_dependence_tester_1): Likewise dra and drb.
5032 (build_classic_dist_vector): Update calls accordingly.
5033 (subscript_dependence_tester): Likewise.
5034 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5035 DDR_COULD_BE_INDEPENDENT_P.
5036 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5037 comp_alias_ddrs instead of may_alias_ddrs.
5038 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5040 (vect_analyze_data_ref_dependence): Use it if
5041 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5042 distance vectors if that fails.
5043 (dependence_distance_ge_vf): New function.
5044 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5045 LOOP_VINFO_MAY_ALIAS_DDRS.
5047 2017-08-04 Richard Biener <rguenther@suse.de>
5050 * fold-const.c (fold_binary_loc): Properly restrict
5051 minus_var0 && minus_var1 case when associating undefined overflow
5054 2017-08-04 Tom de Vries <tom@codesourcery.com>
5056 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5058 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5060 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5061 Don't start diagnostic messages with a capital letter.
5062 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5064 (rs6000_invalid_builtin): Likewise.
5065 (rs6000_trampoline_init): Likewise.
5067 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5070 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5071 after setting changeable df flags.
5073 2017-08-03 Richard Biener <rguenther@suse.de>
5075 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5076 up if the use is in USE - X.
5078 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5080 * toplev.c (dumpfile.h): New include.
5081 (internal_error_reentered): New static function. Use it...
5082 (internal_error_function): ...here to handle reentered internal_error.
5084 2017-08-03 Richard Biener <rguenther@suse.de>
5087 * fold-const.c (split_tree): Add minus_var and minus_con
5088 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5089 here but always use minus_*.
5090 (associate_trees): Assert we never associate with MINUS_EXPR
5091 and NULL first operand. Do not recurse for PLUS_EXPR operands
5092 when associating as MINUS_EXPR either.
5093 (fold_binary_loc): Track minus_var and minus_con.
5095 2017-08-03 Tom de Vries <tom@codesourcery.com>
5098 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5099 ACCEL_COMPILER, apply finish_options on
5100 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5102 2017-08-03 Tom de Vries <tom@codesourcery.com>
5105 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5106 function_entry_patch_area_size > 0.
5108 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5111 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5112 instead of 10??LU, perform unit multiplication in wide_int,
5113 don't change alloc_object_size_limit if the limit is larger
5116 PR tree-optimization/81655
5117 PR tree-optimization/81588
5118 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5119 the case when ranges[i].low and high are 1 for unsigned type with
5123 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5124 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5126 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5128 * tree-vrp.h: Add include guard.
5130 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5133 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5134 (ud2): New insn pattern.
5135 * config/i386/i386.c (ix86_expand_epilogue):
5136 For naked functions, generate ud2 instead of trap insn.
5138 2017-08-02 Marek Polacek <polacek@redhat.com>
5141 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5143 2017-08-02 Tom de Vries <tom@codesourcery.com>
5144 Cesar Philippidis <cesar@codesourcery.com>
5146 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5147 Add missing edge probabilities.
5149 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5151 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5154 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5157 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5158 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5159 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5160 prologue_seq sequences - if any.
5162 2017-08-02 Richard Biener <rguenther@suse.de>
5164 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5165 via vectors if supported, integer extracts via punning if supported
5166 or otherwise vector extracts.
5168 2017-08-02 Richard Biener <rguenther@suse.de>
5170 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5172 (bitmap_insert_into_set): ... this.
5174 2017-08-02 Richard Biener <rguenther@suse.de>
5176 PR tree-optimization/81633
5178 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5180 PR tree-optimization/71752
5181 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5183 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5185 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5186 (machine_function::call_ms2sysv_pad_out): Remove field.
5187 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5188 (ix86_compute_frame_layout): Likewise.
5190 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5193 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5194 attribute with interrupt attribute.
5196 2017-08-01 Andrew Pinski <apinski@cavium.com>
5198 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5199 BIT_INSERT_EXPR's operand 1
5200 to see if the types precision matches.
5202 2017-08-01 Martin Liska <mliska@suse.cz>
5205 * builtins.c (expand_builtin_memcpy_args): Remove.
5206 (expand_builtin_memcpy): Call newly added function
5207 expand_builtin_memory_copy_args.
5208 (expand_builtin_memcpy_with_bounds): Likewise.
5209 (expand_builtin_mempcpy): Remove last argument.
5210 (expand_builtin_mempcpy_with_bounds): Likewise.
5211 (expand_builtin_memory_copy_args): New function created from
5212 expand_builtin_mempcpy_args with small modifications.
5213 (expand_builtin_mempcpy_args): Remove.
5214 (expand_builtin_stpcpy): Remove unused argument.
5215 (expand_builtin): Likewise.
5216 (expand_builtin_with_bounds): Likewise.
5218 2017-08-01 Martin Liska <mliska@suse.cz>
5221 Make mempcpy more optimal (PR middle-end/70140).
5223 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5226 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5227 __builtin_vec_cmpne verify both arguments are compatible vectors
5228 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5229 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5230 move computation of aligned to after checking the argument types.
5234 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5235 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5238 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5239 Doug Rupp <rupp@adacore.com>
5240 Olivier Hainque <hainque@adacore.com>
5242 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5243 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5245 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5246 for TARGET_OS_CPP_BUILTIN.
5247 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5248 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5250 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5251 passing required abi options to the assembler for EABI configurations.
5252 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5253 of .text.hot and .text.unlikely sections for kernel modules when
5254 using ARM style exceptions.
5255 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5256 options. Add EXTRA_CC1_SPEC.
5257 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5259 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5261 (ARM_TARGET2_DWARF_FORMAT): Define.
5262 * config/arm/t-vxworks: Adjust multilib control to removal of the
5263 Diab command line options.
5265 2017-08-01 Martin Liska <mliska@suse.cz>
5267 PR gcov-profile/81561
5268 * gcov.c (unblock): Make unblocking safe as we need to preserve
5269 index correspondence of blocks and block_lists.
5271 2017-08-01 Richard Biener <rguenther@suse.de>
5273 PR tree-optimization/81181
5274 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5275 (compute_antic): ... end of iteration here.
5277 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5279 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5280 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5281 (ftree-slp-vectorize): Likewise.
5282 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5283 can no longer be set independent of flag_tree_loop_vectorize.
5284 * omp-general.c (emp_max_vf): Likewise.
5285 * opts.c (enable_fdo_optimizations): Remove references to
5286 flag_tree_vectorize, these are now implicit.
5287 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5288 and leave it for the options machinery.
5290 2017-08-01 Martin Liska <mliska@suse.cz>
5293 * builtins.c (expand_builtin_memcpy_args): Remove.
5294 (expand_builtin_memcpy): Call newly added function
5295 expand_builtin_memory_copy_args.
5296 (expand_builtin_memcpy_with_bounds): Likewise.
5297 (expand_builtin_mempcpy): Remove last argument.
5298 (expand_builtin_mempcpy_with_bounds): Likewise.
5299 (expand_builtin_memory_copy_args): New function created from
5300 expand_builtin_mempcpy_args with small modifications.
5301 (expand_builtin_mempcpy_args): Remove.
5302 (expand_builtin_stpcpy): Remove unused argument.
5303 (expand_builtin): Likewise.
5304 (expand_builtin_with_bounds): Likewise.
5306 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5309 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5310 print "ds:" only for immediates in generic address space.
5312 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5315 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5316 (ix86_function_ok_for_sibcall): Return false for naked functions.
5318 2017-08-01 Richard Biener <rguenther@suse.de>
5320 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5321 (compute_antic): Seed worklist with exit block predecessors.
5322 * cfganal.c (dfs_find_deadend): For a cycle return the source
5323 of the edge closing it.
5325 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5327 * config/aarch64/aarch64.c
5328 (aarch64_can_const_movi_rtx_p): Move 0 check.
5330 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5332 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5333 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5335 * match.pd: Ditto in address comparison pattern.
5337 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5339 PR tree-optimization/81627
5340 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5341 closed ssa form for store-store chain.
5343 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5345 PR tree-optimization/81620
5346 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5347 for store-store chain.
5349 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5351 PR tree-optimization/81588
5352 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5353 ranges[i].in_p, invert comparison code ccode. For >/>=,
5354 swap rhs1 and rhs2 and comparison code unconditionally,
5355 for </<= don't do that. Don't swap rhs1/rhs2 again if
5356 ranges[i].in_p, instead invert comparison code ccode if
5357 opcode or oe->rank is BIT_IOR_EXPR.
5360 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5361 a direct optab to conversion optab.
5362 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5363 with GET_MODE_INNER as last argument instead of optab_handler.
5364 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5365 vector extraction if possible and optab is available.
5366 * expr.c (store_constructor): Use convert_optab_handler instead
5367 of optab_handler. Use vector initialization from smaller
5368 vectors if possible and optab is available.
5369 * tree-vect-stmts.c (vectorizable_load): Likewise.
5370 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5372 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5373 of vec_init from half-sized vectors with the same element mode.
5374 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5375 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5376 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5377 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5378 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5379 after mode in gen_vec_extract* calls.
5380 (vec_extract<mode>): Renamed to ...
5381 (vec_extract<mode><ssescalarmodelower>): ... this.
5382 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5383 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5384 element mode after mode in gen_vec_init* calls.
5385 (VEC_INIT_HALF_MODE): New mode iterator.
5386 (vec_init<mode>): Renamed to ...
5387 (vec_init<mode><ssescalarmodelower>): ... this.
5388 (vec_init<mode><ssehalfvecmodelower>): New expander.
5389 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5390 (vec_extractv2sfsf): ... this.
5391 (vec_initv2sf): Renamed to ...
5392 (vec_initv2sfsf): ... this.
5393 (vec_extractv2si): Renamed to ...
5394 (vec_extractv2sisi): ... this.
5395 (vec_initv2si): Renamed to ...
5396 (vec_initv2sisi): ... this.
5397 (vec_extractv4hi): Renamed to ...
5398 (vec_extractv4hihi): ... this.
5399 (vec_initv4hi): Renamed to ...
5400 (vec_initv4hihi): ... this.
5401 (vec_extractv8qi): Renamed to ...
5402 (vec_extractv8qiqi): ... this.
5403 (vec_initv8qi): Renamed to ...
5404 (vec_initv8qiqi): ... this.
5405 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5406 (vec_init<mode>): Renamed to ...
5407 (vec_init<mode><VEC_base_l>): ... this.
5408 (vec_extract<mode>): Renamed to ...
5409 (vec_extract<mode><VEC_base_l>): ... this.
5410 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5411 (vec_initv2sfsf): ... this.
5412 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5413 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5414 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5415 element mode after mode in gen_vec_init* calls.
5416 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5417 (vec_init<mode><Vel>): ... this.
5418 (vec_extract<mode>): Renamed to ...
5419 (vec_extract<mode><Vel>): ... this.
5420 * config/aarch64/iterators.md (Vel): New mode attribute.
5421 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5422 Add element mode after mode in gen_vec_extract* calls.
5423 * config/s390/vector.md (non_vec_l): New mode attribute.
5424 (vec_extract<mode>): Renamed to ...
5425 (vec_extract<mode><non_vec_l>): ... this.
5426 (vec_init<mode>): Renamed to ...
5427 (vec_init<mode><non_vec_l>): ... this.
5428 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5429 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5431 * config/arm/iterators.md (V_elem_l): New mode attribute.
5432 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5433 (vec_extract<mode><V_elem_l>): ... this.
5434 (vec_extractv2di): Renamed to ...
5435 (vec_extractv2didi): ... this.
5436 (vec_init<mode>): Renamed to ...
5437 (vec_init<mode><V_elem_l>): ... this.
5438 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5439 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5440 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5441 Add element mode after gen_vec_extract* calls.
5442 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5443 (vec_init<mode><unitmode>): ... this.
5444 (vec_extract<mode>): Renamed to ...
5445 (vec_extract<mode><unitmode>): ... this.
5446 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5447 (vec_init<mode><unitmode>): ... this.
5448 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5449 (vec_initv2sfsf): ... this.
5450 (vec_extractv2sf): Renamed to ...
5451 (vec_extractv2sfsf): ... this.
5452 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5453 Add element mode after gen_vec_extract* calls.
5454 * config/mips/mips.md (unitmode): New mode iterator.
5455 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5456 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5457 * config/spu/spu.md (inner_l): New mode attribute.
5458 (vec_init<mode>): Renamed to ...
5459 (vec_init<mode><inner_l>): ... this.
5460 (vec_extract<mode>): Renamed to ...
5461 (vec_extract<mode><inner_l>): ... this.
5462 * config/sparc/sparc.md (veltmode): New mode iterator.
5463 (vec_init<VMALL:mode>): Renamed to ...
5464 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5465 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5466 (vec_initv2sisi): ... this.
5467 (vec_initv2sf): Renamed to ...
5468 (vec_initv2sfsf): ... this.
5469 (vec_extractv2sf): Renamed to ...
5470 (vec_extractv2sfsf): ... this.
5471 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5472 (vec_init<mode>): Renamed to ...
5473 (vec_init<mode><VEC_base_l>): ... this.
5474 (vec_extract<mode>): Renamed to ...
5475 (vec_extract<mode><VEC_base_l>): ... this.
5476 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5477 (vec_initv2sfsf): ... this.
5478 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5479 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5480 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5481 gen_vec_init* calls.
5483 2017-08-01 Richard Biener <rguenther@suse.de>
5485 PR tree-optimization/81297
5486 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5487 TREE_OVERFLOW from INTEGER_CSTs.
5489 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5491 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5493 2017-07-31 Carl Love <cel@us.ibm.com>
5495 * config/rs6000/rs6000-c: Add support for built-in functions
5496 vector signed char vec_xl_be (signed long long, signed char *);
5497 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5498 vector signed int vec_xl_be (signed long long, signed int *);
5499 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5500 vector signed long long vec_xl_be (signed long long, signed long long *);
5501 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5502 vector signed short vec_xl_be (signed long long, signed short *);
5503 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5504 vector double vec_xl_be (signed long long, double *);
5505 vector float vec_xl_be (signed long long, float *);
5506 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5507 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5508 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5510 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5511 (altivec_expand_builtin): Add switch statement to call
5512 altivec_expand_xl_be for each builtin.
5513 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5514 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5515 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5516 __builtin_vsx_le_be_v16qi.
5517 * doc/extend.texi: Update the built-in documentation file for the
5518 new built-in functions.
5520 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5523 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5525 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5527 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5529 * config.gcc: Add z14.
5530 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5531 CPU model numbers for z13s and z14.
5532 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5534 * config/s390/s390-opts.h (enum processor_type): Rename
5535 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5536 * config/s390/s390.c (processor_table): Add field for CPU name to
5537 be passed to Binutils.
5538 (s390_asm_output_machine_for_arch): Use the new field in
5539 processor_table for Binutils.
5540 (s390_expand_builtin): Replace arch12 with z14.
5541 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5542 (s390_get_sched_attrmask): Likewise.
5543 (s390_get_unit_mask): Likewise.
5544 * config/s390/s390.opt: Add z14 to processor_type enum.
5546 2017-07-31 Martin Jambor <mjambor@suse.cz>
5549 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5550 regardless of optimization level.
5552 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5553 Martin Liska <mliska@suse.cz>
5555 * predict.def: Remove old comment and adjust probability.
5556 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5559 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5562 * config/i386/i386.c (ix86_function_naked): New function.
5563 (ix86_can_use_return_insn_p): Return false for naked functions.
5564 (ix86_expand_prologue): Skip prologue for naked functions.
5565 (ix86_expand_epilogue): Skip epilogue for naked functions
5566 and emit trap instruction.
5567 (ix86_warn_func_return): New function.
5568 (ix86_attribute_table): Add "naked" attribute specification.
5569 (TARGET_WARN_FUNC_RETURN): Define.
5570 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5572 2017-07-31 Martin Liska <mliska@suse.cz>
5574 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5575 (dump_gimple_bb_header): Always dump BB info.
5576 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5578 2017-07-31 Martin Liska <mliska@suse.cz>
5581 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5582 also with current_function_decl non-null equality.
5584 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5587 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5588 change type to the element type, instead add eltype variable and
5589 use it where we are interested in the element type.
5591 PR tree-optimization/81603
5592 * ipa-polymorphic-call.c
5593 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5594 offset arithmetic in offset_int, bail out if the resulting bit offset
5595 doesn't fit into shwi.
5597 2017-07-31 Martin Liska <mliska@suse.cz>
5599 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5600 (gimplify_save_expr): Fix comment.
5602 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5605 * config/i386/i386.c (ix86_function_arg): Update arguments for
5607 (ix86_compute_frame_layout): Set the initial stack offset to
5608 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5609 INCOMING_FRAME_SP_OFFSET.
5610 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5611 stack before exception handler returns.
5612 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5613 the 'ERROR_CODE' for exception handler.
5615 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5617 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5618 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5619 (ASM_OUTPUT_REG_POP): Ditto.
5620 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5621 instead of asm_fprintf to output pure string.
5623 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5625 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5626 to imported_module_or_decl hook.
5627 (debug_nothing_tree_tree_tree_bool): Remove.
5628 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5629 * debug.c (do_nothing_debug_hooks): Use
5630 debug_nothing_tree_tree_tree_bool_bool instead of
5631 debug_nothing_tree_tree_tree_bool.
5632 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5633 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5634 * sdbout.c (sdb_debug_hooks): Likewise.
5635 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5636 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5638 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5639 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5640 attribute, don't add anything.
5642 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5644 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5645 (fold_build2_stat_loc): Likewise.
5646 (fold_build3_stat_loc): Likewise.
5647 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5648 (fold_build1_loc): Remove macro.
5649 (fold_build2_loc): Likewise.
5650 (fold_build3_loc): Likewise.
5652 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5654 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5655 (gimple_build_debug_bind_source_stat): Likewise.
5656 * gimple.h (gimple_build_debug_bind): Remove macro.
5657 (gimple_build_debug_bind_source): Likewise.
5659 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5661 * bitmap.c (bitmap_alloc): Adjust.
5662 (bitmap_gc_alloc): Likewise.
5663 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5665 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5667 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5668 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5669 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5670 (bitmap_gc_alloc_stat): Likewise.
5671 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5673 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5675 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5676 * rtl.h (shallow_copy_rtx): Remove macro.
5678 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5680 * emit-rtl.c (gen_raw_REG): Adjust.
5681 * gengenrtl.c (gendef): Likewise.
5682 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5683 * rtl.h (rtx_alloc): Remove macro.
5685 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5687 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5688 (build_tree_list_stat): Likewise.
5689 * tree.h (build_tree_list): Remove macro.
5690 (build_tree_list_vec): Likewise.
5692 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5694 * tree.c (make_vector_stat): Remove _stat from name.
5695 (build_vector_stat): Likewise.
5696 * tree.h (make_vector_stat): Remove macro.
5697 (build_vector_stat): Likewise.
5699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5701 * tree.h (build_var_debug_value): Remove prototype.
5703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5705 * tree.c (tree_cons_stat): Remove _stat from name.
5706 * tree.h (tree_cons): Remove macro.
5708 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5710 * tree.c (build_vl_exp_stat): Remove _stat from name.
5711 * tree.h (build_vl_exp): Remove macro.
5713 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5715 * tree.c (build_decl_stat): Remove _stat from name.
5716 * tree.h (build_decl): Remove macro.
5718 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5720 * gimple.c (gimple_build_with_ops_stat): Adjust.
5721 (gimple_alloc_stat): Remove _stat from name.
5722 * gimple.h (gimple_alloc): Remove macro.
5724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5726 * tree.c (make_tree_vec_stat): Remove _stat from name.
5727 (grow_tree_vec_stat): Likewise.
5728 * tree.h (make_tree_vec_stat): Adjust prototype.
5729 (grow_tree_vec_stat): Likewise.
5730 (make_tree_vec): Remove macro.
5731 (grow_tree_vec): Likewise.
5733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5735 * fold-const.c (fold_build1_stat_loc): Adjust.
5736 (fold_build2_stat_loc): Likewise.
5737 (fold_build3_stat_loc): Likewise.
5738 * tree.c (build0_stat): Remove _stat from name.
5739 (build1_stat): Likewise.
5740 (build2_stat): Likewise.
5741 (build3_stat): Likewise.
5742 (build4_stat): Likewise.
5743 (build5_stat): Likewise.
5744 * tree.h (build1_loc): Remove macro, and rename _stat function
5746 (build2_loc): Likewise.
5747 (build3_loc): Likewise.
5748 (build4_loc): Likewise.
5749 (build5_loc): Likewise.
5751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5753 * tree.c (make_int_cst_stat): Remove _stat from name.
5754 * tree.h (make_int_cst_stat): Adjust prototype.
5755 (make_int_cst): Remove macro.
5757 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5759 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5760 * tree.h (make_tree_binfo_stat): Adjust prototype.
5761 (make_tree_binfo): Remove.
5763 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5765 * tree.c (copy_node_stat): Rename to copy_node.
5766 (build_distinct_type_copy): Adjust.
5767 * tree.h (copy_node_stat): Adjust prototype.
5768 (copy_node): Remove macro.
5770 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5772 * tree.c (make_node_stat): rename to make_node.
5773 (build_tree_list_stat): Adjust.
5774 (build0_stat): Likewise.
5775 (build2_stat): Likewise.
5776 (build3_stat): Likewise.
5777 (build4_stat): Likewise.
5778 (build5_stat): Likewise.
5779 (build_decl_stat): Likewise.
5780 * tree.h (make_node_stat): Adjust prototype.
5781 (make_node): remove macro.
5783 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5785 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5786 (PPC_FEATURE2_SCV): Likewise.
5787 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5789 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5791 * config/aarch64/aarch64.c
5792 (aarch64_internal_mov_immediate): Add new special pattern.
5793 * config/aarch64/aarch64.md (*movdi_aarch64):
5794 Add reg/32bit const mov case.
5796 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5797 Richard Sandiford <richard.sandiford@linaro.org>
5799 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5800 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5801 Add integer and movi cases.
5802 (movi-split-hf-df-sf split, fp16): New.
5803 (enabled): Added TARGET_FP_F16INST.
5804 * config/aarch64/iterators.md (GPF_HF): New.
5805 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5807 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5809 * config/aarch64/aarch64.c
5810 (aarch64_simd_container_mode): Add prototype.
5811 (aarch64_expand_mov_immediate): Add HI support.
5812 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5813 (aarch64_can_const_movi_rtx_p): New.
5814 (aarch64_preferred_reload_class):
5815 Remove restrictions of using FP registers for certain SIMD operations.
5816 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5817 (aarch64_valid_floating_const): Add integer move validation.
5818 (aarch64_simd_imm_scalar_p): Remove.
5819 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5820 (aarch64_legitimate_constant_p): Expand list of supported cases.
5821 * config/aarch64/aarch64-protos.h
5822 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5823 (aarch64_reinterpret_float_as_int): New.
5824 (aarch64_simd_imm_scalar_p): Remove.
5825 * config/aarch64/constraints.md (Uvi): New.
5826 (Dd): Split into Ds and new Dd.
5827 * config/aarch64/aarch64.md (*movsi_aarch64):
5829 (*movdi_aarch64): Add SIMD mov case.
5831 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5833 * tree-predcom.c: (struct chain): Handle store-store chain in which
5834 stores for elimination only store loop invariant values.
5835 (execute_pred_commoning_chain): Ditto.
5836 (prepare_initializers_chain_store_elim): Ditto.
5837 (prepare_finalizers): Ditto.
5838 (is_inv_store_elimination_chain): New function.
5839 (initialize_root_vars_store_elim_1): New function.
5841 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5843 * tree-predcom.c: Revise general description of the pass.
5844 (enum chain_type): New enum type for store elimination.
5845 (struct chain): New field supporting store elimination.
5846 (struct component): Ditto.
5847 (dump_chain): Dump store-stores chain.
5848 (release_chain): Release resources.
5849 (split_data_refs_to_components): Compute and create component
5850 contains only stores for elimination.
5851 (get_chain_last_ref_at): New function.
5852 (make_invariant_chain): Initialization.
5853 (make_rooted_chain): Specify chain type in parameter and record it.
5854 (add_looparound_copies): Skip for store-stores chain.
5855 (determine_roots_comp): Compute type of chain and pass it to
5857 (initialize_root_vars_store_elim_2): New function.
5858 (finalize_eliminated_stores): New function.
5859 (remove_stmt): Handle store for elimination.
5860 (execute_pred_commoning_chain): Execute predictive commoning on
5862 (determine_unroll_factor): Skip unroll for store-stores chain.
5863 (prepare_initializers_chain_store_elim): New function.
5864 (prepare_initializers_chain): Hanlde store-store chain.
5865 (prepare_finalizers_chain, prepare_finalizers): New function.
5866 (tree_predictive_commoning_loop): Return integer value indicating
5867 if loop is unrolled or lcssa form is corrupted.
5868 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5870 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5872 * tree-predcom.c (initialize_root): Delete.
5873 (execute_pred_commoning_chain): Initialize root vars and replace
5874 reference of non-combined chain directly, rather than call above
5877 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5879 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5880 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5882 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5884 * tree-predcom.c (struct chain): New field init_seq.
5885 (release_chain): Release init_seq.
5886 (prepare_initializers_chain): Record intialization stmts in above
5888 (insert_init_seqs): New function.
5889 (tree_predictive_commoning_loop): Call insert_init_seqs.
5891 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5893 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5895 2017-07-28 Richard Biener <rguenther@suse.de>
5897 * match.pd: Remove superfluous :c.
5898 * genmatch.c (simplify::id): Add member.
5899 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5901 (current_id): New global.
5902 (dt_node::parent): Move from ...
5903 (dt_operand::parent): ... here. Add for_id member.
5904 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5905 (decision_tree::find_node): Relax order requirement when
5906 merging DT_TRUE nodes to ones inbetween the current simplify
5907 and the one we try to merge with. Add diagnostic whenever
5908 we need to enforce pattern order by not merging.
5909 (decision_tree::insert): Set current_id.
5910 (decision_tree::print_node): Dump parent node and for_id.
5911 (parser::last_id): Add member.
5912 (parser::push_simplify): Assign unique id.
5913 (parser::parser): Initialize last_id.
5915 2017-07-28 Martin Liska <mliska@suse.cz>
5918 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5919 gimple_build_debug_bind.
5921 2017-07-28 Richard Biener <rguenther@suse.de>
5923 PR tree-optimization/81502
5924 * match.pd: Add pattern combining BIT_INSERT_EXPR with
5926 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
5928 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
5929 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
5930 for BIT_FIELD_REF args.
5931 * fold-const.c (make_bit_field_ref): Likewise.
5932 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5934 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5937 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
5938 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
5939 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
5940 Or it into SANITIZER_UNDEFINED.
5941 * ubsan.c: Include gimple-fold.h and varasm.h.
5942 (ubsan_expand_ptr_ifn): New function.
5943 (instrument_pointer_overflow): New function.
5944 (maybe_instrument_pointer_overflow): New function.
5945 (instrument_object_size): Formatting fix.
5946 (pass_ubsan::execute): Call instrument_pointer_overflow
5947 and maybe_instrument_pointer_overflow.
5948 * internal-fn.c (expand_UBSAN_PTR): New function.
5949 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
5950 * sanitizer.def (__ubsan_handle_pointer_overflow,
5951 __ubsan_handle_pointer_overflow_abort): New builtins.
5952 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
5953 * internal-fn.def (UBSAN_PTR): New internal function.
5954 * opts.c (sanitizer_opts): Add pointer-overflow.
5955 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
5956 * fold-const.c (build_range_check): Compute pointer range check in
5957 integral type if pointer arithmetics would be needed. Formatting
5960 2017-07-28 Martin Liska <mliska@suse.cz>
5963 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
5964 parameters that are of a variable-length.
5966 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
5968 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
5970 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
5971 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
5972 (CRT_CALL_STATIC_FUNCTION): Likewise.
5973 (ASM_DEFAULT_SPEC): New define.
5974 (ASM_SPEC32): Likewise.
5975 (ASM_SPEC64): Likewise.
5976 (ASM_SPEC_COMMON): Likewise.
5977 (ASM_SPEC): Likewise.
5978 (INVALID_64BIT): Likewise.
5979 (LINK_OS_DEFAULT_SPEC): Likewise.
5980 (LINK_OS_SPEC32): Likewise.
5981 (LINK_OS_SPEC64): Likewise.
5982 (POWERPC_LINUX): Likewise.
5983 (PTRDIFF_TYPE): Likewise.
5984 (RESTORE_FP_PREFIX): Likewise.
5985 (RESTORE_FP_SUFFIX): Likewise.
5986 (SAVE_FP_PREFIX): Likewise.
5987 (SAVE_FP_SUFFIX): Likewise.
5988 (SIZE_TYPE): Likewise.
5989 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
5990 (TARGET_64BIT): Likewise.
5991 (TARGET_64BIT): Likewise.
5992 (TARGET_AIX): Likewise.
5993 (WCHAR_TYPE_SIZE): Likewise.
5994 (WCHAR_TYPE): Undefine.
5995 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
5996 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
5997 (CPP_OS_RTEMS_SPEC): Delete.
5998 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
5999 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6001 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6003 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6005 PR tree-optimization/81578
6006 * tree-parloops.c (build_new_reduction): Bail out if
6007 reduction_code isn't one of the standard OpenMP reductions.
6008 Move the details printing after that decision.
6010 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6012 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6013 related to reload_in_progress.
6014 (splat_input_operand): Likewise.
6015 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6017 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6019 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6020 (TARGET_INSTANTIATE_DECLS): Likewise.
6021 (legitimate_indexed_address_p): Delete reload_in_progress code.
6022 (rs6000_debug_legitimate_address_p): Likewise.
6023 (rs6000_eliminate_indexed_memrefs): Likewise.
6024 (rs6000_emit_le_vsx_store): Likewise.
6025 (rs6000_emit_move_si_sf_subreg): Likewise.
6026 (rs6000_emit_move): Likewise.
6027 (register_to_reg_type): Likewise.
6028 (rs6000_pre_atomic_barrier): Likewise.
6029 (rs6000_machopic_legitimize_pic_address): Likewise.
6030 (rs6000_allocate_stack_temp): Likewise.
6031 (rs6000_address_for_fpconvert): Likewise.
6032 (rs6000_address_for_altivec): Likewise.
6033 (rs6000_secondary_memory_needed_rtx): Delete function.
6034 (rs6000_check_sdmode): Likewise.
6035 (rs6000_alloc_sdmode_stack_slot): Likewise.
6036 (rs6000_instantiate_decls): Likewise.
6037 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6038 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6039 Delete reload_in_progress.
6040 (*vec_reload_and_plus_<mptrsize>): Likewise.
6041 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6042 (vsx_div_v2di): Likewise.
6043 (vsx_udiv_v2di): Likewise.
6045 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6047 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6048 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6049 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6050 (rs6000_debug_reg_global): Delete print of LRA status.
6051 (rs6000_option_override_internal): Delete dead LRA related code.
6052 (rs6000_lra_p): Delete function.
6053 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6055 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6057 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6058 * config/riscv/rtems.h: New file.
6060 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6061 Sudakshina Das <sudi.das@arm.com>
6063 * config/aarch64/aarch64.md
6064 (define_split for and<mode>3nr_compare): Move
6065 non aarch64_logical_operand to a register.
6066 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6067 register immediate operand to a register.
6068 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6070 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6073 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6075 2017-07-27 Richard Biener <rguenther@suse.de>
6077 PR tree-optimization/81573
6078 PR tree-optimization/81494
6079 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6080 multi defuse cycle case.
6082 2017-07-27 Richard Biener <rguenther@suse.de>
6084 PR tree-optimization/81571
6085 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6088 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6090 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6091 earlier and only if MASK_FPU is set. Adjust formatting.
6093 2017-07-27 Martin Liska <mliska@suse.cz>
6095 * opt-functions.awk: Add validation of value of Init.
6096 * optc-gen.awk: Pass new argument.
6098 2017-07-27 Martin Liska <mliska@suse.cz>
6100 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6101 Fix wrong condition.
6103 2017-07-27 Martin Liska <mliska@suse.cz>
6105 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6106 BBs and edges seen by autoFDO.
6108 2017-07-27 Richard Biener <rguenther@suse.de>
6110 PR tree-optimization/81502
6111 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6112 with incompatible but same sized type.
6113 (execute_update_addresses_taken): Likewise.
6115 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6117 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6118 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6120 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6123 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6124 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6125 Change s_operand to memory_operand.
6127 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6129 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6130 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6131 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6132 Emit instructions rather than returning an expression. Handle TFmode
6133 and KFmode by casting to TImode.
6134 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6135 (rs6000_emit_le_vsx_store): Likewise.
6136 * config/rs6000/vsx.md (VSX_TI): New iterator.
6137 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6138 (*vsx_le_undo_permute_<mode>): Likewise.
6139 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6140 emit the split sequence.
6141 (*vsx_le_perm_store_<mode>): Likewise.
6143 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6145 PR tree-optimization/81555
6146 PR tree-optimization/81556
6147 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6148 if true, force CHANGED for the recursive invocation.
6149 (reassociate_bb): Remember original length of ops array, pass
6150 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6152 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6153 attributes for noipa attribute. For naked attribute use
6154 lookup_attribute first before lookup_attribute_spec.
6155 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6157 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6158 for functions with noipa attribute.
6159 (cgraph_externally_visible_p): Return true for functions with noipa
6161 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6162 for functions with noipa attribute.
6163 * doc/extend.texi: Document noipa function attribute.
6164 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6165 also for functions with noipa attribute.
6166 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6168 2017-07-26 Andrew Pinski <apinski@cavium.com>
6170 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6171 vec_unalign_load_cost and vec_unalign_store_cost.
6173 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6175 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6176 -mvsx-small-integer option.
6177 (ISA_3_0_MASKS_IEEE): Likewise.
6178 (OTHER_VSX_VECTOR_MASKS): Likewise.
6179 (POWERPC_MASKS): Likewise.
6180 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6181 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6182 code, only testing for DImode being allowed in non-VSX floating
6184 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6185 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6187 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6188 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6189 TARGET_P8_VECTOR test.
6190 (rs6000_secondary_reload_simple_move): Likewise.
6191 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6192 since TARGET_P9_VECTOR was already tested.
6193 (rs6000_opt_masks): Remove -mvsx-small-integer.
6194 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6195 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6197 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6198 test for TARGET_VEXTRACTUB was used, and that uses
6200 (p9 extract splitter): Likewise.
6201 (vsx_extract_<mode>_di_p9): Likewise.
6202 (vsx_extract_<mode>_store_p9): Likewise.
6203 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6204 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6205 the elimination of TARGET_VSX_SMALL_INTEGER.
6206 (vsx_extract_<mode>_p8): Likewise.
6207 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6208 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6209 (vsx_set_<mode>_p9): Likewise.
6210 (vsx_set_v4sf_p9): Likewise.
6211 (vsx_set_v4sf_p9_zero): Likewise.
6212 (vsx_insert_extract_v4sf_p9): Likewise.
6213 (vsx_insert_extract_v4sf_p9_2): Likewise.
6214 * config/rs6000/rs6000.md (sign extend splitter): Change
6215 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6216 (floatsi<mode>2_lfiwax_mem): Likewise.
6217 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6218 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6219 since a test for TARGET_P9_VECTOR was used.
6220 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6221 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6222 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6223 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6224 TARGET_P8_VECTOR test.
6225 (fix_trunc<mode>si2_stfiwx): Likewise.
6226 (fix_trunc<mode>si2_internal): Likewise.
6227 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6228 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6230 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6231 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6232 TARGET_P8_VECTOR test.
6233 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6234 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6235 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6237 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6238 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6239 since a test for TARGET_P9_VECTOR was used.
6240 (splitter for loading small constants): Likewise.
6242 2017-07-26 Andrew Pinski <apinski@cavium.com>
6244 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6247 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6250 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6251 (fp_valid_at): Likewise.
6253 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6255 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6256 (qdf24xx_addrcost_table): Likewise.
6257 (cortexa57_tunings): Update to use generic_branch_cost.
6258 (cortexa72_tunings): Likewise.
6259 (cortexa73_tunings): Likewise.
6260 (qdf24xx_tunings): Likewise.
6262 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6264 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6265 (thunderx2t99_branch_cost): Likewise.
6266 (cortexa35_tunings): Update to use generic_branch_cost.
6267 (cortexa53_tunings): Likewise.
6268 (cortexa57_tunings): Likewise.
6269 (cortexa72_tunings): Likewise.
6270 (cortexa73_tunings): Likewise.
6271 (thunderx2t99_tunings): Likewise.
6273 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6275 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6276 (sparc_option_override): Honour MASK_FSMULD.
6277 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6278 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6279 * config/sparc/sparc.opt (mfsmuld): New option.
6280 * doc/invoke.texi (mfsmuld): Document option.
6282 2017-07-26 Marek Polacek <polacek@redhat.com>
6285 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6287 2017-07-26 Richard Biener <rguenther@suse.de>
6289 * gimple-match-head.c (do_valueize): Return OP if valueize
6291 (get_def): New helper to get at the def stmt of a SSA name
6293 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6294 do_valueize to get at the def stmt.
6295 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6297 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6300 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6302 2017-07-26 Martin Liska <mliska@suse.cz>
6305 * function.c (expand_function_start): Make expansion of
6306 nonlocal_goto_save_area after parm_birth_insn.
6308 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6310 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6311 from all CPU target flags enable members.
6313 2017-07-26 Richard Biener <rguenther@suse.de>
6315 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6316 (decision_tree::gen): Make 'type' const.
6317 (write_predicate): Likewise.
6319 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6321 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6322 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6323 (rs6000_option_override_internal): Likewise.
6324 (rs6000_expand_vector_set): Likewise.
6325 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6326 (TARGET_UPPER_REGS_SF): Likewise.
6327 (TARGET_UPPER_REGS_DI): Likewise.
6328 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6329 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6330 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6331 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6332 (Splitters for DI constants in Altivec registers): Likewise.
6333 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6334 (vsx_set_v4sf_p9): Likewise.
6335 (vsx_set_v4sf_p9_zero): Likewise.
6336 (vsx_insert_extract_v4sf_p9): Likewise.
6337 (vsx_insert_extract_v4sf_p9_2): Likewise.
6339 2017-07-25 Carl Love <cel@us.ibm.com>
6341 * doc/extend.texi: Update the built-in documentation file for the
6342 existing built-in functions
6343 vector signed char vec_cnttz (vector signed char);
6344 vector unsigned char vec_cnttz (vector unsigned char);
6345 vector signed short vec_cnttz (vector signed short);
6346 vector unsigned short vec_cnttz (vector unsigned short);
6347 vector signed int vec_cnttz (vector signed int);
6348 vector unsigned int vec_cnttz (vector unsigned int);
6349 vector signed long long vec_cnttz (vector signed long long);
6350 vector unsigned long long vec_cnttz (vector unsigned long long);
6352 2017-07-25 Andrew Pinski <apinski@cavium.com>
6354 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6355 accesses where the use is for the first operand of a BIT_INSERT.
6357 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6360 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6361 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6363 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6365 * config/i386/gstabs.h: Delete.
6366 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6368 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6370 * config/i386/i386.c (ix86_decompose_address): Do not check for
6371 register RTX when looking at index_reg or base_reg.
6372 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6374 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6376 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6377 to update EH info here.
6379 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6381 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6383 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6385 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6387 2017-07-25 Torsten Duwe <duwe@suse.de>
6389 * common.opt: Introduce -fpatchable-function-entry
6390 command line option, and its variables function_entry_patch_area_size
6391 and function_entry_patch_area_start.
6392 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6393 including a two-value parser.
6394 * target.def (print_patchable_function_entry): New target hook.
6395 * targhooks.h (default_print_patchable_function_entry): New function.
6396 * targhooks.c (default_print_patchable_function_entry): Likewise.
6397 * toplev.c (process_options): Switch off IPA-RA if
6398 patchable function entries are being generated.
6399 * varasm.c (assemble_start_function): Look at the
6400 patchable-function-entry command line switch and current
6401 function attributes and maybe generate NOP instructions by
6402 calling the print_patchable_function_entry hook.
6403 * doc/extend.texi: Document patchable_function_entry attribute.
6404 * doc/invoke.texi: Document -fpatchable_function_entry
6405 command line option.
6406 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6408 * doc/tm.texi: Re-generate.
6410 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6413 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6414 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6416 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6418 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6420 2017-07-25 Richard Biener <rguenther@suse.de>
6422 PR tree-optimization/81455
6423 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6424 not walk in cycles when looking for guards.
6426 2017-07-25 Richard Biener <rguenther@suse.de>
6428 PR tree-optimization/81529
6429 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6430 when optimizing backedge uses.
6432 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6434 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6436 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6437 to dl_section_ref. On AIX, append an expression to subtract
6438 the size of the section length to dl_section_ref.
6440 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6442 * configure.ac: If any of the config.* scripts fail, exit 1.
6443 * configure: Regenerate.
6445 2017-07-25 Richard Biener <rguenther@suse.de>
6448 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6449 of immediate uses, be more verbose on errors.
6451 2017-07-25 Richard Biener <rguenther@suse.de>
6453 PR tree-optimization/81510
6454 * tree-vect-loop.c (vect_is_simple_reduction): When the
6455 reduction stmt is not inside the loop bail out.
6457 2017-07-25 Richard Biener <rguenther@suse.de>
6459 PR tree-optimization/81303
6460 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6461 profitability check against LOOP_VINFO_NITERSM1.
6463 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6465 * domwalk.c (cmp_bb_postorder): Simplify.
6466 (sort_bbs_postorder): New function. Use it...
6467 (dom_walker::walk): ...here to optimize common cases.
6469 2017-07-25 Martin Liska <mliska@suse.cz>
6472 * ipa-visibility.c (function_and_variable_visibility): Make the
6473 redirection just on target that supports aliasing.
6474 Fix GNU coding style.
6476 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6479 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6481 * config/arm/rtems.h: Likewise.
6482 * config/bfin/rtems.h: Likewise.
6483 * config/i386/rtemself.h: Likewise.
6484 * config/lm32/rtems.h: Likewise.
6485 * config/m32c/rtems.h: Likewise.
6486 * config/m68k/rtemself.h: Likewise.
6487 * config/microblaze/rtems.h: Likewise.
6488 * config/mips/rtems.h: Likewise.
6489 * config/moxie/rtems.h: Likewise.
6490 * config/nios2/rtems.h: Likewise.
6491 * config/powerpcspe/rtems.h: Likewise.
6492 * config/rs6000/rtems.h: Likewise.
6493 * config/rtems.h: Likewise.
6494 * config/sh/rtems.h: Likewise.
6495 * config/sh/rtemself.h: Likewise.
6496 * config/sparc/rtemself.h: Likewise.
6498 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6501 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6502 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6503 * tree-ssa-structalias.c (alias_get_name): Same.
6505 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6508 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6509 instructions if no du chain is found.
6511 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6513 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6515 2017-07-25 Richard Biener <rguenther@suse.de>
6518 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6521 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6523 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6525 (ISA_2_7_MASKS_SERVER): Likewise.
6526 (ISA_3_0_MASKS_IEEE): Likewise.
6527 (OTHER_P8_VECTOR_MASKS): Likewise.
6528 (OTHER_VSX_VECTOR_MASKS): Likewise.
6529 (POWERPC_MASKS): Likewise.
6530 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6531 duplicate list of options.
6532 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6533 explicit -mupper-regs options.
6534 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6535 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6536 alias for -mupper-regs-df.
6537 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6538 (rs6000_init_hard_regno_mode_ok): Likewise.
6539 (rs6000_option_override_internal): Likewise.
6540 (rs6000_opt_masks): Likewise.
6541 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6542 options in terms of whether -mvsx or -mpower8-vector was used.
6543 (TARGET_UPPER_REGS_DI): Likewise.
6544 (TARGET_UPPER_REGS_SF): Likewise.
6545 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6546 -mupper-regs-* options.
6548 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6550 * passes.c (emergency_dump_function): Print some empty lines and a
6551 header before the RTL dump.
6553 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6555 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6557 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6560 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6561 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6563 2017-07-24 Carl Love <cel@us.ibm.com>
6565 * config/rs6000/rs6000-c.c: Add support for built-in functions
6566 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6567 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6568 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6569 vec_extract_fp_from_shortl): Add defines for the two builtins.
6570 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6571 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6573 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6574 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6575 * doc/extend.texi: Update the built-in documentation file for the
6576 new built-in function.
6578 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6581 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6583 * doc/generic.texi: Likewise.
6584 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6585 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6587 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6589 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6590 (aarch64_mls_elt_merge<mode>): Likewise.
6592 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6594 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6595 having __cxa_atexit.
6597 2017-07-23 Michael Collison <michael.collison@arm.com>
6599 * config/arm/arm.c (arm_option_override): Deprecate
6600 use of -mstructure-size-boundary.
6601 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6602 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6604 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6607 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6608 Reduce cost estimate for direct moves.
6610 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6613 * config/i386/i386.c (ix86_option_override_internal): Disable
6614 BMI, BMI2 and TBM instructions for -m16.
6616 2017-07-21 Carl Love <cel@us.ibm.com>
6618 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6619 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6620 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6621 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6622 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6623 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6624 VMULOSW): New enum "unspec" values.
6625 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6626 altivec_vmulosw): New patterns.
6627 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6628 VMULOSW): Add definitions.
6630 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6632 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6633 (qdf24xx): Likewise.
6634 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6635 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6636 (AARCH64_FL_V8_1): Renumber.
6637 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6638 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6639 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6640 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6641 rdma to feature modifiers list.
6643 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6646 * ipa-visibility (function_and_variable_visibility): Convert
6647 recursive PLT call to direct call if appropriate.
6649 2017-07-21 Andrew Pinski <apinski@cavium.com>
6651 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6652 operand 1 to see if the types precision matches.
6653 * fold-const.c (operand_equal_p): Likewise.
6655 2017-07-21 Richard Biener <rguenther@suse.de>
6657 PR tree-optimization/81303
6658 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6659 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6660 (vect_peeling_hash_get_lowest_cost): Adjust.
6661 (vect_enhance_data_refs_alignment): Likewise. Use
6662 vect_get_peeling_costs_all_drs to compute the penalty for no
6663 peeling to match up costs.
6665 2017-07-21 Richard Biener <rguenther@suse.de>
6667 PR tree-optimization/81500
6668 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6669 we didn't identify a reduction path.
6671 2017-07-21 Tom de Vries <tom@codesourcery.com>
6672 Cesar Philippidis <cesar@codesourcery.com>
6674 PR gcov-profile/81442
6675 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6678 2017-07-21 Tom de Vries <tom@codesourcery.com>
6681 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6683 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6684 nvptx_override_options_after_change.
6686 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6688 * dwarf2out.c (output_file_names): Avoid double testing for
6691 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6693 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6695 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6697 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6699 (try_crossjump_to_edge): Do not punt on partitioned functions.
6701 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6703 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6704 Put all BBs reachable only via paths crossing cold region to cold
6706 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6708 2016-07-21 Richard Biener <rguenther@suse.de>
6710 PR tree-optimization/81303
6711 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6712 into account prologue and epilogue iterations when raising
6713 min_profitable_iters to sth at least covering one vector iteration.
6715 2017-07-21 Tamar Christina <tamar.christina@arm.com>
6717 * config/arm/arm.c (arm_test_cpu_arch_dat):
6720 2017-07-20 Nathan Sidwell <nathan@acm.org>
6722 Remove TYPE_METHODS.
6723 * tree.h (TYPE_METHODS): Delete.
6724 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
6725 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
6726 (dbxout_type_methods): Scan TYPE_FIELDS.
6727 (dbxout_type): Don't check TYPE_METHODS here.
6728 * function.c (use_register_for_decl): Always ignore register for
6729 class types when not optimizing.
6730 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
6731 * tree.c (free_lang_data_in_type): Stitch out member functions and
6732 templates from TYPE_FIELDS.
6733 (build_distinct_type_copy, verify_type_variant,
6734 verify_type): Member fns are on TYPE_FIELDS.
6735 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
6736 * tree-pretty-print.c (dump_generic_node): Likewise.
6738 2017-07-20 Jakub Jelinek <jakub@redhat.com>
6741 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
6742 V2TImode and V4TImode.
6743 (ix86_expand_vector_extract): Likewise.
6744 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
6745 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
6746 (ssescalarmode): Handle V4TImode and V2TImode.
6747 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6748 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6749 (VEXTRACTI128_MODE): New mode iterator.
6750 (splitter for *vec_extractv?ti first element): New.
6751 (VEC_INIT_MODE): New mode iterator.
6752 (vec_init<mode>): Consolidate 3 expanders into one using
6753 VEC_INIT_MODE mode iterator.
6755 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6757 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6758 non_spilled_static_chain_regno_p.
6760 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6762 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6764 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6766 * bb-reorder.c (connect_traces): Allow copying of blocks within
6769 2017-07-20 Richard Biener <rguenther@suse.de>
6771 * gimple.h (gimple_phi_result): Add gphi * overload.
6772 (gimple_phi_result_ptr): Likewise.
6773 (gimple_phi_arg): Likewise. Adjust index assert to only
6774 allow actual argument accesses rather than all slots available
6776 (gimple_phi_arg_def): Add gphi * overload.
6777 * tree-phinodes.c (make_phi_node): Initialize only actual
6779 (resize_phi_node): Clear memory not covered by old node,
6780 do not initialize excess argument slots.
6781 (reserve_phi_args_for_new_edge): Initialize new argument slot
6784 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6786 PR tree-optimization/81388
6788 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6790 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6791 by removing computation of may_be_zero.
6793 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6794 Tom de Vries <tom@codesourcery.com>
6797 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6798 when gimple level profile disagrees with what RTL expander did.
6800 2017-07-20 Richard Biener <rguenther@suse.de>
6802 PR tree-optimization/61171
6803 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6804 (vect_analyze_stmt): Add slp instance parameter.
6805 (vectorizable_reduction): Likewise.
6806 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6807 (vect_is_simple_reduction): Deal with chains not detected
6808 as SLP reduction chain, specifically not properly associated
6809 chains containing a mix of plus/minus.
6810 (get_reduction_op): Remove.
6811 (get_initial_defs_for_reduction): Simplify, pass in whether
6812 this is a reduction chain, pass in the SLP node for the PHIs.
6813 (vect_create_epilog_for_reduction): Get the SLP instance as
6815 (vectorizable_reduction): Get the SLP instance as arg.
6816 During analysis remember the SLP node with the PHIs in the
6817 instance. Simplify getting at the vectorized reduction PHIs.
6818 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6819 through SLP instance.
6820 (vect_slp_analyze_operations): Likewise.
6821 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6822 (vect_transform_stmt): Likewise.
6824 2017-07-20 Tom de Vries <tom@codesourcery.com>
6826 PR tree-optimization/81489
6827 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6828 read of phi arg location to before loop that modifies phi.
6830 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6832 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6835 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6838 * except.c (execute): Fix ordering issue.
6840 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6842 PR rtl-optimization/81423
6843 * combine.c (make_compound_operation_int): Don't try to optimize
6844 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6846 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6848 PR rtl-optimization/81423
6849 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6850 with a constant that is -1 in the truncated to mode.
6852 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6854 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6855 (determine_unlikely_bbs): ... here.
6856 * predict.h (propagate_unlikely_bbs_forward): Declare.
6857 * cfgexpand.c (pass_expand::execute): Use it.
6858 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6860 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6861 propagate_unlikely_bbs_forward.
6863 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6866 * except.c (maybe_add_nop_after_section_switch): New function.
6869 2017-07-19 Tom de Vries <tom@codesourcery.com>
6871 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6873 2017-07-19 Tom de Vries <tom@codesourcery.com>
6875 * gimple.h (gimple_phi_arg): Make assert more strict.
6877 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6879 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6880 * config/rs6000/mmintrin.h: New file.
6881 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6883 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6885 PR tree-optimization/81346
6886 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6888 2017-07-19 Tom de Vries <tom@codesourcery.com>
6890 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6892 2017-07-19 Tom de Vries <tom@codesourcery.com>
6894 * config/nvptx/nvptx-modes.def: Add V2DImode.
6895 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6896 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6897 (nvptx_output_mov_insn): Handle lack of mov.b128.
6898 (nvptx_print_operand): Handle 'H' and 'L' codes.
6899 (nvptx_vector_mode_supported): Allow V2DImode.
6900 (nvptx_preferred_simd_mode): New function.
6901 (nvptx_data_alignment): New function.
6902 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6903 nvptx_preferred_simd_mode.
6904 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6906 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6908 2017-07-19 Tom de Vries <tom@codesourcery.com>
6910 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6911 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6912 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6913 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6914 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6915 (mov<VECIM>_insn): New define_insn.
6916 (define_expand "mov<VECIM>): New define_expand.
6918 2017-07-19 Tom de Vries <tom@codesourcery.com>
6920 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6922 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6924 PR tree-optimization/81346
6925 * fold-const.h (fold_div_compare, range_check_type): Declare.
6926 * fold-const.c (range_check_type): New function.
6927 (build_range_check): Use range_check_type.
6928 (fold_div_compare): No longer static, rewritten into
6929 a match.pd helper function.
6930 (fold_comparison): Don't call fold_div_compare here.
6931 * match.pd (X / C1 op C2): New optimization using fold_div_compare
6934 2017-07-19 Nathan Sidwell <nathan@acm.org>
6936 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
6937 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
6938 * tree.c (find_decls_types_r, verify_type): Use
6939 TYPE_{MIN,MAX}_VALUE_RAW.
6940 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
6941 (hash_tree): Likewise.
6942 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6944 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6947 2017-07-18 Tom de Vries <tom@codesourcery.com>
6950 * omp-expand.c (expand_omp_for_static_chunk): Handle
6951 equal-argument loop exit phi.
6953 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
6956 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
6957 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
6958 operand 2 predicate.
6959 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
6960 operand 2 predicate.
6961 (ror,rol -> rorx splitters): Use const_int_operand as
6962 operand 2 predicate.
6964 2017-06-18 Richard Biener <rguenther@suse.de>
6966 PR tree-optimization/81410
6967 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
6968 the gap in the ! slp_perm SLP case after each group.
6970 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6973 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
6976 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6979 * predict.c (set_even_probabilities): Cleanup; do not affect
6980 probabilities that are already known.
6981 (combine_predictions_for_bb): Call even when count is set.
6983 2017-07-18 Nathan Sidwell <nathan@acm.org>
6985 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
6988 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6991 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
6992 optimization for loop niter analysis.
6994 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
6997 * config/avr/avr.c (avr_optimize_casesi): Don't use
6998 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7000 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7002 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7003 body_cost_vec from _vect_peel_extended_info.
7004 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7005 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7008 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7010 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7012 2017-07-18 Richard Biener <rguenther@suse.de>
7014 PR tree-optimization/80620
7015 PR tree-optimization/81403
7016 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7017 info when re-using a VN table entry.
7019 2017-07-18 Richard Biener <rguenther@suse.de>
7021 PR tree-optimization/81418
7022 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7023 vectype_in. Verify that with lane-reducing reduction operations
7024 we have a single def-use cycle.
7026 2017-07-17 Carl Love <cel@us.ibm.com>
7028 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7030 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7031 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7032 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7033 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7034 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7035 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7036 VMULOSW): New enum "unspec" values.
7037 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7038 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7039 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7040 altivec_vmulosw): New patterns.
7041 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7042 VMULOSW): Add definitions.
7043 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7045 * config/alpha/alpha.c: Include predict.h.
7047 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7049 * tree-vrp.c (compare_assert_loc): Fix comparison function
7050 to return predictable results.
7052 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7054 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7057 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7058 * doc/invoke.texi (mexpand-adddi): Update text.
7060 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7062 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7063 that also clobbers the CC register. The old expand code is moved
7065 (*arc_clzsi2): ... here.
7066 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7067 the CC register. The old expand code is moved to ...
7068 (arc_ctzsi2): ... here.
7070 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7072 * config/arc/arc.opt (mindexed-loads): Use initial value
7073 TARGET_INDEXED_LOADS_DEFAULT.
7074 (mauto-modify-reg): Use initial value
7075 TARGET_AUTO_MODIFY_REG_DEFAULT.
7076 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7077 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7078 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7079 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7081 2017-07-17 Martin Liska <mliska@suse.cz>
7084 * opts.c (finish_options): Do not allow -fgnu-tm
7085 w/ -fsanitize={kernel-,}address. Say sorry.
7087 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7090 * tree-loop-distribution.c (classify_partition): Only assert on
7091 numer of iterations.
7092 (merge_dep_scc_partitions): Delete prameter. Update function call.
7093 (distribute_loop): Remove code handling loop with unknown niters.
7094 (pass_loop_distribution::execute): Skip loop with unknown niters.
7096 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7099 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7100 function sort_partitions_by_post_order.
7102 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7104 PR tree-optimization/81374
7105 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7106 the max index of basic blocks, rather than number of basic blocks.
7108 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7110 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7112 (arc_legitimate_pic_operand_p): Likewise.
7113 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7115 (arc_needs_pcl_p): Likewise.
7116 (arc_legitimate_pc_offset_p): Likewise.
7117 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7118 function is also used in constrains.md.
7119 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7120 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7121 PLUS. Only return true/false in known cases, otherwise assert.
7122 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7123 is already called in arc_legitimate_constant_p.
7124 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7126 (LEGITIMATE_PIC_OPERAND_P): Use
7127 arc_raw_symbolic_reference_mentioned_p function.
7128 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7133 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7134 Andrew Burgess <andrew.burgess@embecosm.com>
7136 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7137 (arc_return_address_register): New function.
7138 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7139 (arc_handle_fndecl_attribute): Add naked attribute.
7140 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7141 (TARGET_WARN_FUNC_RETURN): Likewise.
7142 (arc_allocate_stack_slots_for_args): New function.
7143 (arc_warn_func_return): Likewise.
7144 (machine_function): Change type fn_type.
7145 (arc_compute_function_type): Consider new naked function type,
7146 change function return type.
7147 (arc_must_save_register): Adapt to handle new
7148 arc_compute_function_type's return type.
7149 (arc_expand_prologue): Likewise.
7150 (arc_expand_epilogue): Likewise.
7151 (arc_return_address_regs): Delete.
7152 (arc_return_address_register): New function.
7153 (arc_epilogue_uses): Use above function.
7154 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7155 (arc_function_type): Change encoding, add naked type.
7156 (ARC_INTERRUPT_P): Change to handle the new encoding.
7157 (ARC_FAST_INTERRUPT_P): Likewise.
7158 (ARC_NORMAL_P): Define.
7159 (ARC_NAKED_P): Likewise.
7160 (arc_compute_function_type): Delete prototype.
7161 * config/arc/arc.md (in_ret_delay_slot): Use
7162 arc_return_address_register function.
7163 (simple_return): Likewise.
7164 (p_return_i): Likewise.
7166 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7168 PR tree-optimization/81428
7169 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7170 can't be built for those types.
7172 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7174 Remove stuff dead since r239246.
7176 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7177 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7178 (avr_inform_devices): Remove dead stuff.
7180 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7182 * config/arm/arm_neon.h: Fix softp typo.
7184 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7186 PR tree-optimization/81365
7187 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7188 aggregate moves onto bb predecessor edges, make sure there are no
7189 loads that could alias the lhs in between the start of bb and the
7192 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7195 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7196 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7197 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7199 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7201 PR tree-optimization/81396
7202 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7203 (init_symbolic_number): Initialize it to 1.
7204 (perform_symbolic_merge): Add n_ops from both operands into the new
7206 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7207 without base_addr as useless if they need more than one operation.
7208 (bswap_replace): Handle !bswap case for NULL base_addr.
7210 2017-07-17 Tom de Vries <tom@codesourcery.com>
7213 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7216 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7218 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7219 conditional builtin define __FIX_LEON3FT_B2BST.
7221 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7223 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7224 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7227 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7229 PR rtl-optimization/81424
7230 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7231 to remove potential trapping from operands if -fnon-call-exceptions.
7233 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7235 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7236 profile_proability for scalling.
7237 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7239 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7241 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7243 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7245 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7246 fixpoint arithmetics.
7248 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7250 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7251 fixpoint arithmetics.
7253 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7255 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7256 fixpoint arithmetics.
7258 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7260 * profile-count.h (profile_probability::from_reg_br_prob_note,
7261 profile_probability::to_reg_br_prob_note): New functions.
7262 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7263 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7264 * predict.c (probability_reliable_p): Update.
7265 (edge_probability_reliable_p): Update.
7266 (br_prob_note_reliable_p): Update.
7267 (invert_br_probabilities): Update.
7268 (add_reg_br_prob_note): New function.
7269 (combine_predictions_for_insn): Update.
7270 * asan.c (asan_clear_shadow): Update.
7271 * cfgbuild.c (compute_outgoing_frequencies): Update.
7272 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7273 (update_br_prob_note): Update.
7274 (rtl_verify_edges): Update.
7275 (purge_dead_edges): Update.
7276 (fixup_reorder_chain): Update.
7277 * emit-rtl.c (try_split): Update.
7278 * ifcvt.c (cond_exec_process_insns): Update.
7279 (cond_exec_process_if_block): Update.
7280 (dead_or_predicable): Update.
7281 * internal-fn.c (expand_addsub_overflow): Update.
7282 (expand_neg_overflow): Update.
7283 (expand_mul_overflow): Update.
7284 * loop-doloop.c (doloop_modify): Update.
7285 * loop-unroll.c (compare_and_jump_seq): Update.
7286 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7287 * predict.h: Update.
7288 * reorg.c (mostly_true_jump): Update.
7290 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7291 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7292 * config/arc/arc.c: (emit_unlikely_jump): Update.
7293 * config/arm/arm.c: (emit_unlikely_jump): Update.
7294 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7295 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7296 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7297 (ix86_print_operand): Update.
7298 (ix86_split_fp_branch): Update.
7299 (predict_jump): Update.
7300 * config/ia64/ia64.c (ia64_print_operand): Update.
7301 * config/mmix/mmix.c (mmix_print_operand): Update.
7302 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7303 (rs6000_expand_split_stack_prologue): Update.
7304 * config/rs6000/rs6000.c: Update.
7305 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7306 (s390_expand_vec_movstr): Update.
7307 (s390_expand_cs_tdsi): Update.
7308 (s390_expand_split_stack_prologue): Update.
7309 * config/sh/sh.c (sh_print_operand): Update.
7310 (expand_cbranchsi4): Update.
7311 (expand_cbranchdi4): Update.
7312 * config/sparc/sparc.c (output_v9branch): Update.
7313 * config/spu/spu.c (get_branch_target): Update.
7314 (ea_load_store_inline): Update.
7315 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7316 * config/tilepro/tilepro.c: Update.
7318 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7320 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7321 (gimplify_save_expr): Likewise.
7323 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7325 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7327 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7329 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7331 * timevar.def (TV_IPA_FNSUMMARY): Define.
7333 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7335 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7336 to back store errata sensitive sequence from being generated.
7337 (sqrtdf2_fix): Likewise.
7339 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7341 * tree-ssa-threadupdate.c (compute_path_counts,
7342 update_joiner_offpath_counts): Use profile_probability.
7344 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7347 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7349 * config/arm/arm-c.c (arm_cpu_builtins): Define
7350 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7352 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7354 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7355 array entries to represent __ieee128 versions of the
7356 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7357 scalar_extract_sig, and scalar_insert_exp built-in functions.
7358 (altivec_resolve_overloaded_builtin): Add special case handling
7359 for the __builtin_scalar_insert_exp function, as represented by
7360 the P9V_BUILTIN_VEC_VSIEDP constant.
7361 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7362 exponent support for __ieee128 argument.
7363 (VSESQP): Add scalar extract signature support for __ieee128
7365 (VSTDCNQP): Add scalar test negative support for __ieee128
7367 (VSIEQP): Add scalar insert exponent support for __int128 argument
7368 with __ieee128 result.
7369 (VSIEQPF): Add scalar insert exponent support for __ieee128
7370 argument with __ieee128 result.
7371 (VSTDCQP): Add scalar test data class support for __ieee128
7373 (VSTDCNQP): Add overload support for scalar test negative with
7375 (VSTDCQP): Add overload support for scalar test data class
7377 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7379 (UNSPEC_VSX_SIEXPQP): New constant.
7380 (xsxexpqp): New insn for VSX scalar extract exponent quad
7382 (xsxsigqp): New insn for VSX scalar extract significand quad
7384 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7385 precision with floating point argument.
7386 (xststdcqp): New expand for VSX scalar test data class quad
7388 (xststdcnegqp): New expand for VSX scalar test negative quad
7390 (xststdcqp): New insn to match expansions for VSX scalar test data
7391 class quad precision and VSX scalar test negative quad precision.
7392 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7393 special case operand checking to enforce that second operand of
7394 VSX scalar test data class with quad precision argument is a 7-bit
7396 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7397 prototypes and descriptions of __ieee128 versions of
7398 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7399 scalar_test_data_class, and scalar_test_neg built-in functions.
7401 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7403 PR tree-optimization/81162
7404 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7405 replace a negate with an add.
7407 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7409 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7411 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7413 * config/arm/arm-c.c (arm_cpu_builtins): Define
7414 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7416 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7418 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7419 (armv8-r): Set ARM Cortex-R52 as default CPU.
7420 * config/arm/arm-tables.opt: Regenerate.
7421 * config/arm/arm-tune.md: Regenerate.
7422 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7424 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7425 extension for -mcpu=cortex-r52.
7427 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7429 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7430 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7431 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7432 (fp-armv8): Define it as FP_ARMv8 only.
7433 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7434 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7436 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7437 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7438 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7439 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7440 than TARGET_FPU_ARMV8.
7441 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7442 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7443 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7445 * config/arm/neon.md (neon_vrint): Likewise.
7446 (neon_vcvt): Likewise.
7447 (neon_<fmaxmin_op><mode>): Likewise.
7448 (<fmaxmin><mode>3): Likewise.
7449 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7450 * config/arm/predicates.md (arm_cond_move_operator): Check against
7451 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7453 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7455 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7458 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7460 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7461 loop in comment with memset.
7463 2017-07-14 Martin Liska <mliska@suse.cz>
7465 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7466 * dwarf2out.c (is_java): Remove the function.
7467 (output_pubname): Remove usage of the function.
7468 (lower_bound_default): Remove usage of DW_LANG_Java.
7469 (gen_compile_unit_die): Likewise.
7470 * gcc.c: Remove compiler defaults for .java and .zip files.
7471 * gimple-expr.c (remove_suffix): Change as there's no longer
7472 extension than 4-letter one.
7473 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7474 (gimplify_save_expr): Likewise.
7475 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7476 as it's possible even for other languages than Java.
7477 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7478 * lto-opts.c (lto_write_options): Remove reference to Java.
7479 * opts.c (strip_off_ending): Update file extension handling.
7480 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7481 * tree-eh.c (lower_resx): Likewise.
7482 * tree.c (free_lang_data_in_type): Remove dead code.
7483 (find_decls_types_r): Likewise.
7484 (build_common_builtin_nodes): Remove Java from a comment.
7485 (verify_type): Remove dead code.
7486 * varasm.c (assemble_external): Remove Java from a comment.
7488 2017-07-14 Martin Liska <mliska@suse.cz>
7490 * opts.c (finish_options): Add quotes.
7491 (common_handle_option): Likewise.
7493 2017-07-14 Martin Liska <mliska@suse.cz>
7495 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7496 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7498 * dwarf2out.c (lower_bound_default): Do not handle
7500 (gen_compile_unit_die): Likewise.
7501 * gcc.c: Remove default extension binding for GNU Pascal.
7502 * stmt.c: Remove Pascal language from a comment.
7503 * xcoffout.c: Likewise.
7505 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7508 * diagnostic-show-locus.c (fixit_cmp): New function.
7509 (layout::layout): Sort m_fixit_hints.
7510 (column_range::column_range): Assert that the values are valid.
7511 (struct char_span): New struct.
7512 (correction::overwrite): New method.
7513 (struct source_line): New struct.
7514 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7515 calls in terms of classes source_line and char_span, and
7516 correction::overwrite.
7517 (selftest::test_overlapped_fixit_printing_2): New function.
7518 (selftest::diagnostic_show_locus_c_tests): Call it.
7520 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7522 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7523 early if there is no lhs.
7525 2017-07-13 Martin Liska <mliska@suse.cz>
7527 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7528 (gen_reference_type_die): Likewise.
7529 * stor-layout.c: Remove Pascal-related comment.
7531 2017-07-13 Martin Liska <mliska@suse.cz>
7533 * opts.c (finish_options): Add quotes to error messages.
7534 (parse_sanitizer_options): Likewise.
7536 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7538 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7540 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7542 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7544 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7546 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7548 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7550 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7553 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7554 provides the hardware capability bits, define the macro
7555 __BUILTIN_CPU_SUPPORTS__.
7556 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7557 if GLIBC does not provide the hardware capability bits. Add a
7558 gcc_unreachable call if the built-in cpu function is neither
7559 __builtin_cpu_is nor __builtin_cpu_supports.
7560 (rs6000_get_function_versions_dispatcher): Change the warning
7561 that an old GLIBC is used which does not export the capability
7562 bits to be an error.
7563 * doc/extend.texi (target_clones attribute): Document the
7564 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7565 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7566 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7567 the macros defined by GCC if the newer GLIBC is available.
7569 2017-07-12 Jeff Law <law@redhat.com>
7571 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7572 remaining includes slightly.
7573 * config/riscv/riscv-builtins.c: Include profile-count.h.
7575 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7578 * config/avr/avr.c (avr_set_current_function): In diagnostic
7579 messages: Quote keywords and (parts of) identifiers.
7580 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7583 2017-07-12 Carl Love <cel@us.ibm.com>
7585 * config/rs6000/rs6000-c.c: Add support for built-in functions
7586 vector bool char vec_revb (vector bool char);
7587 vector bool short vec_revb (vector short char);
7588 vector bool int vec_revb (vector bool int);
7589 vector bool long long vec_revb (vector bool long long);
7590 * doc/extend.texi: Update the built-in documentation file for the
7591 new built-in functions.
7593 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7595 * config/s390/s390.md: Remove movcc splitter.
7597 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7599 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7600 load/store on condition.
7602 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7605 * config/avr/avr.c (avr_encode_section_info)
7606 [progmem && !TREE_READONLY]: Error if progmem object needs
7609 2017-07-11 Michael Collison <michael.collison@arm.com>
7611 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7614 2017-07-11 Carl Love <cel@us.ibm.com>
7616 * config/rs6000/rs6000-c.c: Add support for builtins
7617 vector unsigned int vec_parity_lsbb (vector signed int);
7618 vector unsigned int vec_parity_lsbb (vector unsigned int);
7619 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7620 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7621 vector unsigned long long vec_parity_lsbb (vector signed long long);
7622 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7623 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7624 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7625 * doc/extend.texi: Update the built-in documentation file for the
7626 new built-in functions.
7628 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7630 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7631 (layout::m_primary_loc): New field.
7632 (layout::layout): Initialize new field. Move location filtering
7633 logic from here to...
7634 (layout::maybe_add_location_range): ...this new method. Add
7635 support for filtering to just the lines already specified by other
7637 (layout::will_show_line_p): New method.
7638 (gcc_rich_location::add_location_if_nearby): New method.
7639 (selftest::test_add_location_if_nearby): New test function.
7640 (selftest::diagnostic_show_locus_c_tests): Call it.
7641 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7644 2017-07-11 Tom de Vries <tom@codesourcery.com>
7646 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7647 (bb_first_real_insn): New function.
7648 (nvptx_single): Add extra initialization of broadcasted condition
7651 2017-07-11 Nathan Sidwell <nathan@acm.org>
7653 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7655 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7657 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7658 Binutils doc as TEXI will mess them up.
7659 * doc/invoke.texi (AVR Options): Same here.
7661 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7663 * config/sparc/sparc.opt (mfix-ut700): New option.
7664 (mfix-gr712rc): Likewise.
7665 (sparc_fix_b2bst): New variable.
7666 * doc/invoke.texi (SPARC options): Document them.
7667 (ARM options): Fix warnings.
7668 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7669 instructions to prevent sequences that can trigger the store-store
7670 errata for certain LEON3FT processors.
7671 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7672 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7673 * config/sparc/sparc.md (fix_b2bst): New attribute.
7674 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7676 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7679 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7681 (*rsqrtsf2_sse): Ditto.
7683 (div<mode>3): Macroize insn from divdf3 and divsf3
7684 using MODEF mode iterator.
7686 2017-07-10 Martin Sebor <msebor@redhat.com>
7688 PR tree-optimization/80397
7689 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7690 instead of testing for equality to INTEGER_TYPE.
7692 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7694 * config.gcc: Remove uclibc from arc target spec.
7696 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7698 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7700 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7703 * lto-wrapper.c (remove_option): New function.
7704 (merge_and_complain): Merge PIC/PIE options more realistically.
7706 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7708 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7712 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7713 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7714 * config.in: Regenerate.
7715 * configure: Regenerate.
7716 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
7717 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
7718 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
7719 (TARGET_GASISR_PROLOGUES): ...target mask.
7720 * common/config/avr/avr-common.c
7721 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
7722 Set -mgas-isr-prologues.
7723 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
7724 INSERT_PASS_BEFORE for it.
7725 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
7726 * config/avr/avr.c (avr_option_override)
7727 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
7728 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
7729 (avr_attribute_table) <no_gccisr>: Add new function attribute.
7730 (avr_set_current_function) <is_no_gccisr>: Init machine field.
7731 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
7733 (make_avr_pass_pre_proep): New function.
7734 (emit_push_sfr) <treg>: Add argument to function and use it
7736 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
7737 and set machine->gasisr.yes.
7738 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
7739 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
7740 __gcc_isr.n_pushed to .L__stack_usage.
7741 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
7742 (avr_asm_final_postscan_insn): ...this new static function.
7743 * config/avr/avr.h (machine_function)
7744 <is_no_gccisr, use_L__stack_usage>: New fields.
7745 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
7746 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
7747 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7748 (gasisr, *gasisr): New expander and insn.
7749 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7750 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7751 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7753 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7755 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7758 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7760 Move jump-tables out of .text again.
7763 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7764 (ASM_OUTPUT_ADDR_VEC): New function.
7765 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7766 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7767 INSN_ADDRESSes as asm comment.
7768 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7769 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7770 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7771 * config/avr/avr.md (*tablejump): Adjust comment.
7772 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7773 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7775 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7776 (avr_output_addr_vec): New proto.
7777 (avr_log_t) <insn_addresses>: New field.
7779 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7782 * config/i386/i386.c (ix86_function_arg_advance): Set
7783 outgoing_args_on_stack to true if there are outgoing arguments
7785 (ix86_function_arg): Likewise.
7786 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7787 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7788 * config/i386/i386.h (machine_function): Add
7789 outgoing_args_on_stack.
7791 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7793 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7795 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7797 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7799 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7800 (REAL_H): Remove $(MACHMODE_H).
7801 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7803 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7804 $(MACHMODE_H) and double-int.h.
7805 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7807 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7810 2017-07-07 Andrew Pinski <apinski@cavium.com>
7812 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7813 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7815 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7817 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7818 Add warning if GCC was not configured to link against a GLIBC that
7819 exports the hardware capability bits.
7820 (make_resolver_func): Make resolver function private and not a
7821 COMDAT function. Create the name with clone_function_name instead
7822 of make_unique_name.
7825 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7826 correct operand in doing the split.
7828 2017-07-07 Carl Love <cel@us.ibm.com>
7830 * config/rs6000/rs6000-c: Add support for built-in function
7831 vector unsigned short vec_pack_to_short_fp32 (vector float,
7833 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7834 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7835 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7836 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7837 (convert_4f32_8i16): Add define_expand.
7838 * doc/extend.texi: Update the built-in documentation file for the
7839 new built-in function.
7841 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7843 * config/sparc/m8.md: New file.
7844 * config/sparc/sparc.md: Include m8.md.
7846 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7848 * config/sparc/sparc.opt: New option -mvis4b.
7849 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7850 (sparc_option_override): Handle VIS4B.
7851 (enum sparc_builtins): Define
7852 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7853 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7854 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7855 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7856 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7857 (check_constant_argument): New function.
7858 (sparc_vis_init_builtins): Define builtins
7859 __builtin_vis_dictunpack{8,16,32},
7860 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7861 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7862 __builtin_vis_fpcmpde{8,16,32}shl and
7863 __builtin_vis_fpcmpur{8,16,32}shl.
7864 (sparc_expand_builtin): Check that the constant operands to
7865 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7866 constant and in range.
7867 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7869 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7870 (SPARC_IMM5_P): Likewise.
7871 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7872 (enabled): Handle vis4b.
7873 (UNSPEC_DICTUNPACK): New unspec.
7874 (UNSPEC_FPCMPSHL): Likewise.
7875 (UNSPEC_FPUCMPSHL): Likewise.
7876 (UNSPEC_FPCMPDESHL): Likewise.
7877 (UNSPEC_FPCMPURSHL): Likewise.
7878 (cpu_feature): New CPU feature `vis4b'.
7879 (dictunpack{8,16,32}): New insns.
7880 (FPCSMODE): New mode iterator.
7881 (fpcscond): New code iterator.
7882 (fpcsucond): Likewise.
7883 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7884 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7885 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7886 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7887 * config/sparc/constraints.md: Define constraints `q' for unsigned
7888 2-bit integer constants and `t' for unsigned 5-bit integer
7890 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7892 (imm5_operand_dictunpack16): Likewise.
7893 (imm5_operand_dictunpack32): Likewise.
7894 (imm2_operand): Likewise.
7895 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7896 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7897 ditunpack* and fpcmp*shl builtins.
7899 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7901 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7902 * config.in: Add HAVE_AS_SPARC6 define.
7903 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7905 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7907 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7908 (CPP_CPU_SPEC): Handle m8.
7909 (ASM_CPU_SPEC): Likewise.
7910 * config/sparc/sparc-opts.h (enum processor_type): Add
7912 * config/sparc/sparc.c (m8_costs): New struct.
7913 (sparc_option_override): Handle TARGET_CPU_m8.
7914 (sparc32_initialize_trampoline): Likewise.
7915 (sparc64_initialize_trampoline): Likewise.
7916 (sparc_issue_rate): Likewise.
7917 (sparc_register_move_cost): Likewise.
7918 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7919 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7920 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7921 (CPP_CPU_SPEC): Handle M8.
7922 (ASM_CPU_SPEC): Likewise.
7923 (AS_M8_FLAG): Define.
7924 * config/sparc/sparc.md: Add m8 to the cpu attribute.
7925 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
7926 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
7928 * configure: Regenerate.
7929 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
7932 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7934 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
7936 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
7937 ("*movdi_insn_sp32"): Do not set v3pipe.
7938 ("*movsi_insn"): Likewise.
7939 ("*movdi_insn_sp64"): Likewise.
7940 ("*movsf_insn"): Likewise.
7941 ("*movdf_insn_sp32"): Likewise.
7942 ("*movdf_insn_sp64"): Likewise.
7943 ("*zero_extendsidi2_insn_sp64"): Likewise.
7944 ("*sign_extendsidi2_insn"): Likewise.
7945 ("*mov<VM32:mode>_insn"): Likewise.
7946 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7947 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7948 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7949 ("<vlop:code><VL:mode>3"): Likewise.
7950 ("*not_<vlop:code><VL:mode>3"): Likewise.
7951 ("*nand<VL:mode>_vis"): Likewise.
7952 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
7953 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
7954 ("one_cmpl<VL:mode>2"): Likewise.
7955 ("faligndata<VM64:mode>_vis"): Likewise.
7956 ("alignaddrsi_vis"): Likewise.
7957 ("alignaddrdi_vis"): Likweise.
7958 ("alignaddrlsi_vis"): Likewise.
7959 ("alignaddrldi_vis"): Likewise.
7960 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7961 ("bmaskdi_vis"): Likewise.
7962 ("bmasksi_vis"): Likewise.
7963 ("bshuffle<VM64:mode>_vis"): Likewise.
7964 ("cmask8<P:mode>_vis"): Likewise.
7965 ("cmask16<P:mode>_vis"): Likewise.
7966 ("cmask32<P:mode>_vis"): Likewise.
7967 ("pdistn<P:mode>_vis"): Likewise.
7968 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7970 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7972 * config/sparc/sparc.md ("subtype"): New insn attribute.
7973 ("*wrgsr_sp64"): Set insn subtype.
7974 ("*rdgsr_sp64"): Likewise.
7975 ("alignaddrsi_vis"): Likewise.
7976 ("alignaddrdi_vis"): Likewise.
7977 ("alignaddrlsi_vis"): Likewise.
7978 ("alignaddrldi_vis"): Likewise.
7979 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7980 ("fexpand_vis"): Likewise.
7981 ("fpmerge_vis"): Likewise.
7982 ("faligndata<VM64:mode>_vis"): Likewise.
7983 ("bshuffle<VM64:mode>_vis"): Likewise.
7984 ("cmask8<P:mode>_vis"): Likewise.
7985 ("cmask16<P:mode>_vis"): Likewise.
7986 ("cmask32<P:mode>_vis"): Likewise.
7987 ("fchksm16_vis"): Likewise.
7988 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
7989 ("fmean16_vis"): Likewise.
7990 ("fp<plusminus_insn>64_vis"): Likewise.
7991 ("<plusminus_insn>v8qi3"): Likewise.
7992 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7993 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
7994 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
7995 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
7996 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
7997 ("*movqi_insn"): Likewise.
7998 ("*movhi_insn"): Likewise.
7999 ("*movsi_insn"): Likewise.
8000 ("movsi_pic_gotdata_op"): Likewise.
8001 ("*movdi_insn_sp32"): Likewise.
8002 ("*movdi_insn_sp64"): Likewise.
8003 ("movdi_pic_gotdata_op"): Likewise.
8004 ("*movsf_insn"): Likewise.
8005 ("*movdf_insn_sp32"): Likewise.
8006 ("*movdf_insn_sp64"): Likewise.
8007 ("*zero_extendhisi2_insn"): Likewise.
8008 ("*zero_extendqihi2_insn"): Likewise.
8009 ("*zero_extendqisi2_insn"): Likewise.
8010 ("*zero_extendqidi2_insn"): Likewise.
8011 ("*zero_extendhidi2_insn"): Likewise.
8012 ("*zero_extendsidi2_insn_sp64"): Likewise.
8013 ("ldfsr"): Likewise.
8014 ("prefetch_64"): Likewise.
8015 ("prefetch_32"): Likewise.
8016 ("tie_ld32"): Likewise.
8017 ("tie_ld64"): Likewise.
8018 ("*tldo_ldub_sp32"): Likewise.
8019 ("*tldo_ldub1_sp32"): Likewise.
8020 ("*tldo_ldub2_sp32"): Likewise.
8021 ("*tldo_ldub_sp64"): Likewise.
8022 ("*tldo_ldub1_sp64"): Likewise.
8023 ("*tldo_ldub2_sp64"): Likewise.
8024 ("*tldo_ldub3_sp64"): Likewise.
8025 ("*tldo_lduh_sp32"): Likewise.
8026 ("*tldo_lduh1_sp32"): Likewise.
8027 ("*tldo_lduh_sp64"): Likewise.
8028 ("*tldo_lduh1_sp64"): Likewise.
8029 ("*tldo_lduh2_sp64"): Likewise.
8030 ("*tldo_lduw_sp32"): Likewise.
8031 ("*tldo_lduw_sp64"): Likewise.
8032 ("*tldo_lduw1_sp64"): Likewise.
8033 ("*tldo_ldx_sp64"): Likewise.
8034 ("*mov<VM32:mode>_insn"): Likewise.
8035 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8036 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8038 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8040 * config/sparc/sparc.md ("type"): New insn type viscmp.
8041 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8043 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8044 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8045 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8046 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8048 ("n7_vis_logical_11cycle"): Likewise.
8049 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8050 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8051 * config/sparc/niagara.md ("niag_vis"): Likewise.
8052 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8053 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8055 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8057 * config/sparc/sparc.md: New instruction type `bmask'.
8058 (bmaskdi_vis): Use the `bmask' type.
8059 (bmasksi_vis): Likewise.
8060 * config/sparc/ultra3.md (us3_array): Likewise.
8061 * config/sparc/niagara7.md (n7_array): Likewise.
8062 * config/sparc/niagara4.md (n4_array): Likewise.
8063 * config/sparc/niagara2.md (niag2_vis): Likewise.
8064 (niag3_vis): Likewise.
8065 * config/sparc/niagara.md (niag_vis): Likewise.
8067 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8069 * ipa-comdats.c: Remove optimize check from gate.
8070 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8071 for functions not optimized.
8072 (ipa_fn_summary_read): Skip optimize check.
8073 (ipa_fn_summary_write): Likewise.
8074 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8076 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8078 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8079 (check_callers): Give up on caller which is not optimized.
8080 (inline_small_functions): Likewise.
8081 (ipa_inline): Do not give up when not optimizing.
8082 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8083 away unoptimizes cdtors.
8084 (whole_program_function_and_variable_visibility): Do
8085 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8086 * ipa.c (process_references): Do not check optimize.
8087 (symbol_table::remove_unreachable_nodes): Update optimize check.
8088 (set_writeonly_bit): Update optimize check.
8089 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8090 (pass_ipa_single_use::gate): Remove.
8092 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8094 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8095 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8096 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8097 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8098 permute_load, permute_store, adjust_extract, adjust_splat,
8099 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8100 replace_swap_with_copy, dump_swap_insn_table,
8101 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8102 recombine_lvx_pattern, recombine_stvx_pattern,
8103 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8104 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8105 to file rs6000-p8swap.c.
8106 * config/rs6000/rs6000-p8swap.c: New file.
8107 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8108 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8109 and rs6000*-*-* targets.
8111 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8113 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8115 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8117 * lto-wrapper.c (merge_and_complain): Do not merge
8118 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8119 fsigned_zeros, ftrapping_math, fwrapv.
8120 (append_compiler_options): Do not track these options.
8121 (append_linker_options): Likewie
8123 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8125 * cgraphunit.c (cgraph_node::finalize_function): When
8126 !flag_toplevel_reorde set no_reorder flag.
8127 (varpool_node::finalize_decl): Likewise.
8128 (symbol_table::compile): Drop no toplevel reorder path.
8130 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8132 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8133 edges; zero probability is not better than uninitialized.
8135 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8137 * asan.h (asan_sanitize_allocas_p): Declare.
8138 * asan.c (asan_sanitize_allocas_p): New function.
8139 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8140 (handle_builtin_alloca): Likewise.
8141 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8142 if !asan_sanitize_allocas_p.
8143 * params.def (asan-instrument-allocas): Add new option.
8144 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8145 * opts.c (common_handle_option): Disable allocas sanitization for
8148 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8150 * asan.c: Include gimple-fold.h.
8151 (get_last_alloca_addr): New function.
8152 (handle_builtin_stackrestore): Likewise.
8153 (handle_builtin_alloca): Likewise.
8154 (asan_emit_allocas_unpoison): Likewise.
8155 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8156 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8157 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8158 (instrument_builtin_call): Pass gimple iterator to
8159 get_mem_refs_of_builtin_call.
8160 (last_alloca_addr): New global.
8161 * asan.h (asan_emit_allocas_unpoison): Declare.
8162 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8163 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8164 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8165 if function calls alloca.
8166 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8167 * gimple-fold.h (replace_call_with_value): Declare.
8168 * internal-fn.c: Include asan.h.
8169 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8170 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8172 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8174 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8175 (C_SELFTEST_FLAGS): New.
8176 (CPP_SELFTEST_FLAGS): New.
8177 (SELFTEST_DEPS): New, from deps of s-selftest.
8178 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8179 (CPP_SELFTEST_DEPS): New.
8180 (selftest): Add dependency on s-selftest-c++.
8181 (s-selftest): Rename to...
8182 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8183 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8184 than SELFTEST_FLAGS.
8185 (selftest-gdb): Rename to...
8186 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8188 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8189 (selftest-valgrind): Rename to...
8190 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8192 (selftest-valgrind): Reintroduce as an alias for
8193 selftest-c-valgrind.
8194 (s-selftest-c++): New.
8195 (selftest-c++-gdb): New.
8196 (selftest-c++-valgrind): New.
8198 2017-07-06 Olivier Hainque <hainque@adacore.com>
8200 * gcc.c (process_command): When deciding if undefined variables
8201 should be ignored when processing specs, accept "gcc -v" as well.
8203 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8205 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8206 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8208 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8210 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8211 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8212 * config/arm/arm-tables.opt: Regenerate.
8213 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8215 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8217 2017-07-06 Carl Love <cel@us.ibm.com>
8219 * ChangeLog: Clean up from mid air collision
8221 2017-07-06 Carl Love <cel@us.ibm.com>
8223 * config/rs6000/rs6000-c.c: Add support for built-in functions
8224 vector signed int vec_subc (vector signed int, vector signed int);
8225 vector signed __int128 vec_subc (vector signed __int128,
8226 vector signed __int128);
8227 vector unsigned __int128 vec_subc (vector unsigned __int128,
8228 vector unsigned __int128);
8229 vector signed int vec_sube (vector signed int, vector signed int,
8231 vector unsigned int vec_sube (vector unsigned int,
8232 vector unsigned int,
8233 vector unsigned int);
8234 vector signed __int128 vec_sube (vector signed __int128,
8235 vector signed __int128,
8236 vector signed__int128);
8237 vector unsigned __int128 vec_sube (vector unsigned __int128,
8238 vector unsigned __int128,
8239 vector unsigned __int128);
8240 vector signed int vec_subec (vector signed int, vector signed int,
8242 vector unsigned int vec_subec (vector unsigned int,
8243 vector unsigned int,
8244 vector unsigned int);
8245 vector signed __int128 vec_subec (vector signed __int128,
8246 vector signed __int128,
8247 vector signed__int128);
8248 vector unsigned __int128 vec_subec (vector unsigned __int128,
8249 vector unsigned __int128,
8250 vector unsigned __int128);
8251 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8252 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8253 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8254 BU_ALTIVEC_OVERLOAD_X definitions.
8255 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8256 * doc/extend.texi: Update the built-in documentation file for the new
8259 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8262 * diagnostic-show-locus.c (layout::layout): Use start and finish
8263 spelling location for the start and finish of each range.
8264 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8265 Add unused aspect param.
8266 * input.c (expand_location_1): Add "aspect" param, and use it
8267 to access the correct part of the location.
8268 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8270 (expand_location_to_spelling_point): Likewise.
8271 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8272 param, and pass it to expand_location_1.
8274 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8276 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8277 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8278 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8279 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8280 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8281 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8282 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8283 _mm_maskz_getmant_ss): New intrinsics.
8284 (__builtin_ia32_getexpss128_mask): Changed to ...
8285 __builtin_ia32_getexpss128_round ... this.
8286 (__builtin_ia32_getexpsd128_mask): Changed to ...
8287 __builtin_ia32_getexpsd128_round ... this.
8288 * config/i386/i386-builtin-types.def
8289 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8290 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8291 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8292 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8293 __builtin_ia32_getmantss_mask_round): New builtins.
8294 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8295 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8296 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8297 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8298 * config/i386/sse.md
8299 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8300 avx512f_sgetexp<mode><mask_scalar_name>
8301 <round_saeonly_scalar_name> ... this.
8302 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8303 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8304 vgetexp<ssescalarmodesuffix>
8305 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8306 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8307 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8308 avx512f_vgetmant<mode><mask_scalar_name>
8309 <round_saeonly_scalar_name> ... this.
8310 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8311 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8312 vgetmant<ssescalarmodesuffix>
8313 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8314 %0<mask_scalar_operand4>, %1, %2
8315 <round_saeonly_scalar_mask_op4>, %3} ... this.
8316 * config/i386/subst.md (mask_scalar_operand4,
8317 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8318 round_saeonly_scalar_nimm_predicate): New subst attributes.
8320 2017-07-06 Julia Koval <julia.koval@intel.com>
8322 * config/i386/i386.c (ix86_erase_embedded_rounding):
8323 Remove code for old rounding pattern.
8325 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8327 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8329 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8331 * doc/sourcebuild.texi (Test Directives, Variants of
8332 dg-require-support): Add documentation for dg-require-stack-check.
8334 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8336 * config/i386/subst.md (mask_scalar, round_scalar,
8337 round_saeonly_scalar): New meta-templates.
8338 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8339 round_scalar_mask_operand3, round_scalar_mask_op3,
8340 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8341 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8342 round_saeonly_scalar_constraint,
8343 round_saeonly_scalar_prefix): New subst attribute.
8344 * config/i386/sse.md
8345 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8346 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8347 <round_scalar_name> ... this.
8348 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8349 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8350 <round_scalar_name> ... this.
8351 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8352 <sse>_vm<code><mode>3<mask_scalar_name>
8353 <round_saeonly_scalar_name> ... this.
8354 (v<plusminus_mnemonic><ssescalarmodesuffix>
8355 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8356 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8357 v<plusminus_mnemonic><ssescalarmodesuffix>
8358 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8359 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8360 (v<multdiv_mnemonic><ssescalarmodesuffix>
8361 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8362 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8363 v<multdiv_mnemonic><ssescalarmodesuffix>
8364 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8365 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8366 (v<maxmin_float><ssescalarmodesuffix>
8367 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8368 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8369 v<maxmin_float><ssescalarmodesuffix>
8370 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8371 %0<mask_scalar_operand3>, %1, %<iptr>2
8372 <round_saeonly_scalar_mask_op3>} ... this.
8374 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8376 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8377 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8379 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8380 Alan Hayward <alan.hayward@arm.com>
8381 David Sherwood <david.sherwood@arm.com>
8383 * combine.c (simplify_if_then_else): Remove "enum" before
8385 * compare-elim.c (can_eliminate_compare): Likewise.
8386 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8388 (aarch64_lookup_simd_builtin_type): Likewise.
8389 (aarch64_simd_builtin_type): Likewise.
8390 (aarch64_init_simd_builtin_types): Likewise.
8391 (aarch64_simd_expand_args): Likewise.
8392 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8394 (aarch64_reverse_mask): Likewise.
8395 (aarch64_simd_emit_reg_reg_move): Likewise.
8396 (aarch64_gen_adjusted_ldpstp): Likewise.
8397 (aarch64_ccmp_mode_to_code): Likewise.
8398 (aarch64_operands_ok_for_ldpstp): Likewise.
8399 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8400 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8402 (aarch64_min_divisions_for_recip_mul): Likewise.
8403 (aarch64_reassociation_width): Likewise.
8404 (aarch64_get_condition_code_1): Likewise.
8405 (aarch64_simd_emit_reg_reg_move): Likewise.
8406 (aarch64_simd_attr_length_rglist): Likewise.
8407 (aarch64_reverse_mask): Likewise.
8408 (aarch64_operands_ok_for_ldpstp): Likewise.
8409 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8410 (aarch64_gen_adjusted_ldpstp): Likewise.
8411 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8413 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8414 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8415 (arm_lookup_simd_builtin_type): Likewise.
8416 (arm_simd_builtin_type): Likewise.
8417 (arm_init_simd_builtin_types): Likewise.
8418 (arm_expand_builtin_args): Likewise.
8419 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8420 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8421 (ft32_setup_incoming_varargs): Likewise.
8422 (ft32_function_arg): Likewise.
8423 (ft32_function_arg_advance): Likewise.
8424 (ft32_pass_by_reference): Likewise.
8425 (ft32_arg_partial_bytes): Likewise.
8426 (ft32_valid_pointer_mode): Likewise.
8427 (ft32_addr_space_pointer_mode): Likewise.
8428 (ft32_addr_space_legitimate_address_p): Likewise.
8429 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8431 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8432 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8433 (iamcu_alignment): Likewise.
8434 (canonicalize_vector_int_perm): Likewise.
8435 (ix86_noce_conversion_profitable_p): Likewise.
8436 (ix86_mpx_bound_mode): Likewise.
8437 (ix86_operands_ok_for_move_multiple): Likewise.
8438 * config/microblaze/microblaze-protos.h
8439 (microblaze_expand_conditional_branch_reg): Likewise.
8440 * config/microblaze/microblaze.c
8441 (microblaze_expand_conditional_branch_reg): Likewise.
8442 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8444 (rs6000_reassociation_width): Likewise.
8445 (rs6000_invalid_binary_op): Likewise.
8446 (fusion_p9_p): Likewise.
8447 (emit_fusion_p9_load): Likewise.
8448 (emit_fusion_p9_store): Likewise.
8449 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8451 (riscv_hard_regno_mode_ok_p): Likewise.
8452 (riscv_address_insns): Likewise.
8453 (riscv_split_symbol): Likewise.
8454 (riscv_legitimize_move): Likewise.
8455 (riscv_function_value): Likewise.
8456 (riscv_hard_regno_nregs): Likewise.
8457 (riscv_expand_builtin): Likewise.
8458 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8459 (riscv_build_integer): Likewise.
8460 (riscv_split_integer): Likewise.
8461 (riscv_legitimate_constant_p): Likewise.
8462 (riscv_cannot_force_const_mem): Likewise.
8463 (riscv_regno_mode_ok_for_base_p): Likewise.
8464 (riscv_valid_base_register_p): Likewise.
8465 (riscv_valid_offset_p): Likewise.
8466 (riscv_valid_lo_sum_p): Likewise.
8467 (riscv_classify_address): Likewise.
8468 (riscv_legitimate_address_p): Likewise.
8469 (riscv_address_insns): Likewise.
8470 (riscv_load_store_insns): Likewise.
8471 (riscv_force_binary): Likewise.
8472 (riscv_split_symbol): Likewise.
8473 (riscv_force_address): Likewise.
8474 (riscv_legitimize_address): Likewise.
8475 (riscv_move_integer): Likewise.
8476 (riscv_legitimize_const_move): Likewise.
8477 (riscv_legitimize_move): Likewise.
8478 (riscv_address_cost): Likewise.
8479 (riscv_subword): Likewise.
8480 (riscv_output_move): Likewise.
8481 (riscv_canonicalize_int_order_test): Likewise.
8482 (riscv_emit_int_order_test): Likewise.
8483 (riscv_function_arg_boundary): Likewise.
8484 (riscv_pass_mode_in_fpr_p): Likewise.
8485 (riscv_pass_fpr_single): Likewise.
8486 (riscv_pass_fpr_pair): Likewise.
8487 (riscv_get_arg_info): Likewise.
8488 (riscv_function_arg): Likewise.
8489 (riscv_function_arg_advance): Likewise.
8490 (riscv_arg_partial_bytes): Likewise.
8491 (riscv_function_value): Likewise.
8492 (riscv_pass_by_reference): Likewise.
8493 (riscv_setup_incoming_varargs): Likewise.
8494 (riscv_print_operand): Likewise.
8495 (riscv_elf_select_rtx_section): Likewise.
8496 (riscv_save_restore_reg): Likewise.
8497 (riscv_for_each_saved_reg): Likewise.
8498 (riscv_register_move_cost): Likewise.
8499 (riscv_hard_regno_mode_ok_p): Likewise.
8500 (riscv_hard_regno_nregs): Likewise.
8501 (riscv_class_max_nregs): Likewise.
8502 (riscv_memory_move_cost): Likewise.
8503 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8504 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8505 (rl78_addr_space_address_mode): Likewise.
8506 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8508 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8509 (rs6000_reassociation_width): Likewise.
8510 (rs6000_invalid_binary_op): Likewise.
8511 (fusion_p9_p): Likewise.
8512 (emit_fusion_p9_load): Likewise.
8513 (emit_fusion_p9_store): Likewise.
8514 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8515 (ok_for_simple_move_operands): Likewise.
8516 (ok_for_simple_move_strict_operands): Likewise.
8517 (ok_for_simple_arith_logic_operands): Likewise.
8518 (visium_legitimize_reload_address): Likewise.
8519 (visium_select_cc_mode): Likewise.
8520 (output_cbranch): Likewise.
8521 (visium_split_double_move): Likewise.
8522 (visium_expand_copysign): Likewise.
8523 (visium_expand_int_cstore): Likewise.
8524 (visium_expand_fp_cstore): Likewise.
8525 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8526 (visium_function_arg): Likewise.
8527 (visium_function_arg_advance): Likewise.
8528 (visium_libcall_value): Likewise.
8529 (visium_setup_incoming_varargs): Likewise.
8530 (visium_legitimate_constant_p): Likewise.
8531 (visium_legitimate_address_p): Likewise.
8532 (visium_legitimize_address): Likewise.
8533 (visium_secondary_reload): Likewise.
8534 (visium_register_move_cost): Likewise.
8535 (visium_memory_move_cost): Likewise.
8536 (prepare_move_operands): Likewise.
8537 (ok_for_simple_move_operands): Likewise.
8538 (ok_for_simple_move_strict_operands): Likewise.
8539 (ok_for_simple_arith_logic_operands): Likewise.
8540 (visium_function_value_1): Likewise.
8541 (rtx_ok_for_offset_p): Likewise.
8542 (visium_legitimize_reload_address): Likewise.
8543 (visium_split_double_move): Likewise.
8544 (visium_expand_copysign): Likewise.
8545 (visium_expand_int_cstore): Likewise.
8546 (visium_expand_fp_cstore): Likewise.
8547 (visium_split_cstore): Likewise.
8548 (visium_select_cc_mode): Likewise.
8549 (visium_split_cbranch): Likewise.
8550 (output_cbranch): Likewise.
8551 (visium_print_operand_address): Likewise.
8552 * expmed.c (flip_storage_order): Likewise.
8553 * expmed.h (emit_cstore): Likewise.
8554 (flip_storage_order): Likewise.
8555 * genrecog.c (validate_pattern): Likewise.
8556 * hsa-gen.c (gen_hsa_addr): Likewise.
8557 * internal-fn.c (expand_arith_overflow): Likewise.
8558 * ira-color.c (allocno_copy_cost_saving): Likewise.
8559 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8560 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8561 (process_invariant_for_inheritance): Likewise.
8562 * lra-eliminations.c (move_plus_up): Likewise.
8563 * omp-low.c (lower_oacc_reductions): Likewise.
8564 * simplify-rtx.c (simplify_subreg): Likewise.
8565 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8566 (TARGET_CHKP_BOUND_MODE): Likewise..
8567 * targhooks.c (default_chkp_bound_mode): Likewise.
8568 (default_setup_incoming_vararg_bounds): Likewise.
8569 * targhooks.h (default_chkp_bound_mode): Likewise.
8570 (default_setup_incoming_vararg_bounds): Likewise.
8571 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8572 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8573 (have_whole_vector_shift): Likewise.
8574 * tree-vect-stmts.c (vectorizable_load): Likewise.
8575 * doc/tm.texi: Regenerate.
8577 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8579 Graceful degrade if Binutils PR21472 is not available.
8582 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8583 .rodata in flash test fails.
8584 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8585 * confgure: Regenerate.
8586 * config.in: Regenerate.
8587 * config/avr/avr.c (avr_asm_named_section)
8588 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8589 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8590 (avr_asm_init_sections): Same.
8592 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8594 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8595 (fma<VH:mode>4_intrinsic): Likewise.
8596 (*fmsub<VCVTF:mode>4): Likewise.
8597 (*fmsub<VH:mode>4_intrinsic): Likewise.
8599 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8602 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8603 Don't depend on "optimize > 0".
8604 (out_movhi_r_mr, out_movqi_mr_r): Same.
8605 (out_movhi_mr_r, out_movqi_r_mr): Same.
8606 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8607 io_address_operand on "optimize > 0".
8609 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8611 * tree-loop-distribution.c: Add general explanantion on the pass.
8612 (generate_loops_for_partition): Mark distributed loop.
8613 (pg_add_dependence_edges): New parameter. Handle alias data
8614 dependence specially and record it in the parameter if asked.
8615 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8616 (init_partition_graph_vertices, add_partition_graph_edge): New.
8617 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8618 (free_partition_graph_vdata, build_partition_graph): New.
8619 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8620 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8621 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8622 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8623 (version_for_distribution_p, finalize_partitions): New.
8624 (distribute_loop): Handle alias data dependence specially. Factor
8625 out loop fusion code as functions and call these functions.
8627 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8629 * tree-loop-distribution.c (classify_partition): New parameter and
8630 better handle reduction statement.
8631 (rdg_build_partitions): Revise comment.
8632 (distribute_loop): Compute statements in all partitions and pass it
8633 to classify_partition.
8635 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8637 * tree-loop-distribution.c (enum partition_type): New.
8638 (struct partition): New field type.
8639 (partition_merge_into): Add parameter. Update partition type.
8640 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8641 (build_rdg_partition_for_vertex): Compute partition type.
8642 (rdg_build_partitions): Dump partition type.
8643 (distribute_loop): Update calls to partition_merge_into.
8645 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8647 * tree-loop-distribution.c (struct ddr_hasher): New.
8648 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8650 (classify_partition): Call get_data_dependence.
8651 (pg_add_dependence_edges): Ditto.
8652 (distribute_loop): Release data dependence hash table.
8654 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8656 * tree-loop-distribution.c (ref_base_address): Delete.
8657 (similar_memory_accesses): Rename ...
8658 (share_memory_accesses): ... to this. Check if partitions access
8659 the same memory reference.
8660 (distribute_loop): Call share_memory_accesses.
8662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8664 * tree-loop-distribution.c (struct partition): New field recording
8666 (partition_alloc, partition_free): Init and release data refs.
8667 (partition_merge_into): Merge data refs.
8668 (build_rdg_partition_for_vertex): Collect data refs for partition.
8669 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8671 (distribute_loop): Remve data refs from vertice data of partition
8674 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8676 * tree-loop-distribution.c (params.h): Include header file.
8677 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8678 (datarefs_vec): New global var.
8679 (create_rdg_vertices): Use datarefs_vec directly.
8680 (free_rdg): Don't free data references.
8681 (build_rdg): Update use. Don't free data references.
8682 (distribute_loop): Compute global variable for data references.
8683 Bail out if there are too many data references.
8685 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8687 * tree-loop-distribution.c (loop_nest): New global var.
8688 (build_rdg): Use loop directly, rather than loop nest.
8689 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8691 (distribute_loop): Compute global variable loop nest. Update use.
8693 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8695 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8696 (partition_merge_into): New parameter. Dump reason for fusion.
8697 (distribute_loop): Update use of partition_merge_into.
8699 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8701 * tree-loop-distribution.c (bb_top_order_index): New.
8702 (bb_top_order_index_size, bb_top_order_cmp): New.
8703 (stmts_from_loop): Use topological order.
8704 (pass_loop_distribution::execute): Compute and release topological
8705 order for basic blocks.
8707 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8709 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8712 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8714 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
8715 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
8716 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
8717 * internal-fn.def (LOOP_DIST_ALIAS): New.
8718 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
8719 (fold_loop_internal_call): ... this.
8720 (vect_loop_dist_alias_call): New function.
8721 (set_uid_loop_bbs): Call fold_loop_internal_call.
8722 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
8725 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8728 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
8729 Require dead FLAGS_REG at the beginning of a peephole.
8731 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8734 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
8735 arguments in the call to __builtin_ia32_sbb_u32.
8736 (_subborrow_u64): Swap _X and _Y arguments in the call to
8737 __builtin_ia32_sbb_u64.
8739 2017-07-04 Jakub Jelinek <jakub@redhat.com>
8742 * tree-vrp.c (compare_assert_loc): Turn into a function template
8743 with stable template parameter. Only test if a->e is NULL,
8744 !a->e == !b->e has been verified already. Use e == NULL or
8745 e != NULL instead of e or ! e tests. If stable is true, don't use
8746 iterative_hash_expr, on the other side allow a or b or both NULL
8747 and sort the NULLs last.
8748 (process_assert_insertions): Sort using compare_assert_loc<false>
8749 instead of compare_assert_loc, later sort using
8750 compare_assert_loc<true> before calling process_assert_insertions_for
8751 in a loop. Use break instead of continue once seen NULL pointer.
8753 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8755 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8756 Cortex-R7 and Cortex-R8 processors.
8758 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8760 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8761 uninitialized while src is not.
8763 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8765 * common/config/arm/arm-common.c: Adjust include path for
8767 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8768 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8769 (arm-cpu-data.h): Likewise.
8770 (arm-cpu-cdata.h): Likewise.
8771 * config/arm/arm-cpu.h: Delete.
8772 * config/arm/arm-cpu-cdata.h: Delete.
8773 * config/arm/arm-cpu-data.h: Delete.
8775 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8777 * config/arm/arm-cpus.in (cortex-a55): New.
8778 (cortex-a75): Likewise.
8779 (cortex-a75.cortex-a55): Likewise.
8780 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8782 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8783 * config/arm/arm-cpu-cdata.h: Regenerate.
8784 * config/arm/arm-cpu-data.h: Regenerate.
8785 * config/arm/arm-cpu.h: Regenerate.
8786 * config/arm/arm-tables.opt: Regenerate.
8787 * config/arm/arm-tune.md: Regenerate.
8789 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8791 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8793 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8795 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8798 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8800 PR tree-optimization/81292
8801 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8802 full_string_p, also call adjust_related_strinfos if the adjustment
8803 is simple, otherwise invalidate related strinfos.
8805 2017-07-04 Martin Liska <mliska@suse.cz>
8808 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8809 newly created variable as DECL_IGNORED_P.
8811 2017-07-04 Martin Liska <mliska@suse.cz>
8814 * ipa-inline.c (inline_small_functions):
8815 Use xstrdup_for_dump.
8817 2017-07-04 Tom de Vries <tom@codesourcery.com>
8819 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8821 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8824 * config/darwin.c (darwin_function_switched_text_sections):
8827 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8829 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8831 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8833 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8835 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8837 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8838 min_profitable_iters, and th as inclusive lower bounds.
8839 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8840 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8841 for min_profitable_iters and min_profitable_estimate.
8842 (vect_transform_loop): Treat th as an inclusive lower bound.
8843 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8845 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8848 * config/darwin.c (darwin_function_switched_text_sections):
8849 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8850 in two pieces, and suppress the use of buf.
8852 2017-07-03 Nathan Sidwell <nathan@acm.org>
8854 * hash-table.h (hash_table_mod1): Fix indentation.
8856 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8859 * predict.c (force_edge_cold): Be more careful about propagation
8861 * profile-count.h (profile_probability::guessed,
8862 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8864 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8866 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8868 * doc/invoke.texi (rcpc architecture extension): Document it.
8870 2017-07-03 Richard Biener <rguenther@suse.de>
8872 PR tree-optimization/60510
8873 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8874 the scalar reduction PHI and use it.
8875 (vectorizable_reduction): Properly guard the single_defuse_cycle
8876 path for non-SLP reduction chains where we cannot use it.
8877 Rework reduc_def/index and vector type deduction. Rework
8878 vector operand gathering during reduction op code-gen.
8879 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8880 chains dissolve the chain and leave it to non-SLP reduction
8883 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8885 * tree-data-ref.h (dr_alignment): Declare.
8886 * tree-data-ref.c (dr_alignment): New function.
8887 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8888 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8890 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8892 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8894 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8895 and base_misalignment fields.
8896 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8897 * tree-data-ref.c: Include builtins.h.
8898 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8899 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8900 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8901 * tree-vect-data-refs.c: Include tree-cfg.h.
8902 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8903 fields instead of calculating an alignment here.
8904 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8905 innermost_loop_behavior fields.
8907 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8909 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8911 (DR_STEP_ALIGNMENT): New macro.
8912 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8913 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8914 (create_data_ref): Print it.
8915 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8916 to tell whether the step preserves vector (mis)alignment.
8917 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8918 Move the check for an integer step and generalise to all INTEGER_CST.
8919 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8920 Print the outer step alignment.
8922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8924 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
8925 with offset_alignment.
8926 (DR_ALIGNED_TO): Delete.
8927 (DR_OFFSET_ALIGNMENT): New macro.
8928 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
8929 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
8930 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
8931 (create_data_ref): Likewise.
8932 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8933 (vect_analyze_data_refs): Likewise.
8934 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
8935 creating dummy innermost behavior.
8937 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8939 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
8940 with a "innermost_loop_behavior *" and refeence tree.
8941 * tree-data-ref.c (dr_analyze_innermost): Likewise.
8942 (create_data_ref): Update call accordingly.
8943 * tree-predcom.c (find_looparound_phi): Likewise.
8945 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8947 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
8948 fields with dr_wrt_vec_loop.
8949 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
8950 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
8951 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
8952 (vect_dr_behavior): New function.
8953 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8954 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
8955 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
8956 track whether the step preserves the misalignment.
8957 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8958 Use vect_dr_behavior.
8959 (vect_setup_realignment): Update call accordingly.
8960 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
8961 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
8962 call to vect_create_addr_base_for_vector_ref.
8963 (vect_create_cond_for_align_checks): Likewise.
8964 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
8965 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
8966 (vect_recog_mask_conversion_pattern): Likewise.
8967 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
8968 (new_stmt_vec_info): Remove redundant zeroing.
8970 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8972 * common/config/arm/arm-common.c (arm_be8_option): New function.
8973 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
8974 (ISA_ARMv6): Add isa_bit_be8.
8975 * config/arm/arm.h (arm_be8_option): Add prototype.
8976 (BE8_SPEC_FUNCTION): New define.
8977 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
8978 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
8979 (mlittle-endian): Similarly.
8980 (mbe8, mbe32): New options.
8981 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
8982 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
8984 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8986 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
8988 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8990 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
8991 (cleanup_tree_cfg_bb): Use it.
8992 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
8994 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
8996 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8999 * loop-doloop.c (add_test): Update profile.
9001 2017-07-03 Martin Liska <mliska@suse.cz>
9004 * sanopt.c (rewrite_usage_of_param): New function.
9005 (sanitize_rewrite_addressable_params): Likewise.
9006 (pass_sanopt::execute): Call rewrite_usage_of_param.
9008 2017-07-03 Richard Biener <rguenther@suse.de>
9010 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9011 back to using VIEW_CONVERT_EXPR.
9013 2017-07-03 Martin Liska <mliska@suse.cz>
9016 * doc/extend.texi: Document when a resolver function is
9017 generated for target_clones.
9019 2017-07-03 Martin Liska <mliska@suse.cz>
9021 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9022 and shadow memory of auto variables which are subject of
9023 use-after-scope sanitization.
9024 (asan_expand_mark_ifn): Add do set only when is_poison.
9026 2016-07-03 Richard Biener <rguenther@suse.de>
9028 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9030 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9031 (vectorizable_reduction): Perform reduction PHI creation when
9032 visiting a reduction PHI and adjust and simplify code generation
9033 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9034 (vect_transform_loop): Visit reduction PHIs.
9035 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9036 defs into the SLP tree.
9037 (vect_build_slp_tree): Reduction defs terminate the recursion.
9038 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9040 (vect_get_vec_defs_for_stmt_copy): Export.
9041 (vect_get_vec_defs): Likewise.
9042 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9044 (vect_get_vec_defs_for_stmt_copy): Declare.
9045 (vect_get_vec_defs): Likewise.
9047 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9049 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9050 parameter with a "loop" parameter and use it instead of the
9051 loop containing DR_STMT. Don't check simple_iv when doing
9052 BB analysis. Describe the two analysis modes in the comment.
9054 2017-07-03 Tom de Vries <tom@codesourcery.com>
9056 PR tree-optimization/69468
9057 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9058 (find_same_succ_bb): Handle ignore_edge_flags.
9060 2017-07-03 Tom de Vries <tom@codesourcery.com>
9062 PR tree-optimization/81192
9063 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9065 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9067 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9069 2017-07-03 Tom de Vries <tom@codesourcery.com>
9071 PR tree-optimization/81192
9072 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9073 BB_SAME_SUCC (bb) == NULL.
9075 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9077 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9080 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9082 * dumpfile.c: Include profile-count.h
9083 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9085 (insert_cond_bb): Update profile.
9086 * tree-cfg.h (insert_cond_bb): Update prototype.
9087 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9088 * tree-dump.c: Do not include tree-cfg.
9090 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9092 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9094 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9096 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9099 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9101 * tree-complex.c (expand_complex_div_wide): update profile.
9103 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9104 Alan Hayward <alan.hayward@arm.com>
9105 David Sherwood <david.sherwood@arm.com>
9107 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9108 (CORETYPES_H): New define.
9109 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9110 (insn-modes-inline.h, s-modes-inline-h): New rules.
9111 (generated_files): Add insn-modes-inline.h.
9112 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9113 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9114 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9115 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9116 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9117 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9118 (build/gencodes.o, build/genconditions.o): Likewise.
9119 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9120 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9121 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9122 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9123 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9124 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9125 * coretypes.h: Include everything up to real.h for generators.
9126 Include insn-modes.h first. Include wide-int-print.h after
9127 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9128 * machmode.h: Don't include insn-modes.h here.
9129 * function-tests.c: Remove includes of signop.h, machmode.h,
9130 double-int.h and wide-int.h.
9132 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9134 * optc-save-gen.awk: Likewise.
9135 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9136 * godump.c: Remove include of wide-int-print.h.
9137 * pretty-print.h: Likewise.
9138 * wide-int-print.cc: Likewise.
9139 * wide-int.cc: Likewise.
9140 * hash-map-tests.c: Remove include of signop.h.
9141 * hash-set-tests.c: Likewise.
9142 * rtl-tests.c: Likewise.
9143 * mkconfig.sh: Remove include of machmode.h.
9144 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9146 (emit_insn_modes_inline_h): ...this new function. Emit the code
9147 into an insn-modes-inline.h header file, adding appropriate
9148 include guards and end comments.
9149 (emit_insn_modes_c_header): Remove include of machmode.h.
9150 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9152 (main): Handle -i flag and call emit_insn_modes_inline_h when
9155 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9157 * tree-ssa-strlen.c (strinfo): Rename the length field to
9158 nonzero_chars. Add a full_string_p field.
9159 (compare_nonzero_chars, zero_length_string_p): New functions.
9160 (get_addr_stridx): Add an offset_out parameter.
9161 Use compare_nonzero_chars.
9162 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9163 (new_strinfo): Update after above changes to strinfo.
9164 (set_endptr_and_length): Set full_string_p.
9165 (get_string_length): Update after above changes to strinfo.
9166 (unshare_strinfo): Update call to new_strinfo.
9167 (maybe_invalidate): Likewise.
9168 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9169 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9170 as a uhwi instead of an shwi. Update after above changes to
9171 strinfo and new_strinfo.
9172 (zero_length_string): Assert that chainsi contains full strings.
9173 Use zero_length_string_p. Update call to new_strinfo.
9174 (adjust_related_strinfos): Update after above changes to strinfo.
9175 Copy full_string_p from origsi.
9176 (adjust_last_stmt): Use zero_length_string_p.
9177 (handle_builtin_strlen): Update after above changes to strinfo and
9178 new_strinfo. Install the lhs as the string length if the previous
9179 entry didn't describe a full string.
9180 (handle_builtin_strchr): Update after above changes to strinfo
9182 (handle_builtin_strcpy): Likewise.
9183 (handle_builtin_strcat): Likewise.
9184 (handle_builtin_malloc): Likewise.
9185 (handle_pointer_plus): Likewise.
9186 (handle_builtin_memcpy): Likewise. Track nonzero characters
9187 that aren't necessarily followed by a nul terminator.
9188 (handle_char_store): Likewise.
9190 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9192 PR tree-optimization/80769
9193 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9194 for malloc and calloc. Document the new invariant that all related
9195 strinfos have delayed lengths or none do.
9196 (verify_related_strinfos): Move earlier in file.
9197 (set_endptr_and_length): New function, split out from...
9198 (get_string_length): ...here. Also set the lengths of related
9200 (zero_length_string): Assert that chainsi has known (rather than
9202 (adjust_related_strinfos): Likewise.
9204 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9206 PR tree-optimization/81136
9207 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9208 assert that two references with the same misalignment have the same
9209 compile-time misalignment if those compile-time misalignments
9212 2017-07-01 Andi Kleen <ak@linux.intel.com>
9214 * print-tree.c (print_node): Print all attributes.
9216 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9218 * cfg.c (scale_bbs_frequencies): New function.
9219 * cfg.h (scale_bbs_frequencies): Declare it.
9220 * cfgloopanal.c (single_likely_exit): Cleanup.
9221 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9223 (scale_loop_profile): Likewise.
9224 (loop_version): Likewise.
9225 (create_empty_loop_on_edge): Update.
9226 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9227 scale_loop_frequencies, scale_loop_profile, loopify,
9228 loop_version): Update prototypes.
9229 * modulo-sched.c (sms_schedule): Update.
9230 * predict.c (unlikely_executed_edge_p): Also check probability.
9231 (probably_never_executed_edge_p): Fix typo.
9232 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9233 * tree-parloops.c (gen_parallel_loop): Update.
9234 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9235 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9236 * tree-ssa-loop-split.c (split_loop): Update.
9237 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9238 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9239 (vect_loop_versioning): Update.
9240 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9242 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9244 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9246 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9248 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9249 to keep profile consistent.
9251 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9253 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9254 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9255 * profile-count.h (max_safe_multiplier): Make unsigned.
9256 (profile_count::guessed_zero): New.
9258 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9260 * bb-reorder.c (fix_up_crossing_landing_pad,
9261 fix_crossing_conditional_branches): Use make_single_succ_edge
9262 to keep profile consistent.
9264 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9266 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9269 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9272 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9273 the right scopes, make sure cond_jump isn't preserved between multiple
9274 iterations. Search for fallthru edge whenever there are 3+ edges and
9275 use find_fallthru_edge for it.
9277 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9279 Patch by Alexander Monakov <amonakov@ispras.ru>
9280 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9281 probabilities consistently.
9283 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9285 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9286 emit_cmp_and_jump_insns.
9288 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9291 * tree-inline.c (expand_call_inline): Combine profile statuses.
9293 2017-06-30 Andrew Pinski <apinski@cavium.com>
9295 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9296 fold_stmt returned true.
9298 2017-06-30 Nathan Sidwell <nathan@acm.org>
9300 * ggc.h (empty_string): Delete.
9301 * cfgexpand.c (expand_asm_stmt): Use plain "".
9302 * optabs.c (expand_asm_memory_barrier): Likewise.
9303 * stringpool.c (empty_string): Delete.
9304 (digit_vector, digit_string): Delete.
9305 (ggc_alloc_string): Use plain "", don't optimize single digit
9306 strings. Use ggc_alloc_atomic.
9308 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9310 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9311 comparison set and one other set, use the cost of the non-comparison
9314 2017-06-30 Nathan Sidwell <nathan@acm.org>
9316 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9319 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9321 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9322 loops. Remove now unneeded calls to gimple_switch_set_label() that
9323 just set removed labels to NULL_TREE.
9325 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9327 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9328 (set_range_info): ...here. Only call set_range_info_raw if domain
9330 (set_nonzero_bits): Call set_range_info_raw.
9331 * tree-ssanames.h (set_range_info_raw): New.
9333 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9336 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9337 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9338 of nonimmediate_operand and <store_mask_constraint> instead of m
9339 for the input operand. For V8FI iterator, always split if input
9340 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9341 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9342 <store_mask_predicate> instead of register_operand and
9343 <store_mask_constraint> instead of v for the input operand. Make
9344 sure both operands aren't MEMs for if not <mask_applied>.
9346 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9348 * lto-wrapper.c (copy_file) Close both file descriptors before
9351 2017-06-30 Martin Liska <mliska@suse.cz>
9354 * multiple_target.c (create_dispatcher_calls): Make ifunc
9355 also for function that don't have calls or are not referenced.
9357 2017-06-30 Richard Biener <rguenther@suse.de>
9359 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9360 analyze the first scalar stmt. Move vector type computation
9361 for the BB case here from ...
9362 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9363 live operation processing in the SLP case properly.
9365 2017-06-30 Richard Biener <rguenther@suse.de>
9367 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9369 2017-06-30 Martin Liska <mliska@suse.cz>
9372 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9373 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9375 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9377 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9380 2017-06-30 Richard Biener <rguenther@suse.de>
9382 PR tree-optimization/81249
9383 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9384 condition reduction result to original scalar type.
9386 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9388 * profile-count.h (enum profile_quality): Fix typos and whitespace
9391 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9393 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9394 type for branch probabilities.
9396 2017-06-29 Julian Brown <julian@codesourcery.com>
9397 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9399 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9400 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9401 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9402 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9404 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9406 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9407 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9408 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9409 CC usage from generic code to here.
9410 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9411 CC usage into the target macros.
9413 2017-06-29 Maya Rashish <coypu@sdf.org>
9415 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9418 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9420 * arm/arm-builtins.c: Include profile-count.h
9421 * except.c (sjlj_emit_function_enter): Use
9422 profile_probability::unlikely.
9424 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9426 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9427 and tocrel_offset be pointer args rather than implicitly using
9429 (legitimate_constant_pool_address_p, rs6000_emit_move,
9430 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9431 tocrel_offset and use in toc_relative_expr_p call.
9432 (print_operand, print_operand_address): Use static tocrel_base_oac
9433 and tocrel_offset_oac.
9434 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9437 2017-06-29 Maya Rashish <coypu@sdf.org>
9439 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9441 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9443 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9444 objects, take into account only the alignment of 'op0' and 'mode1' if
9447 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9449 * ccmp.c (ccmp_tree_comparison_p): New function.
9450 (ccmp_candidate_p): Update to use above function.
9451 (get_compare_parts): New function.
9452 (expand_ccmp_next): Update to use new functions.
9453 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9455 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9456 take mode as argument.
9457 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9458 * expr.c (expand_expr_real_1): Pass mode as argument.
9460 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9462 * combine.c (combine_instructions): Print insns to dump_file, together
9465 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9467 * asan.c (asan_emit_stack_protection): Update.
9468 (create_cond_insert_point): Update.
9469 * auto-profile.c (afdo_propagate_circuit): Update.
9470 * basic-block.h (struct edge_def): Turn probability to
9471 profile_probability.
9472 (EDGE_FREQUENCY): Update.
9473 * bb-reorder.c (find_traces_1_round): Update.
9474 (better_edge_p): Update.
9475 (sanitize_hot_paths): Update.
9476 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9477 (make_single_succ_edge): Update.
9478 (check_bb_profile): Update.
9479 (dump_edge_info): Update.
9480 (update_bb_profile_for_threading): Update.
9481 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9482 probabilitycount to 0.
9483 * cfgbuild.c (compute_outgoing_frequencies): Update.
9484 * cfgcleanup.c (try_forward_edges): Update.
9485 (outgoing_edges_match): Update.
9486 (try_crossjump_to_edge): Update.
9487 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9488 (expand_gimple_tailcall): Update.
9489 (construct_init_block): Use make_single_succ_edge.
9490 (construct_exit_block): Use make_single_succ_edge.
9491 * cfghooks.c (verify_flow_info): Update.
9492 (redirect_edge_succ_nodup): Update.
9493 (split_edge): Update.
9494 (account_profile_record): Update.
9495 * cfgloopanal.c (single_likely_exit): Update.
9496 * cfgloopmanip.c (scale_loop_profile): Update.
9497 (set_zero_probability): Remove.
9498 (duplicate_loop_to_header_edge): Update.
9499 * cfgloopmanip.h (loop_version): Update prototype.
9500 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9501 (force_nonfallthru_and_redirect): Update.
9502 (update_br_prob_note): Update.
9503 (rtl_verify_edges): Update.
9504 (purge_dead_edges): Update.
9505 (rtl_lv_add_condition_to_bb): Update.
9506 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9507 * cgraphunit.c (init_lowered_empty_function): Update.
9508 (cgraph_node::expand_thunk): Update.
9509 * cilk-common.c: Include profile-count.h
9510 * dojump.c (inv): Remove.
9511 (jumpifnot): Update.
9512 (jumpifnot_1): Update.
9513 (do_jump_1): Update.
9515 (do_jump_by_parts_greater_rtx): Update.
9516 (do_compare_rtx_and_jump): Update.
9517 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9518 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9519 * dwarf2cfi.c: Include profile-count.h
9520 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9521 (sjlj_emit_dispatch_table): Likewise.
9522 * explow.c: Include profile-count.h
9523 * expmed.c (emit_store_flag_force): Update.
9524 (do_cmp_and_jump): Update.
9525 * expr.c (compare_by_pieces_d::generate): Update.
9526 (compare_by_pieces_d::finish_mode): Update.
9527 (emit_block_move_via_loop): Update.
9528 (store_expr_with_bounds): Update.
9529 (store_constructor): Update.
9530 (expand_expr_real_2): Update.
9531 (expand_expr_real_1): Update.
9532 * expr.h (try_casesi, try_tablejump): Update prototypes.
9533 * gimple-pretty-print.c (dump_probability): Update.
9534 (dump_profile): New.
9535 (dump_gimple_label): Update.
9536 (dump_gimple_bb_header): Update.
9537 * graph.c (draw_cfg_node_succ_edges): Update.
9538 * hsa-gen.c (convert_switch_statements): Update.
9539 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9540 (find_if_case_1): Update.
9541 (find_if_case_2): Update.
9542 * internal-fn.c (expand_arith_overflow_result_store): Update.
9543 (expand_addsub_overflow): Update.
9544 (expand_neg_overflow): Update.
9545 (expand_mul_overflow): Update.
9546 (expand_vector_ubsan_overflow): Update.
9547 * ipa-cp.c (good_cloning_opportunity_p): Update.
9548 * ipa-split.c (split_function): Use make_single_succ_edge.
9549 * ipa-utils.c (ipa_merge_profiles): Update.
9550 * loop-doloop.c (add_test): Update.
9551 (doloop_modify): Update.
9552 * loop-unroll.c (compare_and_jump_seq): Update.
9553 (unroll_loop_runtime_iterations): Update.
9554 * lra-constraints.c (lra_inheritance): Update.
9555 * lto-streamer-in.c (input_cfg): Update.
9556 * lto-streamer-out.c (output_cfg): Update.
9557 * mcf.c (adjust_cfg_counts): Update.
9558 * modulo-sched.c (sms_schedule): Update.
9559 * omp-expand.c (expand_omp_for_init_counts): Update.
9560 (extract_omp_for_update_vars): Update.
9561 (expand_omp_ordered_sink): Update.
9562 (expand_omp_for_ordered_loops): Update.
9563 (expand_omp_for_generic): Update.
9564 (expand_omp_for_static_nochunk): Update.
9565 (expand_omp_for_static_chunk): Update.
9566 (expand_cilk_for): Update.
9567 (expand_omp_simd): Update.
9568 (expand_omp_taskloop_for_outer): Update.
9569 (expand_omp_taskloop_for_inner): Update.
9570 * omp-simd-clone.c (simd_clone_adjust): Update.
9571 * optabs.c (expand_doubleword_shift): Update.
9572 (expand_abs): Update.
9573 (emit_cmp_and_jump_insn_1): Update.
9574 (expand_compare_and_swap_loop): Update.
9575 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9576 * predict.c (predictable_edge_p): Update.
9577 (edge_probability_reliable_p): Update.
9578 (set_even_probabilities): Update.
9579 (combine_predictions_for_insn): Update.
9580 (combine_predictions_for_bb): Update.
9581 (propagate_freq): Update.
9582 (estimate_bb_frequencies): Update.
9583 (force_edge_cold): Update.
9584 * profile-count.c (profile_count::dump): Add missing space into dump.
9585 (profile_count::debug): Add newline.
9586 (profile_count::differs_from_p): Explicitly convert to unsigned.
9587 (profile_count::stream_in): Update.
9588 (profile_probability::dump): New member function.
9589 (profile_probability::debug): New member function.
9590 (profile_probability::differs_from_p): New member function.
9591 (profile_probability::differs_lot_from_p): New member function.
9592 (profile_probability::stream_in): New member function.
9593 (profile_probability::stream_out): New member function.
9594 * profile-count.h (profile_count_quality): Rename to ...
9595 (profile_quality): ... this one.
9596 (profile_probability): New.
9597 (profile_count): Update.
9598 * profile.c (compute_branch_probabilities): Update.
9599 * recog.c (peep2_attempt): Update.
9600 * sched-ebb.c (schedule_ebbs): Update.
9601 * sched-rgn.c (find_single_block_region): Update.
9602 (compute_dom_prob_ps): Update.
9603 (schedule_region): Update.
9604 * sel-sched-ir.c (compute_succs_info): Update.
9605 * stmt.c (struct case_node): Update.
9606 (do_jump_if_equal): Update.
9607 (get_outgoing_edge_probs): Update.
9608 (conditional_probability): Update.
9609 (emit_case_dispatch_table): Update.
9610 (expand_case): Update.
9611 (expand_sjlj_dispatch_table): Update.
9612 (emit_case_nodes): Update.
9613 * targhooks.c: Update.
9614 * tracer.c (better_p): Update.
9615 (find_best_successor): Update.
9616 * trans-mem.c (expand_transaction): Update.
9617 * tree-call-cdce.c: Update.
9618 * tree-cfg.c (gimple_split_edge): Upate.
9619 (move_sese_region_to_fn): Upate.
9620 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9621 * tree-eh.c (lower_resx): Upate.
9622 (cleanup_empty_eh_move_lp): Upate.
9623 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9624 * tree-inline.c (copy_edges_for_bb): Update.
9625 (copy_cfg_body): Update.
9626 * tree-parloops.c (gen_parallel_loop): Update.
9627 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9628 (gimple_gen_time_profiler): Update.
9629 * tree-ssa-dce.c (remove_dead_stmt): Update.
9630 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9631 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9632 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9633 (unloop_loops): Update.
9634 (try_peel_loop): Update.
9635 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9636 * tree-ssa-loop-split.c (connect_loops): Update.
9637 (split_loop): Update.
9638 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9639 (hoist_guard): Update.
9640 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9641 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9642 (value_replacement): Update.
9643 * tree-ssa-reassoc.c (branch_fixup): Update.
9644 * tree-ssa-tail-merge.c (replace_block_by): Update.
9645 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9646 (create_edge_and_update_destination_phis): Update.
9647 (compute_path_counts): Update.
9648 (recompute_probabilities): Update.
9649 (update_joiner_offpath_counts): Update.
9650 (freqs_to_counts_path): Update.
9651 (duplicate_thread_path): Update.
9652 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9653 (struct switch_conv_info): Update.
9654 (gen_inbound_check): Update.
9655 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9656 (vect_do_peeling): Update.
9657 (vect_loop_versioning): Update.
9658 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9659 (optimize_mask_stores): Update.
9660 * ubsan.c (ubsan_expand_null_ifn): Update.
9661 * value-prof.c (gimple_divmod_fixed_value): Update.
9662 (gimple_divmod_fixed_value_transform): Update.
9663 (gimple_mod_pow2): Update.
9664 (gimple_mod_pow2_value_transform): Update.
9665 (gimple_mod_subtract): Update.
9666 (gimple_mod_subtract_transform): Update.
9667 (gimple_ic): Update.
9668 (gimple_stringop_fixed_value): Update.
9669 (gimple_stringops_transform): Update.
9670 * value-prof.h: Update.
9672 2017-06-29 Carl Love <cel@us.ibm.com>
9674 * config/rs6000/rs6000-c.c: Add support for built-in functions
9675 vector signed int vec_signed (vector float);
9676 vector signed long long vec_signed (vector double);
9677 vector signed int vec_signed2 (vector double, vector double);
9678 vector signed int vec_signede (vector double);
9679 vector signed int vec_signedo (vector double);
9680 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9681 instruction generator.
9682 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9683 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9684 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9686 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9687 vunsignede_v2df): Add define_expands.
9688 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9689 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9690 VEC_UNSIGNEDO): Add definitions.
9691 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9692 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9693 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9694 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9695 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9696 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9697 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9698 vec_unsignede, vec_unsignedo): Add builtin defines.
9699 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9701 * doc/extend.texi: Update the built-in documentation file for the
9702 new built-in functions.
9704 2017-06-29 Richard Biener <rguenther@suse.de>
9706 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9707 reduction chains to LOOP_VINFO_REDUCTIONS.
9708 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9709 SLP reductions after processing reduction chains.
9711 2017-06-29 Nathan Sidwell <nathan@acm.org>
9713 * builtins.c (fold_builtin_FUNCTION): Use
9714 lang_hooks.decl_printable_name.
9716 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
9719 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
9720 with only one label.
9721 * stmt.c (expand_case): Assert NCASES is greater than one.
9723 2017-06-29 Richard Biener <rguenther@suse.de>
9725 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
9727 (group_case_labels): Likewise.
9728 (find_taken_edge): Push sanity checking on val to workers...
9729 (find_taken_edge_cond_expr): ... here
9730 (find_taken_edge_switch_expr): ... and here, handle cases
9731 with just a default label.
9732 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
9733 (group_case_labels): Likewise.
9734 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
9735 group_case_labels does anything cleanup the CFG again.
9737 2017-06-29 Bin Cheng <bin.cheng@arm.com>
9739 PR tree-optimization/81196
9740 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
9741 exit condition comparing two IVs.
9743 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
9745 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
9746 profile to the dummy entry at the end of the list of architectures.
9747 * config/arm/arm-cpu-cdata.h: Regenerated.
9749 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9750 Michael Collison <michael.collison@arm.com>
9753 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9755 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9756 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9757 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9758 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9759 of shift when the shift amount is masked with constant equal to
9760 the size of the mode.
9761 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9764 2017-06-29 Martin Liska <mliska@suse.cz>
9766 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9768 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9771 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9774 (edge_growth_cache_entry::edge_growth_cache_entry): New
9776 (reset_edge_growth_cache): Update to use constructor.
9778 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9780 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9781 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9782 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9784 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9786 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9787 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9789 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9791 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9792 (*-linux-uclibc*): Add t-uclibc tmake_file.
9793 * config/t-musl: New.
9794 * config/t-uclibc: New.
9796 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9798 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9800 (gen_comm_data): Emit architectural setting of arch_prof.
9801 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9803 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9804 (armv8-m.base, armv8-m.main): Likewise.
9805 * arm-protos.h (arm_build_target): Add profile field.
9806 (arch_option): Likewise.
9807 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9809 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9810 arm_active_target.profile.
9812 2017-06-28 Richard Biener <rguenther@suse.de>
9815 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9816 TYPE_OVERFLOW_WRAPS.
9817 * match.pd (negate_expr_p): Likewise.
9818 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9819 fold_build2, not fold_binary.
9821 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9823 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9824 Convert memory address to Pmode.
9825 (aarch64_print_operand): Assert MEM operands are always Pmode.
9827 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9830 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9831 Remove redundant if.
9832 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9833 * config/arm/aarch-common-protos.h
9834 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9835 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9837 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9840 * multiple_target.c (create_dispatcher_calls): Set the default
9841 clone to be static, not public.
9843 2017-06-28 Richard Biener <rguenther@suse.de>
9845 * tree-vect-loop.c (vectorizable_reduction): Move special
9846 cond reduction IV var creation ...
9847 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9848 parameter. Use STMT_VINFO_VECTYPE.
9849 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9852 2017-06-28 Martin Liska <mliska@suse.cz>
9855 * ipa-visibility.c (non_local_p): Handle visibility.
9857 2017-06-28 Martin Liska <mliska@suse.cz>
9860 * common.opt: Add IntegerRange to various options.
9861 * opt-functions.awk (integer_range_info): New function.
9862 * optc-gen.awk: Add integer_range_info to cl_options struct.
9863 * opts-common.c (decode_cmdline_option): Handle
9864 CL_ERR_INT_RANGE_ARG.
9865 (cmdline_handle_error): Likewise.
9866 * opts.c (print_filtered_help): Show valid interval in
9867 when --help is provided.
9868 * opts.h (struct cl_option): Add range_min and range_max fields.
9869 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9871 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9873 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9874 (x * C EQ/NE y * C): New transformation.
9876 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9878 * genmultilib (combination_space): Accept '+' in option names.
9880 2017-06-28 Martin Liska <mliska@suse.cz>
9883 * asan.c (instrument_derefs): Bail out inner references
9884 that are hard register variables.
9886 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9889 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9890 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9892 2017-06-28 Richard Biener <rguenther@suse.de>
9894 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9895 (vect_get_slp_defs): Adjust.
9896 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9898 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9900 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9901 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9902 (vectorizable_reduction): Adjust.
9903 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9905 (vect_get_slp_defs): Likewise.
9906 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9907 (vectorizable_bswap): Adjust.
9908 (vectorizable_call): Likewise.
9909 (vectorizable_conversion): Likewise.
9910 (vectorizable_assignment): Likewise.
9911 (vectorizable_shift): Likewise.
9912 (vectorizable_operation): Likewise.
9913 (vectorizable_store): Likewise.
9914 (vectorizable_condition): Likewise.
9915 (vectorizable_comparison): Likewise.
9917 2017-06-28 Michael Collison <michael.collison@arm.com>
9920 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9922 (arm_gen_movmemqi): Removed unused variable 'i'.
9923 Convert 'for' loop into 'while' loop.
9924 (arm_expand_prologue): Remove last unnecessary set of insn.
9925 (thumb_pop): Remove unused variable 'pushed_words'.
9926 (thumb_exit): Remove last unnecessary set of regs_to_pop.
9928 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9930 * config/s390/predicates.md: Use s390_rel_address_ok_p.
9931 * config/s390/s390-protos.h: Add prototype of
9932 s390_rel_address_ok_p.
9933 * config/s390/s390.c (s390_got_symbol): New function.
9934 (s390_rel_address_ok_p): New function.
9935 (legitimize_pic_address): Use s390_rel_address_ok_p.
9936 (s390_load_got): Use s390_got_symbol.
9937 (s390_option_override): Issue error if
9938 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
9939 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
9941 * config/s390/s390.opt: New option mpic-data-is-text-relative.
9943 2017-06-27 Andrew Pinski <apinski@cavium.com>
9945 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
9946 (X * copysign (1.0, X)): New pattern.
9947 (X * copysign (1.0, -X)): New pattern.
9948 (copysign (-1.0, CST)): New pattern.
9950 2017-06-27 Joseph Myers <joseph@codesourcery.com>
9952 * genmultilib (combination_space): Remove variable.
9953 Validate reuse rules against regular expression for any sequence
9954 of multilib options in any order.
9956 2017-06-27 Michael Collison <michael.collison@arm.com>
9958 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
9959 call aarch64_split_simd_combine.
9960 * (aarch64_combine_internal<mode>): Delete pattern.
9961 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
9962 Allow register and subreg operands.
9964 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9966 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
9967 specific need, just fallback on defaults.
9968 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
9970 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9971 Olivier Hainque <hainque@adacore.com>
9973 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
9975 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
9976 targets. Pick a default if no particular attempt applied.
9977 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
9980 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9982 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
9983 (x86_64-wrs-vxworks7): Likewise.
9985 2017-06-27 Marek Polacek <polacek@redhat.com>
9988 * ubsan.c (instrument_null): Check get_base_address's result for null.
9990 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9992 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
9994 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9996 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
9997 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
9998 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
9999 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10000 New function types.
10001 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10002 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10003 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10004 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10005 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10006 BUILT_IN_FEUPDATEENV): New builtins.
10007 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10008 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10009 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10010 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10012 (builtin_structptr_types): Adjust size.
10013 * tree.c (builtin_structptr_types): Add four entries.
10015 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10016 Olivier Hainque <hainque@adacore.com>
10018 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10019 (TLS_SYM): New local macro, forcing reference to __tls__ on
10020 link command lines for VxWorks 7 RTPs, triggering initialization
10022 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10023 OS features TLS support, true for RTPs on VxWorks 7.
10024 * config/vxworks.c (vxworks_override_options): Setup emutls
10027 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10029 * predict.c (test_prediction_value_range): Use -1U instead of -1
10030 to avoid narrowing conversion warning.
10031 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10032 to avoid narrowing conversion warning.
10033 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10035 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10037 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10039 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10040 64bit configurations.
10041 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10042 (SIZE_TYPE): Likewise.
10043 * config/vxworks.c (vxworks_emutls_var_fields): Use
10044 long_unsigned_type_node instead of unsigned_type_node as the offset
10045 field type, which is "pointer" mode in emutls.c.
10047 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10050 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10052 PR middle-end/81207
10053 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10054 gimple_vuse copying separately from gimple_vdef copying.
10056 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10058 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10061 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10062 Olivier Hainque <hainque@adacore.com>
10064 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10065 for all vxworks7 targets.
10066 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10067 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10068 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10069 variations for VX6/VX7 and 32/64bits later on in ...
10070 (VXWORKS_LIB_SPEC): Leverage new macros.
10071 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10072 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10074 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10076 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10077 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10079 2017-06-26 Carl Love <cel@us.ibm.com>
10081 * config/rs6000/rs6000-c.c: Add support for built-in functions
10082 vector bool char vec_reve (vector bool char);
10083 vector signed char vec_reve (vector signed char);
10084 vector unsigned char vec_reve (vector unsigned char);
10085 vector bool int vec_reve (vector bool int);
10086 vector signed int vec_reve (vector signed int);
10087 vector unsigned int vec_reve (vector unsigned int);
10088 vector bool long long vec_reve (vector bool long long);
10089 vector signed long long vec_reve (vector signed long long);
10090 vector unsigned long long vec_reve (vector unsigned long long);
10091 vector bool short vec_reve (vector bool short);
10092 vector signed short vec_reve (vector signed short);
10093 vector double vec_reve (vector double);
10094 vector float vec_reve (vector float);
10095 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10096 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10097 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10098 (altivec_vreve): New pattern.
10099 * config/rs6000/altivec.h (vec_reve): New define.
10100 * doc/extend.texi (vec_rev): Update the built-in documentation file
10101 for the new built-in functions.
10103 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10105 PR tree-optimization/71815
10106 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10108 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10110 (slsr_process_phi): Likewise.
10111 (replace_uncond_cands_and_profitable_phis): Don't replace a
10112 multiply candidate with a stride of 1 (copy or cast).
10113 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10115 (lowest_cost_path): Likewise.
10116 (total_savings): Likewise.
10118 2017-06-26 Richard Biener <rguenther@suse.de>
10121 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10122 Use def_builtin_pure for all gather builtins.
10124 2017-06-26 Richard Biener <rguenther@suse.de>
10126 PR tree-optimization/81203
10127 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10128 non-dominating BBs.
10130 2017-06-26 Marek Polacek <polacek@redhat.com>
10133 * doc/invoke.texi: Document -Wmultistatement-macros.
10135 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10137 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10138 arm_neon_ok_no_float_abi effective target.
10140 2017-06-26 Richard Biener <rguenther@suse.de>
10142 PR tree-optimization/80928
10143 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10144 (copy_bbs): Set BB_DUPLICATED flag early.
10145 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10147 (execute_on_shrinking_pred): Likewise.
10148 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10149 BB_DUPLICATED blocks.
10150 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10151 iterate over all PHIs considering removal of *gsi.
10153 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10155 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10158 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10160 * config/rs6000/rs6000-string.c: (expand_block_clear,
10161 do_load_for_compare, select_block_compare_mode,
10162 compute_current_alignment, expand_block_compare,
10163 expand_strncmp_align_check, expand_strn_compare,
10164 expand_block_move, rs6000_output_load_multiple)
10165 Move functions related to string/block move/compare
10166 to a separate file.
10167 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10168 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10169 for this function which is now used in two files.
10170 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10171 * config.gcc: Add rs6000-string.o to extra_objs for
10172 targets powerpc*-*-* and rs6000*-*-*.
10174 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10177 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10178 32-bit, since indexed is not valid for DImode.
10179 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10180 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10181 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10182 (define_peephole2 for Altivec d-form store): Likewise.
10185 * multiple_target.c (create_dispatcher_calls): Only create the
10186 dispatcher call if the function is the default clone of a
10187 versioned function.
10189 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10191 PR middle-end/80902
10192 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10193 a call, force the call to not be a tail call.
10195 2017-06-23 Jeff Law <law@redhat.com>
10197 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10200 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10202 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10203 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10204 handling for early expansion of vector shifts (sl,sr,sra,rl).
10205 (builtin_function_type): Add vector shift right instructions
10206 to the unsigned argument list.
10208 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10210 rtl-optimizatoin/79286
10211 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10212 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10213 trap. PIC register plus a const unspec without offset can never trap.
10215 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10217 * tree.h (builtin_structptr_type): New type.
10218 (builtin_structptr_types): Declare new array.
10219 * tree.c (builtin_structptr_types): New array.
10220 (free_lang_data, build_common_tree_nodes): Use it.
10222 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10225 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10228 2017-06-22 Matt Turner <mattst88@gmail.com>
10230 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10231 Lake models to skylake case. Assume skylake for unknown
10232 models with clflushopt.
10234 2017-06-22 Jeff Law <law@redhat.com>
10236 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10237 frame sizes that do not satisfy aarch64_uimm12_shift.
10239 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10241 * profile-count.h (apply_probability,
10242 apply_scale, probability_in): Fix checks for zero.
10244 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10246 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10247 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10249 2016-06-22 Richard Biener <rguenther@suse.de>
10251 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10252 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10253 REDUC_MAX_EXPR support.
10254 (vectorizable_reduction): Likewise.
10255 (vect_create_epilog_for_reduction): Likewise.
10257 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10259 * match.pd (A / (1 << B) -> A >> B): New.
10260 * generic-match-head.c: Include optabs-tree.h.
10261 * gimple-match-head.c: Likewise.
10262 * optabs-tree.h (target_supports_op_p): New.
10263 * optabs-tree.c (target_supports_op_p): New.
10265 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10267 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10268 $gcc_cv_ld --help output.
10269 (gcc_cv_ld_demangle): Likewise.
10270 (gcc_cv_ld_eh_frame_hdr): Likewise.
10271 (gcc_cv_ld_pie): Likewise.
10272 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10273 (gcc_cv_ld_buildid): Likewise.
10274 (gcc_cv_ld_sysroot): Likewise.
10275 (ld_bndplt_support): Likewise.
10276 (ld_pushpopstate_support): Likewise.
10277 * configure: Regenerate.
10278 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10280 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10283 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10284 operands indexes to avoid gap between operands and match_dups.
10286 2017-06-21 Andrew Pinski <apinski@cavium.com>
10288 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10289 Increment Arith_shift and Arith_shift_reg by 1.
10290 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10292 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10293 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10294 (aarch64_strip_extend): Add new argument and test for it.
10295 (aarch64_cheap_mult_shift_p): New function.
10296 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10297 add a cost if it is true.
10298 Update calls to aarch64_strip_extend.
10299 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10301 2017-06-21 Andrew Pinski <apinski@cavium.com>
10303 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10305 (thunderxt88): Likewise.
10306 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10307 (thunderx_prefetch_tune): New variable.
10308 (thunderx2t99_prefetch_tune): Update for the correct values.
10309 (thunderxt88_tunings): New variable.
10310 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10311 generic_prefetch_tune.
10312 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10314 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10316 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10317 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10318 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10319 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10320 (aarch64_atomic_cas<mode>, GPI): Likewise.
10322 2017-06-21 Martin Liska <mliska@suse.cz>
10324 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10325 statements on cold and hot labels.
10326 * predict.c (tree_estimate_probability_bb): Remove the
10327 prediction from this place.
10329 2017-06-21 Martin Liska <mliska@suse.cz>
10331 PR tree-optimization/79489
10332 * gimplify.c (maybe_add_early_return_predict_stmt): New
10334 (gimplify_return_expr): Call the function.
10335 * predict.c (tree_estimate_probability_bb): Remove handling
10337 * predict.def: Update comment about early return predictor.
10338 * gimple-predict.h (is_gimple_predict): New function.
10339 * predict.def: Change default value of early return to 66.
10340 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10342 * passes.def: Put pass_strip_predict_hints to the beginning of
10345 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10347 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10348 FUNCTION_DECL declarations.
10349 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10351 (dwaf2out_decl): Likewise.
10352 * godump.c (go_early_global_decl): Skip call to the real debug hook
10353 for FUNCTION_DECL declarations.
10354 * passes.c (rest_of_decl_compilation): Skip call to the
10355 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10356 -fdump-go-spec is passed.
10358 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10360 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10361 Reorder for compactness.
10362 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10363 (def_builtin_pure, def_builtin_pure2): New functions.
10364 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10366 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10368 * match.pd (nop_convert): New predicate.
10369 ((A +- CST1) +- CST2): Allow some NOP conversions.
10371 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10374 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10375 with ctors/dtors if GOVD_SHARED is set.
10377 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10379 * config/aarch64/aarch64.md (movti_aarch64):
10380 Emit mov rather than orr.
10381 (movtf_aarch64): Likewise.
10382 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10383 Emit mov rather than orr.
10385 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10387 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10388 Swap alternatives, make integer dup more expensive.
10390 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10392 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10393 Return true for non-tls symbols.
10395 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10397 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10398 (cortex-a75): Likewise.
10399 (cortex-a75.cortex-a55): Likewise.
10400 * config/aarch64/aarch64-tune.md: Regenerate.
10401 * doc/invoke.texi (-mtune): Document new values for -mtune.
10403 2017-06-21 Tom de Vries <tom@codesourcery.com>
10405 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10406 stack_size feature.
10407 (Effective-Target Keywords, Other attributes): Suggest using
10408 dg-add-options stack_size feature to get stack limit in stack_size
10409 effective target documentation.
10411 2017-06-21 Julian Brown <julian@codesourcery.com>
10412 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10414 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10415 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10416 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10418 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10419 attribute type list for neon_multiply.
10420 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10421 attribute type list for neon_multiply.
10422 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10423 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10424 attribute type list for neon_multiply.
10425 * config/arm/types.md (crypto_pmull): Add.
10426 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10427 attribute type list.
10429 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10431 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10434 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10436 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10437 to make sure not to dereference a NULL cost_classes_ptr pointer.
10439 2017-06-20 Carl Love <cel@us.ibm.com>
10441 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10442 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10443 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10444 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10445 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10446 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10447 VMULOSW): New enum "unspec" values.
10448 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10449 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10450 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10451 altivec_vmulosw): New patterns.
10452 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10453 VMULOSW): Add definitions.
10455 2017-06-20 Julia Koval <julia.koval@intel.com>
10457 * config/i386/i386.c: Fix rounding expand for new pattern.
10458 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10460 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10462 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10463 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10465 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10467 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10470 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10472 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10473 architecture, then by implementer ID.
10474 * config/aarch64/aarch64-tune.md: Regenerate.
10476 2017-06-20 Richard Biener <rguenther@suse.de>
10478 PR middle-end/81097
10479 * fold-const.c (split_tree): Fold to type before negating.
10481 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10483 * diagnostic-show-locus.c
10484 (selftest::test_fixit_deletion_affecting_newline): New function.
10485 (selftest::diagnostic_show_locus_c_tests): Call it.
10487 2017-06-20 Andreas Schwab <schwab@suse.de>
10490 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10493 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10495 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10496 __ARM_FEATURE_COPROC according to support.
10498 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10500 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10501 Rewritten to avoid overflow for > 32-bit pointers.
10504 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10505 by removing enum keyword.
10506 (ubsan_type_descriptor): Likewise. Formatting fix.
10509 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10510 splitter): Require TARGET_SSE2 in the condition.
10512 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10515 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10516 for doing vector set of SFmode on ISA 3.0.
10517 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10518 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10520 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10521 SFmode value into a V4SF variable that was extracted from another
10522 V4SF variable without converting the element to double precision
10523 and back to single precision vector format.
10524 (vsx_insert_extract_v4sf_p9_2): Likewise.
10526 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10528 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10529 in UWHI to avoid undefined overflow.
10532 * ubsan.h (enum ubsan_encode_value_phase): New.
10533 (ubsan_encode_value): Change second argument to
10534 enum ubsan_encode_value_phase with default value of
10535 UBSAN_ENCODE_VALUE_GENERIC.
10536 * ubsan.c (ubsan_encode_value): Change second argument to
10537 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10538 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10539 create_tmp_var_raw instead of create_tmp_var and use a
10541 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10542 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10543 ubsan_encode_value callers.
10546 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10547 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10548 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10550 2017-06-19 Richard Biener <rguenther@suse.de>
10552 PR middle-end/81118
10553 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10554 estimates if we changed anything.
10556 2017-06-19 Richard Biener <rguenther@suse.de>
10558 PR tree-optimization/80887
10559 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10560 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10561 simplified lookups, then reset mprts_hook.
10562 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10564 (try_to_simplify): Likewise.
10566 2017-06-19 Martin Liska <mliska@suse.cz>
10569 * gimplify.c (gimplify_switch_expr):
10570 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10572 2017-06-19 Martin Liska <mliska@suse.cz>
10574 * doc/install.texi: Document that PGO runs in 4 stages.
10576 2017-06-19 Martin Liska <mliska@suse.cz>
10579 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10580 to dispatcher function name.
10581 * multiple_target.c (replace_function_decl): New function.
10582 (create_dispatcher_calls): Redirect both edges and references.
10584 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10586 * profile-count.c (profile_count::dump): Dump quality.
10587 (profile_count::differs_from_p): Update for unsigned val.
10588 * profile-count.h (profile_count_quality): New enum.
10589 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10591 2017-06-19 Richard Biener <rguenther@suse.de>
10593 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10594 struct function as arg.
10595 (estimate_numbers_of_iterations): Export overload with loop arg.
10596 (free_numbers_of_iterations_estimates_loop): Use an overload of
10597 free_numbers_of_iterations_estimates instead.
10598 * tree-cfg.c (remove_bb): Adjust.
10599 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10600 * tree-parloops.c (gen_parallel_loop): Likewise.
10601 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10603 (tree_unroll_loops_completely): Likewise.
10604 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10605 Use an overload instead and export.
10606 (estimated_loop_iterations): Adjust.
10607 (max_loop_iterations): Likewise.
10608 (likely_max_loop_iterations): Likewise.
10609 (estimate_numbers_of_iterations): Take struct function as arg
10611 (loop_exits_before_overflow): Adjust.
10612 (free_numbers_of_iterations_estimates_loop): Use an overload.
10613 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10614 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10616 2017-06-19 Richard Biener <rguenther@suse.de>
10619 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10620 RANGE_EXPR conservatively.
10622 2017-06-16 Carl Love <cel@us.ibm.com>
10624 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10625 definitions for vec_float, vec_float2, vec_floato,
10626 vec_floate built-ins.
10627 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10628 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10630 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10631 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10632 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10633 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10634 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10635 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10636 vec_floato): Add builtin defines.
10637 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10638 Update the built-in documentation file for the new built-in
10641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10643 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10644 (mthumb): Mark as the negative of -marm.
10646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10648 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10650 (ARM Options, -mtune): Document that this accepts the same
10651 extension options as -mcpu.
10652 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10656 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10657 permitted extensions.
10659 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10661 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10662 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10663 (armv8-m.main): Add option +nodsp.
10664 * config/arm/arm-cpu-cdata.h: Regenerated.
10666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10668 * config/arm/t-fuchsia: New file.
10669 * config.gcc (arm*-*-fuchsia*): Use it.
10671 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10673 * config/arm/t-symbian: Rewrite for new option infrastructure.
10675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10677 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10678 (MULTILIB_REQUIRED): Likewise.
10680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10682 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10683 (MULTILIB_RESUE): Likewise.
10684 (MULTILIB_MATCHES): Likewise.
10685 (MULTLIB_REQUIRED): Likewise.
10687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10689 * config/arm/t-rtems: Rewrite for new option framework.
10691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10693 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10694 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10695 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10696 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10697 * config/arm/t-multilib: ... here.
10698 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10699 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10700 armv7-a and armv8*-a when A-profile libraries have not been built.
10701 * config/arm/t-rmprofile: Rewrite.
10703 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10705 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10706 with a backslash. Remove the backslash after substituting unescaped
10708 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10712 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10713 the driver to pass through the default setting of -mfloat-abi.
10714 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
10716 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
10717 (all_feat_combs): New rule.
10718 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
10720 * config/arm/t-aprofile: Rewrite.
10722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10724 * config/arm/arm.h (FPUTYPE_AUTO): Define.
10725 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
10726 fpu is not specified by the user/command-line.
10727 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
10728 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
10729 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
10730 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
10731 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
10732 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
10734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10736 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
10737 * config/arm/t-arm-elf: Rewritten.
10739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10741 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
10742 have some floating-point instructions.
10743 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
10744 (TARGET_MAYBE_HARD_FLOAT): New macro.
10745 * config/arm/arm-builtins.c (arm_init_builtins): Use
10746 TARGET_MAYBE_HARD_FLOAT.
10747 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10749 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10751 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10752 (configargs.h): Include it.
10753 (arm_print_hint_for_fpu_option): New function.
10754 (arm_parse_fpu_option): New function.
10755 (candidate_extension): New class.
10756 (arm_canon_for_multilib): New function.
10757 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10758 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10759 (ARCH_CANONICAL_SPECS): New macro.
10760 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10764 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10765 are set after handling multilib fragments. Set target_cpu_default2
10768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10770 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10772 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10773 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10775 (check_arch): Likewise.
10776 * config/arm/arm.c (arm_configure_build_target): Handle
10777 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10778 options in the default.
10780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10782 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10783 when an option is an alias of another.
10784 * config/arm/parsecpu.awk (optalias): New parser token.
10785 (gen_comm_data): Mark non-alias options as such. Emit entries
10786 for extension aliases.
10787 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10788 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10789 (armv6kz, armv6zk, armv6t2): Likewise.
10790 (armv7): Make vfpv3-d16 an alias.
10791 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10793 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10794 Sort in canonical order.
10795 (armv8-a): Sort in canonical order.
10796 (armv8.1-a, armv8.2-a): Likewise.
10797 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10799 (cortex-a9): Sort in canonical order.
10800 * config/arm/arm.c (selftests.h): Include it.
10801 (arm_test_cpu_arch_data): New function.
10802 (arm_run_self_tests): New function.
10803 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10804 (targetm): Move declaration to the end of the file.
10805 * arm-cpu-cdata.h: Regenerated.
10807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10809 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10810 call to target_mode_check describing the type of option passed.
10811 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10812 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10813 arm_parse_cpu_option_name to match parameters against list of
10815 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10816 arm_arch_core_flags data structure.
10817 * config/arm/arm-cpu_cdata.h: Regenerated.
10819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10821 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10823 (arm_print_hint_for_cpu_option): Likewise.
10824 (arm_print_hint_for_arch_option): Likewise.
10825 (arm_parse_cpu_option_name): Likewise.
10826 (arm_parse_arch_option_name): Likewise.
10827 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10828 of entries in the all_fpus list.
10829 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10830 (arm_parse_cpu_option_name): Declare.
10831 (arm_parse_arch_option_name): Declare.
10832 (arm_parse_option_features): Declare.
10833 (arm_intialize_isa): Declare.
10834 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10836 (gen_comm_data): ... here. Make definitions non-static.
10837 * config/arm/arm-cpu-data.h: Regenerated.
10838 * config/arm/arm-cpu-cdata.h: Regenerated.
10840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10842 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10843 (cpu_arch_extension): New structure.
10844 (cpu_arch_option, arch_option, cpu_option): New structures.
10845 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10846 architecture types.
10847 (gen_data): Generate new format data tables.
10848 * config/arm/arm.c (cpu_tune): New structure.
10849 (cpu_option, processors): Delete.
10850 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10851 (arm_print_hint_for_cpu_option): ... this and ...
10852 (arm_print_hint_for_arch_option): ... this.
10853 (arm_parse_arch_cpu_name): Delete. Replace with ...
10854 (arm_parse_cpu_option_name): ... this and ...
10855 (arm_parse_arch_option_name): ... this.
10856 (arm_unrecognized_feature): Change type of target parameter to
10858 (arm_parse_arch_cpu_features): Delete. Replace with ...
10859 (arm_parse_option_features): ... this.
10860 (arm_configure_build_target): Rework to use new configuration data
10862 (arm_print_tune_info): Rework for new configuration data tables.
10863 * config/arm/arm-cpu-data.h: Regenerated.
10864 * config/arm/arm-cpu.h: Regenerated.
10866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10868 * Makefile.in (OBJS): Move sbitmap.o from here ...
10869 (OBJS-libcommon): ... to here.
10871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10873 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10874 (ISA_ALL_CRYPTO): New macro.
10875 (ISA_ALL_SIMD): New macro
10876 (ISA_ALL_FP): New macro.
10877 * config/arm/arm.c (fpu_bitlist): Update initializer.
10878 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10880 (arm9e): Add fpu. Add option for nofp
10881 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10882 (arm926ej-s, arm1026ej-s): Likewise.
10883 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10884 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10885 neon-fp16, neon-vfpv4, nofp and nosimd.
10886 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10887 (cortex-a8): Add fpu. Add option for nofp.
10888 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10889 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10890 (cortex-r4f): Add fpu.
10891 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10892 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10894 (cortex-r8): Likewise.
10895 (cortex-m4): Add fpu. Add option for nofp.
10896 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10897 (cortex-a17.cortex-a7): Likewise.
10898 (cortex-a32): Add fpu. Add options for crypto and nofp.
10899 (cortex-a35, cortex-a53): Likewise.
10900 (cortex-a57): Add fpu. Add option for crypto.
10901 (cortex-a72, cortex-a73): Likewise.
10902 (exynos-m1): Likewise.
10903 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10904 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10905 (cortex-m33): Add fpu. Add option for nofp.
10906 * config/arm/arm-cpu-cdata.h: Regenerated
10907 * config/arm/arm-cpu-data.h: Regenerated.
10909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10911 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10912 (armv5te, armv5tej): Likewise.
10913 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10914 (armv7): Add options fp and vfpv3-d16.
10915 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10916 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10918 (armv7ve): Likewise.
10919 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10920 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10921 (armv8-a): Add nocrypto option.
10922 (armv8.1-a, armv8.2-a): Likewise.
10923 (armv8-m.main): add options fp, fp.dp and nofp.
10925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10927 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
10929 (armv8-a+crc): Delete.
10930 (armv8.1-a): Add options simd, crypto and nofp.
10931 (armv8.2-a): Add options fp16, simd, crypto and nofp.
10932 (armv8.2-a+fp16): Delete.
10933 (armv8-m.main): Add option dsp.
10934 (armv8-m.main+dsp): Delete.
10935 (cortex-a8): Add fpu. Add nofp option.
10936 (cortex-a9): Add fpu. Add nofp and nosimd options.
10937 * config/arm/parsecpu.awk (gen_data): Generate option tables and
10938 link to main cpu and architecture data structures.
10939 (gen_comm_data): Only put isa attributes from the main architecture
10941 (option): New statement for architecture and CPU entries.
10942 * arm.c (struct cpu_option): New structure.
10943 (struct processors): Add entry for options.
10944 (arm_unrecognized_feature): New function.
10945 (arm_parse_arch_cpu_name): Ignore any characters after the first
10947 (arm_parse_arch_cpu_feature): New function.
10948 (arm_configure_build_target): Separate out any CPU and architecture
10949 features and parse separately. Don't error out if -mfpu=auto is
10950 used with only an architecture string.
10951 (arm_print_asm_arch_directives): New function.
10952 (arm_file_start): Call it.
10953 * config/arm/arm-cpu-cdata.h: Regenerated.
10954 * config/arm/arm-cpu-data.h: Likewise.
10955 * config/arm/arm-tables.opt: Likewise.
10957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10959 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
10960 assembler when it is not -mfpu=auto.
10962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10964 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
10965 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
10966 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
10967 (ASM_CPU_SPEC): Rewrite.
10968 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
10969 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
10970 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
10971 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
10972 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
10973 copied string is NUL-terminated. Also strip any characters prefixed
10975 (arm_rewrite_selected_arch): New function.
10976 (arm_rewrite_march): New function.
10978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10980 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
10981 (x_arm_cpu_string, x_arm_tune_string): Likewise.
10982 (march, mcpu, mtune): Convert to string-based options.
10983 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
10984 (arm_parse_arch_cpu_name): New function.
10985 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
10986 identify selected architecture or CPU.
10987 (arm_option_save): New function.
10988 (TARGET_OPTION_SAVE): Redefine.
10989 (arm_option_restore): Restore string options.
10990 (arm_option_print): Print string options.
10992 2017-06-16 Martin Sebor <msebor@redhat.com>
10994 PR tree-optimization/80933
10995 PR tree-optimization/80934
10996 * builtins.c (fold_builtin_3): Do not handle bcmp here.
10997 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
10998 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
10999 (gimple_fold_builtin): Call them.
11001 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11003 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11004 as unlikely; update profile.
11006 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11008 * predict.c (force_edge_cold): Handle declaring edges impossible
11011 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11013 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11015 (try_unroll_loop_completely): Fix reporting.
11017 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11019 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11021 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11024 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11025 if given a non-constant argument for an intrinsic which requires a
11028 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11030 * profile.c (compare_freqs): New function.
11031 (branch_prob): Sort edge list.
11032 (find_spanning_tree): Assume that the list is priority sorted.
11034 2017-06-16 Richard Biener <rguenther@suse.de>
11036 PR tree-optimization/81090
11037 * passes.def (pass_record_bounds): Remove.
11038 * tree-pass.h (make_pass_record_bounds): Likewise.
11039 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11040 make_pass_record_bounds): Likewise.
11041 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11042 not free niter estimates at the beginning but at the end.
11043 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11045 2017-06-16 Richard Biener <rguenther@suse.de>
11047 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11048 initializer to workaround ICE in host GCC 4.8.
11050 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11052 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11054 (clone_inlined_nodes): Update.
11056 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11058 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11059 prefetch settings, and enable prefetching by default at -O3.
11061 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11063 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11064 Set flag_prefetch_loop_arrays according to tuning data.
11066 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11068 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11069 New tune structure.
11070 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11071 [Unrelated to main purpose of the patch] Place the pointer field last
11072 to enable type checking errors when tune structure are wrongly merged.
11073 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11074 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11075 (thunderx2t99_prefetch_tune): New tune constants.
11076 (tune_params *_tunings): Update all tunings (no functional change).
11077 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11078 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11079 from tunings structures.
11081 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11084 * ubsan.c (instrument_null): Add T argument, use it instead
11085 of computing it based on IS_LHS.
11086 (instrument_object_size): Likewise.
11087 (pass_ubsan::execute): Adjust instrument_null and
11088 instrument_object_size callers to pass gimple_get_lhs or
11089 gimple_assign_rhs1 result to it. Use instrument_null instead of
11090 calling get_base_address and instrument_mem_ref. Handle
11091 aggregate call arguments for object-size sanitization.
11093 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11095 PR tree-optimization/81089
11096 * tree-vrp.c (is_masked_range_test): Validate operands of
11099 2017-06-15 Martin Sebor <msebor@redhat.com>
11102 * dumpfile.c (dump_register): Avoid calling memset to initialize
11103 a class with a default ctor.
11104 * gcc.c (struct compiler): Remove const qualification.
11105 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11106 * hash-table.h: Ditto.
11107 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11109 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11110 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11112 * params.h (struct param_info): Make struct members non-const.
11113 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11114 with default initialization.
11115 * vec.h (vec_copy_construct, vec_default_construct): New helper
11117 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11118 with vec_copy_construct.
11119 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11120 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11121 * doc/invoke.texi (-Wclass-memaccess): Document.
11123 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11125 * emit-rtl.h (is_leaf): Update comment about local
11126 register allocator.
11128 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11131 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11132 for a variable to have a section before checking if the section has a
11134 Set section to.persistent if persistent attribute is set.
11135 Warn if .persistent attribute is used on an automatic variable.
11137 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11139 PR rtl-optimization/80474
11140 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11142 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11144 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11147 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11149 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11150 (casesi): Emit bounds checking as RTL.
11151 (casesi_internal_mips16_<mode>): Remove bounds checking.
11153 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11155 * config/xtensa/xtensa.c (xtensa_option_override): Append
11156 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11157 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11158 xtensa_doloop_hooks): Define unconditionally.
11159 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11161 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11162 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11163 for it in xtensa_option_override.
11164 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11165 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11167 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11169 * doc/cppopts.texi: Document '-' special value to -MF.
11171 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11173 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11174 (cortex_a53_fconst): Likewise.
11175 (cortex_a53_fpmul): Likewise.
11176 (cortex_a53_f_load_64): Likewise.
11177 (cortex_a53_f_load_many): Likewise.
11178 (cortex_a53_advsimd_alu): Likewise.
11179 (cortex_a53_advsimd_alu_q): Likewise.
11180 (cortex_a53_advsimd_mul): Likewise.
11181 (cortex_a53_advsimd_mul_q): Likewise.
11182 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11183 Add missing fmul, r2f_cvt and fconst cases.
11185 2017-06-14 Richard Biener <rguenther@suse.de>
11187 PR middle-end/81088
11188 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11190 (fold_binary_loc): When associating do not treat pre-existing
11191 TREE_OVERFLOW on literal constants as a reason to allow
11192 TREE_OVERFLOW on associated literal constants.
11194 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11196 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11197 (MASK_FEATURES): New macro.
11198 * config/sparc/sparc.c (sparc_option_override): Remove the special
11199 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11201 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11203 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11204 a division of 0 if non-call exceptions are enabled.
11206 2017-06-14 Andrew Pinski <apinski@cavium.com>
11207 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11210 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11211 Improve vector initialization code gen for only variable case.
11213 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11215 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11217 2017-06-14 Richard Biener <rguenther@suse.de>
11219 PR tree-optimization/81083
11220 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11223 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11225 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11226 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11227 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11228 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11229 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11230 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11232 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11234 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11235 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11237 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11239 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11241 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11243 * config/rs6000/t-rtems: Don't handle SPE.
11245 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11247 * config/rs6000/t-linux: Don't handle SPE.
11249 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11251 * config/rs6000/eabispe.h: Delete file.
11253 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11255 * config/rs6000/t-spe: Delete file.
11257 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11259 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11260 (rs6000_legitimate_offset_address_p): Return false for anything in
11261 V2SImode or V2SFmode.
11263 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11265 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11266 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11267 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11269 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11270 (rs6000_legitimate_offset_address_p): Ditto.
11271 (rs6000_emit_move): Ditto.
11272 (rs6000_init_builtins): Remove V4HI_type_node.
11274 2017-06-13 Martin Liska <mliska@suse.cz>
11277 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11278 (gate_asan): Likewise.
11279 * asan.h (asan_no_sanitize_address_p): Remove the function.
11280 (sanitize_flags_p): New function.
11281 * builtins.def: Fix coding style.
11282 * common.opt: Use renamed enum value.
11283 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11284 * doc/extend.texi: Document no_sanitize attribute.
11285 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11286 to SANITIZE_UNDEFINED_NONDEFAULT.
11287 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11288 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11289 Use sanitize_flags_p.
11290 * gimplify.c (gimplify_function_tree): Likewise.
11291 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11292 * opts.c (parse_no_sanitize_attribute): New function.
11293 (common_handle_option): Use renamed enum value.
11294 * opts.h (parse_no_sanitize_attribute): Declare.
11295 * tree.c (sanitize_flags_p): New function.
11296 * tree.h: Declared here.
11297 * tsan.c: Use sanitize_flags_p.
11298 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11299 (instrument_mem_ref): Likewise.
11300 (instrument_bool_enum_load): Likewise.
11301 (do_ubsan_in_current_function): Remove the function.
11302 (pass_ubsan::execute): Use sanitize_flags_p.
11303 * ubsan.h: Remove do_ubsan_in_current_function
11304 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11305 (dump_function_to_file): Use it here.
11307 2017-06-13 Martin Jambor <mjambor@suse.cz>
11309 PR tree-optimization/80803
11310 PR tree-optimization/81063
11311 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11312 (propagate_subaccesses_across_link): Enqueue subtree whenever
11313 necessary instead of relying on the caller.
11315 2017-06-13 Martin Jambor <mjambor@suse.cz>
11317 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11318 that have a first_link.
11319 (sort_and_splice_var_accesses): Do not check first_link before
11321 (subtree_mark_written_and_enqueue): Likewise.
11322 (propagate_all_subaccesses): Likewise and do not stop at first
11323 parent with a first_link.
11325 2017-06-13 Martin Jambor <mjambor@suse.cz>
11327 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11330 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11332 * match.pd: New pattern.
11334 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11336 * tree-vrp.c (is_masked_range_test): New function.
11337 (register_edge_assert_for): Determine ranges for
11340 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11342 PR tree-optimization/67328
11343 * fold-const.c (maskable_range_p): New function.
11344 (build_range_check): Generate bittests if possible.
11346 2017-06-13 Martin Liska <mliska@suse.cz>
11348 * gimple-pretty-print.c (dump_probability): Add new argument.
11349 (dump_edge_probability): Dump both probability and count.
11350 (dump_gimple_label): Likewise.
11351 (dump_gimple_bb_header): Likewise.
11353 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11356 * config/avr/avr-devices.c: Fix indentation.
11357 * config/avr/gen-avr-mmcu-specs.c: Dito.
11359 2017-06-13 Richard Biener <rguenther@suse.de>
11361 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11362 instead get vector type from stmt_info.
11363 (vectorizable_reduction): Adjust. Remove dead code.
11365 2017-06-13 Richard Biener <rguenther@suse.de>
11367 PR middle-end/81065
11368 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11369 case of C * (x * C2 + C3).
11370 (fold_addr_of_array_ref_difference): Properly fold index difference.
11372 2017-06-12 David S. Miller <davem@davemloft.net>
11375 * config/sparc/sparc.md (return expander): Emit frame blockage if
11376 function uses alloca.
11378 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11380 * combine.c (make_field_assignment): Check len rather than the mode
11381 precision when calling force_to_mode.
11383 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11385 Support multilibs and devices that see flash in RAM address range.
11388 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11389 (avr_mcu_t) <flash_pm_offset>: New field.
11390 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11391 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11392 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11393 (AVR_TINY_PM_OFFSET): Remove macro.
11394 * config/avr/avr.opt (-mshort-calls): New option.
11395 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11396 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11397 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11398 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11399 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11400 instead of avr_arch->have_jmp_call.
11401 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11402 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11403 avr_arch->flash_pm_offset to define.
11404 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11405 new field flash_pm_offset. Add entry for avrxmega3.
11406 (avr_texinfo): Add entry for avrxmega3.
11407 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11408 attiny212, attiny214,
11409 attiny412, attiny414, attiny416, attiny417,
11410 attiny814, attiny816, attiny817,
11411 attiny1614, attiny1616, attiny1617,
11412 attiny3214, attiny3216, attiny3217.
11413 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11414 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11415 (avr_print_operand_address) [AVR_TINY]: Same.
11416 (avr_asm_init_sections) <readonly_data_section>: Only patch
11417 callback if avr_arch->flash_pm_offset = 0.
11418 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11419 for rodata if avr_arch->flash_pm_offset != 0.
11420 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11421 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11422 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11423 (m_options): Append opt_rcall.
11424 (m_dirnames): Append dir_rcall.
11425 * config/avr/t-multilib: Regenerate.
11427 * configure.ac [target=avr]: Check whether avrxmega3 default
11428 linker description file works as needed.
11429 * configure: Regenerate.
11430 * doc/avr-mmcu.texi: Regenerate.
11431 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11432 <__AVR_ARCH__>: Document avrxmega3 and 103.
11433 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11434 <__AVR_SHORT_CALLS__>: Document it.
11435 <__AVR_PM_BASE_ADDRESS__>: Document it.
11436 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11437 (AVR Variable Attributes) <progmem>: Document this is
11438 not needed for avrxmega3.
11439 (AVR Named Address Spaces) <__flash>: Dito.
11441 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11443 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11445 2017-06-12 Doug Rupp <rupp@adacore.com>
11447 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11448 Append vxworks-stdint.h to the tm_file list.
11449 * config/vxworks-stdint.h: New file.
11451 2017-06-12 Martin Liska <mliska@suse.cz>
11453 PR tree-optimization/81041
11454 * tree-profile.c (gimple_gen_ic_func_profiler):
11455 Create an extra BB in profile-generate
11456 (gimple_gen_time_profiler): Likewise.
11458 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11460 PR tree-optimization/81003
11461 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11462 (update_range_test): Use it instead of force_gimple_operand_gsi.
11464 2017-06-12 Richard Biener <rguenther@suse.de>
11466 PR tree-optimization/81053
11467 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11468 with backedge value not defined in loop. Simplify def stmt
11471 2017-06-11 Tom de Vries <tom@codesourcery.com>
11474 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11476 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11477 nvptx_cannot_force_const_mem.
11479 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11481 * opts.c (finish_options): Move test for flag_split_stack after
11482 it has been initialized.
11484 2017-06-11 Jason Merrill <jason@redhat.com>
11486 * tree.h (id_equal): New.
11487 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11488 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11489 instead of strcmp of IDENTIFIER_POINTER.
11491 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11493 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11494 (mark_all_inlined_calls_cdtor): Fix formating.
11495 (inline_transform): Rescale profile before inlining.
11497 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11499 * cgraph.h (cgraph_edge::clone): Update prototype.
11500 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11501 (cgraph_node::create_clone): Update.
11502 (cgraph_node::create_version_clone): Update.
11503 * tree-inline.c (copy_bb): Update.
11504 (expand_call_inline): Update.
11506 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11508 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11509 factored out from ...
11510 (rs6000_emit_prologue): ... here.
11512 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11514 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11515 factored out from ...
11516 (rs6000_emit_prologue): ... here.
11518 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11520 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11522 (handle_missing_profiles): Fix computation of tp_first_run.
11523 (counts_to_freqs): Do not touch freqs when count is 0.
11525 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11527 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11530 2017-06-10 Tom de Vries <tom@codesourcery.com>
11532 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11533 attributes): Document signal effective target.
11535 2017-06-10 Tom de Vries <tom@codesourcery.com>
11537 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11538 Document effective target stack_size.
11540 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11542 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11543 to the edit_context if they can be auto-applied.
11545 2017-06-9 Ian Lance Taylor <iant@golang.org>
11547 * opts.c (finish_options): If -fsplit-stack, disable implicit
11548 -forder-blocks-and-partition.
11549 * doc/invoke.texi (Optimize Options): Document that when using
11550 -fsplit-stack -forder-blocks-and-partition is not implicitly
11553 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11555 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11556 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11557 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11558 * builtins.def (abort, trap, unreachable): Declare cold.
11559 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11560 * tree-core.h (ECF_COLD): New.
11561 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11562 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11564 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11566 * predict.c (unlikely_executed_stmt_p): Cleanup.
11568 2017-06-09 Richard Biener <rguenther@suse.de>
11570 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11571 model if the ref is always written to.
11573 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11575 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11577 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11579 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11582 2017-06-09 Tom de Vries <tom@codesourcery.com>
11585 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11586 "target cannot support label values" when encountering LABEL_REF.
11588 2017-06-09 Martin Liska <mliska@suse.cz>
11590 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11591 (gimple_gen_ic_func_profiler): Emit direct comparison
11592 of __gcov_indirect_call_callee with NULL.
11593 (gimple_gen_time_profiler): Change probability from
11594 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11596 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11598 * profile.c (edge_gcov_counts): Turn to pointer.
11599 (compute_branch_probabilities, compute_branch_probabilities): Update.
11600 (branch_prob): Do not clear edge_gcov_count.
11601 * profile.h (edge_gcov_counts): Turn to pointer.
11602 (edge_gcov_count): Update.
11604 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11606 * gimple.h (gimple_check_failed): Mark cold.
11608 2017-06-09 Richard Biener <rguenther@suse.de>
11610 PR tree-optimization/66623
11611 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11612 refactor check_reduction into two parts, properly computing
11613 whether we have to check reduction validity for outer loop
11616 2017-06-09 Richard Biener <rguenther@suse.de>
11618 PR tree-optimization/79483
11619 * graphite-scop-detection.c (order): New global.
11620 (get_order): Compute bb to order mapping that satisfies code
11621 generation constraints.
11622 (cmp_pbbs): New helper.
11623 (build_scops): Start domwalk at entry block, sort generated
11626 2017-06-09 Richard Biener <rguenther@suse.de>
11628 PR middle-end/81007
11629 * ipa-polymorphic-call.c
11630 (ipa_polymorphic_call_context::restrict_to_inner_class):
11631 Skip FIELD_DECLs with error_mark_node type.
11632 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11635 2017-06-09 Martin Liska <mliska@suse.cz>
11637 * predict.c (struct branch_predictor): New struct.
11638 (test_prediction_value_range): New test.
11639 (predict_c_tests): New function.
11640 * selftest-run-tests.c (selftest::run_tests): Run the function.
11641 * selftest.h: Declare new tests.
11643 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11646 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11647 gen_add3_insn did not fail.
11648 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11649 r0, construct that number in a temporary reg and add that reg to r0.
11650 If asked to put the result in r0 as well, fail.
11652 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11654 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11655 for early expansion of vec_eqv.
11657 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11659 PR middle-end/81005
11660 * ubsan.c (instrument_null): Avoid pointless code temporary.
11661 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11663 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11667 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11670 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11673 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11675 * predict.c (unlikely_executed_edge_p): Move ahead.
11676 (probably_never_executed_edge_p): Use it.
11678 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11680 PR middle-end/79988
11681 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11682 gimple_call_builtin_p call.
11684 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11686 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11687 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11688 rtl_check_failed_type2, rtl_check_failed_code1,
11689 rtl_check_failed_code2, rtl_check_failed_code_mode,
11690 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11691 rtvec_check_failed_bounds, rtl_check_failed_flag,
11692 _fatal_insn_not_found, _fatal_insn): Likewise.
11693 * tree.h (tree_contains_struct_check_failed,
11694 tree_check_failed, tree_not_check_failed,
11695 tree_class_check_failed, tree_range_check_failed,
11696 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11697 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11698 tree_operand_check_failed, omp_clause_check_failed,
11699 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11702 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11704 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11705 flag_branch_probabilities.
11706 * ipa-inline.c (edge_badness): Likewise.
11707 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11708 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11709 * predict.c (maybe_hot_frequency_p): Likewise.
11710 (probably_never_executed): Likewise.
11711 * sched-ebb.c (schedule_ebbs): Likewise.
11712 * sched-rgn.c (find_single_block_region): Likewise.
11713 * tracer.c (tail_duplicate): Likewise.
11715 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11717 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
11718 longer requires x_flag_profile_use.
11720 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11722 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
11723 instead of flag_reorder_blocks_and_partition.
11724 * dbxout.c (dbxout_function_end): Likewise.
11725 * dwarf2out.c (gen_subprogram_die): Likewise.
11726 * haifa-sched.c (sched_create_recovery_edges): Likewise.
11727 * hw-doloop.c (reorg_loops): Likewise.
11728 * varasm.c (assemble_start_function,
11729 assemble_end_function): Likewise.
11730 (decide_function_section): Do not check for
11731 flag_reorder_blocks_and_partition.
11733 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11735 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11737 (chkp_get_hard_register_fake_addr_expr): Ditto.
11738 (chkp_build_addr_expr): Add check for hard reg case.
11739 (chkp_parse_array_and_component_ref): Ditto.
11740 (chkp_find_bounds_1): Ditto.
11741 (chkp_process_stmt): Don't generate bounds store for
11744 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11746 * predict.c (maybe_hot_bb_p): Do not check profile status.
11747 (maybe_hot_edge_p): Likewise.
11748 (probably_never_executed): Check for zero counts even if profile
11750 (unlikely_executed_edge_p): New function.
11751 (unlikely_executed_stmt_p): New function.
11752 (unlikely_executed_bb_p): New function.
11753 (set_even_probabilities): Use unlikely predicates.
11754 (combine_predictions_for_bb): Likewise.
11755 (predict_paths_for_bb): Likewise.
11756 (predict_paths_leading_to_edge): Likewise.
11757 (determine_unlikely_bbs): New function.
11758 (estimate_bb_frequencies): Use it.
11759 (compute_function_frequency): Use zero counts even if profile is
11761 * profile-count.h: Fix typo.
11763 2017-08-08 Julia Koval <julia.koval@intel.com>
11765 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11766 _mm512_mask_cvtsepi16_storeu_epi8,
11767 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11768 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11769 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11770 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11771 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11772 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11773 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11774 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11775 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11776 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11777 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11778 __builtin_ia32_pmovuswb256mem_mask,
11779 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11780 __builtin_ia32_pmovwb512mem_mask): New builtins.
11782 2017-08-08 Julia Koval <julia.koval@intel.com>
11784 PR target/73350,80862
11785 * config/i386/subst.md (round): Fix round pattern.
11786 * config/i386/i386.c (ix86_erase_embedded_rounding):
11787 Fix erasing rounding for the fixed pattern.
11789 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11791 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11793 2017-06-08 Martin Liska <mliska@suse.cz>
11795 PR gcov-profile/80911
11796 * gcov.c (block_info::block_info): New constructor.
11798 2017-06-07 Carl Love <cel@us.ibm.com>
11800 * config/rs6000/rs6000-c: The return type of the following
11801 built-in functions was implemented as int not long long. Fix sign
11802 of return value for the unsigned version of vec_mulo and vec_mule.
11803 vector unsigned long long vec_bperm (vector unsigned long long,
11804 vector unsigned char)
11805 vector signed long long vec_mule (vector signed int,
11807 vector unsigned long long vec_mule (vector unsigned int,
11808 vector unsigned int)
11809 vector signed long long vec_mulo (vector signed int,
11811 vector unsigned long long vec_mulo (vector unsigned int,
11812 vector unsigned int)
11813 * doc/extend.texi: Fix the documentation for the built-in
11816 2017-06-07 Carl Love <cel@us.ibm.com>
11819 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11822 2017-06-07 Carl Love <cel@us.ibm.com>
11824 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11825 support, Generate doublehv for signed int/float for BE case only.
11827 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11829 * doc/invoke.texi (mcx16): Rewrite.
11831 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11833 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11834 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11835 *mov<mode>_softfloat, and an anonymous splitter): Use
11836 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11838 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11840 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11842 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11843 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11844 (rs6000_debug_reg_global): Adjust.
11845 (rs6000_init_hard_regno_mode_ok): Adjust.
11846 (rs6000_dbx_register_number): Adjust.
11847 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11848 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11849 Remove SPE_ACC and SPEFSCR.
11850 (REG_ALLOC_ORDER): Ditto.
11851 (FRAME_POINTER_REGNUM): Change to 111.
11852 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11853 (REG_CLASS_NAMES): Ditto.
11854 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11855 (REGISTER_NAMES): Ditto.
11856 (ADDITIONAL_REG_NAMES): Ditto.
11857 (rs6000_reg_names): Ditto.
11858 * config/rs6000/rs6000.md: Renumber some register number
11861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11863 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11865 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11866 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11868 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11869 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11870 Delete the SPE high registers.
11871 (REG_ALLOC_ORDER): Ditto.
11872 (enum reg_class): Remove SPE_HIGH_REGS.
11873 (REG_CLASS_NAMES): Ditto.
11874 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11875 (REGISTER_NAMES): Ditto.
11876 (rs6000_reg_names): Ditto.
11877 * doc/tm.texi.in: Remove SPE as example.
11878 * doc/tm.texi: Regenerate.
11880 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11882 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11883 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11884 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11885 * config/rs6000/rs6000.md (type): Remove "brinc".
11887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11889 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11890 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11891 * config/rs6000/linuxspe.h: Delete file.
11892 * config/rs6000/rs6000.md: Don't include spe.md.
11893 * config/rs6000/spe.h: Delete file.
11894 * config/rs6000/spe.md: Delete file.
11895 * config/rs6000/t-rs6000: Remove spe.md.
11897 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11899 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11900 (reg_or_none500mem_operand): Delete.
11901 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11902 instead of reg_or_none500mem_operand.
11904 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11906 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11907 handling of SPE flags.
11908 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11910 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11912 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11914 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11915 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11916 paired_divv2sf3): Similar.
11917 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11918 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11919 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11921 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11922 Rename the paired_* instruction patterns.
11923 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
11925 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
11926 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
11927 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
11928 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
11929 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
11930 PAIRED_VECTOR_MODE.
11931 (struct machine_function): Delete field spe_insn_chain_scanned_p.
11932 (spe_func_has_64bit_regs_p): Delete.
11933 (spe_expand_predicate_builtin): Delete.
11934 (spe_expand_evsel_builtin): Delete.
11935 (TARGET_DWARF_REGISTER_SPAN): Do not define.
11936 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
11937 (invalid_e500_subreg): Delete.
11938 (rs6000_legitimize_address): Always force_reg op2 as well, for
11939 paired single memory accesses.
11940 (rs6000_member_type_forces_blk): Delete.
11941 (rs6000_spe_function_arg): Delete.
11942 (rs6000_expand_unop_builtin): Delete SPE handling.
11943 (rs6000_expand_binop_builtin): Ditto.
11944 (spe_expand_stv_builtin): Delete.
11945 (bdesc_2arg_spe): Delete.
11946 (spe_expand_builtin): Delete.
11947 (spe_expand_predicate_builtin): Delete.
11948 (spe_expand_evsel_builtin): Delete.
11949 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
11950 (spe_init_builtins): Delete.
11951 (spe_func_has_64bit_regs_p): Delete.
11952 (savres_routine_name): Delete "info" parameter. Adjust callers.
11953 (rs6000_emit_stack_reset): Ditto.
11954 (rs6000_dwarf_register_span): Delete.
11955 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
11956 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
11957 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
11959 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
11961 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
11962 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
11963 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
11964 mulv2sf3, divv2sf3): Delete expanders.
11966 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11968 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
11970 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11972 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
11973 * config/rs6000/rs6000.c: Ditto.
11975 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11977 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
11978 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
11979 comparison_operator.
11981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11983 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
11984 * config/rs6000/rs6000.opt: Ditto.
11985 * config/rs6000/t-rtems: Ditto.
11987 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11989 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
11990 TARGET_E500_SINGLE by 0, simplify.
11991 * config/rs6000/rs6000.c: Ditto.
11992 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
11993 (spe_build_register_parallel): Delete.
11994 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
11995 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
11996 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
11997 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
11998 (E500_CONVERT): Delete.
11999 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12001 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12003 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12004 * config/rs6000/dfp.md: Ditto.
12005 (negdd2, *negdd2_fpr): Merge.
12006 (absdd2, *absdd2_fpr): Merge.
12007 (negtd2, *negtd2_fpr): Merge.
12008 (abstd2, *abstd2_fpr): Merge.
12009 * config/rs6000/e500.h: Delete file.
12010 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12011 TARGET_FPRS by 1 and simplify.
12012 * config/rs6000/rs6000-c.c: Ditto.
12013 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12014 TARGET_DF_SPE by 0.
12015 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12017 * config/rs6000/rs6000.md: Ditto.
12018 (floatdidf2, *floatdidf2_fpr): Merge.
12019 (move_from_CR_gt_bit): Delete.
12020 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12021 (E500_CR_IOR_COMPARE): Delete.
12022 (All patterns that require !TARGET_FPRS): Delete.
12023 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12025 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12027 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12029 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12031 * graphds.c (add_edge): Intitialize edge's attached data.
12032 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12033 pointer parameter. Call pointed function on each edge during
12034 graph traversing. Skip traversing the edge when the function
12036 (graphds_dfs, graphds_scc): Ditto.
12037 (for_each_edge): New parameter. Pass the new parameter to callback
12039 * graphds.h (skip_edge_callback): New function pointer type.
12040 (graphds_dfs, graphds_scc): New function pointer parameter.
12041 (graphds_edge_callback, for_each_edge): New parameter.
12043 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12045 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12046 out code checking if runtime alias check is possible to below ...
12047 Call the new function.
12048 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12049 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12051 2017-06-07 Marek Polacek <polacek@redhat.com>
12054 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12055 TYPE_OVERFLOW_WRAPS check.
12057 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12059 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12060 if versioning is required.
12061 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12062 peeling with the check for versioning.
12064 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12066 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12067 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12068 Set true to new parameter if new ssa variable is defined.
12069 (vect_gen_vector_loop_niters): Refactor. Set range information
12070 for the new vector loop bound variable.
12071 (vect_do_peeling): Ditto.
12073 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12075 * tree-affine.c (ssa.h): Include header file.
12076 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12077 has wrapping overflow behavior.
12079 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12081 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12083 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12085 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12086 (tree_to_aff_combination): ... here.
12088 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12090 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12091 reg_pressure model function.
12092 (ivopts_global_cost_for_size): Delete.
12093 (determine_set_costs, iv_ca_recount_cost): Call new model function
12094 ivopts_estimate_reg_pressure.
12096 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12098 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12099 expensive than udiv. Remove floating point cases from mod.
12101 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12103 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12104 Increase idiv cost.
12106 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12108 * config/aarch64/aarch64.md
12109 (copysignsf3): Fix mask generation.
12111 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12113 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12115 (class dump_manager): Add register_dumps method.
12116 * dumpfile.c: Include langhooks.h.
12117 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12118 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12119 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12120 (dump_manager::dump_register): Start with 512 entries instead of 32.
12121 (dump_manager::register_dumps): New method.
12122 * toplev.c (general_init): Instead of invoking register_dumps
12123 langhook, invoke register_dumps method on the dump manager.
12124 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12127 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12129 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12130 Say that # requires an associated define_split to exist, and that
12131 the define_split must be suitable for use after register allocation.
12133 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12135 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12136 (compute_outgoing_frequencies): Also initialize zero counts.
12137 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12138 around loops; preserve more of profile when nothing changes.
12140 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12142 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12144 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12145 * config/arm/arm-cpu-cdata.h: Regenerate.
12146 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12147 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12148 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12149 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12150 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12152 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12154 * config/arm/t-rmprofile: Likewise.
12155 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12157 2017-06-06 David S. Miller <davem@davemloft.net>
12160 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12161 blockage if function uses alloca.
12163 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12165 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12166 New "uid" fields to hold pretty-print IDs of group and ref.
12167 Memory references are now identified as <group_id>:<ref_id>
12168 instead of using [random] addresses.
12169 (dump_mem_details): Simplify, no functional change.
12170 (dump_mem_ref): Simplify and make output more concise.
12171 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12172 (find_or_create_group): Initialize group uid.
12173 (record_ref): Initialize ref uid. Improve debug output.
12174 (prune_group_by_reuse, should_issue_prefetch_p,)
12175 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12176 (mark_nontemporal_store, determine_loop_nest_reuse):
12177 Improve debug output.
12179 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12181 * dbgcnt.def (prefetch): New debug counter.
12182 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12183 (schedule_prefetches): Stop issueing prefetches if debug counter
12186 2017-06-06 Tom de Vries <tom@codesourcery.com>
12188 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12189 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12191 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12193 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12194 Use aarch64_reg_or_zero predicate for operand 4.
12195 (aarch64_compare_and_swap<mode> define_insn_and_split):
12196 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12197 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12199 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12201 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12202 (arm_compute_save_core_reg_mask): This.
12203 (thumb1_compute_save_reg_mask): Rename into ...
12204 (thumb1_compute_save_core_reg_mask): This.
12205 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12206 (arm_compute_frame_layout): Likewise.
12208 2017-06-06 Richard Biener <rguenther@suse.de>
12210 PR tree-optimization/80974
12211 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12212 keep or clear leaders SSA info.
12214 2017-06-06 Tom de Vries <tom@codesourcery.com>
12216 * config/nvptx/nvptx.c (split_mode_p): New function.
12217 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12219 2017-06-06 Tom de Vries <tom@codesourcery.com>
12221 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12223 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12226 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12229 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12231 * shrink-wrap.c (handle_simple_exit): Update profile.
12232 (try_shrink_wrapping): Upate profile.
12234 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12236 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12237 (tree_guess_outgoing_edge_probabilities): New.
12238 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12239 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12241 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12243 * ipa-split.c (split_function): Initialize return bb profile.
12245 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12247 * profile.c (compute_branch_probabilities): Also initialize
12248 EXIT_BLOCK profile.
12250 2017-06-06 Richard Biener <rguenther@suse.de>
12252 PR tree-optimization/80928
12253 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12254 (vect_analyze_loop_operations): Properly guard analysis for
12256 (vect_transform_loop): Likewise.
12257 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12258 (vect_model_induction_cost): Do not cost for pure SLP.
12259 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12260 of induction in inner loop vectorization.
12261 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12262 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12263 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12265 (vect_analyze_slp_cost_1): Cost induction.
12266 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12267 (vect_get_slp_vect_defs): Likewise.
12268 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12269 (vect_transform_stmt): Handle SLP reductions.
12270 * tree-vectorizer.h (vectorizable_induction): Adjust.
12272 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12274 * config/rs6000/rs6000.c (make_resolver_func): Update
12275 init_lowered_empty_function call.
12277 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12279 * doc/invoke.texi: Document the -fprofile-abs-path option.
12280 * common.opt (fprofile-abs-path): New option.
12281 * gcov-io.h (gcov_write_filename): Declare.
12282 * gcov-io.c (gcov_write_filename): New function.
12283 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12284 * profile.c (output_location): Likewise.
12286 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12288 * shring-wrap.c: Revert accidental commit.
12290 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12292 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12294 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12296 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12298 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12299 profile in callgraph edge.
12300 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12301 (apply_scale): Likewise.
12302 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12303 Also scale profile when inlining function with zero profile.
12304 (initialize_cfun): Update exit block profile even when it is zero.
12305 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12306 when profile is read.
12308 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12310 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12311 (CLONE_*): New constants to define the processors we can generate
12312 code for with the target_clone attribute.
12313 (rs6000_clone_map): New array to identify which clone processors
12314 the current program is running on.
12315 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12316 target_clone attribute.
12317 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12318 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12319 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12320 (cpu_expand_builtin): Add support for target_clone attribute.
12321 (rs6000_valid_attribute_p): Allow "default" attribute.
12322 (get_decl_name): New debug function to simplify printing the
12323 current function name in debugging statements.
12324 (rs6000_clone_priority): New functions to support the target_clone
12325 attribute, and be able to generate code to switch between ISA 2.05
12326 through ISA 3.0 (power6 through power9).
12327 (rs6000_compare_version_priority): Likewise.
12328 (rs6000_get_function_versions_dispatcher): Likewise.
12329 (make_resolver_func): Likewise.
12330 (add_condition_to_bb): Likewise.
12331 (dispatch_function_versions): Likewise.
12332 (rs6000_generate_version_dispatcher_body): Likewise.
12333 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12334 (fusion_gpr_load_p): Fix a spacing issue.
12335 * doc/extend.texi (Common Function Attributes): Document that the
12336 PowerPC supports the target_clone attribute.
12338 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12340 * config/arm/arm.h: explain F symbol found in description of ARM
12341 register allocation in its legend.
12343 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12345 * config/mips/frame-header-opt.c: Include profile-count.h.
12346 * config/riscv/riscv.c: Include profile-count.h
12348 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12350 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12352 (sm_set_flag_if_changed): Add bbs field.
12353 (execute_sm_if_changed_flag_set): Pass BBS.
12354 (execute_sm): Update.
12356 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12358 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12361 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12363 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12364 (peephole2): New peephole2 to emit the above.
12365 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12367 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12369 * config/aarch64/aarch64.c (define_peephole2 above
12370 *sub_<shift>_<mode>): New peephole.
12372 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12374 * config/i386/i386.c (make_resolver_func): Update.
12375 * Makefile.in: Add profile-count.h and profile-count.o
12376 * auto-profile.c (afdo_indirect_call): Update to new API.
12377 (afdo_set_bb_count): Update.
12378 (afdo_propagate_edge): Update.
12379 (afdo_propagate_circuit): Update.
12380 (afdo_calculate_branch_prob): Update.
12381 (afdo_annotate_cfg): Update.
12382 * basic-block.h: Include profile-count.h
12383 (struct edge_def): Turn count to profile_count.
12384 (struct basic_block_def): Likewie.
12385 (REG_BR_PROB_BASE): Move to profile-count.h
12386 (RDIV): Move to profile-count.h
12387 * bb-reorder.c (max_entry_count): Turn to profile_count.
12388 (find_traces): Update.
12389 (rotate_loop):Update.
12390 (connect_traces):Update.
12391 (sanitize_hot_paths):Update.
12392 * bt-load.c (migrate_btr_defs): Update.
12393 * cfg.c (RDIV): Remove.
12394 (init_flow): Use alloc_block.
12395 (alloc_block): Uninitialize count.
12396 (unchecked_make_edge): Uninitialize count.
12397 (check_bb_profile): Update.
12398 (dump_edge_info): Update.
12399 (dump_bb_info): Update.
12400 (update_bb_profile_for_threading): Update.
12401 (scale_bbs_frequencies_int): Update.
12402 (scale_bbs_frequencies_gcov_type): Update.
12403 (scale_bbs_frequencies_profile_count): New.
12404 * cfg.h (update_bb_profile_for_threading): Update.
12405 (scale_bbs_frequencies_profile_count): Declare.
12406 * cfgbuild.c (compute_outgoing_frequencies): Update.
12407 (find_many_sub_basic_blocks): Update.
12408 * cfgcleanup.c (try_forward_edges): Update.
12409 (try_crossjump_to_edge): Update.
12410 * cfgexpand.c (expand_gimple_tailcall): Update.
12411 (construct_exit_block): Update.
12412 * cfghooks.c (verify_flow_info): Update.
12413 (dump_bb_for_graph): Update.
12414 (split_edge): Update.
12415 (make_forwarder_block): Update.
12416 (duplicate_block): Update.
12417 (account_profile_record): Update.
12418 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12419 (get_estimated_loop_iterations): Update.
12420 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12421 (single_likely_exit): Update.
12422 * cfgloopmanip.c (scale_loop_profile): Update.
12424 (set_zero_probability): Update.
12425 (lv_adjust_loop_entry_edge): Update.
12426 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12427 (purge_dead_edges): Update.
12428 (rtl_account_profile_record): Update.
12429 * cgraph.c (cgraph_node::create): Uninitialize count.
12430 (symbol_table::create_edge): Uninitialize count.
12431 (cgraph_update_edges_for_call_stmt_node): Update.
12432 (cgraph_edge::dump_edge_flags): Update.
12433 (cgraph_node::dump): Update.
12434 (cgraph_edge::maybe_hot_p): Update.
12435 * cgraph.h: Include profile-count.h
12436 (create_clone), create_edge, create_indirect_edge): Update.
12437 (cgraph_node): Turn count to profile_count.
12438 (cgraph_edge0: Likewise.
12439 (make_speculative, clone): Update.
12440 (create_edge): Update.
12441 (init_lowered_empty_function): Update.
12442 * cgraphclones.c (cgraph_edge::clone): Update.
12443 (duplicate_thunk_for_node): Update.
12444 (cgraph_node::create_clone): Update.
12445 * cgraphunit.c (cgraph_node::analyze): Update.
12446 (cgraph_node::expand_thunk): Update.
12447 * final.c (dump_basic_block_info): Update.
12448 * gimple-streamer-in.c (input_bb): Update.
12449 * gimple-streamer-out.c (output_bb): Update.
12450 * graphite.c (print_global_statistics): Update.
12451 (print_graphite_scop_statistics): Update.
12452 * hsa-brig.c: Include basic-block.h.
12453 * hsa-dump.c: Include basic-block.h.
12454 * hsa-gen.c (T sum_slice): Update.
12455 (convert_switch_statements):Update.
12456 * hsa-regalloc.c: Include basic-block.h.
12457 * ipa-chkp.c (chkp_produce_thunks): Update.
12458 * ipa-cp.c (struct caller_statistics): Update.
12459 (init_caller_stats): Update.
12460 (gather_caller_stats): Update.
12461 (ipcp_cloning_candidate_p): Update.
12462 (good_cloning_opportunity_p): Update.
12463 (get_info_about_necessary_edges): Update.
12464 (dump_profile_updates): Update.
12465 (update_profiling_info): Update.
12466 (update_specialized_profile): Update.
12467 (perhaps_add_new_callers): Update.
12468 (decide_about_value): Update.
12469 (ipa_cp_c_finalize): Update.
12470 * ipa-devirt.c (struct odr_type_warn_count): Update.
12471 (struct decl_warn_count): Update.
12472 (struct final_warning_record): Update.
12473 (possible_polymorphic_call_targets): Update.
12474 (ipa_devirt): Update.
12475 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12476 * ipa-icf.c (sem_function::merge): Update.
12477 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12478 * ipa-inline.c (compute_uninlined_call_time): Update.
12479 (compute_inlined_call_time): Update.
12480 (want_inline_small_function_p): Update.
12481 (want_inline_self_recursive_call_p): Update.
12482 (edge_badness): Update.
12483 (lookup_recursive_calls): Update.
12484 (recursive_inlining): Update.
12485 (inline_small_functions): Update.
12486 (dump_overall_stats): Update.
12487 (dump_inline_stats): Update.
12488 * ipa-profile.c (ipa_profile_generate_summary): Update.
12489 (ipa_propagate_frequency): Update.
12490 (ipa_profile): Update.
12491 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12492 * ipa-utils.c (ipa_merge_profiles): Update.
12493 * loop-doloop.c (doloop_modify): Update.
12494 * loop-unroll.c (report_unroll): Update.
12495 (unroll_loop_runtime_iterations): Update.
12496 * lto-cgraph.c (lto_output_edge): Update.
12497 (lto_output_node): Update.
12498 (input_node): Update.
12499 (input_edge): Update.
12500 (merge_profile_summaries): Update.
12501 * lto-streamer-in.c (input_cfg): Update.
12502 * lto-streamer-out.c (output_cfg): Update.
12503 * mcf.c (create_fixup_graph): Update.
12504 (adjust_cfg_counts): Update.
12505 (sum_edge_counts): Update.
12506 * modulo-sched.c (sms_schedule): Update.
12507 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12508 * predict.c (maybe_hot_count_p): Update.
12509 (probably_never_executed): Update.
12510 (dump_prediction): Update.
12511 (combine_predictions_for_bb): Update.
12512 (propagate_freq): Update.
12513 (handle_missing_profiles): Update.
12514 (counts_to_freqs): Update.
12515 (rebuild_frequencies): Update.
12516 (force_edge_cold): Update.
12517 * predict.h: Include profile-count.h
12518 (maybe_hot_count_p, counts_to_freqs): UPdate.
12519 * print-rtl-function.c: Do not include cfg.h
12520 * print-rtl.c: Include basic-block.h
12521 * profile-count.c: New file.
12522 * profile-count.h: New file.
12523 * profile.c (is_edge_inconsistent): Update.
12524 (correct_negative_edge_counts): Update.
12525 (is_inconsistent): Update.
12526 (set_bb_counts): Update.
12527 (read_profile_edge_counts): Update.
12528 (compute_frequency_overlap): Update.
12529 (compute_branch_probabilities): Update; Initialize and deinitialize
12531 (branch_prob): Update.
12532 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12533 (edge_gcov_count): New.
12534 (bb_gcov_count): New.
12535 * shrink-wrap.c (try_shrink_wrapping): Update.
12536 * tracer.c (better_p): Update.
12537 * trans-mem.c (expand_transaction): Update.
12538 (ipa_tm_insert_irr_call): Update.
12539 (ipa_tm_insert_gettmclone_call): Update.
12540 * tree-call-cdce.c: Update.
12541 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12542 (gimple_duplicate_sese_tail): Update.
12543 (gimple_account_profile_record): Update.
12544 (execute_fixup_cfg): Update.
12545 * tree-inline.c (copy_bb): Update.
12546 (copy_edges_for_bb): Update.
12547 (initialize_cfun): Update.
12548 (freqs_to_counts): Update.
12549 (copy_cfg_body): Update.
12550 (expand_call_inline): Update.
12551 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12552 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12553 (try_unroll_loop_completely): Update.
12554 (try_peel_loop): Update.
12555 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12556 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12557 * tree-ssa-loop-split.c (connect_loops): Update.
12558 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12559 * tree-ssa-reassoc.c (branch_fixup): Update.
12560 * tree-ssa-tail-merge.c (replace_block_by): Update.
12561 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12562 (compute_path_counts): Update.
12563 (update_profile): Update.
12564 (recompute_probabilities): Update.
12565 (update_joiner_offpath_counts): Update.
12566 (estimated_freqs_path): Update.
12567 (freqs_to_counts_path): Update.
12568 (clear_counts_path): Update.
12569 (ssa_fix_duplicate_block_edges): Update.
12570 (duplicate_thread_path): Update.
12571 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12572 (struct switch_conv_info): Update.
12573 * tree-tailcall.c (decrease_profile): Update.
12574 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12575 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12576 * value-prof.c (check_counter): Update.
12577 (gimple_divmod_fixed_value): Update.
12578 (gimple_mod_pow2): Update.
12579 (gimple_mod_subtract): Update.
12580 (gimple_ic_transform): Update.
12581 (gimple_stringop_fixed_value): Update.
12582 * value-prof.h (gimple_ic): Update.
12584 2017-06-02 Carl Love <cel@us.ibm.com>
12586 * config/rs6000/rs6000-c: Add support for built-in functions
12587 vector double vec_doublee (vector signed int);
12588 vector double vec_doublee (vector unsigned int);
12589 vector double vec_doublee (vector float);
12590 vector double vec_doubleh (vector signed int);
12591 vector double vec_doubleh (vector unsigned int);
12592 vector double vec_doubleh (vector float);
12593 vector double vec_doublel (vector signed int);
12594 vector double vec_doublel (vector unsigned int);
12595 vector double vec_doublel (vector float);
12596 vector double vec_doubleo (vector signed int);
12597 vector double vec_doubleo (vector unsigned int);
12598 vector double vec_doubleo (vector float);.
12599 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12600 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12602 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12603 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12604 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12606 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12607 vec_doublel, vec_doubleh.
12608 * doc/extend.texi: Update the built-in documentation file for the
12609 new built-in functions.
12611 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12614 * ipa-inline-analysis.c (free_growth_caches): Set
12615 edge_removal_hook_holder to NULL after removing it.
12617 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12619 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12620 comparision with zero.
12622 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12624 for early expansion of vec_min and vec_max builtins.
12625 (builtin_function_type): Add min/max unsigned variants to those
12626 identified as having unsigned arguments.
12628 2017-06-02 Olivier Hainque <hainque@adacore.com>
12630 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12632 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12634 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12635 Use VALL_F16 iterator rather than VALL.
12637 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12639 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12640 Emit CBNZ inside loop when doing a strong exchange and comparing
12641 against zero. Generate the CC flags after the loop.
12643 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12645 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12646 (dl_section_ref): New.
12647 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12648 On AIX, append an expression to subtract the size of the
12649 section length to dl_section_ref.
12651 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12654 for early expansion of vector absolute builtins.
12656 2017-06-02 Richard Biener <rguenther@suse.de>
12658 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12659 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12661 2017-06-02 Richard Biener <rguenther@suse.de>
12663 PR tree-optimization/80948
12664 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12665 stmt order as well.
12667 2017-06-02 Richard Biener <rguenther@suse.de>
12669 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12671 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12672 for inductions as relevant.
12674 2017-06-02 Richard Biener <rguenther@suse.de>
12676 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12677 (vectorizable_induction): ... this. Remove dead code.
12679 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12681 * builtins. (expand_builtin_alloca): Remove second parameter and
12682 infer its value from the first parameter instead.
12683 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12685 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12687 PR rtl-optimization/80903
12688 * loop-doloop.c (add_test): Unshare sequence.
12690 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12692 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12694 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12696 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12698 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12699 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12700 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12701 (xlogue_layout::get_stub_rtx): Make static.
12702 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12703 (xlogue_layout::compute_stub_managed_regs): Rename to...
12704 (xlogue_layout::count_stub_managed_regs): ...this.
12705 (xlogue_layout::is_stub_managed_reg): New function.
12706 (xlogue_layout::m_stub_names): Rename to...
12707 (xlogue_layout::s_stub_names): ...this, make static.
12708 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12709 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12710 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12711 xlogue_layout::s_stub_names): Instantiate statics.
12712 (stub_managed_regs): Remove.
12713 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12714 (disable_call_ms2sysv_xlogues): Rename to...
12715 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
12716 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
12718 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
12719 change after reload_completed.
12720 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
12722 (ix86_expand_prologue): Likewise.
12723 (ix86_expand_epilogue): Likewise.
12724 (ix86_expand_split_stack_prologue): Likewise.
12725 (ix86_compute_frame_layout): Remove frame parameter ...
12726 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
12727 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
12729 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
12730 (ix86_frame): Move from here ...
12731 * config/i386/i386.h (ix86_frame): ... to here.
12732 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
12733 complete ix86_frame data structure instead. Remove some_ld_name.
12735 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12737 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
12738 symbols that hold a DECL_VALUE_EXPR.
12740 2017-06-01 Martin Jambor <mjambor@suse.cz>
12742 PR tree-optimization/80898
12743 * tree-sra.c (process_subtree_disqualification): Removed.
12744 (disqualify_candidate): Do not acll
12745 process_subtree_disqualification.
12746 (subtree_mark_written_and_enqueue): New function.
12747 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12748 RHS has been disqualified and re-queue LHS if necessary. Apart
12749 from that, ignore disqualified RHS.
12751 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12753 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12754 address fetch for z10 or later.
12756 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12758 * config/arc/arc.md (tst_movb): Add guard when splitting.
12760 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12762 * config/arc/arc.c (arc_can_eliminate): Test against
12763 arc_frame_pointer_needed.
12765 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12767 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12768 to prevent store reordering.
12769 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12770 (type): Add block type.
12771 (stack_tie): Define special instruction to be used in
12774 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12776 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12777 constraint. It is not valid for the pattern.
12778 (noncommutative_binary_comparison): Likewise.
12780 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12782 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12785 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12787 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12788 be used by the reg-alloc.
12790 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12792 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12793 reg-alloc when having mul64 or mul32x16 instructions.
12794 (mulsidi3): Likewise.
12795 (umulsidi3): Likewise.
12796 (mulsi32x16): New pattern.
12797 (mulsi64): Likewise.
12798 (mulsidi64): Likewise.
12799 (umulsidi64): Likewise.
12800 (MUL32x16_REG): Define.
12801 (mul64_600): Use MUL32x16_REG.
12802 (mac64_600): Likewise.
12803 (umul64_600): Likewise.
12804 (umac64_600): Likewise.
12806 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12808 * config/arc/arc.md (mulsi3_700): Make it commutative.
12810 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12812 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12814 (*sign_extendsidi2_insn): Likewise for movstosw.
12816 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12818 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12819 the type of the input discriminant value. Convert the
12820 discriminant value of signedness vary.
12822 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12824 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12825 Add to -Wall section.
12827 2017-06-01 Richard Biener <rguenther@suse.de>
12829 PR middle-end/66313
12830 * fold-const.c (fold_plusminus_mult_expr): If the factored
12831 factor may be zero use a wrapping type for the inner operation.
12832 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12833 and handle moved defs.
12834 (process_assignment): Properly guard the unary op case. Return a
12835 tri-state indicating that moving the stmt before the call may allow
12836 to continue. Pass through to_move.
12837 (find_tail_calls): Handle moving unrelated defs before
12840 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12843 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12844 splitter result in the canonical way.
12846 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12848 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12849 also for 32bit target. Update insn attributes.
12850 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12852 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12854 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12855 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12856 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12857 (_mm_maskz_min_ss): New intrinsics.
12859 2017-05-31 Martin Liska <mliska@suse.cz>
12861 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12862 Change comment style to one we normally use.
12863 (vectorizable_reduction): Likewise.
12864 (vectorizable_induction): Likewise.
12865 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12866 (vectorizable_call): Likewise.
12867 (vectorizable_simd_clone_call): Likewise.
12868 (vectorizable_conversion): Likewise.
12869 (vectorizable_assignment): Likewise.
12870 (vectorizable_shift): Likewise.
12871 (vectorizable_operation): Likewise.
12872 (vectorizable_store): Likewise.
12873 (vectorizable_load): Likewise.
12874 * tree-vectorizer.h: Likewise.
12876 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12878 * passes.c (emergency_dump_function): New.
12879 * tree-pass.h (emergency_dump_function): Declare.
12880 * plugin.c (plugins_internal_error_function): Remove.
12881 * plugin.h (plugins_internal_error_function): Remove declaration.
12882 * toplev.c (internal_error_function): New static function. Use it...
12883 (general_init): ...here.
12885 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12887 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12888 (arc_rtx_costs): Add costs for new patterns.
12889 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12890 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12892 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12894 * tree-ssa-strlen.c (get_next_strinfo): New function.
12895 (get_stridx_plus_constant): Use it.
12896 (zero_length_string): Likewise.
12897 (adjust_related_strinfos): Likewise.
12898 (adjust_last_stmt): Likewise.
12900 2017-05-31 Richard Biener <rguenther@suse.de>
12903 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12904 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12906 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12908 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12909 loop_vinfo argument and use of dependence distance vectors.
12910 Check instead whether the two references differ only in their
12911 initial value and assume that they have the same alignment if the
12912 difference is a multiple of the vector alignment.
12913 (vect_analyze_data_refs_alignment): Update call accordingly.
12915 2017-05-31 Martin Liska <mliska@suse.cz>
12918 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12920 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12922 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
12923 (create_intersect_range_checks): Move from ...
12924 * tree-data-ref.c (create_intersect_range_checks_index)
12925 (create_intersect_range_checks): ... to here.
12926 (create_runtime_alias_checks): New function factored from ...
12927 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
12928 here. Call above function.
12929 * tree-data-ref.h (create_runtime_alias_checks): New function.
12931 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12933 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
12934 segment length for dr_b and compute it in wide_int.
12936 2017-05-31 Richard Biener <rguenther@suse.de>
12938 PR tree-optimization/80906
12939 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
12940 and pass through iv_map.
12941 (copy_bb_and_scalar_dependences): Adjust.
12942 (translate_pending_phi_nodes): Likewise.
12943 (copy_loop_close_phi_args): Handle code-generating IVs instead
12946 2017-05-30 David Malcolm <dmalcolm@redhat.com>
12948 * diagnostic-color.c (color_dict): Add "type-diff".
12949 (parse_gcc_colors): Update comment.
12950 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12951 -fdiagnostics-show-template-tree and -fno-elide-type.
12952 (GCC_COLORS): Add type-diff to example.
12954 (-fdiagnostics-show-template-tree): New.
12955 (-fno-elide-type): New.
12956 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
12957 the pp_format_decoder callback. Call any m_format_postprocessor's
12959 (pretty_printer::pretty_printer): Initialize
12960 m_format_postprocessor.
12961 (pretty_printer::~pretty_printer): Delete any
12962 m_format_postprocessor.
12963 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
12964 (class format_postprocessor): New class.
12965 (struct pretty_printer::format_decoder): Document the new parameters.
12966 (struct pretty_printer::m_format_postprocessor): New field.
12967 * tree-diagnostic.c (default_tree_printer): Update for new
12968 bool and const char ** params.
12969 * tree-diagnostic.h (default_tree_printer): Likewise.
12971 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
12973 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
12974 (lwa_operand): Delete rs6000_gen_cell_microcode test.
12975 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12976 rs6000_gen_cell_microcode code.
12977 (rs6000_final_prescan_insn): Delete.
12978 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
12979 "warn-cell-microcode" entries.
12980 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
12981 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
12982 throughout. Change cc_reg_not_micro_cr0_operand to
12983 cc_reg_not_cr0_operand throughout.
12984 (*extendhi<mode>2_noload): Delete.
12985 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
12986 (mwarn-cell-microcode): Delete.
12987 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
12988 -mgen-cell-microcode and -mwarn-cell-microcode.
12990 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
12993 * config/i386/constraints.md (Yd): New constraint.
12995 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
12996 and (?Yd, r) alternatives. Update insn attributes.
12997 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
12998 and (?*Yd, r) alternatives. Update insn attributes.
12999 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13001 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13003 * gimplify.c (gimplify_modify_expr): Don't create a
13004 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13007 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13009 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13011 2017-05-30 Richard Biener <rguenther@suse.de>
13013 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13014 and reduc_def fields.
13015 (STMT_VINFO_REDUC_TYPE): New define.
13016 (STMT_VINFO_REDUC_DEF): Likewise.
13017 (vect_force_simple_reduction): Adjust prototype.
13018 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13019 (vect_is_simple_reduction): Remove check_reduction argument.
13020 (vect_force_simple_reduction): Adjust and set
13021 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13022 (vectorizable_reduction): Do not re-do reduction analysis
13023 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13024 * tree-parloops.c (gather_scalar_reductions): Adjust.
13026 2017-05-30 Richard Biener <rguenther@suse.de>
13028 PR middle-end/80901
13029 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13032 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13034 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13035 Introduce unknown_misalignment parameter and remove vf.
13036 (vect_peeling_hash_get_lowest_cost):
13037 Pass unknown_misalignment parameter.
13038 (vect_enhance_data_refs_alignment):
13039 Fix unsupportable data ref treatment.
13041 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13043 * tree-vect-data-refs.c (vect_get_data_access_cost):
13044 Workaround for SLP handling.
13045 (vect_enhance_data_refs_alignment):
13046 Compute costs for doing no peeling at all, compare to the best
13047 peeling costs so far and avoid peeling if cheaper.
13049 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13051 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13052 Return peeling info and set costs to zero for unlimited cost
13054 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13055 with unknown misalignment. Compute and costs for unknown
13056 misalignment, compare them to the costs for known misalignment
13057 and choose the cheapest for peeling.
13059 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13061 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13062 (vect_get_peeling_costs_all_drs): Create function.
13063 (vect_peeling_hash_get_lowest_cost):
13064 Use vect_get_peeling_costs_all_drs.
13065 (vect_peeling_supportable): Create function.
13066 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13068 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13070 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13071 DR_HAS_NEGATIVE_STEP.
13072 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13073 (vect_enhance_data_refs_alignment): Use.
13074 (vect_duplicate_ssa_name_ptr_info): Use.
13075 * tree-vectorizer.h (dr_misalignment): Use.
13076 (known_alignment_for_access_p): Use.
13078 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13081 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13083 (has_section_name): New function.
13085 2017-05-30 Martin Liska <mliska@suse.cz>
13088 * auto-profile.c (get_function_decl_from_block): Fix
13091 2017-05-30 Richard Biener <rguenther@suse.de>
13093 PR middle-end/80876
13094 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13096 2017-05-30 Martin Liska <mliska@suse.cz>
13098 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13099 * dumpfile.h (struct dump_file_info): Remove ctors.
13101 2017-05-30 Martin Liska <mliska@suse.cz>
13103 * predict.def: Fix GNU coding style.
13105 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13107 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13108 Mark 'to' argument with ATTRIBUTE_UNUSED.
13110 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13112 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13113 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13114 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13117 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13119 * doc/install.texi (Options specification): Restore entry of
13120 --enable-sjlj-exceptions.
13122 2017-05-27 Michael Eager <eager@eagercon.com>
13125 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13127 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13129 * config/microblaze/microblaze.h
13130 (FIXED_REGISTERS): Update in macro.
13131 (CALL_USED_REGISTERS): Update in macro.
13133 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13135 * doc/install.texi: Add links to macOS binary distributions.
13137 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13141 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13143 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13145 2017-05-26 Martin Liska <mliska@suse.cz>
13147 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13149 2017-05-26 Martin Liska <mliska@suse.cz>
13151 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13152 always leading ';; '.
13153 (dump_bb_info): Likewise.
13154 (brief_dump_cfg): Likewise.
13155 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13156 * dumpfile.c: Remove usage of TDF_VERBOSE.
13157 * dumpfile.h (enum dump_kind): Likewise.
13158 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13159 * print-tree.c (debug_verbose): Remove.
13160 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13161 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13162 * tree-diagnostic.c (default_tree_printer): Replace
13163 TDF_DIAGNOSTIC with TDF_SLIM.
13165 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13167 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13168 in parameter loop, rather than loop_vinfo.
13169 (create_intersect_range_checks): Ditto.
13170 (vect_create_cond_for_alias_checks): Update call to above functions.
13172 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13174 PR tree-optimization/80815
13175 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13176 for merging runtime alias checks. Handle negative DR_STEPs.
13178 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13180 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13182 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13183 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13184 out code pruning runtime alias checks.
13185 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13186 factored out from above.
13187 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13189 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13191 (prune_runtime_alias_test_list): New decalaration.
13193 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13195 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13196 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13197 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13198 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13199 (operator==, comp_dr_with_seg_len_pair): Ditto.
13200 (vect_prune_runtime_alias_test_list): Ditto.
13202 2017-05-26 Martin Liska <mliska@suse.cz>
13205 * params.def: Bound partial-inlining-entry-probability param.
13207 2017-05-26 Marek Polacek <polacek@redhat.com>
13210 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13213 2017-05-26 Richard Biener <rguenther@suse.de>
13215 PR tree-optimization/80842
13216 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13219 2017-05-26 Richard Biener <rguenther@suse.de>
13221 PR tree-optimization/80844
13222 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13224 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13226 * doc/md.texi (Machine Constraints): Update x86 family
13227 machine constraints section to match 'config/i386/constraints.md'.
13229 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13231 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13233 2017-05-25 Nathan Sidwell <nathan@acm.org>
13235 * doc/invoke.texi (--enable-languages): Update documentation.
13237 2017-05-25 Martin Liska <mliska@suse.cz>
13239 * dumpfile.c: Add TDF_FOLDING.
13240 * dumpfile.h (enum dump_kind): Likewise.
13241 * genmatch.c (dt_simplify::gen_1): Use it.
13243 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13245 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13247 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13249 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13250 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13252 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13254 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13255 * match.pd (X == C): Rewrite it here.
13256 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13257 with_certain_nonzero_bits2): New predicates.
13258 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13260 2017-05-24 Nathan Sidwell <nathan@acm.org>
13262 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13265 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13268 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13270 * config/powerpcspe: New port. Files are copied from the rs6000
13271 port, with "rs6000" in filenames replaced by "powerpcspe".
13273 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13275 PR rtl-optimization/80754
13276 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13278 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13280 * config/sparc/sparc.md (length): Return the correct value for -mflat
13281 sibcalls to match output_sibcall.
13283 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13287 * config/rs6000/rs6000.c (struct machine_function): Add new field
13289 (rs6000_get_separate_components): Init that field, use it.
13290 (rs6000_components_for_bb): Use the field.
13292 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13294 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13296 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13298 PR middle-end/80823
13299 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13301 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13304 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13305 address against address_operand predicate.
13306 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13308 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13310 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13311 parameters passed indirectly.
13313 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13315 * config/i386/i386.md (*movdi_internal): Remove SSE4
13316 alternative 18 (?r, *v). Update insn attributes.
13317 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13318 Update insn attributes.
13319 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13320 Update insn attributes.
13321 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13322 alternative 1 (r, v). Remove isa attribute.
13323 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13324 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13325 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13327 2017-05-23 Tom de Vries <tom@codesourcery.com>
13329 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13332 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13334 * cgraphunit.c (symbol_table::process_new_functions): Update.
13335 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13336 (inline_generate_summary): Rename to ...
13337 (ipa_fn_summary_generate): ... this one.
13338 (inline_read_summary): Rename to ...
13339 (ipa_fn_summary_read): ... this one.
13340 (inline_write_summary): Rename to ...
13341 (ipa_fn_summary_write): ... this one.
13342 (inline_free_summary): Rename to ...
13343 (ipa_free_fn_summary): ... this one.
13344 (pass_data_local_fn_summary, pass_local_fn_summary,
13345 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13346 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13347 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13348 make_pass_ipa_fn_summary): New.
13349 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13350 inline_write_summary, inline_free_summary): Remove.
13351 (ipa_free_fn_summary) : New.
13352 * ipa-inline.c (ipa_inline): Update.
13353 (pass_ipa_inline): Do not generate summaries.
13354 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13356 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13357 and add pass_ipa_fn_summary.
13358 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13360 (make_pass_inline_parameters): Remove.
13362 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13364 * omp-low.c (struct omp_context): Remove "default_kind" member.
13367 * omp-offload.c (execute_oacc_device_lower): Remove the
13368 parallelism dimensions function attributes for unparallelized
13369 OpenACC kernels constructs.
13371 2017-05-23 Martin Liska <mliska@suse.cz>
13373 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13375 (cgraph_edge::make_speculative): Likewise.
13376 (cgraph_edge::resolve_speculation): Likewise.
13377 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13378 (cgraph_node::dump): Likewise.
13379 * cgraph.h: Likewise.
13380 * cgraphunit.c (analyze_functions): Likewise.
13381 (symbol_table::compile): Likewise.
13382 * ipa-cp.c (print_all_lattices): Likewise.
13383 (determine_versionability): Likewise.
13384 (initialize_node_lattices): Likewise.
13385 (ipcp_verify_propagated_values): Likewise.
13386 (estimate_local_effects): Likewise.
13387 (update_profiling_info): Likewise.
13388 (create_specialized_node): Likewise.
13389 (perhaps_add_new_callers): Likewise.
13390 (decide_about_value): Likewise.
13391 (decide_whether_version_node): Likewise.
13392 (identify_dead_nodes): Likewise.
13393 (ipcp_store_bits_results): Likewise.
13394 * ipa-devirt.c (dump_targets): Likewise.
13395 (ipa_devirt): Likewise.
13396 * ipa-icf.c (sem_item::dump): Likewise.
13397 (sem_function::equals): Likewise.
13398 (sem_variable::equals): Likewise.
13399 (sem_item_optimizer::read_section): Likewise.
13400 (sem_item_optimizer::execute): Likewise.
13401 (congruence_class::dump): Likewise.
13402 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13403 (dump_inline_summary): Likewise.
13404 (estimate_node_size_and_time): Likewise.
13405 (inline_analyze_function): Likewise.
13406 * ipa-inline-transform.c (inline_call): Likewise.
13407 * ipa-inline.c (report_inline_failed_reason): Likewise.
13408 (want_early_inline_function_p): Likewise.
13409 (edge_badness): Likewise.
13410 (update_edge_key): Likewise.
13411 (inline_small_functions): Likewise.
13412 * ipa-profile.c (ipa_profile): Likewise.
13413 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13414 (ipa_make_edge_direct_to_target): Likewise.
13415 (remove_described_reference): Likewise.
13416 (ipa_impossible_devirt_target): Likewise.
13417 (propagate_controlled_uses): Likewise.
13418 (ipa_print_node_params): Likewise.
13419 (ipcp_transform_function): Likewise.
13420 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13421 (propagate_pure_const): Likewise.
13422 * ipa-reference.c (generate_summary): Likewise.
13423 (read_write_all_from_decl): Likewise.
13424 (propagate): Likewise.
13425 (ipa_reference_read_optimization_summary): Likewise.
13426 * ipa-utils.c (ipa_merge_profiles): Likewise.
13427 * ipa.c (walk_polymorphic_call_targets): Likewise.
13428 (symbol_table::remove_unreachable_nodes): Likewise.
13429 (ipa_single_use): Likewise.
13430 * passes.c (execute_todo): Likewise.
13431 * predict.c (drop_profile): Likewise.
13432 * symtab.c (symtab_node::get_dump_name): New function.
13433 (symtab_node::dump_name): Likewise.
13434 (symtab_node::dump_asm_name): Likewise.
13435 (symtab_node::dump_references): Likewise.
13436 (symtab_node::dump_referring): Likewise.
13437 (symtab_node::dump_base): Likewise.
13438 (symtab_node::debug_symtab): Likewise.
13439 * tree-sra.c (convert_callers_for_node): Likewise.
13440 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13441 * value-prof.c (init_node_map): Likewise.
13443 2017-05-23 Martin Liska <mliska@suse.cz>
13445 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13446 and symtab_node::debug_symtab to symbol_table::debug.
13447 * cgraphunit.c (analyze_functions): Use the renamed function.
13448 (symbol_table::compile): Likewise.
13449 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13450 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13451 * passes.c (execute_todo): Likewise.
13452 * symtab.c (symbol_table::dump): New function.
13453 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13455 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13457 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13458 that nonconst implies exec.
13460 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13462 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13463 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13464 (inline_edge_summary_vec): Turn into ...
13465 (ipa_call_summaries): ... this one.
13466 (redirect_to_unreachable, edge_set_predicate,
13467 evaluate_properties_for_edge, inline_summary_alloc,
13468 reset_ipa_call_summary, reset_inline_summary,
13469 inline_summary_t::duplicate): Update.
13470 (inline_edge_duplication_hook): Turn to ...
13471 (ipa_call_summary_t::duplicate): ... this one.
13472 (inline_edge_removal_hook): Turn to ...
13473 (ipa_call_summary_t::remove): ... this one.
13474 (dump_inline_edge_summary): Turn to ...
13475 (dump_ipa_call_summary): ... this one.
13476 (estimate_function_body_sizes): Update.
13477 (inline_update_callee_summaries): Update.
13478 (remap_edge_change_prob): Update.
13479 (remap_edge_summaries): Update.
13480 (inline_merge_summary): Update.
13481 (do_estimate_edge_time): Update.
13482 (inline_generate_summary): Update.
13483 (inline_read_section): Update.
13484 (inline_read_summary): Update.
13485 (inline_free_summary): Update.
13486 * ipa-inline.c (can_inline_edge_p): Update.
13487 (compute_inlined_call_time): Update.
13488 (want_inline_small_function_p): Update.
13489 (edge_badness): Update.
13490 (early_inliner): Update.
13491 * ipa-inline.h (inline_edge_summary): Turn to ...
13492 (ipa_call_summary): ... this one.
13493 (ipa_call_summary_t): New class.
13494 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13495 (ipa_call_summaries): New.
13496 (inline_edge_summary): Remove.
13497 (estimate_edge_growth): Update.
13498 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13499 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13500 * ipa-split.c (execute_split_functions): Update.
13501 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13503 2017-05-23 Tom de Vries <tom@codesourcery.com>
13505 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13506 attributes): Document rdrand effective target.
13508 2017-05-23 Tom de Vries <tom@codesourcery.com>
13510 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13511 attributes): Sort alphabetically.
13513 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13515 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13517 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13520 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13521 V2DF/V2DI splat into two separate patterns, one that handles
13522 registers, and the other that only handles memory. Drop support
13523 for splatting from a GPR on ISA 2.07 and then splitting the
13524 splat into direct move and splat.
13525 (vsx_splat_<mode>_reg): Likewise.
13526 (vsx_splat_<mode>_mem): Likewise.
13528 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13530 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13532 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13534 PR middle-end/80809
13535 * omp-low.c (finish_taskreg_remap): New function.
13536 (finish_taskreg_scan): If unit size of ctx->record_type
13537 is non-constant, unshare the size expression and replace
13538 decls in it with possible outer var refs.
13540 PR middle-end/80809
13541 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13542 GOVD_SHARED rather than GOVD_PRIVATE with it.
13543 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13544 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13546 PR middle-end/80853
13547 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13548 as last argument to build_outer_var_ref for pointer bases of array
13549 section reductions.
13551 2017-05-19 Martin Sebor <msebor@redhat.com>
13553 * print-tree.c (print_node): Print DECL_READ_P flag.
13555 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13557 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13558 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13559 * cgraph.c: Likewise.
13560 * cgraphunit.c: Likewise.
13561 * gengtype.c: Likewise.
13562 * ipa-cp.c: Likewise.
13563 * ipa-devirt.c: Likewise.
13564 * ipa-icf.c: Likewise.
13565 * ipa-predicate.c: Likewise.
13566 * ipa-profile.c: Likewise.
13567 * ipa-prop.c: Likewise.
13568 * ipa-split.c: Likewise.
13570 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13571 edge_predicate_pool, dump_inline_hints,
13572 inline_summary::account_size_time, redirect_to_unreachable,
13573 edge_set_predicate, set_hint_predicate,
13574 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13575 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13576 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13577 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13578 ipa_call_summary_t::remove, initialize_growth_caches,
13579 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13580 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13581 mark_modified, unmodified_parm_1, unmodified_parm,
13582 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13583 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13584 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13585 will_be_nonconstant_predicate, record_modified_bb_info,
13586 get_minimal_bb, record_modified, param_change_prob,
13587 phi_result_unknown_predicate, predicate_for_phi_result,
13588 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13589 estimate_function_body_sizes, compute_inline_parameters,
13590 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13591 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13592 inline_update_callee_summaries, remap_edge_change_prob,
13593 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13594 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13595 inline_analyze_function, inline_summary_t::insert,
13596 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13597 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13598 inline_free_summary): Move to ipa-fnsummary.h
13599 (predicate_t): Remove.
13600 * ipa-fnsummary.c: New file.
13601 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13602 (enum inline_hints_vals, inline_hints, agg_position_info,
13603 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13604 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13605 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13606 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13607 inline_read_summary, inline_write_summary, inline_free_summary,
13608 inline_analyze_function, initialize_inline_failed,
13609 inline_merge_summary, inline_update_overall_summary,
13610 compute_inline_parameters): Move to ipa-fnsummary.h
13611 * ipa-fnsummary.h: New file.
13612 * ipa-inline-transform.h: Include ipa-inline.h.
13613 * ipa-inline.c: LIkewise.
13615 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13617 * ipa-inline.c (edge_badness): Use inlined_time instead of
13618 inline_summaries->get.
13620 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13622 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13624 2017-05-22 Nathan Sidwell <nathan@acm.org>
13626 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13627 (fdump-lang): Document 'raw' option.
13628 * dumpfile.h (TDI_tu): Delete.
13629 * dumpfile.c (dump_files): Remove translation-unit.
13630 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13632 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13634 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13635 command option from $(AWK) call.
13636 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13637 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13638 [FORMAT]: Remove handling of variable.
13639 * config/avr/t-multilib: Regenerate.
13641 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13643 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13645 (dump_inline_summary): Do not print self_time.
13646 (estimate_function_body_sizes): Do not set self_time.
13647 (compute_inline_parameters): Likewise.
13648 (inline_read_section, inline_write_summary): Do not stream self_time.
13649 * ipa-inline.h (inline_summary): Drop self_time.
13651 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13653 * ipa-inline-analysis.c (account_size_time): Rename to ...
13654 (inline_summary::account_size_time): ... this one.
13655 (reset_ipa_call_summary): Turn to ...
13656 (ipa_call_summary::reset): ... this one.
13657 (reset_inline_summary): Turn to ...
13658 (inline_summary::reset): ... this one.
13659 (inline_summary_t::remove): Update.
13660 (inline_summary_t::duplicate): Update.
13661 (ipa_call_summary_t::remove): Update.
13662 (dump_inline_summary): Update.
13663 (estimate_function_body_sizes): Update.
13664 (compute_inline_parameters): Update.
13665 (estimate_node_size_and_time): Update.
13666 (inline_merge_summary): Update.
13667 (inline_update_overall_summary): Update.
13668 (inline_read_section): Update.
13669 (inline_write_summary): Update.
13670 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13671 add account_size_time and reset member functions.
13672 (ipa_call_summary): Add reset function.
13673 * ipa-predicate.h (predicate::operator &): Constify.
13675 2017-05-22 Richard Biener <rguenther@suse.de>
13677 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13679 2017-05-19 Jason Merrill <jason@redhat.com>
13681 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13683 2017-05-19 Marek Polacek <polacek@redhat.com>
13686 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13687 TYPE_OVERFLOW_WRAPS checks.
13689 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13691 * tree-core.h (enum omp_clause_default_kind): Add
13692 "OMP_CLAUSE_DEFAULT_PRESENT".
13693 * tree-pretty-print.c (dump_omp_clause): Handle it.
13694 * gimplify.c (enum gimplify_omp_var_data): Add
13695 "GOVD_MAP_FORCE_PRESENT".
13696 (gimplify_adjust_omp_clauses_1): Map it to
13697 "GOMP_MAP_FORCE_PRESENT".
13698 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13700 * gimplify.c (oacc_default_clause): Clarify.
13702 2017-05-19 Nathan Sidwell <nathan@acm.org>
13704 LANG_HOOK_REGISTER_DUMPS
13705 * toplev.c (general_init): Call register dump lang hook.
13706 * doc/invoke.texi: Document -fdump-lang option family.
13707 * dumpfile.c (dump_files): Remove class dump here.
13708 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13709 * dumpfile.h (tree_dump_index): Remove TDI_class.
13710 * langhooks-def.h (lhd_register_dumps): Declare.
13711 (LANG_HOOKS_REGISTER_DUMPS): Define.
13712 (LANG_HOOKS_INITIALIZER): Add it.
13713 * langhooks.c (lhd_register_dumps): Define.
13714 * langhooks.h (struct lang_hooks): Add register_dumps.
13716 2017-05-19 Nathan Sidwell <nathan@acm.org>
13718 * context.h (context::set_passes): New.
13719 * context.c (context::context): Do not create pass manager.
13720 * toplev.c (general_init): Create pass manager here.
13722 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
13724 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
13725 use this splitter if two add or or instructions would also work for
13726 the constant we want to generate.
13728 2017-05-19 Richard Biener <rguenther@suse.de>
13731 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
13732 predicate evaluation.
13734 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13736 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
13738 * ipa-inline.c (want_inline_small_function_p): Do not cast to
13741 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13743 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13744 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13745 (inline_edge_summary_vec): Turn into ...
13746 (ipa_call_summaries): ... this one.
13747 (redirect_to_unreachable, edge_set_predicate,
13748 evaluate_properties_for_edge, inline_summary_alloc,
13749 reset_ipa_call_summary, reset_inline_summary,
13750 inline_summary_t::duplicate): Update.
13751 (inline_edge_duplication_hook): Turn to ...
13752 (ipa_call_summary_t::duplicate): ... this one.
13753 (inline_edge_removal_hook): Turn to ...
13754 (ipa_call_summary_t::remove): ... this one.
13755 (dump_inline_edge_summary): Turn to ...
13756 (dump_ipa_call_summary): ... this one.
13757 (estimate_function_body_sizes): Update.
13758 (inline_update_callee_summaries): Update.
13759 (remap_edge_change_prob): Update.
13760 (remap_edge_summaries): Update.
13761 (inline_merge_summary): Update.
13762 (do_estimate_edge_time): Update.
13763 (inline_generate_summary): Update.
13764 (inline_read_section): Update.
13765 (inline_read_summary): Update.
13766 (inline_free_summary): Update.
13767 * ipa-inline.c (can_inline_edge_p): Update.
13768 (compute_inlined_call_time): Update.
13769 (want_inline_small_function_p): Update.
13770 (edge_badness): Update.
13771 (early_inliner): Update.
13772 * ipa-inline.h (inline_edge_summary): Turn to ...
13773 (ipa_call_summary): ... this one.
13774 (ipa_call_summary_t): New class.
13775 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13776 (ipa_call_summaries): New.
13777 (inline_edge_summary): Remove.
13778 (estimate_edge_growth): Update.
13779 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13780 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13781 * ipa-split.c (execute_split_functions): Update.
13782 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13784 2017-05-19 Richard Biener <rguenther@suse.de>
13786 PR middle-end/80764
13787 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13789 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13791 * config/rs6000/rs6000.c (struct machine_function): Add field
13792 fpr_is_wrapped_separately.
13793 (rs6000_get_separate_components): Use 64 components. Handle the
13794 new FPR components.
13795 (rs6000_components_for_bb): Handle the FPR components.
13796 (rs6000_emit_prologue_components): Handle the FPR components.
13797 (rs6000_emit_epilogue_components): Handle the FPR components.
13798 (rs6000_set_handled_components): Handle the FPR components.
13799 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13800 that are already separately shrink-wrapped.
13801 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13802 that are already separately shrink-wrapped.
13804 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13807 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13810 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13811 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13812 cases where the register allocator uses a move and an offsettable
13813 memory operation to/from a FPR register on ISA 2.06/2.07.
13814 (define_peephole2 for Altivec d-form store): Likewise.
13816 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13819 * config/i386/mmx.md (*mov<mode>_internal): Enable
13820 alternatives 11, 12, 13 and 14 also for 32bit targets.
13821 Remove alternatives 15, 16, 17 and 18.
13822 * config/i386/sse.md (vec_concatv2di): Change
13823 alternative (!x, *y) to (x, ?!*Yn).
13825 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13827 * dumpfile.h (enum dump_kind): Remove stray comma.
13829 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13831 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13832 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13833 predicate::num_conditions
13834 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13835 (CHANGED): turn into predicate::changed.
13836 (agg_position_info): Move to ipa-predicate.h
13837 (add_condition, predicate::add_clause, predicate::operator &=,
13838 predicate::or_with, predicate::evaluate, predicate::probability,
13839 dump_condition, dump_clause, predicate::dump,
13840 predicate::remap_after_duplication, predicate::remap_after_inlining,
13841 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13842 (evaluate_conditions_for_known_args): Update.
13843 (set_cond_stmt_execution_predicate): Update.
13844 * ipa-inline.h: Include ipa-predicate.h
13845 (condition, inline_param_summary, conditions, agg_position_info,
13846 predicate): Move to ipa-predicate.h
13847 * ipa-predicate.c: New file.
13848 * ipa-predicate.h: New file.
13850 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13852 * final.c (leaf_function_p): Check we are not in a sequence.
13854 2017-05-18 Martin Liska <mliska@suse.cz>
13856 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13857 * dumpfile.c (dump_register): Use new enum dump_kind.
13858 (get_dump_file_name): Likewise.
13859 (dump_enable_all): Likewise.
13860 (dump_switch_p_1): Likewise.
13861 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13862 * dumpfile.h (enum dump_kind): New enum type.
13863 (struct dump_file_info): Create constructor and
13864 format fields and comments.
13865 * passes.c (pass_manager::register_one_dump_file):
13867 * statistics.c (statistics_early_init): Likewise.
13868 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13869 TDF_TREE with TDF_SLIM.
13870 (gather_memory_references_ref): Likewise.
13872 2017-05-18 Martin Liska <mliska@suse.cz>
13874 * vec.h (struct vnull): Use it.
13876 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13878 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13879 (true_predicate, false_predicate, true_predicate_p,
13880 false_predicate_p): Remove.
13881 (single_cond_predicate, not_inlined_predicate): Turn to member function
13883 (add_condition): Update.
13884 (add_clause): Turn to...
13885 (predicate::add_clause): ... this one; update; allow passing NULL
13887 (and_predicates): Turn to ...
13888 (predicate::operator &=): ... this one.
13889 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13890 (or_predicates): Turn to ...
13891 (predicate::or_with): ... this one.
13892 (evaluate_predicate): Turn to ...
13893 (predicate::evaluate): ... this one.
13894 (predicate_probability): Turn to ...
13895 (predicate::probability): ... this one.
13896 (dump_condition): Update.
13897 (dump_predicate): Turn to ...
13898 (predicate::dump): ... this one.
13899 (account_size_time): Update.
13900 (edge_set_predicate): Update.
13901 (set_hint_predicate): UPdate.
13902 (evaluate_conditions_for_known_args): Update.
13903 (evaluate_properties_for_edge): Update.
13904 (remap_predicate_after_duplication): Turn to...
13905 (predicate::remap_after_duplication): ... this one.
13906 (remap_hint_predicate_after_duplication): Update.
13907 (inline_summary_t::duplicate): UPdate.
13908 (dump_inline_edge_summary): Update.
13909 (dump_inline_summary): Update.
13910 (set_cond_stmt_execution_predicate): Update.
13911 (set_switch_stmt_execution_predicate): Update.
13912 (compute_bb_predicates): Update.
13913 (will_be_nonconstant_expr_predicate): Update.
13914 (will_be_nonconstant_predicate): Update.
13915 (phi_result_unknown_predicate): Update.
13916 (predicate_for_phi_result): Update.
13917 (array_index_predicate): Update.
13918 (estimate_function_body_sizes): Update.
13919 (estimate_node_size_and_time): Update.
13920 (estimate_ipcp_clone_size_and_time): Update.
13921 (remap_predicate): Rename to ...
13922 (predicate::remap_after_inlining): ... this one.
13923 (remap_hint_predicate): Update.
13924 (inline_merge_summary): Update.
13925 (inline_update_overall_summary): Update.
13926 (estimate_size_after_inlining): Update.
13927 (read_predicate): Rename to ...
13928 (predicate::stream_in): ... this one.
13929 (read_inline_edge_summary): Update.
13930 (write_predicate): Rename to ...
13931 (predicate::stream_out): ... this one.
13932 (write_inline_edge_summary): Update.
13933 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
13934 (clause_t): Turn to uint32_t
13935 (predicate): Turn to class; implement constructor and operators
13937 (size_time_entry): Update.
13938 (inline_summary): Update.
13939 (inline_edge_summary): Update.
13941 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
13943 * fold-const.c (fold_binary_loc): Move transformation...
13944 * match.pd (C - X CMP X): ... here.
13946 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
13948 * config/sparc/sparc.c (sparc_option_override): Set function
13949 alignment for -mcpu=niagara7 to 64 to match the I$ line.
13950 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
13952 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
13954 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
13956 2017-05-18 Marek Polacek <polacek@redhat.com>
13959 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
13960 (pass_ubsan::execute): Call gimple_assign_single_p instead of
13961 gimple_assign_load_p.
13963 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
13965 PR middle-end/80692
13966 * real.c (do_compare): Give decimal_do_compare preference over
13967 comparing just the signs.
13969 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
13971 * doc/md.texi (Canonicalization of Instructions): Describe the
13972 canonical form of instructions that inherently set a condition
13975 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
13977 PR middle-end/80775
13978 * tree-cfg.c: Move deletion of unreachable case statements to after
13979 the merging of consecutive case labels.
13981 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
13983 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
13984 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
13985 restoring of callee-saved registers.
13987 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
13989 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
13990 * config/visium/visium.c (single_set_and_flags): Likewise.
13991 * config/visium/visium.md (Substitutions): Likewise.
13993 2017-05-17 Martin Liska <mliska@suse.cz>
13995 * cfg.c: Introduce dump_flags_t type and
13996 use it instead of int type.
13998 * cfghooks.c: Likewise.
13999 * cfghooks.h (struct cfg_hooks): Likewise.
14000 * cfgrtl.c: Likewise.
14001 * cfgrtl.h: Likewise.
14002 * cgraph.c (cgraph_node::get_body): Likewise.
14003 * coretypes.h: Likewise.
14004 * domwalk.c: Likewise.
14005 * domwalk.h: Likewise.
14006 * dumpfile.c (struct dump_option_value_info): Likewise.
14007 (dump_enable_all): Likewise.
14008 (dump_switch_p_1): Likewise.
14009 (opt_info_switch_p): Likewise.
14010 * dumpfile.h (enum tree_dump_index): Likewise.
14011 (struct dump_file_info): Likewise.
14012 * genemit.c: Likewise.
14013 * generic-match-head.c: Likewise.
14014 * gengtype.c (open_base_files): Likewise.
14015 * gimple-pretty-print.c: Likewise.
14016 * gimple-pretty-print.h: Likewise.
14017 * graph.c (print_graph_cfg): Likewise.
14018 * graphite-scop-detection.c (dot_all_sese): Likewise.
14019 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14020 * loop-unroll.c (report_unroll): Likewise.
14021 * passes.c (pass_manager::register_one_dump_file): Likewise.
14022 * print-tree.c: Likewise.
14023 * statistics.c: Likewise.
14024 * tree-cfg.c: Likewise.
14025 * tree-cfg.h: Likewise.
14026 * tree-dfa.c: Likewise.
14027 * tree-dfa.h: Likewise.
14028 * tree-dump.c (dump_function): Likewise.
14029 * tree-dump.h (struct dump_info): Likewise.
14030 * tree-pretty-print.c: Likewise.
14031 * tree-pretty-print.h: Likewise.
14032 * tree-ssa-live.c: Likewise.
14033 * tree-ssa-live.h: Likewise.
14034 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14035 * tree-vect-loop.c: Likewise.
14036 * tree-vect-slp.c: Likewise.
14038 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14039 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14041 PR tree-optimization/80457
14042 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14043 of all arguments to a statement as scalar_to_vec operations.
14044 (vectorizable_call): Adjust call to vect_model_simple_cost for
14046 (vectorizable_conversion): Likewise.
14047 (vectorizable_assignment): Likewise.
14048 (vectorizable_shift): Likewise.
14049 (vectorizable_operation): Likewise.
14050 (vectorizable_comparison): Likewise.
14051 (vect_is_simple_cond): Record the def types for operands.
14052 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14053 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14054 for statement argument count.
14056 2017-05-16 Carl Love <cel@us.ibm.com>
14058 * config/rs6000/rs6000-c: Add support for built-in functions
14059 vector unsigned long long vec_bperm (vector unsigned long long,
14060 vector unsigned char)
14061 vector signed long long vec_mule (vector signed int,
14063 vector unsigned long long vec_mule (vector unsigned int,
14064 vector unsigned int)
14065 vector signed long long vec_mulo (vector signed int,
14067 vector unsigned long long vec_mulo (vector unsigned int,
14068 vector unsigned int)
14069 vector signed char vec_sldw (vector signed char,
14070 vector signed char,
14072 vector unsigned char vec_sldw (vector unsigned char,
14073 vector unsigned char,
14075 vector signed short vec_sldw (vector signed short,
14076 vector signed short,
14078 vector unsigned short vec_sldw (vector unsigned short,
14079 vector unsigned short,
14081 vector signed int vec_sldw (vector signed int,
14084 vector unsigned int vec_sldw (vector unsigned int,
14085 vector unsigned int,
14087 vector signed long long vec_sldw (vector signed long long,
14088 vector signed long long,
14090 vector unsigned long long vec_sldw (vector unsigned long long,
14091 vector unsigned long long,
14093 * config/rs6000/rs6000-c: Add support for built-in functions
14094 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14095 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14096 * doc/extend.texi: Update the built-in documentation for the
14097 new built-in functions.
14099 2017-05-16 Marek Polacek <polacek@redhat.com>
14103 * tree.c (save_expr): Don't fold the expression.
14105 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14107 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14108 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14109 and (?*y,m). Update insn attributes.
14111 2017-05-16 Martin Liska <mliska@suse.cz>
14113 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14114 flags argument of print_gimple_stmt, print_gimple_expr,
14115 print_generic_stmt and print_generic_expr.
14116 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14117 * coretypes.h: Likewise.
14118 * except.c (dump_eh_tree): Likewise.
14119 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14120 * gimple-pretty-print.h: Likewise.
14121 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14122 (backprop::push_to_worklist): Likewise.
14123 (backprop::pop_from_worklist): Likewise.
14124 (backprop::process_use): Likewise.
14125 (backprop::intersect_uses): Likewise.
14126 (note_replacement): Likewise.
14127 * gimple-ssa-store-merging.c
14128 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14129 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14130 (pass_store_merging::execute): Likewise.
14131 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14132 (ssa_base_cand_dump_callback): Likewise.
14133 (dump_incr_vec): Likewise.
14134 (replace_refs): Likewise.
14135 (replace_mult_candidate): Likewise.
14136 (create_add_on_incoming_edge): Likewise.
14137 (create_phi_basis): Likewise.
14138 (insert_initializers): Likewise.
14139 (all_phi_incrs_profitable): Likewise.
14140 (introduce_cast_before_cand): Likewise.
14141 (replace_one_candidate): Likewise.
14142 * gimplify.c (gimplify_expr): Likewise.
14143 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14144 (set_rename): Likewise.
14145 (rename_uses): Likewise.
14146 (copy_loop_phi_nodes): Likewise.
14147 (add_close_phis_to_merge_points): Likewise.
14148 (copy_loop_close_phi_args): Likewise.
14149 (copy_cond_phi_args): Likewise.
14150 (graphite_copy_stmts_from_block): Likewise.
14151 (translate_pending_phi_nodes): Likewise.
14152 * graphite-poly.c (print_pdr): Likewise.
14153 (dump_gbb_cases): Likewise.
14154 (dump_gbb_conditions): Likewise.
14155 (print_scop_params): Likewise.
14156 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14157 (build_cross_bb_scalars_use): Likewise.
14158 (gather_bbs::before_dom_children): Likewise.
14159 * hsa-dump.c (dump_hsa_immed): Likewise.
14160 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14161 (get_replacement_map): Likewise.
14162 * ipa-inline-analysis.c (dump_condition): Likewise.
14163 (estimate_function_body_sizes): Likewise.
14164 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14165 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14166 * ipa-prop.c (ipa_dump_param): Likewise.
14167 (ipa_print_node_jump_functions_for_edge): Likewise.
14168 (ipa_modify_call_arguments): Likewise.
14169 (ipa_modify_expr): Likewise.
14170 (ipa_dump_param_adjustments): Likewise.
14171 (ipa_dump_agg_replacement_values): Likewise.
14172 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14173 * ipa-pure-const.c (check_stmt): Likewise.
14174 (pass_nothrow::execute): Likewise.
14175 * ipa-split.c (execute_split_functions): Likewise.
14176 * omp-offload.c (dump_oacc_loop_part): Likewise.
14177 (dump_oacc_loop): Likewise.
14178 * trans-mem.c (tm_log_emit): Likewise.
14179 (tm_memopt_accumulate_memops): Likewise.
14180 (dump_tm_memopt_set): Likewise.
14181 (dump_tm_memopt_transform): Likewise.
14182 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14183 (print_loop): Likewise.
14184 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14185 (chkp_gather_checks_info): Likewise.
14186 (chkp_get_check_result): Likewise.
14187 (chkp_remove_check_if_pass): Likewise.
14188 (chkp_use_outer_bounds_if_possible): Likewise.
14189 (chkp_reduce_bounds_lifetime): Likewise.
14190 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14191 (chkp_mark_completed_bounds): Likewise.
14192 (chkp_register_incomplete_bounds): Likewise.
14193 (chkp_mark_invalid_bounds): Likewise.
14194 (chkp_maybe_copy_and_register_bounds): Likewise.
14195 (chkp_build_returned_bound): Likewise.
14196 (chkp_get_bound_for_parm): Likewise.
14197 (chkp_build_bndldx): Likewise.
14198 (chkp_get_bounds_by_definition): Likewise.
14199 (chkp_generate_extern_var_bounds): Likewise.
14200 (chkp_get_bounds_for_decl_addr): Likewise.
14201 * tree-chrec.c (chrec_apply): Likewise.
14202 * tree-data-ref.c (dump_data_reference): Likewise.
14203 (dump_subscript): Likewise.
14204 (dump_data_dependence_relation): Likewise.
14205 (analyze_overlapping_iterations): Likewise.
14206 * tree-inline.c (expand_call_inline): Likewise.
14207 (tree_function_versioning): Likewise.
14208 * tree-into-ssa.c (dump_defs_stack): Likewise.
14209 (dump_currdefs): Likewise.
14210 (dump_names_replaced_by): Likewise.
14211 (dump_update_ssa): Likewise.
14212 (update_ssa): Likewise.
14213 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14214 * tree-parloops.c (build_new_reduction): Likewise.
14215 (try_create_reduction_list): Likewise.
14216 (ref_conflicts_with_region): Likewise.
14217 (oacc_entry_exit_ok_1): Likewise.
14218 (oacc_entry_exit_single_gang): Likewise.
14219 * tree-pretty-print.h: Likewise.
14220 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14221 (get_scalar_evolution): Likewise.
14222 (add_to_evolution): Likewise.
14223 (get_loop_exit_condition): Likewise.
14224 (analyze_evolution_in_loop): Likewise.
14225 (analyze_initial_condition): Likewise.
14226 (analyze_scalar_evolution): Likewise.
14227 (instantiate_scev): Likewise.
14228 (number_of_latch_executions): Likewise.
14229 (gather_chrec_stats): Likewise.
14230 (final_value_replacement_loop): Likewise.
14231 (scev_const_prop): Likewise.
14232 * tree-sra.c (dump_access): Likewise.
14233 (disqualify_candidate): Likewise.
14234 (create_access): Likewise.
14235 (reject): Likewise.
14236 (maybe_add_sra_candidate): Likewise.
14237 (create_access_replacement): Likewise.
14238 (analyze_access_subtree): Likewise.
14239 (analyze_all_variable_accesses): Likewise.
14240 (sra_modify_assign): Likewise.
14241 (initialize_constant_pool_replacements): Likewise.
14242 (find_param_candidates): Likewise.
14243 (decide_one_param_reduction): Likewise.
14244 (replace_removed_params_ssa_names): Likewise.
14245 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14246 * tree-ssa-copy.c (dump_copy_of): Likewise.
14247 (copy_prop_visit_cond_stmt): Likewise.
14248 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14249 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14250 (record_equivalences_from_stmt): Likewise.
14251 * tree-ssa-dse.c (compute_trims): Likewise.
14252 (delete_dead_call): Likewise.
14253 (delete_dead_assignment): Likewise.
14254 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14255 (forward_propagate_into_cond): Likewise.
14256 (pass_forwprop::execute): Likewise.
14257 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14258 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14260 (move_computations_worker): Likewise.
14261 (execute_sm): Likewise.
14262 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14263 (remove_exits_and_undefined_stmts): Likewise.
14264 (remove_redundant_iv_tests): Likewise.
14265 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14266 (adjust_iv_update_pos): Likewise.
14267 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14268 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14269 (value_replacement): Likewise.
14270 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14271 * tree-ssa-pre.c (print_pre_expr): Likewise.
14272 (get_representative_for): Likewise.
14273 (create_expression_by_pieces): Likewise.
14274 (insert_into_preds_of_block): Likewise.
14275 (eliminate_insert): Likewise.
14276 (eliminate_dom_walker::before_dom_children): Likewise.
14277 (eliminate): Likewise.
14278 (remove_dead_inserted_code): Likewise.
14279 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14280 * tree-ssa-reassoc.c (get_rank): Likewise.
14281 (eliminate_duplicate_pair): Likewise.
14282 (eliminate_plus_minus_pair): Likewise.
14283 (eliminate_not_pairs): Likewise.
14284 (undistribute_ops_list): Likewise.
14285 (eliminate_redundant_comparison): Likewise.
14286 (update_range_test): Likewise.
14287 (optimize_range_tests_var_bound): Likewise.
14288 (optimize_vec_cond_expr): Likewise.
14289 (rewrite_expr_tree): Likewise.
14290 (rewrite_expr_tree_parallel): Likewise.
14291 (linearize_expr): Likewise.
14292 (break_up_subtract): Likewise.
14293 (linearize_expr_tree): Likewise.
14294 (attempt_builtin_powi): Likewise.
14295 (attempt_builtin_copysign): Likewise.
14296 (transform_stmt_to_copy): Likewise.
14297 (transform_stmt_to_multiply): Likewise.
14298 (dump_ops_vector): Likewise.
14299 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14300 (print_scc): Likewise.
14301 (set_ssa_val_to): Likewise.
14302 (visit_reference_op_store): Likewise.
14303 (visit_use): Likewise.
14304 (sccvn_dom_walker::before_dom_children): Likewise.
14305 (run_scc_vn): Likewise.
14306 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14308 (expr_hash_elt::print): Likewise.
14309 (const_and_copies::pop_to_marker): Likewise.
14310 (const_and_copies::record_const_or_copy_raw): Likewise.
14311 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14312 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14313 (dump_predicates): Likewise.
14314 (find_uninit_use): Likewise.
14315 (warn_uninitialized_phi): Likewise.
14316 (pass_late_warn_uninitialized::execute): Likewise.
14317 * tree-ssa.c (verify_vssa): Likewise.
14318 (verify_ssa): Likewise.
14319 (maybe_optimize_var): Likewise.
14320 * tree-vrp.c (dump_value_range): Likewise.
14321 (dump_all_value_ranges): Likewise.
14322 (dump_asserts_for): Likewise.
14323 (register_edge_assert_for_2): Likewise.
14324 (vrp_visit_cond_stmt): Likewise.
14325 (vrp_visit_switch_stmt): Likewise.
14326 (vrp_visit_stmt): Likewise.
14327 (vrp_visit_phi_node): Likewise.
14328 (simplify_cond_using_ranges_1): Likewise.
14329 (fold_predicate_in): Likewise.
14330 (evrp_dom_walker::before_dom_children): Likewise.
14331 (evrp_dom_walker::push_value_range): Likewise.
14332 (evrp_dom_walker::pop_value_range): Likewise.
14333 (execute_early_vrp): Likewise.
14335 2017-05-16 Richard Biener <rguenther@suse.de>
14337 * dwarf2out.c (loc_list_from_tree_1): Do not create
14338 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14340 2017-05-16 Richard Biener <rguenther@suse.de>
14342 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14344 (note_variable_value_in_expr): If we resolved the decl ref
14345 do not push to the stack.
14347 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14349 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14350 operations in fast-math mode.
14351 (vaddq_f16): Likewise.
14352 (vmul_f16): Likewise.
14353 (vmulq_f16): Likewise.
14354 (vsub_f16): Likewise.
14355 (vsubq_f16): Likewise.
14356 * config/arm/neon.md (add<mode>3): New.
14358 (fma:<VH:mode>3): New. Also remove outdated comment.
14361 2017-05-16 Martin Liska <mliska@suse.cz>
14365 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14366 (odr_types_equivalent_p): Likewise.
14368 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14370 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14372 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14375 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14376 non-interunit SSE move alternatives with '?'.
14377 (zero-extendsidi peephole2): New peephole to skip intermediate
14378 general register in SSE zero-extend sequence.
14380 2017-05-15 Jeff Law <law@redhat.com>
14382 * reorg.c (relax_delay_slots): Create a new variable to hold
14383 the temporary target rather than clobbering TARGET_LABEL.
14385 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14386 missing argument to extract_bit_field call.
14387 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14389 2017-05-15 Martin Liska <mliska@suse.cz>
14392 * gcc.c (process_command): Do not allow empty argument of -o option.
14394 2017-05-15 Renlin Li <renlin.li@arm.com>
14396 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14397 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14398 * config/aarch64/constraints.md (Usf): Add long call check.
14399 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14400 (call_value): Likewise.
14401 (sibcall): Likewise.
14402 (sibcall_value): Likewise.
14404 (call_value_insn): New.
14405 (sibcall_insn): Update rtx pattern.
14406 (sibcall_value_insn): Likewise.
14407 (call_internal): Remove.
14408 (call_value_internal): Likewise.
14409 (sibcall_internal): Likewise.
14410 (sibcall_value_internal): Likewise.
14411 (call_reg): Likewise.
14412 (call_symbol): Likewise.
14413 (call_value_reg): Likewise.
14414 (call_value_symbol): Likewise.
14416 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14419 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14421 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14423 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14424 compatible with CCGOCmode and with CCZmode.
14426 2017-05-14 Martin Sebor <msebor@redhat.com>
14428 PR middle-end/77671
14429 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14430 (gimple_fold_builtin_snprintf): Same.
14431 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14432 (gimple_fold_builtin_snprintf): Same.
14433 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14434 of character types.
14435 (is_call_safe): New function.
14436 (try_substitute_return_value): Call it.
14437 (try_simplify_call): New function.
14438 (pass_sprintf_length::handle_gimple_call): Call it.
14440 2017-05-14 Martin Sebor <msebor@redhat.com>
14442 PR middle-end/80669
14443 * builtins.c (expand_builtin_stpncpy): Simplify.
14445 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14447 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14448 * config/i386/i386.h
14449 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14450 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14451 (struct machine_function): Add new members call_ms2sysv,
14452 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14453 (struct machine_frame_state): New fields sp_realigned and
14454 sp_realigned_offset.
14455 * config/i386/i386.c
14456 (enum xlogue_stub): New enum.
14457 (enum xlogue_stub_sets): New enum.
14458 (class xlogue_layout): New class.
14459 (struct ix86_frame): New fields stack_realign_allocate_offset,
14460 stack_realign_offset and outlined_save_offset. Modify comments to
14461 detail stack layout when using out-of-line stubs.
14462 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14463 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14464 -mcall-ms2sysv-xlogues.
14465 (stub_managed_regs): New static variable.
14466 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14467 registers managed by out-of-line stub.
14468 (disable_call_ms2sysv_xlogues): New function.
14469 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14470 m->call_ms2sysv when appropriate and compute frame layout for
14472 (sp_valid_at, fp_valid_at): New inline functions.
14473 (choose_basereg): New function.
14474 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14476 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14477 Use align parameter of choose_baseaddr to generated aligned SSE movs
14479 (pro_epilogue_adjust_stack): Modify to track
14480 machine_frame_state::sp_realigned.
14481 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14482 (ix86_nsaved_sseregs): Likewise.
14483 (ix86_emit_save_regs): Likewise.
14484 (ix86_emit_save_regs_using_mov): Likewise.
14485 (ix86_emit_save_sse_regs_using_mov): Likewise.
14486 (get_scratch_register_on_entry): Likewise.
14487 (gen_frame_set): New function.
14488 (gen_frame_load): Likewise.
14489 (gen_frame_store): Likewise.
14490 (emit_outlined_ms2sysv_save): Likewise.
14491 (emit_outlined_ms2sysv_restore): Likewise.
14492 (ix86_expand_prologue): Modify stack re-alignment code and call
14493 emit_outlined_ms2sysv_save when appropriate.
14494 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14495 parameter rtx_insn *insn, which allows the function to be used to only
14496 generate the notes.
14497 (ix86_expand_epilogue): Modify validity checks of frame and stack
14498 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14499 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14500 * config/i386/predicates.md
14501 (save_multiple): New predicate.
14502 (restore_multiple): Likewise.
14503 * config/i386/sse.md
14504 (save_multiple<mode>): New pattern.
14505 (save_multiple_realign<mode>): Likewise.
14506 (restore_multiple<mode>): Likewise.
14507 (restore_multiple_and_return<mode>): Likewise.
14508 (restore_multiple_leave_return<mode>): Likewise.
14509 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14511 2017-05-14 Julia Koval <julia.koval@intel.com>
14513 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14514 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14515 (__builtin_ia32_xsetbv): New builtins.
14516 * config/i386/i386.c (ix86_expand_special_args_builtin):
14518 (ix86_expand_builtin): Special expand for new intrinsics.
14519 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14520 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14521 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14523 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14525 * cfganal.c (inverted_post_order_compute): Change argument type
14527 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14528 * df-core.c (rest_of_handle_df_initialize): Adjust.
14529 (rest_of_handle_df_finish): Likewise.
14530 (df_analyze_1): Likewise.
14531 (df_analyze): Likewise.
14532 (loop_inverted_post_order_compute): Change argument to be a vec *.
14533 (df_analyze_loop): Adjust.
14534 (df_get_n_blocks): Likewise.
14535 (df_get_postorder): Likewise.
14536 * df.h (struct df_d): Change field to be a vec.
14537 * lcm.c (compute_laterin): Adjust.
14538 (compute_available): Likewise.
14539 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14540 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14541 * tree-ssa-pre.c (compute_antic): Likewise.
14543 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14545 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14546 (depth_first_search::depth_first_search): Change structure init
14547 function to this constructor.
14548 (depth_first_search::add_bb): Rename function to this member.
14549 (depth_first_search::execute): Likewise.
14550 (flow_dfs_compute_reverse_finish): Adjust.
14552 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14554 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14555 (longest_simple_path): Likewise.
14556 * shrink-wrap.c (spread_components): Likewise.
14557 (disqualify_problematic_components): Likewise.
14558 (emit_common_heads_for_components): Likewise.
14559 (emit_common_tails_for_components): Likewise.
14560 (insert_prologue_epilogue_for_components): Likewise.
14562 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14564 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14567 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14569 * df-core.c (df_set_blocks): Start using auto_bitmap.
14570 (df_compact_blocks): Likewise.
14571 * df-problems.c (df_rd_confluence_n): Likewise.
14572 * df-scan.c (df_insn_rescan_all): Likewise.
14573 (df_process_deferred_rescans): Likewise.
14574 (df_update_entry_block_defs): Likewise.
14575 (df_update_exit_block_uses): Likewise.
14576 (df_entry_block_bitmap_verify): Likewise.
14577 (df_exit_block_bitmap_verify): Likewise.
14578 (df_scan_verify): Likewise.
14579 * lra-constraints.c (lra_constraints): Likewise.
14580 (undo_optional_reloads): Likewise.
14581 (lra_undo_inheritance): Likewise.
14582 * lra-remat.c (calculate_gen_cands): Likewise.
14583 (do_remat): Likewise.
14584 * lra-spills.c (assign_spill_hard_regs): Likewise.
14585 (spill_pseudos): Likewise.
14586 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14587 (bitmap_set_subtract_values): Likewise.
14589 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14591 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14592 management with auto_bitmap.
14593 (fix_inter_tick): Likewise.
14594 (fix_recovery_deps): Likewise.
14595 * ira.c (add_store_equivs): Likewise.
14596 (find_moveable_pseudos): Likewise.
14597 (split_live_ranges_for_shrink_wrap): Likewise.
14598 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14599 (rtx_reuse_manager::seen_def_p): Likewise.
14600 (rtx_reuse_manager::set_seen_def): Likewise.
14601 * print-rtl.h (class rtx_reuse_manager): Likewise.
14603 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14605 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14607 (migrate_btr_def): Likewise.
14608 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14609 * df-core.c (loop_post_order_compute): Likewise.
14610 (loop_inverted_post_order_compute): Likewise.
14611 * hsa-common.h: Likewise.
14612 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14613 * init-regs.c (initialize_uninitialized_regs): Likewise.
14614 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14615 (inline_small_functions): Likewise.
14616 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14617 * ira.c (combine_and_move_insns): Likewise.
14618 (build_insn_chain): Likewise.
14619 * loop-invariant.c (find_invariants): Likewise.
14620 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14621 * predict.c (tree_predict_by_opcode): Likewise.
14622 (predict_paths_leading_to): Likewise.
14623 (predict_paths_leading_to_edge): Likewise.
14624 (estimate_loops_at_level): Likewise.
14625 (estimate_loops): Likewise.
14626 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14627 (spread_components): Likewise.
14628 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14629 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14630 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14631 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14632 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14633 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14634 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14635 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14636 (mark_threaded_blocks): Likewise.
14637 (thread_through_all_blocks): Likewise.
14638 * tree-ssa.c (verify_ssa): Likewise.
14639 (execute_update_addresses_taken): Likewise.
14640 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14642 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14644 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14646 (post_order_compute): Likewise.
14647 (inverted_post_order_compute): Likewise.
14648 (pre_and_rev_post_order_compute_fn): Likewise.
14650 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14652 * genrecog.c (int_set::int_set): Explicitly construct our
14653 auto_vec base class.
14654 * vec.h (auto_vec::auto_vec): New constructor.
14656 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14658 * bitmap.h (class auto_bitmap): New constructor taking
14659 bitmap_obstack * argument.
14661 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14663 * bitmap.h (class auto_bitmap): Change type of m_bits to
14664 bitmap_head, and adjust ctor / dtor and member operators.
14666 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14668 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14669 when returned register mode doesn't match original mode.
14671 2017-05-12 Jeff Law <law@redhat.com>
14672 Jakub Jelinek <jakub@redhat.com>
14674 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14675 we look for cc setter after the compare-elim changes.
14676 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14677 within the vector to match what compare-elim now expects.
14678 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14679 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14681 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14682 after the compare-elim changes.
14683 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14684 the vector to match what compare-elim now expects.
14685 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14686 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14687 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14688 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14689 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14691 * config/visium/visium.c (single_set_and_flags): Fix where
14692 we look for cc setter after the compare-elim changes.
14693 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14694 with the vector to match what compare-elim now expects.
14695 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14696 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14697 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14698 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14699 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14700 (neg<mode>2_insn_set_overflow): Likewise.
14702 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14704 PR middle-end/79794
14705 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14706 maybe_expand_insn call, set ops[0].target. If still set after call,
14707 set alt_rtl. Add extra arg to recursive calls.
14708 (extract_bit_field): Add alt_rtl argument. Pass to
14710 * expmed.h (extract_bit_field): Fix prototype.
14711 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14712 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14713 to extract_bit_field_calls.
14714 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
14715 Pass alt_rtl to extract_bit_field calls.
14716 * calls.c (store_unaligned_arguments_into_psuedos)
14717 load_register_parameters): Pass extra NULL to extract_bit_field calls.
14718 * optabs.c (maybe_legitimize_operand): Clear op->target when call
14720 * optabs.h (struct expand_operand): Add target bitfield.
14722 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14724 * compare-elim.c (try_eliminate_compare): Canonicalize
14725 operation with embedded compare to
14726 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
14727 (set (reg) (operation)].
14729 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
14731 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14734 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
14735 cost of adding a carry flag for ADC instruction.
14736 [case MINUS]: Ignore the cost of subtracting a carry flag
14737 for SBB instruction.
14739 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
14741 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
14743 * config/rs6000/bmiintrin.h: New file.
14744 * config/rs6000/bmi2intrin.h: New file.
14745 * config/rs6000/x86intrin.h: New file.
14747 2017-05-12 Jeff Law <law@redhat.com>
14749 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14752 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14754 PR middle-end/80707
14755 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14757 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14759 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14760 early expansion of vector divide builtins.
14761 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14762 builtins identified as having unsigned arguments.
14764 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14766 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14767 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14768 expansion of vector logical operations (and, andc, or, xor,
14771 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14773 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14774 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14776 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14778 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14779 early GIMPLE expansion of vector multiplies.
14781 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14783 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14784 TARGET_HAVE_MOVT conditional.
14785 (movt splitter): Likewise.
14787 2017-05-12 Richard Biener <rguenther@suse.de>
14789 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14790 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14791 Fold all stmts not inplace.
14793 2017-05-12 Richard Biener <rguenther@suse.de>
14795 PR tree-optimization/80713
14796 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14797 inserted_exprs bit for not removed stmts.
14799 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14801 PR middle-end/69921
14802 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14803 parallelized" attribute for parallelized OpenACC kernels.
14804 * omp-offload.c (execute_oacc_device_lower): Use it.
14806 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14807 Set "oacc kernels" attribute.
14808 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14809 parameter. Adjust all users.
14810 (oacc_fn_attrib_kernels_p): Remove function.
14811 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14812 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14813 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14814 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14815 assert "oacc kernels" attribute is set.
14817 2017-05-11 Carl Love <cel@us.ibm.com>
14819 * config/rs6000/rs6000-c: Add support for built-in functions
14820 vector unsigned char vec_popcnt (vector signed char)
14821 vector unsigned char vec_popcnt (vector unsigned char)
14822 vector unsigned short vec_popcnt (vector signed short)
14823 vector unsigned short vec_popcnt (vector unsigned short)
14824 vector unsigned int vec_popcnt (vector signed int)
14825 vector unsigned int vec_popcnt (vector unsigned int)
14826 vector unsigned long long vec_popcnt (vector signed long long)
14827 vector unsigned long long vec_popcnt (vector unsigned long long)
14828 vector signed long long vec_slo (vector signed long long,
14829 vector signed char)
14830 vector signed long long vec_slo (vector signed long long,
14831 vector unsigned char)
14832 vector unsigned long long vec_slo (vector unsigned long long,
14833 vector signed char)
14834 vector unsigned long long vec_slo (vector unsigned long long,
14835 vector unsigned char)
14836 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14837 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14838 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14839 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14840 * doc/extend.texi: Update the built-in documentation file for the
14841 new built-in functions.
14843 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14845 * attribs.h (sorted_attr_string): Move machine independent
14846 functions for target clone support from the i386 port to common
14847 code. Rename ix86_function_versions to common_function_versions.
14848 Rename make_name to make_unique_name.
14849 (common_function_versions): Likewise.
14850 (make_unique_name): Likewise.
14851 (make_dispatcher_decl): Likewise.
14852 (is_function_default_version): Likewise.
14853 * attribs.c (attr_strcmp): Likewise.
14854 (sorted_attr_string): Likewise.
14855 (common_function_versions): Likewise.
14856 (make_unique_name): Likewise.
14857 (make_dispatcher_decl): Likewise.
14858 (is_function_default_version): Likewise.
14859 * config/i386/i386.c (attr_strcmp): Likewise.
14860 (sorted_attr_string): Likewise.
14861 (ix86_function_versions): Likewise.
14862 (make_name): Likewise.
14863 (make_dispatcher_decl): Likewise.
14864 (is_function_default_version): Likewise.
14865 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14867 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14870 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14871 Account for direct move costs for vec_construct of integer
14874 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14877 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14878 (UNSPEC_STX_ATOMIC): Ditto.
14879 (loaddi_via_sse): New insn.
14880 (storedi_via_sse): Ditto.
14881 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14882 Update corresponding peephole2 patterns.
14883 (atomic_storedi_fpu): Ditto.
14885 2017-05-11 Julia Koval <julia.koval@intel.com>
14887 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14888 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14890 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14891 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14892 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14894 2017-05-11 Nathan Sidwell <nathan@acm.org>
14896 * graphite-poly.c: Include dumpfile.h.
14898 * dumpfle.h (dump_function): Declare here ...
14899 * tree-dump.h (dump_function): ... not here.
14900 * dumpfile.c: #include tree-cfg.h.
14901 (dump_function): Move here from ...
14902 * tree-dump.c (dump_function): ... here.
14903 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14904 * graphite-poly.c: Don't include tree-dump.h.
14905 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14906 * print-tree.c: Likewise.
14907 * stor-layout.c: Likewise.
14908 * tree-nested.c: Likewise.
14910 * dumpfile.c (dump_start): Use TDF_FLAGS.
14911 (dump_enable_all): Fix TDF_KIND check thinko.
14913 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14915 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14916 array entries to represent two legal parameterizations of the
14917 overloaded __builtin_cmpb function, as represented by the
14918 P6_OV_BUILTIN_CMPB constant.
14919 (altivec_resolve_overloaded_builtin): Add special case handling
14920 for the __builtin_cmpb function, as represented by the
14921 P6_OV_BUILTIN_CMPB constant.
14922 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
14923 (BU_P6_64BIT_2): New macro.
14924 (BU_P6_OVERLOAD_2): New macro
14925 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
14926 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
14927 (CMPB): Add overload support to represent both 32-bit and 64-bit
14928 compare-bytes function.
14929 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14930 support for TARGET_CMPB.
14931 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
14932 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14933 documentation of the __builtin_cmpb overloaded built-in function.
14935 2017-05-11 Richard Biener <rguenther@suse.de>
14937 PR tree-optimization/80705
14938 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
14939 bases are not vectorizable.
14941 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14943 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
14944 when counting register pressure.
14946 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14948 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
14949 (iv_ca_more_deps): Renamed to ...
14950 (iv_ca_compare_deps): ... this.
14951 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
14953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14955 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
14957 (determine_group_iv_costs): ... here.
14958 (find_inv_vars_cb): Record inv var if it's not recorded before.
14960 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14962 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
14963 (get_shiftadd_cost): Ditto.
14965 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14967 * tree-ssa-address.c: Include header file.
14968 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
14970 (add_to_parts): Refactor.
14971 (addr_to_parts): New parameter. Update use of move_hint_to_base.
14972 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
14975 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14977 PR tree-optimization/53090
14978 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
14980 (extract_cond_operands): Detect condition with IV on both sides
14981 and return COMP_IV_EXPR_2.
14982 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
14983 (rewrite_use_compare): Simplify by removing call to function
14984 extract_cond_operands.
14986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14988 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
14989 (extract_cond_operands): Detect condition comparing against non-
14990 invariant bound and return appropriate enum value.
14991 (find_interesting_uses_cond): Update use of extract_cond_operands.
14992 Handle its return value accordingly.
14993 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
14995 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14997 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
14998 nonlinear iv_use computation in loop invariant sensitive way.
15000 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15002 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15003 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15005 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15007 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15008 (dump_cand): Support iv_cand.inv_exprs.
15009 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15011 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15014 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15016 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15018 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15019 as local function. Include necessary header files.
15020 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15024 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15026 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15028 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15029 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15030 RSHIFT_EXPR and BIT_NOT_EXPR.
15032 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15034 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15035 (adjust_setup_cost): New parameter supporting round up adjustment.
15036 (struct address_cost_data): Delete.
15037 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15038 (split_address_cost, ptr_difference_cost): Delete.
15039 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15040 (struct ainc_cost_data): New struct.
15041 (get_address_cost_ainc): New function.
15042 (get_address_cost, get_computation_cost): Reimplement.
15043 (determine_group_iv_cost_address): Record inv_expr for all uses of
15045 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15046 (iv_ca_has_deps): Reimplemented to ...
15047 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15049 (iv_ca_extend): Call iv_ca_more_deps.
15051 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15053 * tree-ssa-address.c (struct mem_address): Move to header file.
15054 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15055 * tree-ssa-address.h (struct mem_address): Move from C file.
15056 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15060 * tree-affine.h (aff_combination_type): New interface.
15061 (aff_combination_zero_p): Remove static.
15062 (aff_combination_const_p): New interface.
15063 (aff_combination_singleton_var_p): New interfaces.
15065 2017-05-11 Richard Biener <rguenther@suse.de>
15067 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15068 Skip unreachable blocks and destinations.
15069 (eliminate): Move stmt removal and fixup ...
15070 (fini_eliminate): ... here. Skip inserted exprs.
15071 (pass_pre::execute): Move fini_pre after fini_eliminate.
15072 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15073 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15074 PRE to get rid of dead code that has invalid SSA form and
15075 split critical edges again.
15077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15079 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15081 2017-05-11 Richard Biener <rguenther@suse.de>
15083 * passes.c (execute_function_todo): Verify loops if they are
15084 said to be up-to-date.
15085 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15086 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15088 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15091 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15092 handle calling assemble_external ourself.
15095 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15096 modes with zero size. Enhance comment.
15098 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15100 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15101 built-ins for vec_xl and vec_xst with short and char pointer
15104 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15106 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15107 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15108 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15109 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15110 (_mm_maskz_min_round_ss): New intrinsics.
15111 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15112 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15113 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15114 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15115 (__builtin_ia32_minss_mask_round): New builtins.
15116 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15117 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15118 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15120 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15121 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15123 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15126 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15128 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15129 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15130 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15131 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15132 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15133 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15134 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15135 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15136 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15137 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15138 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15139 (__builtin_ia32_mulss_mask_round): New builtins.
15140 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15141 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15142 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15144 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15145 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15147 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15150 2017-05-10 Julia Koval <julia.koval@intel.com>
15152 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15153 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15154 (_mm256_setr_m128i): New intrinsics.
15156 2017-05-10 Julia Koval <julia.koval@intel.com>
15158 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15159 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15160 (_mm_maskz_rcp14_ss): New intrinsics.
15161 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15162 (__builtin_ia32_rcp14ss_mask): New builtins.
15163 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15165 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15167 PR tree-optimization/51513
15168 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15169 (assert_unreachable_fallthru_edge_p): Use it.
15170 (group_case_labels_stmt): Likewise.
15171 * tree-cfg.h: Prototype it.
15172 * stmt.c: Include cfghooks.h and tree-cfg.h.
15173 (emit_case_dispatch_table) <gap_label>: New local variable.
15174 Use it to fill dispatch table gaps.
15175 Test for default_label before updating probabilities.
15176 (expand_case) <default_label>: Remove unneeded initialization.
15177 Test for unreachable default case statement and remove its edge.
15178 Set default_label accordingly.
15179 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15181 2017-05-10 Carl Love <cel@us.ibm.com>
15183 * config/rs6000/rs6000-c: Add support for built-in functions
15184 vector signed char vec_neg (vector signed char)
15185 vector signed short int vec_neg (vector short int)
15186 vector signed int vec_neg (vector signed int)
15187 vector signed long long vec_neg (vector signed long long)
15188 vector float vec_neg (vector float)
15189 vector double vec_neg (vector double)
15190 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15192 * config/rs6000/altivec.h: Add define for vec_neg
15193 * doc/extend.texi: Update the built-in documentation for the
15194 new built-in functions.
15196 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15198 PR tree-optimization/77644
15199 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15201 2017-05-10 Nathan Sidwell <nathan@acm.org>
15203 * dumpfile.h (TDI_lang_all): New.
15204 (TDF_KIND): New. Renumber others
15205 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15207 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15209 (get_dump_file_name): Adjust suffix generation.
15210 (dump_enable_all): Use TDF_KIND.
15211 * doc/invoke.texi (-fdump-lang-all): Document.
15213 * dumpfile.h: Tabify.
15215 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15218 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15219 Move member access before delete.
15221 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15223 * tree-inline.c (expand_call_inline): Split block at stmt
15226 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15229 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15230 are now unused after splitting mov{sf,sd}_hardfloat.
15231 (f32_lr2): Likewise.
15232 (f32_lm): Likewise.
15233 (f32_lm2): Likewise.
15234 (f32_li): Likewise.
15235 (f32_li2): Likewise.
15236 (f32_lv): Likewise.
15237 (f32_sr): Likewise.
15238 (f32_sr2): Likewise.
15239 (f32_sm): Likewise.
15240 (f32_sm2): Likewise.
15241 (f32_si): Likewise.
15242 (f32_si2): Likewise.
15243 (f32_sv): Likewise.
15244 (f32_dm): Likewise.
15245 (f32_vsx): Likewise.
15246 (f32_av): Likewise.
15247 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15248 For movsf, order stores so the VSX stores occur before the GPR
15249 store which encourages the register allocator to use a traditional
15250 FPR instead of a GPR. For movsd, order the stores so that the GPR
15251 store comes before the VSX stores to allow the power6 to work.
15252 This is due to the power6 not having a 32-bit integer store
15253 instruction from a FPR.
15254 (movsf_hardfloat): Likewise.
15255 (movsd_hardfloat): Likewise.
15257 2017-05-09 Martin Sebor <msebor@redhat.com>
15259 PR translation/80280
15260 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15263 PR translation/80280
15264 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15265 data member added in r247778.
15266 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15268 2017-05-09 Nathan Sidwell <nathan@acm.org>
15270 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15272 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15275 2017-05-09 Marek Polacek <polacek@redhat.com>
15277 * doc/invoke.texi: Fix typo.
15279 2017-05-09 Richard Biener <rguenther@suse.de>
15281 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15282 (vrp_val_is_min): Likewise.
15283 (set_value_range_to_value): Likewise.
15284 (set_value_range_to_nonnegative): Likewise.
15285 (gimple_assign_nonzero_p): Likewise.
15286 (gimple_stmt_nonzero_p): Likewise.
15287 (vrp_int_const_binop): Likewise. Remove unreachable case.
15288 (adjust_range_with_scev): Adjust comments.
15289 (compare_range_with_value): Likewise.
15290 (extract_range_from_phi_node): Likewise.
15291 (test_for_singularity): Likewise.
15293 2017-05-09 Richard Biener <rguenther@suse.de>
15295 * tree-vrp.c (get_single_symbol): Add assert that we don't
15296 get overflowed constants as invariant part.
15297 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15298 checks. Use wi::cmp instead of recursing for integer constants.
15299 (compare_values): Just ignore whether we assumed undefined
15300 overflow instead of failing the compare.
15301 (extract_range_for_var_from_comparison_expr): Add comment before the
15302 TREE_NO_WARNING sets.
15303 (test_for_singularity): Likewise.
15304 (extract_range_from_comparison): Do not disable optimization
15305 when we assumed undefined overflow.
15306 (extract_range_basic): Remove init of unused var.
15308 2017-05-09 Richard Biener <rguenther@suse.de>
15310 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15311 (extract_range_from_multiplicative_op_1): Adjust.
15312 (extract_range_from_binary_expr_1): Use int_const_binop.
15314 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15317 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15318 rs6000_store_data_bypass_p in seven define_bypass directives and
15319 in several comments.
15320 * config/rs6000/rs6000-protos.h: Add prototype for
15321 rs6000_store_data_bypass_p function.
15322 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15323 function implements slightly different (rs6000-specific) semantics
15324 than store_data_bypass_p, returning false rather than aborting
15325 with assertion error when arguments do not satisfy the
15326 requirements of store data bypass.
15327 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15328 rs6000_store_data_bypass_p.
15330 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15332 * config/xtensa/xtensa-protos.h
15333 (xtensa_initial_elimination_offset): New declaration.
15334 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15335 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15336 macro definition, add case for FRAME_POINTER_REGNUM when
15337 FRAME_GROWS_DOWNWARD.
15338 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15339 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15340 xtensa_initial_elimination_offset.
15342 2017-05-08 Nathan Sidwell <nathan@acm.org>
15344 * doc/invoke.texi: Alphabetize -fdump options.
15346 2017-05-08 Martin Sebor <msebor@redhat.com>
15348 PR translation/80280
15349 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15351 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15353 * target.def (compute_frame_layout): New optional target hook.
15354 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15355 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15356 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15358 * reload1.c (verify_initial_elim_offsets): Likewise.
15359 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15360 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15361 (arm_get_frame_offsets): Split up into this ...
15362 (arm_compute_frame_layout): ... and this function.
15364 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15366 * config/aarch64/constraints.md (Usa): New constraint.
15367 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15369 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15371 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15372 with_multilib_list after it has been checked.
15374 2017-05-08 Richard Biener <rguenther@suse.de>
15376 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15377 (bitmap_set_subtract_values): Likewise.
15379 2017-05-08 Richard Biener <rguenther@suse.de>
15381 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15382 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15384 (gimple_stmt_nonzero_warnv_p): Rename to ...
15385 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15387 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15388 (extract_range_basic): Adjust, do not disable propagation on
15389 strict overflow sensitive simplification.
15390 (vrp_visit_cond_stmt): Likewise.
15392 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15394 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15395 body size unconditionally.
15397 2017-05-07 Jeff Law <law@redhat.com>
15400 2017-05-06 Jeff Law <law@redhat.com>
15401 PR tree-optimization/78496
15402 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15405 PR tree-optimization/78496
15406 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15407 (simplify_stmt_using_ranges): Call it.
15408 (vrp_dom_walker::before_dom_children): Extract equivalences
15409 from an ASSERT_EXPR with an equality comparison against a
15412 2017-05-06 Jeff Law <law@redhat.com>
15414 PR tree-optimization/78496
15415 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15418 PR tree-optimization/78496
15419 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15420 (simplify_stmt_using_ranges): Call it.
15421 (vrp_dom_walker::before_dom_children): Extract equivalences
15422 from an ASSERT_EXPR with an equality comparison against a
15425 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15427 * lra-constraints.c (lra_copy_reg_equiv): New function.
15428 (split_reg): Use it to copy equivalence information from the
15429 original register to the spill register.
15431 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15433 PR rtl-optimization/75964
15434 * simplify-rtx.c (simplify_const_relational_operation): Remove
15435 invalid handling of comparisons of integer ABS.
15437 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15439 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15440 initialize to zero.
15441 (init_regs): Remove declaration.
15442 (function_arg_advance_32): Initialize error_p as boolean variable.
15444 2017-05-05 Nathan Sidwell <nathan@acm.org>
15446 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15447 lines. Use for (;;).
15449 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15451 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15452 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15453 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15454 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15455 VF=2 that require versioning.
15457 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15459 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15462 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15464 * diagnostic.h (diagnostic_override_option_index): Convert from
15465 macro to inline function.
15467 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15469 * diagnostic.c (last_module_changed_p): New function.
15470 (set_last_module): New function.
15471 (diagnostic_report_current_module): Convert macro usage to
15472 the above functions.
15473 * diagnostic.h (diagnostic_context::last_module): Strengthen
15474 from const line_map * to const line_map_ordinary *.
15475 (diagnostic_last_module_changed): Delete macro.
15476 (diagnostic_set_last_module): Delete macro.
15478 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15480 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15481 with diagnostic_report_diagnostic.
15482 (diagnostic_n_impl_richloc): Likewise.
15483 * diagnostic.h (report_diagnostic): Delete macro.
15484 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15485 with diagnostic_report_diagnostic.
15486 * substring-locations.c (format_warning_va): Likewise.
15488 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15490 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15491 save/restor of format_spec. Move option-printing code to...
15492 (print_option_information): ...this new function, and
15493 reimplement by simply printing to the pretty_printer,
15494 rather than appending to the format string.
15496 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15498 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15499 handling logic into...
15500 (update_effective_level_from_pragmas): ...this new function.
15502 2017-05-04 Andrew Waterman <andrew@sifive.com>
15504 * config/riscv/riscv.opt (mstrict-align): New option.
15505 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15506 (SLOW_UNALIGNED_ACCESS): Define.
15507 (riscv_slow_unaligned_access): Declare.
15508 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15510 (riscv_slow_unaligned_access): New variable.
15511 (rocket_tune_info): Set slow_unaligned_access to true.
15512 (optimize_size_tune_info): Set slow_unaligned_access to false.
15513 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15514 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15515 (riscv_option_override): Set riscv_slow_unaligned_access.
15516 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15518 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15520 * config/riscv/riscv.md: Unify indentation.
15522 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15527 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15529 (extendsi<mode>2): Add support for doing sign extension via
15530 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15531 don't have ISA 3.0 instructions.
15532 (extendsi<mode>2 splitter): Likewise.
15533 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15534 generate the normal insns since SImode can now go in vector
15535 registers. Disallow the special UNSPECs needed for previous
15536 machines to hide SImode being used. Add new insns
15537 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15538 (fix_trunc<mode>si2_stfiwx): Likewise.
15539 (fix_trunc<mode>si2_internal): Likewise.
15540 (fixuns_trunc<mode>si2): Likewise.
15541 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15542 (fctiw<u>z_<mode>_smallint): Likewise.
15543 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15544 of floating point to 32-bit integer from doing a direct move to
15545 the GPR registers to do a store.
15546 (fctiwz_<mode>): Break long line.
15548 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15550 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15551 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15552 (addr_list, addr_offset_valid_p): New.
15553 (split_address_groups): Check offset validity with above function.
15554 (gt-tree-ssa-loop-ivopts.h): Include header file.
15556 2017-05-05 Nathan Sidwell <nathan@acm.org>
15558 * config.gcc (arm*-*-*): Add missing 'fi'.
15560 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15562 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15563 included in -fopt-info does not matter.
15564 * doc/optinfo.texi (-fopt-info): Fix description of default
15565 behavour. Explicitly say order of options included in -fopt-info
15568 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15570 * config.gcc: Allow combinations of aprofile and rmprofile values for
15571 --with-multilib-list.
15572 * config/arm/t-multilib: New file.
15573 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15574 variables. Remove setting of ISA and floating-point ABI in
15575 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15576 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15577 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15578 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15580 * config/arm/t-rmprofile: Likewise except for the matches changes.
15581 * doc/install.texi (--with-multilib-list): Document the combination of
15582 aprofile and rmprofile values and warn about pitfalls in doing that.
15584 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15586 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15587 (movdi_aarch64): Likewise.
15589 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15591 PR tree-optimization/80632
15592 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15594 (build_arrays): Initialize it for virtual phis.
15595 (fix_phi_nodes): Use it for virtual phis.
15597 PR tree-optimization/80558
15598 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15599 [x, y] op z into [x op, y op z] for op & or | if conditions
15602 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15603 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15606 * config/arm/arm.md (use_literal_pool): Remove.
15607 (64-bit immediate split): No longer takes cost into consideration
15608 if arm_disable_literal_pool is enabled.
15609 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15610 used when arm_disable_literal_pool is enabled.
15611 (arm_max_const_double_inline_cost): Remove use of
15612 arm_disable_literal_pool.
15613 (push_minipool_fix): Add assert.
15614 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15615 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15616 (no_literal_pool_sf_immediate): New.
15618 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15620 PR tree-optimization/80613
15621 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15622 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15624 2017-05-05 Richard Biener <rguenther@suse.de>
15626 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15628 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15630 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15631 of this flag from insn conditions due to removal from r247495.
15633 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15635 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15637 (arm_early_store_addr_dep_ptr): Likewise.
15638 * config/arm/aarch-common-protos.h
15639 (arm_early_load_addr_dep_ptr): Add prototype.
15640 (arm_early_store_addr_dep_ptr): Likewise.
15641 * config/arm/cortex-a53.md: Add new bypasses.
15643 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15645 * tree.c (next_type_uid): Change type to unsigned.
15646 (type_hash_canon): Decrement back next_type_uid if
15647 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15648 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15651 2017-05-04 Martin Sebor <msebor@redhat.com>
15653 * builtins.c: Fix a trivial typo in a comment.
15655 PR middle-end/79234
15656 * builtins.c (check_sizes): Adjust to handle reading past the end.
15657 Avoid printing excessive upper bound of ranges. Use %E to print
15658 tree nodes instead of converting them to %wu.
15659 (expand_builtin_memchr): New function.
15660 (compute_dest_size): Rename...
15661 (compute_objsize): ...to this.
15662 (expand_builtin_memcpy): Adjust.
15663 (expand_builtin_mempcpy): Adjust.
15664 (expand_builtin_strcat): Adjust.
15665 (expand_builtin_strcpy): Adjust.
15666 (check_strncat_sizes): Adjust.
15667 (expand_builtin_strncat): Adjust.
15668 (expand_builtin_strncpy): Adjust and simplify.
15669 (expand_builtin_memset): Adjust.
15670 (expand_builtin_bzero): Adjust.
15671 (expand_builtin_memcmp): Adjust.
15672 (expand_builtin): Handle memcmp.
15673 (maybe_emit_chk_warning): Check strncat just once.
15675 2017-05-04 Martin Sebor <msebor@redhat.com>
15677 PR preprocessor/79214
15678 PR middle-end/79222
15679 PR middle-end/79223
15680 * builtins.c (check_sizes): Add inlining context and issue
15681 warnings even when -Wno-system-headers is set.
15682 (check_strncat_sizes): Same.
15683 (expand_builtin_strncat): Same.
15684 (expand_builtin_memmove): New function.
15685 (expand_builtin_stpncpy): Same.
15686 (expand_builtin): Handle memmove and stpncpy.
15688 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15690 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15691 which is not used any more.
15693 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15695 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15697 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15699 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15700 (cortexa53_tunings): Likewise.
15701 (cortexa57_tunings): Likewise.
15702 (cortexa72_tunings): Likewise.
15703 (cortexa73_tunings): Likewise.
15705 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15707 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15708 Set loop alignment to 8.
15710 2017-05-04 Martin Sebor <msebor@redhat.com>
15712 PR translation/80280
15713 * builtins.c (expand_builtin_object_size): Add missing quoting to
15714 %D and like directives.
15715 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
15716 (hsa_type_for_tree_type): Same.
15717 (verify_function_arguments): Same.
15718 * symtab.c (symbol_table::change_decl_assembler_name): Same.
15719 * varasm.c (get_section): Same.
15722 2017-05-04 Martin Sebor <msebor@redhat.com>
15724 PR translation/80280
15725 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
15727 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15729 * config/aarch64/aarch64.c (generic_addrcost_table):
15730 Change HI/TI mode setting.
15732 2017-05-04 Martin Jambor <mjambor@suse.cz>
15734 PR tree-optimization/80622
15735 * tree-sra.c (comes_initialized_p): New function.
15736 (build_accesses_from_assign): Only set write lazily when
15737 comes_initialized_p is false.
15738 (analyze_access_subtree): Use comes_initialized_p.
15739 (propagate_subaccesses_across_link): Assert !comes_initialized_p
15740 instead of testing for PARM_DECL.
15742 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15744 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
15745 constraint on operand 0 to allow more general addressing modes.
15746 Adjust output template.
15747 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15749 * config/aarch64/aarch64-protos.h
15750 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15751 * config/aarch64/constraints.md (Dp): New address constraint.
15752 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15755 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15757 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15758 update use of estimate_ipcp_clone_size_and_time.
15759 (estimate_local_effects): Update use of
15760 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15761 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15762 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15763 Return nonspecialized time.
15765 2017-05-04 Richard Biener <rguenther@suse.de>
15767 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15768 for the last VUSE which def dominates the PHI. Directly call
15770 (get_continuation_for_phi_1): Remove.
15772 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15774 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15775 to explain the use of truncating division. Cap the number of
15776 iterations to the maximum given by nb_iterations_upper_bound,
15779 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15781 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15782 * configure: Regenerate.
15783 * config.in: Regenerate.
15784 * config/i386/driver-mingw32.c: new file.
15785 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15786 * config.host: Link driver-mingw32.o on MinGW host.
15787 * doc/install.texi: Document new --enable-mingw-wildcard configure
15790 2017-05-04 Marek Polacek <polacek@redhat.com>
15792 PR tree-optimization/80612
15793 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15795 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15796 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15798 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15799 (movt splitter): Likewise.
15800 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15801 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15802 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15803 block for Thumb-1 with MOVT.
15804 (thumb2_legitimate_address_p): Move code block ...
15805 (can_avoid_literal_pool_for_label_p): ... into this new function.
15806 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15808 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15809 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15810 "M-profile targets with the MOVT instruction".
15812 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15814 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15815 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15816 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15818 2017-05-04 Martin Liska <mliska@suse.cz>
15820 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15821 variable cond_code.
15823 2017-05-04 Richard Biener <rguenther@suse.de>
15825 * tree.c (array_at_struct_end_p): Handle arrays at struct
15826 end with flexarrays more conservatively. Refactor and treat
15827 arrays of arrays or aggregates more strict. Fix
15828 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15829 * tree.c (array_at_struct_end_p): Adjust prototype.
15830 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15831 * gimple-fold.c (get_range_strlen): Likewise.
15832 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15834 2017-05-04 Richard Biener <rguenther@suse.de>
15836 PR tree-optimization/31130
15837 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15839 (supports_overflow_infinity): Likewise.
15840 (is_negative_overflow_infinity): Likewise.
15841 (is_positive_overflow_infinity): Likewise.
15842 (is_overflow_infinity): Likewise.
15843 (stmt_overflow_infinity): Likewise.
15844 (overflow_infinity_range_p): Likewise.
15845 (usable_range_p): Remove as always returning true.
15846 (make_overflow_infinity): Remove.
15847 (negative_overflow_infinity): Likewise.
15848 (positive_overflow_infinity): Likewise.
15849 (avoid_overflow_infinity): Likewise.
15850 (set_value_range): Adjust accordingly.
15851 (set_value_range_to_nonnegative): Likewise, remove now unused
15852 overflow_infinity arg.
15853 (vrp_operand_equal_p): Adjust.
15854 (update_value_range): Likewise.
15855 (range_int_cst_singleton_p): Likewise.
15856 (operand_less_p): Likewise.
15857 (compare_values_warnv): Likewise.
15858 (extract_range_for_var_from_comparison_expr): Likewise.
15859 (vrp_int_const_binop): Likewise.
15860 (zero_nonzero_bits_from_vr): Likewise.
15861 (extract_range_from_multiplicative_op_1): Likewise.
15862 (extract_range_from_binary_expr_1): Likewise.
15863 (extract_range_from_unary_expr): Likewise.
15864 (extract_range_from_comparison): Likewise.
15865 (extract_range_basic): Likewise.
15866 (adjust_range_with_scev): Likewise.
15867 (compare_ranges): Likewise.
15868 (compare_range_with_value): Likewise.
15869 (dump_value_range): Likewise.
15870 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15872 (simplify_cond_using_ranges): Adjust.
15874 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15876 * brig-builtins.def: Added a builtin for class_f64.
15877 * builtin-types.def: Added a builtin type needed by class_f64.
15879 2017-05-03 Jason Merrill <jason@redhat.com>
15881 * timevar.def: Add TV_CONSTEXPR.
15883 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15885 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15887 2017-05-03 Martin Jambor <mjambor@suse.cz>
15889 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15890 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15891 * ipa-cp.c (ipcp_propagate_stage): Do not call
15892 ipa_update_after_lto_read.
15893 * ipa-inline.c (ipa_inline): Likewise.
15895 2017-05-03 Martin Jambor <mjambor@suse.cz>
15897 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15898 tag. Added a default constructor and a destructor.
15899 (ipa_edge_args_sum_t): New class;
15900 (ipa_edge_args_sum): Declare.
15901 (ipa_edge_args_vector): Remove declaration.
15902 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15903 (ipa_free_edge_args_substructures): Remove declaration.
15904 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15905 (ipa_edge_args_info_available_for_edge_p): Likewise.
15906 * ipa-prop.c (ipa_edge_args_vector): Removed.
15907 (edge_removal_hook_holder): Likewise.
15908 (edge_duplication_hook_holder): Likewise.
15909 (ipa_edge_args_sum): New variable.
15910 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15911 ipa_edge_args_vector.
15912 (ipa_free_edge_args_substructures): Likewise.
15913 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15914 ipa_edge_args_vector.
15915 (ipa_edge_removal_hook): Turned into method
15916 ipa_edge_args_sum_t::remove.
15917 (ipa_edge_duplication_hook): Turned into method
15918 ipa_edge_args_sum_t::duplicate.
15919 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15920 registering edge hooks.
15921 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15922 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
15923 ipa_edge_args_sum instead of ipa_edge_args_vector.
15924 * ipa-profile.c (ipa_profile): Likewise.
15926 2017-05-03 Martin Jambor <mjambor@suse.cz>
15928 * symbol-summary.h (function_summary): New method exists.
15929 (function_summary::symtab_removal): Deallocate through release.
15930 (call_summary): New class.
15931 (gt_ggc_mx): New overload.
15932 (gt_pch_nx): Likewise.
15933 (gt_pch_nx): Likewise.
15935 2017-05-03 Jeff Law <law@redhat.com>
15937 PR tree-optimization/78496
15938 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
15939 from simplify_cond_using_ranges. Split off code to walk
15940 backwards through casts into ...
15941 (simplify_cond_using_ranges_2): New function.
15942 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
15943 (execute_vrp): After identifying jump threads, call
15944 simplify_cond_using_ranges_2.
15946 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
15949 * ipa-inline.h (inline_summary): Add ctor.
15950 (create_ggc): Do not use ggc_cleared_alloc.
15952 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
15953 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15955 * gcc.c (handle_braces): Support escaping in switch matching
15957 * doc/invoke.texi (Spec Files): Document it.
15958 Remove superfluous @code markup in items.
15960 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15962 * diagnostic-show-locus.c (struct column_range): New struct.
15963 (get_affected_columns): New function.
15964 (get_printed_columns): New function.
15965 (struct correction): New struct.
15966 (correction::ensure_capacity): New function.
15967 (correction::ensure_terminated): New function.
15968 (struct line_corrections): New struct.
15969 (line_corrections::~line_corrections): New dtor.
15970 (line_corrections::add_hint): New function.
15971 (layout::print_trailing_fixits): Reimplement in terms of the new
15973 (selftest::test_overlapped_fixit_printing): New function.
15974 (selftest::diagnostic_show_locus_c_tests): Call it.
15976 2017-05-03 Nathan Sidwell <nathan@acm.org>
15978 Canonicalize canonical type hashing
15979 * tree.h (type_hash_canon_hash): Declare.
15980 * tree.c (type_hash_list, attribute_hash_list): Move into
15981 type_hash_canon_hash.
15982 (build_type_attribute_qual_variant): Break out hash code calc into
15983 type_hash_canon_hash.
15984 (type_hash_canon_hash): New. Generic type hash computation.
15985 (build_range_type_1, build_array_type_1, build_function_type,
15986 build_method_type_directly, build_offset_type, build_complex_type,
15987 make_vector_type): Call it.
15989 2017-05-03 Richard Biener <rguenther@suse.de>
15991 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15992 When all DRs have unknown misaligned do not always peel
15993 when there is a store but apply the same costing model as if
15994 there were only loads.
15996 2017-05-03 Richard Biener <rguenther@suse.de>
15999 PR tree-optimization/80492
16000 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16001 compare_base_decls returning dont-know properly.
16003 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16005 * config/arm/iterators.md (CCSI): New mode iterator.
16006 (arch): New mode attribute.
16007 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16008 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16009 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16010 code iterator for success result mode.
16011 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16012 the corresponding new insn generators.
16014 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16017 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16018 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16020 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16022 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16023 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16024 (DDR_A): Wrap DDR argument in brackets.
16025 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16026 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16027 (DDR_REVERSED_P): Likewise.
16029 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16031 PR tree-optimization/79472
16032 * tree-switch-conversion.c (struct switch_conv_info): Add
16033 contiguous_range and default_case_nonstandard fields.
16034 (collect_switch_conv_info): Compute contiguous_range and
16035 default_case_nonstandard fields, don't clear final_bb if
16036 contiguous_range and only the default case doesn't have the required
16038 (check_all_empty_except_final): Set default_case_nonstandard instead
16039 of failing if contiguous_range and the default case doesn't have empty
16041 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16042 and only the default case doesn't have the required constants. Skip
16044 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16045 if default_case_nonstandard.
16046 (build_constructors): Build constant 1 just once. Assert that default
16047 values aren't inserted in between cases if contiguous_range. Skip
16049 (build_arrays): Skip virtual phis.
16050 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16051 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16052 Handle virtual phis.
16053 (gen_inbound_check): Handle default_case_nonstandard case.
16054 (process_switch): Adjust check_final_bb caller. Call
16055 gather_default_values with the first non-default case instead of
16056 default case if default_case_nonstandard.
16058 2017-05-02 Nathan Sidwell <nathan@acm.org>
16060 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16061 check. Fix formatting.
16063 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16065 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16066 errors when comparing specialized and unspecialized times.
16068 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16070 * diagnostic-show-locus.c
16071 (layout::should_print_annotation_line_p): Make private.
16072 (layout::print_annotation_line): Make private.
16073 (layout::annotation_line_showed_range_p): Make private.
16074 (layout::show_ruler): Make private.
16075 (layout::print_source_line): Make private. Pass in line and
16076 line_width, rather than calling location_get_source_line. Drop
16078 (layout::print_leading_fixits): New method.
16079 (layout::print_any_fixits): Rename to...
16080 (layout::print_trailing_fixits): ...this, and make private.
16081 Don't print newline fixits.
16082 (diagnostic_show_locus): Move logic for printing one row into...
16083 (layout::print_line): ...this new function. Move the
16084 location_get_source_line call and error-handling from
16085 print_source_line to here. Call print_leading_fixits, and rename
16086 print_any_fixits to print_trailing_fixits.
16087 (selftest::test_fixit_insert_containing_newline): Update now that
16088 newlines are partially supported.
16089 (selftest::test_fixit_insert_containing_newline_2): New test.
16090 (selftest::test_fixit_replace_containing_newline): Update comments.
16091 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16092 * edit-context.c (class added_line): New class.
16093 (class edited_line): Describe newline handling in comment.
16094 (edited_line::actually_edited_p): New method.
16095 (edited_line::print_content): Delete redundant decl.
16096 (edited_line::m_predecessors): New field.
16097 (edited_file::print_content): Call edited_line::print_content.
16098 (edited_file::print_diff): Update to support newlines.
16099 (edited_file::print_diff_hunk): Likewise.
16100 (edited_file::print_run_of_changed_lines): New function.
16101 (edited_file::print_diff_line): Convert to...
16102 (print_diff_line): ...this.
16103 (edited_file::get_effective_line_count): New function.
16104 (edited_line::edited_line): Initialize new field m_predecessors.
16105 (edited_line::~edited_line): Clean up m_predecessors.
16106 (edited_line::apply_fixit): Handle newlines.
16107 (edited_line::get_effective_line_count): New function.
16108 (edited_line::print_content): New function.
16109 (edited_line::print_diff_lines): New function.
16110 (selftest::test_applying_fixits_insert_containing_newline): New
16112 (selftest::test_applying_fixits_replace_containing_newline): New
16114 (selftest::insert_line): New function.
16115 (selftest::test_applying_fixits_multiple_lines): Add example of
16117 (selftest::edit_context_c_tests): Call the new tests.
16119 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16121 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16122 parameter cand. Update dump information.
16123 (get_computation_cost): Update uses.
16125 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16127 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16128 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16129 (get_computation_at, rewrite_use_address): Update use of
16130 get_computation_aff.
16132 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16134 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16135 (get_computation): Delete.
16136 (get_computation_cost): Implement like get_computation_cost_at.
16137 Use get_computation_at.
16138 (get_computation_cost_at): Delete.
16139 (rewrite_use_nonlinear_expr): Use get_computation_at.
16140 (rewrite_use_compare, remove_unused_ivs): Ditto.
16142 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16144 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16146 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16148 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16149 (ivopts_global_cost_for_size): Rename parameter and update uses.
16150 (iv_ca_recount_cost): Update uses.
16151 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16152 candidates seperately in n_invs and n_cands.
16153 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16155 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16157 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16158 (find_inv_vars_cb): New.
16159 (find_depends): Renamed to ...
16160 (find_inv_vars): ... this.
16161 (add_candidate_1, force_var_cost): Call find_inv_vars.
16162 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16164 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16166 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16167 inv_vars. Add inv_exprs.
16168 (struct iv_cand): Rename depends_on to inv_vars.
16169 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16170 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16171 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16172 (dump_cand): Dump inv_vars.
16173 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16174 (record_invariant, find_depends, add_candidate_1): Ditto.
16175 (set_group_iv_cost, force_var_cost): Ditto.
16176 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16177 (get_computation_cost_at, get_computation_cost): Ditto.
16178 (determine_group_iv_cost_generic): Ditto.
16179 (determine_group_iv_cost_address): Ditto.
16180 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16181 (determine_group_iv_costs): Ditto.
16182 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16183 (iv_ca_set_remove_invariants): Renamed to ...
16184 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16185 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16186 (iv_ca_set_add_invariants): Renamed to ...
16187 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16188 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16189 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16190 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16191 (create_new_ivs): Remove useless dump.
16193 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16195 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16197 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16198 (iv_ca_set_no_cp, create_new_iv): Ditto.
16200 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16202 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16204 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16206 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16207 function tree_check2.
16209 2017-05-02 Martin Liska <mliska@suse.cz>
16211 * doc/gcov.texi: Add missing preposition.
16212 * gcov.c (function_info::function_info): Properly fill up
16213 all member variables.
16215 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16217 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16219 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16221 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16223 2017-05-02 Martin Liska <mliska@suse.cz>
16226 * lto-streamer-in.c (lto_read_tree_1): Remove
16227 LTO_STREAMER_DEBUG.
16228 * lto-streamer.c (struct tree_hash_entry): Likewise.
16229 (struct tree_entry_hasher): Likewise.
16230 (tree_entry_hasher::hash): Likewise.
16231 (tree_entry_hasher::equal): Likewise.
16232 (lto_streamer_init): Likewise.
16233 (lto_orig_address_map): Likewise.
16234 (lto_orig_address_get): Likewise.
16235 (lto_orig_address_remove): Likewise.
16236 * lto-streamer.h: Likewise.
16237 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16238 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16240 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16242 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16243 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16244 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16245 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16246 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16247 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16248 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16249 (mm_maskz_sub_ss): New intrinsics.
16250 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16251 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16252 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16253 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16254 (__builtin_ia32_subss_mask_round): New builtins.
16255 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16256 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16257 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16259 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16260 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16262 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16265 2017-05-02 Martin Jambor <mjambor@suse.cz>
16267 PR tree-optimization/78687
16268 * tree-sra.c (access): New field parent.
16269 (process_subtree_disqualification): New function.
16270 (disqualify_candidate): Call it.
16271 (build_accesses_from_assign): Reset write flag if creating an
16273 (build_access_subtree): Fill in parent field and also prpagate
16274 down grp_write flag.
16275 (create_artificial_child_access): New parameter set_grp_write, set
16276 grp_write to its value.
16277 (propagate_subaccesses_across_link): Also propagate grp_write flag
16279 (propagate_all_subaccesses): Push the closest parent back to work
16280 queue if add_access_to_work_queue returned true.
16282 2017-05-02 Richard Biener <rguenther@suse.de>
16284 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16285 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16286 -fstrict-overflow documentation.
16287 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16288 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16289 flag_strict_overflow.
16290 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16291 * lto-opts.c (lto_write_options): Do not stream it.
16292 * lto-wrapper.c (merge_and_complain): Do not handle it.
16293 * opts.c (default_options_table): Do not set -fstrict-overflow.
16294 (finish_options): Likewise do not clear it when sanitizing.
16295 * simplify-rtx.c (simplify_const_relational_operation): Do not
16296 test flag_strict_overflow.
16298 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16300 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16301 using enabled attribute.
16302 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16303 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16304 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16305 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16306 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16307 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16308 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16309 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16310 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16311 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16313 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16315 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16317 2017-05-02 Richard Biener <rguenther@suse.de>
16319 PR tree-optimization/80591
16321 2017-04-10 Richard Biener <rguenther@suse.de>
16323 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16326 2017-05-02 Richard Biener <rguenther@suse.de>
16328 PR tree-optimization/80549
16329 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16330 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16331 headers if they do not have a preheader.
16333 2017-05-02 Martin Liska <mliska@suse.cz>
16336 * common.opt: Fix typo.
16337 * doc/invoke.texi: Likewise.
16339 2017-05-01 Jan Beulich <jbeulich@suse.com>
16341 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16342 swapping, add (x,x,m,x,n) alternative.
16344 2017-05-01 Nathan Sidwell <nathan@acm.org>
16346 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16347 unnecessary unadjusted_alignment check.
16349 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16352 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16354 * gimplify.c (gimplify_cilk_detach): New function.
16355 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16356 * tree-core.h: Document EXPR_CILK_SPAWN.
16357 * tree.h (EXPR_CILK_SPAWN): Define.
16359 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16361 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16362 to use new fixit_hint representation, using the "replace" logic.
16363 (get_line_span_for_fixit_hint): Likewise.
16364 (layout::print_any_fixits): Likewise.
16365 (selftest::test_one_liner_many_fixits): Rename to...
16366 (selftest::test_one_liner_many_fixits_1): ...this, and update
16367 comment and expected output to reflect that the multiple fix-it
16368 hints are now consolidated into one insertion.
16369 (selftest::test_one_liner_many_fixits_2): New test.
16370 (selftest::test_diagnostic_show_locus_one_liner): Update for
16372 (selftest::test_fixit_consolidation): Update for fix-it API
16374 * diagnostic.c (print_parseable_fixits): Likewise.
16375 * edit-context.c (edited_line::m_line_events): Convert from
16376 auto_vec <line_event *> to auto_vec <line_event>.
16377 (class line_event): Convert from abstract base class to a concrete
16378 class, taking over the role of replace_event.
16379 (class insert_event): Delete.
16380 (class replace_event): Rename to class line_event. Convert to
16382 (edit_context::add_fixits): Reimplement.
16383 (edit_context::apply_insert): Delete.
16384 (edit_context::apply_replace): Rename to...
16385 (edit_context::apply_fixit): ...this. Convert to half-open range.
16386 (edited_file::apply_insert): Delete.
16387 (edited_file::apply_replace): Rename to...
16388 (edited_file::apply_fixit): ...this.
16389 (edited_line::~edited_line): Drop deletion of events.
16390 (edited_line::apply_insert): Delete.
16391 (edited_line::apply_replace): Rename to...
16392 (edited_line::apply_fixit): ...this. Convert to half-open range.
16393 Update for change to type of m_line_events.
16394 * edit-context.h (edit_context::apply_insert): Delete.
16395 (edit_context::apply_replace): Rename to...
16396 (edit_context::apply_fixit): ...this.
16398 2017-05-01 Martin Sebor <msebor@redhat.com>
16400 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16403 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16406 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16407 __get_cpuid_max returns 0.
16408 (__get_cpuid_count): Ditto.
16410 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16412 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16413 replacement expression is another instance of one of its arguments.
16415 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16418 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16419 check for stack push/pop autoinc.
16420 * config/i386/i386.c (ix86_agi_dependent): Return false
16421 if the only reason why modified_in_p returned true is that
16422 addr is SP based and set_insn is a push or pop.
16424 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16426 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16429 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16432 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16433 (account_size_time): Use two predicates - exec_pred and
16435 (evaluate_conditions_for_known_args): Compute both clause and
16437 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16438 (inline_summary_t::duplicate): Update.
16439 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16441 (compute_inline_parameters): Likewise.
16442 (estimate_edge_size_and_time): Update caluclation of time.
16443 (estimate_node_size_and_time): Compute both time and nonspecialized
16445 (estimate_ipcp_clone_size_and_time): Update.
16446 (inline_merge_summary): Update.
16447 (do_estimate_edge_time): Update.
16448 (do_estimate_edge_size): Update.
16449 (do_estimate_edge_hints): Update.
16450 (inline_read_section, inline_write_summary): Stream both new predicates.
16451 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16453 (compute_inlined_call_time): Cleanup.
16454 (big_speedup_p): Update.
16455 (edge_badness): Update.
16456 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16457 (size_time_entry): Replace predicate by exec_predicate and
16458 nonconst_predicate.
16459 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16460 (estimate_edge_time): Return also nonspec_time.
16461 (reset_edge_growth_cache): Update.
16463 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16465 PR rtl-optimization/80491
16466 * ifcvt.c (noce_process_if_block): When looking for x setter
16467 with missing else_bb, don't check only the insn right before
16468 cond_earliest, but look for the last insn that x is modified in
16469 within the same bb.
16471 PR rtl-optimization/80491
16472 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16474 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16476 PR tree-optimization/80487
16477 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16479 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16481 PR tree-optimization/79697
16482 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16483 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16484 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16486 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16487 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16489 2017-04-28 Martin Sebor <msebor@redhat.com>
16491 PR tree-optimization/80523
16492 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16493 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16495 (maybe_warn, format_directive, parse_directive): Use new functions.
16496 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16498 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16500 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16502 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16504 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16505 target_header_dir): Set correctly.
16506 * configure: Regenerated.
16507 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16508 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16509 instead of SYSTEM_HEADER_DIR.
16511 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16513 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16514 (estimate_local_effects): Likewise.
16515 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16516 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16517 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16518 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16519 do_estimate_edge_time, estimate_edge_time): Likewise.
16520 * ipa-inline-analysis.c (estimate_node_size_and_time,
16521 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16522 (estimate_time_after_inlining): Remove.
16524 2017-04-28 Martin Liska <mliska@suse.cz>
16526 * doc/gcov.texi: Enhance documentation of gcov.
16528 2017-04-28 Martin Liska <mliska@suse.cz>
16530 * doc/gcov.texi: Sort options in alphabetic order.
16531 * doc/gcov-dump.texi: Likewise.
16532 * doc/gcov-tool.texi: Likewise.
16533 * gcov.c (print_usage): Likewise.
16534 * gcov-dump.c (print_usage): Likewise.
16535 * gcov-tool.c (print_merge_usage_message): Likewise.
16536 (print_rewrite_usage_message): Likewise.
16537 (print_overlap_usage_message): Likewise.
16539 2017-04-28 Martin Liska <mliska@suse.cz>
16541 PR gcov-profile/53915
16542 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16544 2017-04-28 Martin Liska <mliska@suse.cz>
16546 PR gcov-profile/79891
16547 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16548 is marked by compiler as living on a line.
16549 (get_cycles_count): Remove usage of the union.
16550 (output_intermediate_file): Likewise.
16551 (find_source): Fix GNU coding style.
16552 (accumulate_line_counts): Remove old non-all block mode.
16553 (output_lines): Remove usage of the union.
16554 * profile.c (output_location): Include all BBs, even if
16555 belonging to a same line (and file) as a previous BB.
16557 2017-04-28 Martin Liska <mliska@suse.cz>
16559 * gcov.c (process_args): Handle new argument 'w'.
16560 (read_graph_file): Assign ID to BBs.
16561 (output_branch_count): Display BB # if verbose flag is set.
16562 (output_lines): Likewise for arcs.
16563 (print_usage): Add '--verbose' option help.
16564 * doc/gcov.texi: Document --verbose (-w) option.
16566 2017-04-28 Martin Liska <mliska@suse.cz>
16568 * gcov.c (struct block_location_info): New struct.
16569 (process_file): Fill up the new structure.
16570 (read_graph_file): Replace usage of encoding by the newly added
16572 (add_line_counts): Likewise.
16573 (accumulate_line_counts): Remove usage of the union.
16574 (function_info::function_info): New function.
16575 (function_info::~function_info): Likewise.
16576 (process_file): Call delete instead of release_function.
16577 (release_function): Release the function.
16578 (release_structures): Call delete instead of release_function.
16579 (solve_flow_graph): Replace usage of num_blocks.
16580 (find_exception_blocks): Likewise.
16581 (output_lines): Fix GNU coding style.
16583 2017-04-28 Martin Liska <mliska@suse.cz>
16586 * coverage.c (coverage_remove_note_file): New function.
16587 * coverage.h: Declare the function.
16588 * toplev.c (finalize): Clean if an error has been seen.
16590 2017-04-28 Martin Liska <mliska@suse.cz>
16592 PR gcov-profile/80031
16593 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16594 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16595 * gcov.c (read_graph_file): Read just number of blocks.
16596 * profile.c (branch_prob): Do not stream 0 flags per a basic
16599 2017-04-28 Martin Liska <mliska@suse.cz>
16601 * gcov-dump.c (tag_*): Add new argument to declarations.
16602 (dump_gcov_file): Likewise.
16603 (tag_blocks): Add and use new argument depth.
16604 (tag_arcs): Likewise.
16605 (tag_lines): Likewise.
16606 (tag_counters): Likewise.
16607 (tag_summary): Likewise.
16608 (dump_working_sets): Use depth to do a proper indentation.
16610 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16613 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16614 * symtab.c (symtab_node::debug_symtab): Move definition here.
16616 2017-04-28 Richard Biener <rguenther@suse.de>
16618 * lto-streamer.h (LTO_major_version): Bump to 7.
16620 2017-04-28 Richard Biener <rguenther@suse.de>
16622 * tree-vrp.c (assert_info): New struct.
16623 (add_assert_info): New helper.
16624 (register_edge_assert_for_2): Refactor to add asserts to a vector
16626 (register_edge_assert_for_1): Likewise.
16627 (register_edge_assert_for): Likewise.
16628 (finish_register_edge_assert_for): New helper actually registering
16629 asserts where live on edge.
16630 (find_conditional_asserts): Adjust.
16631 (find_switch_asserts): Likewise.
16632 (evrp_dom_walker::try_find_new_range): Generalize.
16633 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16635 2017-04-27 Marek Polacek <polacek@redhat.com>
16638 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16639 arg10 and arg11 to itype.
16641 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16643 * doc/extend.texi (Object Size Checking): Improve grammar.
16645 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16648 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16649 that the logic for permitting reciprocal estimates matches that
16652 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16655 * tree.c (type_cache_hasher::equal): Only compare
16656 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16657 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16658 non-aggregate element types.
16659 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16660 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16662 2017-04-27 Richard Biener <rguenther@suse.de>
16664 PR middle-end/80533
16665 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16666 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16667 keeping a reference to a trailing array.
16669 2017-04-27 Richard Biener <rguenther@suse.de>
16671 PR middle-end/80539
16672 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16673 being in loop-closed SSA form conservatively.
16674 (chrec_fold_multiply_poly_poly): Likewise.
16676 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16678 PR middle-end/79665
16679 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16680 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16682 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16685 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16686 (aarch64_function_arg_alignment): Return unsigned int again, but still
16687 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16688 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16689 Don't emit -Wpsabi note.
16690 (aarch64_function_arg_boundary): Likewise.
16691 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16694 2017-04-26 Nathan Sidwell <nathan@acm.org>
16696 * tree.h (crc32_unsigned_n): Declare.
16697 (crc32_unsigned, crc32_unsigned): Make inline.
16698 * tree.c (crc32_unsigned_bits): Replace with ...
16699 (crc32_unsigned_n): ... this.
16700 (crc32_unsigned, crc32_byte): Remove.
16701 (crc32_string): Remove unnecessary braces.
16703 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16705 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16706 * ipa-inline-analysis.c (MAX_TIME): Remove.
16707 (account_size_time): Use sreal for time.
16708 (dump_inline_summary): Update.
16709 (estimate_function_body_sizes): Update.
16710 (estimate_edge_size_and_time): Update.
16711 (estimate_calls_size_and_time): Update.
16712 (estimate_node_size_and_time): Update.
16713 (inline_merge_summary): Update.
16714 (inline_update_overall_summary): Update.
16715 (estimate_time_after_inlining): Update.
16716 (inline_read_section): Update.
16717 (inline_write_summary): Update.
16718 * ipa-inline.c (compute_uninlined_call_time): Update.
16719 (compute_inlined_call_time): Update.
16720 (recursive_inlining): Update.
16721 (inline_small_functions): Update.
16722 (dump_overall_stats): Update.
16723 * ipa-inline.h: Include sreal.h.
16724 (size_time_entry): Turn time to sreal.
16725 (inline_summary): Turn self_time nad time to sreal.
16727 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16729 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
16731 (sreal::stream_out, sreal::stream_in): New.
16732 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
16734 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16736 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
16739 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
16742 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
16744 (dimode_scalar_chain::compute_convert_gain): Ditto.
16745 (dimode_scalar_chain::make_vector_copies): Ditto.
16746 (dimode_scalar_chain::convert_reg): Ditto.
16747 (dimode_scalar_chain::convert_insn): Ditto.
16748 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16749 (VI248_AVX512BW_1): New mode iterator.
16750 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16751 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16754 2017-04-25 Martin Sebor <msebor@redhat.com>
16756 PR tree-optimization/80497
16757 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16758 constants are representable in HOST_WIDE_INT.
16759 (parse_directive): Ditto.
16761 2017-04-25 Martin Sebor <msebor@redhat.com>
16764 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16765 (new_zero_array): Adjust signature.
16766 (dom_info::dom_init): Used unsigned rather that size_t.
16767 (dom_info::dom_info): Same.
16769 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16770 Jakub Jelinek <jakub@redhat.com>
16773 * config/arm/arm.c: Include gimple.h.
16774 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16775 returns negative, increment ncrn only if it returned positive.
16776 (arm_needs_doubleword_align): Return int instead of bool,
16777 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16778 members, but if there is any such non-FIELD_DECL
16779 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16780 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16781 returns negative, increment nregs only if it returned positive.
16782 (arm_setup_incoming_varargs): Likewise.
16783 (arm_function_arg_boundary): Emit -Wpsabi note if
16784 arm_needs_doubleword_align returns negative, return
16785 DOUBLEWORD_ALIGNMENT only if it returned positive.
16787 2017-04-25 Marek Polacek <polacek@redhat.com>
16790 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16791 first argument to type.
16793 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16796 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16797 type checks to test for compatibility instead of equality.
16799 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16800 Jakub Jelinek <jakub@redhat.com>
16803 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16805 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16806 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16807 the alignment computation, but return their maximum in warn_alignment.
16808 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16809 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16811 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16812 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16815 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16817 * config/arc/simdext.md (dmpyh): Fix typo.
16819 2017-04-25 Richard Biener <rguenther@suse.de>
16821 PR tree-optimization/80492
16822 * alias.c (compare_base_decls): Handle registers with asm
16823 specification conservatively.
16824 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16825 compare_base_decls returning dont-know properly.
16827 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16829 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16830 (legitimate_offset_address_p): New function.
16831 (arc_legitimate_address_p): Use above function.
16833 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16835 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16837 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16839 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16840 ACCH registers whenever they are available.
16842 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16844 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16845 double regs fix when not used.
16847 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16849 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16851 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16852 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16854 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16856 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16857 when emitting short ADD instructions.
16859 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16861 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16863 (cmpsi_cc_c_insn): Likewise.
16864 (cbranchsi4_scratch): Compute proper instruction length using
16865 compact_hreg_operand.
16866 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16868 2017-04-25 Richard Biener <rguenther@suse.de>
16870 PR middle-end/80509
16871 * passes.c (pass_manager::pass_manager): Initialize
16872 m_name_to_pass_map.
16874 2017-04-25 Richard Biener <rguenther@suse.de>
16876 PR tree-optimization/79201
16877 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16879 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16882 * config/s390/vector.md: Split MEM->GPR vector moves for
16883 non-s_operand addresses.
16885 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16888 * config/s390/predicates.md (reload_const_wide_int_operand): New
16890 * config/s390/s390.md ("movti"): Remove d/P alternative.
16891 ("movti_bigconst"): New pattern definition.
16893 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16896 * s390-protos.h (s390_expand_cs_hqi): Removed.
16897 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16898 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16899 modes as well as CCZ1mode and CCZmode.
16900 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16901 signature of s390_emit_compare_and_swap.
16902 (s390_expand_cs_hqi): Likewise, make static.
16903 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16904 compare-and-swap, in some cases.
16905 (s390_expand_cs): Wrapper function.
16906 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16908 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16909 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16910 patterns for small and large integers. Forbid symref memory operands.
16911 Move expander to s390.c. Require cc register.
16912 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16913 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16914 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16915 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16916 symref memory operands. Remove CC mode and call s390_match_ccmode
16918 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16920 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16922 * config/s390/s390.md (define_peephole2): New peephole to help
16923 combining the load-and-test pattern with volatile memory.
16925 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16927 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
16928 with CCZmode for TARGET_Z196.
16930 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16932 PR rtl-optimization/80501
16933 * combine.c (make_compound_operation_int): Set subreg_code to SET
16934 even for AND with mask of the sign bit of mode.
16936 PR rtl-optimization/80500
16937 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
16938 sum's initial value.
16940 2017-04-25 Julian Brown <julian@codesourcery.com>
16941 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16943 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
16945 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
16947 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
16949 2017-04-25 Julian Brown <julian@codesourcery.com>
16950 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16952 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
16953 (thunderx2t99_sha): New Reservation.
16955 2017-04-25 Julian Brown <julian@codesourcery.com>
16956 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16958 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
16959 type for 1-element load.
16961 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
16963 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
16965 2017-04-24 Martin Jambor <mjambor@suse.cz>
16967 PR tree-optimization/80293
16968 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
16969 char arrays not totally scalarizable if it is false.
16970 (analyze_all_variable_accesses): Pass correct value in the new
16971 parameter. Add a statistics counter.
16973 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
16975 PR middle-end/79931
16976 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
16978 2017-04-24 Richard Biener <rguenther@suse.de>
16980 PR tree-optimization/80494
16981 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
16982 out for complex types.
16984 2017-04-24 Richard Biener <rguenther@suse.de>
16986 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
16987 * tree-ssa-sccvn.c (print_scc): Print SCC size.
16988 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
16989 (DFS): Adjust and never fail.
16990 (sccvn_dom_walker::fail): Remove.
16991 (sccvn_dom_walker::before_dom_children): Adjust.
16992 (run_scc_vn): Likewise and never fail.
16993 * tree-ssa-pre.c (pass_pre::execute): Adjust.
16994 (pass_fre::execute): Likewise.
16996 2017-04-24 Richard Biener <rguenther@suse.de>
16998 PR tree-optimization/79725
16999 * tree-ssa-sink.c (statement_sink_location): Return whether
17000 failure reason was zero uses. Move that check later.
17001 (sink_code_in_bb): Deal with zero uses by removing the stmt
17004 2017-04-24 Richard Biener <rguenther@suse.de>
17007 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17008 pointer-based references.
17010 2017-04-24 Richard Biener <rguenther@suse.de>
17013 * pass_manager.h (pass_manager::operator new): Remove.
17014 (pass_manager::operator delete): Likewise.
17015 * passes.c (pass_manager::operator new): Remove.
17016 (pass_manager::operator delete): Likewise.
17017 (pass_manager::pass_manager): Zero individual pass members.
17019 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17022 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17023 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17024 Check "XEXP (src, 1)" operand here.
17025 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17026 Check "XEXP (src, 1)" operand here.
17027 (dimode_scalar_chain::make_vector_copies): Detect count register
17028 of a shift instruction. Zero extend count register from QImode
17029 to DImode to satisfy vector shift pattern count operand predicate.
17030 Substitute vector shift count operand with a DImode copy.
17031 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17034 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17036 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17037 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17038 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17039 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17040 (UNSPEC_NOREX_MEM): Remove definition.
17042 2017-04-21 Richard Biener <rguenther@suse.de>
17044 PR tree-optimization/79547
17045 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17046 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17047 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17048 without any constraints.
17050 2017-04-21 Richard Biener <rguenther@suse.de>
17052 PR tree-optimization/78847
17053 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17055 2017-04-21 Richard Biener <rguenther@suse.de>
17057 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17058 (build_distinct_type_copy): Likewise.
17059 (build_variant_type_copy): Likewise.
17060 * tree.c (build_qualified_type): Pass down mem-stat info.
17061 (build_distinct_type_copy): Likewise.
17062 (build_variant_type_copy): Likewise.
17064 2017-04-21 Richard Biener <rguenther@suse.de>
17066 PR tree-optimization/80237
17067 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17069 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17070 for a simplified result.
17072 2016-04-21 Richard Biener <rguenther@suse.de>
17074 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17075 sth as strict as a simple_iv but a chrec without symbols and an
17076 operand defined in the loop we are peeling (and not some subloop).
17077 (propagate_constants_for_unrolling): Propagate all constants.
17079 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17082 * config/i386/i386.c (print_reg): Remove assert for disalowed
17083 regno values, call output_operand_lossage instead.
17085 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17088 * config/i386/constraints.md (Yc): New register constraint.
17089 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17090 Use Yc constraint for alternative 2 of operand 0. Remove
17091 preferred_for_speed attribute.
17093 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17095 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17096 lastprivate clauses in SIMT case.
17098 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17100 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17102 2017-04-20 Richard Biener <rguenther@suse.de>
17104 PR tree-optimization/57796
17105 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17106 as N scalar stores.
17107 (vect_model_load_cost): Cost gathers as N scalar loads.
17109 2017-04-20 Richard Biener <rguenther@suse.de>
17111 * ggc-page.c (ggc_allocated_p): Rename to ...
17112 (safe_lookup_page_table_entry): ... this and return the lookup
17114 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17116 2017-04-20 Richard Biener <rguenther@suse.de>
17118 PR tree-optimization/80453
17119 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17120 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17121 from the conditions.
17122 (vn_phi_eq): Pass them down.
17123 (vn_phi_lookup): Record them.
17124 (vn_phi_insert): Likewise.
17126 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17128 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17129 uninitialized variable warning to avoid buffer overrun.
17131 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17134 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17135 is suppressed for '{ 0 }' in C.
17137 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17139 * BASE-VER: Set to 8.0.0.
17141 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17143 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17144 priority .init_array and .fini_array section with SECTION_NOTYPE
17147 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17149 PR middle-end/80423
17150 * tree.h (build_array_type): Add typeless_storage default argument.
17151 * tree.c (type_cache_hasher::equal): Also compare
17152 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17153 (build_array_type): Add typeless_storage argument, set
17154 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17156 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17157 (build_array_type): Likewise. Add typeless_storage argument.
17159 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17160 Jakub Jelinek <jakub@redhat.com>
17162 PR tree-optimization/80426
17163 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17164 operation on symbolic operands, also compute the overflow for the
17165 invariant part when the operation degenerates into a negation.
17167 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17170 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17171 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17174 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17176 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17179 * config/avr/avr.c (tree.h): Include it.
17180 (cgraph.h): Include it.
17181 (avr_encode_section_info): Don't warn for uninitialized progmem
17182 variable if it's just an alias.
17184 2017-04-19 Richard Biener <rguenther@suse.de>
17187 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17188 when needed by AutoPGO.
17190 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17193 * doc/lto.texi: Remove an extra 'that'.
17195 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17197 PR rtl-optimization/80429
17198 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17199 are only used in debug insns.
17201 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17202 Vladimir Makarov <vmakarov@redhat.com>
17204 * config/sparc/predicates.md (input_operand): Add comment. Return
17205 true for any memory operand when LRA is in progress.
17206 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17208 2017-04-18 Jeff Law <law@redhat.com>
17211 * mips.md ({return,simple_return}_internal): Do not overwrite
17214 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17216 PR tree-optimization/80443
17217 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17218 instead of adding 1, subtract -1 and similarly instead of subtracting
17221 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17223 PR rtl-optimization/80357
17224 * haifa-sched.c (tmp_bitmap): New variable.
17225 (model_recompute): Handle duplicate use records.
17226 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17227 (free_global_sched_pressure_data): Free it.
17229 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17232 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17233 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17234 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17235 instead of SYSTEM_HEADER_DIR.
17237 2017-04-18 Jeff Law <law@redhat.com>
17239 PR middle-end/80422
17240 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17241 predecessors after walking up the insn chain.
17243 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17246 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17247 sizetype type into debug info.
17249 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17252 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17253 unneeded test for TARGET_UPPER_REGS_SF.
17254 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17256 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17259 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17260 instead of gsi_after_labels.
17262 2017-04-18 Jeff Law <law@redhat.com>
17264 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17268 2017-04-13 Jeff Law <law@redhat.com>
17269 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17270 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17272 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17275 * config/avr/avr.c (intl.h): Include it.
17276 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17278 2017-04-18 Martin Liska <mliska@suse.cz>
17280 PR gcov-profile/78783
17281 * gcov-tool.c (gcov_output_files): Validate that destination
17282 file is either removed by the tool or by a user.
17284 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17285 Guy Benyei <guybe@mellanox.com>
17287 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17288 block, and do not negate it, the stored id is already negative.
17290 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17292 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17294 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17297 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17298 masks of options that should be turned off if the VSX vector
17299 options are turned off.
17300 (OTHER_P8_VECTOR_MASKS): Likewise.
17301 (OTHER_VSX_VECTOR_MASKS): Likewise.
17302 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17303 rs6000_disable_incompatible_switches to validate no type switches
17305 (rs6000_incompatible_switch): New function to disallow turning on
17306 other vector options if -mno-vsx, -mno-power8-vector, or
17307 -mno-power9-vector are specified.
17309 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17311 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17313 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17315 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17316 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17317 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17318 (ARG_POINTER_CFA_OFFSET): Likewise.
17320 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17322 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17323 conditions to take advantage of various optimizations.
17325 2017-04-13 Jeff Law <law@redhat.com>
17327 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17328 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17329 (zero_extendsidi2_dext): Likewise.
17331 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17334 * fold-const.c (fold_ternary_loc): Revert
17335 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17338 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17340 PR rtl-optimization/80343
17341 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17342 new scratch pseudo.
17344 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17347 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17348 to ubsan_encode_value.
17350 2017-04-13 Jeff Law <law@redhat.com>
17352 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17353 appearing in DEBUG_INSNs.
17355 2017-04-13 Martin Liska <mliska@suse.cz>
17357 PR gcov-profile/80413
17358 * gcov-io.c (gcov_write_string): Copy to buffer just when
17359 allocated size is greater than zero.
17361 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17364 * dwarf2out.c (decls_for_scope): Ignore declarations of
17365 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17367 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17370 * ipa-visibility.c (non_local_p): Fix typos.
17371 (localize_node): When localizing symbol in same comdat group,
17372 dissolve the group only when we know external symbols are going
17374 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17376 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17378 PR tree-optimization/79390
17379 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17380 order does not result in usable sequence, retry with reversed operand
17386 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17387 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17388 op0 instead of fold_convert_loc (loc, type, arg0).
17390 2017-04-12 Jeff Law <law@redhat.com>
17392 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17393 has a delay slot in the generated code.
17395 * config/cris/cris.md (cris_preferred_reload_class): Return
17396 GENNONACR_REGS rather than GENERAL_REGS.
17398 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17401 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17402 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17403 signedness of the result type.
17405 2017-04-12 Richard Biener <rguenther@suse.de>
17406 Jeff Law <law@redhat.com>
17408 PR tree-optimization/80359
17409 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17410 trim stores to TARGET_MEM_REFs.
17412 2017-04-12 Richard Biener <rguenther@suse.de>
17414 PR tree-optimization/79390
17415 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17416 threading case even more.
17418 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17421 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17422 for quad_address_p for TImode, instead of just not indexed_address.
17424 2017-04-12 Richard Biener <rguenther@suse.de>
17425 Bernd Edlinger <bernd.edlinger@hotmail.de>
17427 PR middle-end/79671
17428 * alias.c (component_uses_parent_alias_set_from): Handle
17429 TYPE_TYPELESS_STORAGE.
17430 (get_alias_set): Likewise.
17431 * tree-core.h (tree_type_common): Add typeless_storage flag.
17432 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17433 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17434 for types containing members with TYPE_TYPELESS_STORAGE.
17435 (place_field): Likewise.
17436 (layout_type): Likewise for ARRAY_TYPE.
17437 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17438 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17439 TYPE_TYPELESS_STORAGE.
17440 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17442 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17445 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17446 first argument to type.
17448 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17452 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17453 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17454 (rs6000_expand_binop_builtin): Likewise.
17455 (rs6000_expand_ternop_builtin): Likewise; also add missing
17456 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17457 vshasigma built-ins.
17458 * doc/extend.texi: Document that vec_xxpermdi's third argument
17459 must be a constant.
17461 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17463 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17464 Use shift_const cost parameter when calculating gain of STV shifts.
17466 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17468 PR rtl-optimization/70478
17469 * lra-constraints.c (process_alt_operands): Check memory for
17470 disfavoring memory insn operand.
17472 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17474 PR middle-end/80100
17475 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17476 left shift in unsigned HOST_WIDE_INT type.
17478 PR rtl-optimization/80385
17479 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17480 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17483 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17484 if they have any depend clauses.
17486 2017-04-11 Martin Liska <mliska@suse.cz>
17489 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17490 * ipa-split.c (split_function): Create a local comdat symbol
17491 if caller is in a comdat group.
17493 2017-04-11 Martin Liska <mliska@suse.cz>
17496 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17499 2017-04-11 Martin Sebor <msebor@redhat.com>
17501 PR middle-end/80364
17502 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17503 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17505 (directive::set_width, directive::set_precision, format_character):
17507 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17510 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17513 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17514 conflict, set target->arch_name instead of target->cpu_name.
17516 2017-04-11 Richard Biener <rguenther@suse.de>
17518 PR tree-optimization/80374
17519 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17520 build_zero_cst, remove fold_convertible_p check again.
17522 2017-04-11 Martin Liska <mliska@suse.cz>
17525 * ubsan.c (instrument_object_size): Do not instrument register
17528 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17531 * config/i386/i386-builtin-types.def
17532 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17533 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17534 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17535 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17536 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17537 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17538 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17539 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17540 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17541 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17542 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17543 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17544 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17545 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17546 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17547 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17548 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17549 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17550 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17551 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17552 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17553 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17554 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17555 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17556 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17557 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17558 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17559 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17560 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17561 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17562 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17563 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17564 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17565 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17566 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17567 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17568 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17569 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17570 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17571 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17572 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17573 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17574 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17575 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17576 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17577 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17579 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17580 flag to second_arg_count, handle 4 argument function type _COUNT
17581 aliases, handle second_arg_count on second argument rather than last.
17583 2017-04-10 Jeff Law <law@redhat.com>
17585 PR tree-optimization/80374
17586 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17587 record anything if we can not convert integer_zero_node to the
17590 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17593 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17594 Enhance special handling given to the TARGET_P9_MINMAX option in
17595 relation to certain other options.
17597 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17599 PR tree-optimization/80153
17600 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17601 remove POINTER_PLUS_EXPR's base part directly, rather than through
17604 2017-04-10 Richard Biener <rguenther@suse.de>
17605 Bin Cheng <bin.cheng@arm.com>
17607 PR tree-optimization/80153
17608 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17609 the first element of pointer type aff_tree. Build result expr in
17611 (add_elt_to_tree): Convert to type unconditionally. Remove other
17612 fold_convert calls.
17613 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17614 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17616 2017-04-10 Richard Biener <rguenther@suse.de>
17618 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17621 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17623 PR rtl-optimization/70478
17624 * lra-constraints.c (curr_small_class_check): New.
17625 (update_and_check_small_class_inputs): New.
17626 (process_alt_operands): Update curr_small_class_check. Disfavor
17627 alternative insn memory operands. Check available regs for small
17630 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17633 * config/mips/mips.opt (-mvirt): Update description.
17634 * doc/invoke.texi (-mvirt): Likewise.
17636 2017-04-10 Richard Biener <rguenther@suse.de>
17638 PR middle-end/80362
17639 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17640 looking for NEGATE_EXPR in -A / -B to A / B folding.
17642 2017-04-10 Martin Liska <mliska@suse.cz>
17644 PR gcov-profile/80224
17645 * gcov.c (print_usage): Fix usage string.
17646 (get_gcov_intermediate_filename): Remove.
17647 (output_gcov_file): Use both for normal and intermediate format.
17648 (generate_results): Do not initialize special file for
17649 intermediate format.
17651 2017-04-10 Richard Biener <rguenther@suse.de>
17653 PR tree-optimization/80304
17654 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17657 2017-04-10 Nathan Sidwell <nathan@acm.org>
17660 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17661 (rs6000_init_builtins): Use it.
17663 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17665 * config/arm/arm.md (<mrc>): Add mode to SET source.
17666 (<mrrc>): Likewise.
17668 2017-04-10 Richard Biener <rguenther@suse.de>
17670 PR middle-end/80344
17671 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17673 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17676 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17677 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17678 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17679 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17680 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17681 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17682 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17683 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17684 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17685 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17686 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17687 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17688 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17689 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17690 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17691 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17692 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17693 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17694 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17695 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17696 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17697 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17698 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17700 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17702 PR rtl-optimization/70478
17703 * lra-constraints.c: Reverse the last patch.
17705 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17707 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17708 Add comment for WCHAR_T.
17710 2017-04-08 Martin Liska <mliska@suse.cz>
17713 2017-04-07 Martin Liska <mliska@suse.cz>
17716 * ipa-split.c (split_function): Add function part to a same comdat
17719 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17722 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
17724 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
17726 * rs6000/rs6000.c (vec_load_pendulum): Rename...
17727 (vec_pairing): ...to this.
17728 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
17729 (rs6000_sched_init): Adjust for name change.
17730 (struct rs6000_sched_context): Likewise.
17731 (rs6000_init_sched_context): Likewise.
17732 (rs6000_set_sched_context): Likewise.
17734 2017-04-07 Jakub Jelinek <jakub@redhat.com>
17740 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
17742 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
17743 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
17744 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
17746 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17748 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17750 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17752 PR rtl-optimization/70703
17753 * ira-color.c (update_conflict_hard_regno_costs): Use
17754 int64_t instead of HOST_WIDE_INT.
17756 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17758 PR rtl-optimization/70478
17759 * lra-constraints.c (process_alt_operands): Disfavor alternative
17760 insn memory operands.
17762 2017-04-07 Jeff Law <law@redhat.com>
17764 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17765 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17767 2017-04-07 Martin Liska <mliska@suse.cz>
17770 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17771 Show error message instead of an ICE.
17773 2017-04-07 Martin Liska <mliska@suse.cz>
17776 * ipa-split.c (split_function): Add function part to a same comdat
17779 2017-04-07 Richard Biener <rguenther@suse.de>
17781 PR middle-end/80341
17782 * tree.c (get_unwidened): Also handle ! for_type case for
17784 * convert.c (do_narrow): Split out from ...
17785 (convert_to_integer_1): ... here. Do not pass final truncation
17786 type to get_unwidened for TRUNC_DIV_EXPR.
17788 2017-04-07 Richard Biener <rguenther@suse.de>
17790 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17792 (aff_combination_const): Adjust.
17793 (aff_combination_scale): Likewise.
17794 (aff_combination_add_elt): Likewise.
17795 (aff_combination_add_cst): Likewise.
17796 (aff_combination_convert): Likewise.
17797 (add_elt_to_tree): Likewise. Remove unused argument.
17798 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17800 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17802 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17804 * config/arm/arm.c (arm_default_short_enums): Use
17805 ARM_DEFAULT_SHORT_ENUMS.
17806 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17808 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17811 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17812 members with redundant out-of-class redeclaration.
17814 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17817 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17818 * config/i386/i386.md (*zero_extendsidi2):
17819 Add (?*x,*x) and (?*v,*v) alternatives.
17821 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17824 * config/i386/i386.c (ix86_expand_builtin)
17825 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17826 mode from insn data. Convert operands to insn operand mode.
17827 Copy operands that don't satisfy insn predicate to a register.
17829 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17831 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17834 2017-04-06 Richard Biener <rguenther@suse.de>
17836 PR tree-optimization/80334
17837 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17838 preserve alignment of accesses.
17840 2017-04-06 Richard Biener <rguenther@suse.de>
17842 PR tree-optimization/80262
17843 * tree-sra.c (build_ref_for_offset): Preserve address-space
17845 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17846 Drop useless address-space information on MEM_REF offsets.
17848 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17850 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17852 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17854 PR rtl-optimization/70703
17855 * ira-color.c (update_conflict_hard_regno_costs): Use
17856 HOST_WIDE_INT instead of long.
17858 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17861 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17862 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17864 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17865 for x86_64 target. Handle -m3dnowa option.
17867 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17869 PR rtl-optimization/70703
17870 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17871 (update_conflict_hard_regno_costs): Use long instead of unsigned
17872 arithmetic for cost calculation.
17874 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17875 Bernd Edlinger <bernd.edlinger@hotmail.de>
17878 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17881 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17884 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17885 ptr_mode with Pmode throughout.
17886 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17887 into probe_stack_range and use DImode.
17889 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17892 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17893 call_eh_return is true.
17895 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17897 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17898 Initialize last_match_fntype_index.
17900 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17903 * tree-nvr.c: Include internal-fn.h.
17904 (pass_return_slot::execute): Ignore internal calls without
17907 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17908 Richard Biener <rguenther@suse.de>
17911 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17912 captures used multiple times, except for the last use.
17913 * generic-match-head.c: Include gimplify.h.
17915 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17917 PR tree-optimization/79390
17918 * target.h (struct noce_if_info): Declare.
17919 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17920 * target.def (noce_conversion_profitable_p): New target hook.
17921 * ifcvt.h (struct noce_if_info): New type, moved from ...
17922 * ifcvt.c (struct noce_if_info): ... here.
17923 (noce_conversion_profitable_p): Renamed to ...
17924 (default_noce_conversion_profitable_p): ... this. No longer
17926 (noce_try_store_flag_constants, noce_try_addcc,
17927 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
17928 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
17929 instead of noce_conversion_profitable_p.
17930 * config/i386/i386.c: Include ifcvt.h.
17931 (ix86_option_override_internal): Don't override
17932 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
17933 (ix86_noce_conversion_profitable_p): New function.
17934 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
17935 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
17936 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
17937 * doc/tm.texi: Regenerated.
17939 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17941 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
17944 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17947 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
17948 instructions for small multiply cores.
17950 2017-04-04 Jeff Law <law@redhat.com>
17952 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
17954 (mips_expand_vec_perm_const): Initialize elements in orig_perm
17955 that are not set by the loop over the elements.
17957 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17960 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
17961 int mode, convert_modes it to mode as unsigned, otherwise use
17962 lowpart_subreg to mode rather than SImode.
17963 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
17964 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
17965 Use DImode instead of SImode for the shift count operand.
17966 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
17969 2017-04-04 Richard Biener <rguenther@suse.de>
17971 PR middle-end/80281
17972 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
17973 arithmetic done for the negate or the plus. Simplify.
17974 (A - (-B) -> A + B): Likewise.
17975 * fold-const.c (split_tree): Make sure to not negate pointers.
17977 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
17979 PR rtl-optimization/60818
17980 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
17981 a compare of comparisons with the thing compared if this results
17982 in a different machine mode.
17984 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
17986 * alias.c (base_alias_check): Fix typo in comment.
17987 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
17988 * cgraphunit.c (symbol_table::compile): Likewise.
17989 * collect2.c (maybe_run_lto_and_relink): Likewise.
17990 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
17991 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
17992 * config/avr/avr.c (avr_map_op_t): Likewise.
17993 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
17994 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
17995 * config/epiphany/epiphany.md (movcc): Likewise.
17996 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
17997 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
17999 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18000 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18001 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18002 * config/sh/sh.c (sh_rtx_costs): Likewise.
18003 * fold-const.c (fold_truth_andor): Likewise.
18004 * genautomata.c (collapse_flag): Likewise.
18005 * gengtype.h (struct type::u::s): Likewise.
18006 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18007 * input.c (FORMAT_AMOUNT): Likewise.
18008 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18009 (known_aggs_to_agg_replacement_list): Likewise.
18010 * ipa-inline-analysis.c: Likewise.
18011 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18012 * ipa-polymorphic-call.c
18013 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18014 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18015 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18017 * modulo-sched.c (apply_reg_moves): Likewise.
18018 * omp-expand.c (build_omp_regions_1): Likewise.
18019 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18020 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18021 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18022 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18023 * value-prof.c: Likewise.
18024 * var-tracking.c (val_reset): Likewise.
18026 2017-04-03 Richard Biener <rguenther@suse.de>
18028 PR tree-optimization/80275
18029 * fold-const.c (split_address_to_core_and_offset): Handle
18032 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18034 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18035 descriptors is at least equal to that of functions.
18037 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18039 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18041 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18044 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18045 (mov<IMOD4:mode>): New expander.
18046 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18048 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18050 PR rtl-optimization/79405
18051 * fwprop.c (propagations_left): New variable.
18052 (forward_propagate_into): Decrement it.
18053 (fwprop_init): Initialize it.
18054 (fw_prop): If the variable has reached zero, stop propagating.
18055 (fwprop_addr): Ditto.
18057 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18060 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18061 a FUNCTION_DECL, pass it as decl instead of origin to
18064 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18066 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18069 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18072 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18073 TARGET_VSX_SMALL_INTEGER.
18075 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18077 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18078 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18080 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18082 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18083 extraction from odd-numbered MSA register.
18085 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18087 PR middle-end/80173
18088 * expmed.c (store_bit_field_1): Don't attempt to create
18089 a word subreg out of hard registers wider than word if they
18090 have HARD_REGNO_NREGS of 1 for their mode.
18092 PR middle-end/80163
18093 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18094 conversions to integer types wider than word and pointer.
18097 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18098 (rtx_equal_for_cselib_p): Pass 0 to it.
18099 * cselib.c (cselib_hasher::equal): Likewise.
18100 (rtx_equal_for_cselib_1): Add depth argument. If depth
18101 is 128, don't look up VALUE locs and punt. Increment
18102 depth in recursive calls when walking VALUE locs.
18104 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18106 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18107 (make_gcov_file_name): Use the canonical path name for generating
18109 (read_line): Fix handling of files with ascii null bytes.
18111 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18113 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18114 to initialise a vector register instead
18115 of using a const_int.
18117 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18119 PR translation/80189
18120 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18121 diagnostic messages.
18123 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18126 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18127 (dfp_diex_<mode>): Update mode of operand 1.
18128 * doc/extend.texi (dxex, dxexq): Document change to return type.
18129 (diex, diexq): Document change to argument type.
18131 2017-03-30 Martin Jambor <mjambor@suse.cz>
18134 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18135 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18136 it reflects the signature changes performed at the callee side.
18137 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18138 to cgraph_build_function_type_skip_args.
18139 (build_function_decl_skip_args): Adjust call to the above function.
18141 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18144 * config/i386/sse.md
18145 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18146 register as dest whenever it is a MEM not rtx_equal_p to the
18147 corresponding dup operand, and when forcing into reg move the
18148 reg into the memory afterwards.
18149 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18150 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18151 for the force_reg mode.
18152 (avx512vl_vextractf128<mode>): Use register as dest either
18153 always when a MEM, or when it is a MEM not rtx_equal_p to the
18154 corresponding dup operand, or even not when it is a CONST_VECTOR
18155 depending on the mode and lo vs. hi.
18156 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18158 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18159 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18160 Likewise. Require that operands[2] is even.
18161 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18162 Remove extraneous parens. Require that operands[2] is a multiple
18164 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18165 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18166 disallow memory then.
18168 2017-03-30 Richard Biener <rguenther@suse.de>
18170 PR tree-optimization/77498
18171 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18172 to non-constants over backedges.
18174 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18176 PR rtl-optimization/80233
18177 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18178 as last_combined_insn. Do not test for BARRIER_P separately.
18180 2017-03-29 Andreas Schwab <schwab@suse.de>
18183 * calls.c (prepare_call_address): Convert funexp to Pmode before
18184 copying to temp reg.
18186 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18188 PR tree-optimization/80158
18189 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18190 Handle possible future case of more than one alternate
18192 (replace_rhs_if_not_dup): Likewise.
18193 (replace_one_candidate): Likewise.
18195 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18197 PR rtl-optimization/80193
18198 * ira.c (ira): Do not check allocation for LRA.
18200 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18202 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18203 (nvptx_output_simt_exit): Declare.
18204 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18205 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18206 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18207 (nvptx_declare_function_name): ...here. Emit declaration of local
18208 memory space buffer for omp_simt_enter insn.
18209 (nvptx_output_unisimt_switch): New.
18210 (nvptx_output_softstack_switch): New.
18211 (nvptx_output_simt_enter): New.
18212 (nvptx_output_simt_exit): New.
18213 * config/nvptx/nvptx.h (struct machine_function): New fields
18214 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18215 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18216 (UNSPECV_SIMT_EXIT): Ditto.
18217 (omp_simt_enter_insn): New insn.
18218 (omp_simt_enter): New expansion.
18219 (omp_simt_exit): New insn.
18220 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18222 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18223 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18224 (expand_GOMP_SIMT_EXIT): New.
18225 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18226 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18227 (GOMP_SIMT_EXIT): Ditto.
18228 * target-insns.def (omp_simt_enter): New insn.
18229 (omp_simt_exit): Ditto.
18230 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18232 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18233 (lower_rec_input_clauses): Likewise.
18234 (lower_lastprivate_clauses): Handle SIMT privatization.
18236 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18237 (ompdevlow_adjust_simt_enter): New.
18238 (find_simtpriv_var_op): New.
18239 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18240 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18242 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18243 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18244 (copy_decl_for_dup_finish): Ditto.
18246 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18248 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18251 * config/i386/i386.c (ix86_option_override_internal): Always
18252 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18254 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18256 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18258 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18260 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18261 mark new edge's irreducible flag accordign to it.
18262 (vect_do_peeling): Check loop preheader edge's irreducible flag
18263 and pass it to function slpeel_add_loop_guard.
18265 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18267 PR tree-optimization/80218
18268 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18269 Update block frequencies and counts.
18271 2017-03-28 Richard Biener <rguenther@suse.de>
18273 PR tree-optimization/78644
18274 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18275 of a simplification result we may not use it at all.
18277 2017-03-28 Richard Biener <rguenther@suse.de>
18280 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18281 without arguments, generate default definition of a SSA name.
18283 2017-03-28 Richard Biener <rguenther@suse.de>
18285 PR middle-end/80222
18286 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18287 TYPE_REF_CAN_ALIAS_ALL references.
18288 * fold-const.c (fold_indirect_ref_1): Likewise.
18290 2017-03-28 Martin Liska <mliska@suse.cz>
18293 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18294 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18296 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18297 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18299 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18300 (EXTRA_SPECS): Define.
18301 (SUBTARGET_EXTRA_SPECS): Likewise.
18302 (SUBTARGET_CPP_SPEC): Likewise.
18303 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18304 SUBTARGET_EXTRA_SPECS.
18305 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18307 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18309 * config/arc/simdext.md (vst64_insn): Update pattern.
18310 (vld32wh_insn): Likewise.
18311 (vld32wl_insn): Likewise.
18312 (vld64_insn): Likewise.
18313 (vld32_insn): Likewise.
18315 2017-03-28 Marek Polacek <polacek@redhat.com>
18318 * fold-const.c (fold_comparison): Use protected_set_expr_location
18319 instead of SET_EXPR_LOCATION.
18321 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18323 * tree.c (add_expr): Avoid name lookup warning.
18325 2017-03-27 Jeff Law <law@redhat.com>
18327 PR tree-optimization/80216
18328 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18329 function name. Limit recursion depth.
18330 (record_temporary_equivalences): Corresponding changes.
18332 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18334 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18337 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18340 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18342 * cfgcleanup.c (reg_note_cfa_p): New array.
18343 (insns_have_identical_cfa_notes): New function.
18344 (old_insns_match_p): Don't cross-jump in between /f
18345 and non-/f instructions. If both i1 and i2 are frame related,
18346 verify all CFA notes, their order and content.
18348 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18351 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18352 HImode and SImode with zero extend to DImode to one insn.
18353 (bswap<mode>2_extenddi): Likewise.
18354 (bswapsi2_extenddi): Likewise.
18355 (bswaphi2_extendsi): Likewise.
18356 (bswaphi2): Combine bswap HImode and SImode into one insn.
18357 Separate memory insns from swapping register.
18358 (bswapsi2): Likewise.
18359 (bswap<mode>2): Likewise.
18360 (bswaphi2_internal): Delete, no longer used.
18361 (bswapsi2_internal): Likewise.
18362 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18363 store, and gpr<-gpr swap insns.
18364 (bswap<mode>2_store): Likewise.
18365 (bswaphi2_reg): Register only splitter, combine with the splitter.
18366 (bswaphi2 splitter): Likewise.
18367 (bswapsi2_reg): Likewise.
18368 (bswapsi2 splitter): Likewise.
18369 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18370 the insns into load, store, and register/register insns.
18371 (bswapdi2_ldbrx): Likewise.
18372 (bswapdi2_load): Likewise.
18373 (bswapdi2_store): Likewise.
18374 (bswapdi2_reg): Likewise.
18376 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18378 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18379 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18381 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18384 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18386 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18387 special handling for target option conflicts between dform
18388 options (-mpower9-dform, -mpower9-dform-vector,
18389 -mpower9-dform-scalar) and -mno-direct-move.
18391 2017-03-27 Richard Biener <rguenther@suse.de>
18393 PR tree-optimization/80181
18394 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18396 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18398 * config/arc/predicates.md (move_double_src_operand): Replace the
18399 call to move_double_src_operand with a call to address_operand.
18401 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18403 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18404 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18405 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18407 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18409 * config/arc/predicates.md (long_immediate_loadstore_operand):
18410 Consider scaled addresses cases.
18412 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18414 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18415 restored when in interrupt.
18416 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18417 doesn't have delay slot.
18419 2017-03-27 Richard Biener <rguenther@suse.de>
18422 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18423 inlined thunk clones.
18425 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18428 * asan.c (instrument_derefs): Copy over last operand from
18429 original COMPONENT_REF to the new COMPONENT_REF with
18430 DECL_BIT_FIELD_REPRESENTATIVE.
18431 * ubsan.c (instrument_object_size): Likewise.
18433 2017-03-27 Richard Biener <rguenther@suse.de>
18435 PR tree-optimization/80170
18436 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18437 sure DR/SCEV didnt fold in constants we do not see when looking
18438 at the reference base alignment.
18440 2017-03-27 Richard Biener <rguenther@suse.de>
18442 PR middle-end/80171
18443 * gimple-fold.c (fold_ctor_reference): Properly guard against
18444 NULL return value from canonicalize_constructor_val.
18446 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18449 * config/i386/i386.c (ix86_expand_builtin)
18450 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18451 flags reg setting and flags reg using instructions.
18452 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18453 clobbering instructions to zero extend op2.
18455 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18457 * doc/install.texi (Configuration) <--with-aix-soname>:
18458 Update link to AIX ld.
18460 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18462 PR rtl-optimization/80160
18463 PR rtl-optimization/80159
18464 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18465 reg_alternate_class into account.
18467 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18470 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18471 to consider in curr_insn_transform.
18473 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18475 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18476 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18477 and emit_mode_inner.
18479 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18481 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18482 argument to the overloaded builtin variants. Use the new flag to
18483 deprecate certain builtin variants.
18484 * config/s390/s390-builtin-types.def: Add new builtin types.
18485 * config/s390/s390-builtins.h: Support new flags field for
18486 overloaded builtins.
18487 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18488 (s390_macro_to_expand): Enable vector float data type.
18489 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18490 builtins by incrementing the __VEC__ version number.
18491 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18493 (s390_resolve_overloaded_builtin): Emit error messages depending
18494 on the builtin flags.
18495 * config/s390/s390.c (s390_expand_builtin): Support additional
18496 flags argument. Change error message to match the messages
18497 emitted in s390-c.c.
18498 * config/s390/s390.md: New UNSPEC_* constants.
18499 (op_type): Add new instruction types.
18500 * config/s390/vecintrin.h: Add new builtins and test data class
18502 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18503 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18504 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18505 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18506 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18507 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18509 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18510 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18511 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18512 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18514 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18515 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18516 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18517 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18519 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18520 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18521 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18522 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18523 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18524 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18525 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18527 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18528 ("vec_scatter_element<V_HW_4:mode>_DI")
18529 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18530 ("vec_fpint<mode>", "vflls")
18531 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18532 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18533 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18534 ("*vec_cmphe<mode>_cc"): ... these.
18536 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18537 mode constant instead of magic value.
18539 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18541 * config/s390/s390.c (s390_expand_vec_compare): Support other
18542 vector floating point modes than just V2DF.
18543 (s390_expand_vcond): Likewise.
18544 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18545 (s390_cannot_change_mode_class): Prevent mode changes between TF
18546 and V1TF in vector registers.
18547 * config/s390/s390.md (DF, SF): New mode attributes.
18548 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18549 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18550 SFmode support for VRs.
18551 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18553 (VFT, VF_HW): New mode iterators.
18554 (vw, sdx): New mode attributes.
18555 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18556 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18557 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18558 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18559 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18560 also the new vector floating point modes. Renaming to ...
18562 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18563 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18564 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18565 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18566 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18567 ("vec_unordered<mode>"): ... these.
18569 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18570 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18571 ("*vec_extendv2df"): New insn definitions.
18573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18575 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18576 ("mulditi3_2", "*muldi3_sign"): New patterns.
18577 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18578 rename the pattern definition.
18580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18582 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18584 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18586 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18588 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18589 instruction if possible.
18590 * config/s390/vector.md (vec_halfnumelts): New mode
18592 ("*vec_vllezlf<mode>"): New pattern.
18594 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18596 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18597 ("popcountv4si2", "popcountv2di2"): Rename to ...
18598 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18599 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18601 ("popcount<mode>2_vxe"): New pattern.
18603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18605 * common/config/s390/s390-common.c (processor_flags_table): Add
18607 * config.gcc: Add arch12.
18608 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18609 Default to arch12 for unknown CPU model numbers.
18610 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18611 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18612 PROCESSOR_max sanity check.
18613 * config/s390/s390-opts.h (enum processor_type): Add
18615 * config/s390/s390.c (processor_table): Add arch12.
18616 (s390_expand_builtin): Add check for B_VXE flag.
18617 (s390_issue_rate): Add PROCESSOR_ARCH12.
18618 (s390_get_sched_attrmask): Likewise.
18619 (s390_get_unit_mask): Likewise.
18620 (s390_sched_score): Enable z13 scheduling for arch12.
18621 (s390_sched_reorder): Likewise.
18622 (s390_sched_variable_issue): Likewise.
18623 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18625 (s390_tune_attr): Use z13 scheduling also for arch12.
18626 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18627 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18628 (TARGET_VXE_P): New macros.
18629 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18630 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18631 * config/s390/s390.opt: Add arch12 as processor_type.
18633 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18635 * config/s390/s390.md
18636 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18637 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18638 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18640 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18641 Rename expanders to ...
18643 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18644 ("fixuns_truncdddi2_emu"): ... these.
18646 ("fixuns_trunc<mode>si2_emu"): New expander.
18648 ("*fixuns_truncdfdi2_z13"): Rename to ...
18649 ("*fixuns_truncdfdi2_vx"): ... this.
18651 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18653 * config/s390/2964.md: Remove the single element vector compare
18654 instructions which are no longer used.
18655 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18657 (s390_canonicalize_comparison): Remove handling of DFmode
18659 (s390_expand_vec_compare_scalar): Remove function.
18660 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18661 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18663 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18665 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18667 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18669 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18670 will anyway by matched by mov<mode>_64dfp.
18672 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18674 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18675 vlef/vstef. Add missing operand to vleif.
18677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18679 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18680 pair for all vector types with 64 bit elements.
18681 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18682 * config/s390/vector.md (V_HW_64): ... here.
18683 (V_128_NOSINGLE): New mode iterator.
18684 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18685 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18686 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18687 ("*vec_load_pairv2di"): Change to ...
18688 ("*vec_load_pair<mode>"): ... this one.
18690 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18692 * config/s390/constraints.md: Add comments.
18693 (jKK): Reject element sizes > 8 bytes.
18694 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18696 * config/s390/s390.md: Add the s_operand checks formerly in
18697 s390_split_ok_p to various splitters where they are still
18699 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18700 for 128 bit vectors. Plus two splitters.
18702 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18704 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18707 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18710 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18711 error if the boundary argument is not constant.
18713 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18715 PR rtl-optimization/80112
18716 * loop-doloop.c (doloop_condition_get): Don't check condition
18717 if cmp isn't SET with IF_THEN_ELSE src.
18719 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18721 PR tree-optimization/80158
18722 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
18723 replacing a candidate statement, also replace it for the
18724 candidate's alternate interpretation.
18725 (replace_rhs_if_not_dup): Likewise.
18726 (replace_one_candidate): Likewise.
18728 2017-03-24 Richard Biener <rguenther@suse.de>
18730 PR tree-optimization/80167
18731 * graphite-isl-ast-to-gimple.c
18732 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
18734 (translate_isl_ast_to_gimple::get_rename): Likewise.
18736 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
18738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
18739 handling of certain combinations of target options, including the
18740 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
18741 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
18743 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18746 * config/arm/arm.md (*load_multiple): Add reload_completed to
18747 matching condition.
18749 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18750 Richard Biener <rguenth@suse.de>
18752 PR tree-optimization/79908
18753 PR tree-optimization/80136
18754 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18755 been cast away, gimplify_and_add suffices.
18757 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18759 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18761 2017-03-23 Richard Biener <rguenther@suse.de>
18763 PR tree-optimization/80032
18764 * gimplify.c (gimple_push_cleanup): Forced unconditional
18765 cleanups still have to go to the conditional_cleanups
18768 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18770 PR tree-optimization/80072
18771 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18773 (next_operand_entry_id): Change type to unsigned int.
18774 (sort_by_operand_rank): Make sure to return the right return value
18775 even if unsigned fields are bigger than INT_MAX.
18776 (struct oecount): Change cnt and id type to unsigned int.
18777 (oecount_hasher::equal): Formatting fix.
18778 (oecount_cmp): Make sure to return the right return value
18779 even if unsigned fields are bigger than INT_MAX.
18780 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18783 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18784 TREE_READONLY on result if writing it more than once.
18787 * doc/invoke.texi (-fsanitize=thread): Document that with
18788 -fnon-call-exceptions atomics are not able to throw
18792 * tsan.c: Include tree-eh.h.
18793 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18794 maybe_clean_or_replace_eh_stmt where needed.
18795 (instrument_memory_accesses): Add cfg_changed argument.
18796 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18797 if it returned true.
18798 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18800 PR rtl-optimization/63191
18801 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18802 wrapper function, moved the whole old content into ...
18803 (ix86_delegitimize_address_1): ... this. New inline function.
18804 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18805 true as last argument instead of ix86_delegitimize_address.
18807 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18809 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18810 cortexa57_branch_cost.
18812 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18814 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18816 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18819 * doc/md.texi (Constraints): Document wA constraint.
18820 * config/rs6000/constraints.md (wA): New.
18821 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18822 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18823 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18824 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18826 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18829 * gimplify.c (is_oacc_declared): New function.
18830 (oacc_default_clause): Use it to set default flags for acc declared
18831 variables inside parallel regions.
18832 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18833 declared variables.
18834 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18835 declare attribute to any decl as necessary.
18837 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18840 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18841 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18842 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18843 (arm_arch_lpae): This.
18844 * config/arm/arm.c (arm_arch7ve): Rename into ...
18845 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18846 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18849 2017-03-22 Martin Liska <mliska@suse.cz>
18852 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18853 error message instead of an ICE.
18855 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18857 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18859 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18861 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18866 2017-03-21 Martin Sebor <msebor@redhat.com>
18868 * doc/extend.texi: Use "cannot" instead of "can't."
18869 * doc/hostconfig.texi: Same.
18870 * doc/install.texi: Same.
18871 * doc/invoke.texi: Same.
18872 * doc/loop.texi: Same.
18873 * doc/md.texi: Same.
18874 * doc/objc.texi: Same.
18875 * doc/rtl.texi: Same.
18876 * doc/tm.texi: Same.
18877 * doc/tm.texi.in: Same.
18878 * doc/trouble.texi: Same.
18880 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18883 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18884 (collect_checksum_attributes): Set it.
18885 (die_checksum_ordered): Use it.
18887 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18889 PR tree-optimization/79908
18890 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18891 change: For a VA_ARG whose LHS has been cast away, use
18892 force_gimple_operand to construct the side effects.
18894 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18896 PR translation/80001
18897 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18898 more amenable to translation.
18899 (oacc_loop_auto_partitions): Likewise.
18901 2017-03-21 Marek Polacek <polacek@redhat.com>
18902 Martin Sebor <msebor@redhat.com>
18904 PR tree-optimization/80109
18905 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18906 on INTEGRAL_TYPE_P.
18908 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18909 Segher Boessenkool <segher@kernel.crashing.org>
18912 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18913 check reg_used_between_p between insn and one of succ or succ2
18914 depending on if succ is artificial insn not inserted into insn
18917 2017-03-21 Martin Liska <mliska@suse.cz>
18919 PR gcov-profile/80081
18920 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18921 * doc/gcc.texi: Include gcov-dump stuff.
18922 * doc/gcov-dump.texi: New file.
18924 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
18926 PR rtl-optimization/79150
18927 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
18928 conditional jump, if the jump is the last insn of the loop.
18930 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18931 Richard Biener <rguenth@suse.de>
18933 PR tree-optimization/79908
18934 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18935 been cast away, use force_gimple_operand to construct the side
18938 2017-03-21 Martin Liska <mliska@suse.cz>
18940 PR libfortran/79956
18941 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
18944 2017-03-21 Brad Spengler <spender@grsecurity.net>
18947 * plugin.c (htab_hash_plugin): New function.
18948 (add_new_plugin): Use it and adjust.
18949 (parse_plugin_arg_opt): Adjust.
18950 (init_one_plugin): Likewise.
18952 2017-03-21 Richard Biener <rguenther@suse.de>
18954 PR tree-optimization/80032
18955 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
18956 if set force the cleanup to happen unconditionally.
18957 (gimplify_target_expr): Push inserted clobbers with force_uncond
18958 to avoid them being removed by control-dependent DCE.
18960 2017-03-21 Richard Biener <rguenther@suse.de>
18962 PR tree-optimization/80122
18963 * tree-inline.c (copy_bb): Do not expans va-arg packs or
18964 va_arg_pack_len when the inlined call stmt requires pack
18966 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
18968 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18971 * tsan.c (instrument_builtin_call): If the memory model argument
18972 is not a constant, assume it is valid.
18975 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
18978 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18980 PR rtl-optimization/79910
18981 * combine.c (can_combine_p): Do not allow combining an I0 or I1
18982 if its dest is used by an insn before I2 (other than the combined
18983 insns themselves, which are properly handled already).
18985 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18988 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18990 * combine.c (record_used_regs): New static function.
18991 (try_combine): Handle situations where there is an additional
18992 instruction between I2 and I3 which needs to have a LOG_LINK
18996 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18998 * combine.c (try_combine): Delete redundant i1 test. Call
18999 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19001 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19004 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19005 alternatives 13/14.
19007 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19009 PR tree-optimization/80054
19010 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19011 the optimization if a PHI or any of its arguments is not dominated
19012 by the candidate's basis. Use gphi* rather than gimple* as
19014 (replace_profitable_candidates): Clean up a gimple* variable that
19015 should be a gphi* variable.
19017 2017-03-20 Martin Sebor <msebor@redhat.com>
19020 * doc/extend.texi (attribute constructor): Document present limitation.
19022 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19025 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19026 __POWER9_VECTOR__ #ifdef control, change template definition to
19027 use Power9-specific built-in function.
19028 (vec_any_eq): Likewise.
19029 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19030 to control outcomes from this test.
19031 (vector_ae_<mode>p): For VEC_F modes, likewise.
19033 2017-03-20 Ian Lance Taylor <iant@google.com>
19035 * config/i386/i386.c (ix86_function_regparm): Save an extra
19036 register for -fsplit-stack with DECL_STATIC_CHAIN.
19038 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19041 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19042 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19044 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19046 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19048 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19051 2017-03-20 Marek Polacek <polacek@redhat.com>
19054 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19056 2017-03-20 Richard Biener <rguenther@suse.de>
19058 PR tree-optimization/80113
19059 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19060 allocate extra SSA name for PHI def.
19061 (add_close_phis_to_outer_loops): Likewise.
19062 (add_close_phis_to_merge_points): Likewise.
19063 (copy_loop_close_phi_args): Likewise.
19064 (copy_cond_phi_nodes): Likewise.
19066 2017-03-20 Martin Liska <mliska@suse.cz>
19068 PR middle-end/79753
19069 * tree-chkp.c (chkp_build_returned_bound): Do not build
19070 returned bounds for a LHS that's not a BOUNDED_P type.
19072 2017-03-20 Martin Liska <mliska@suse.cz>
19076 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19077 COMPLEX_CST and VECTOR_CST.
19079 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19082 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19083 target operand. A new splitter adds the clobber statement in case
19084 the target operand is dead anyway.
19086 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19088 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19089 to age-old versions of binutils and glibc.
19091 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19093 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19095 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19097 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19099 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19101 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19102 requirement for binutils 2.13.
19104 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19106 * combine.c (try_combine): Delete redundant i1 test. Call
19107 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19109 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19111 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19112 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19114 <riscv64-*-elf>: Re-arrange section
19115 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19116 <riscv32-*-linux>: Likewise.
19117 <riscv64-*-elf>: Likewise
19118 <riscv64-*-linux>: Likewise.
19120 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19123 * aarch64.opt(verbose-cost-dump): Fix typo.
19125 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19128 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19129 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19131 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19133 * reload.c (find_reloads): When reloading a nonoffsettable address,
19134 use RELOAD_OTHER for it and its address reloads.
19136 PR rtl-optimization/79910
19137 * combine.c (record_used_regs): New static function.
19138 (try_combine): Handle situations where there is an additional
19139 instruction between I2 and I3 which needs to have a LOG_LINK
19142 2017-03-17 Jeff Law <law@redhat.com>
19144 PR tree-optimization/71437
19145 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19146 conditional in the hash table first.
19147 (vrp_dom_walker::before_dom_children): Extract condition from
19148 ASSERT_EXPR. Record condition, its inverion and any implied
19149 conditions as well.
19151 2017-03-17 Marek Polacek <polacek@redhat.com>
19152 Markus Trippelsdorf <markus@trippelsdorf.de>
19154 PR tree-optimization/80079
19155 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19158 2017-03-17 Richard Biener <rguenther@suse.de>
19160 PR middle-end/80075
19161 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19162 Properly verify the LHS before the RHS possibly claims to be
19164 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19167 2017-03-17 Martin Jambor <mjambor@suse.cz>
19169 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19170 (List of -O2 options): Likewise.
19171 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19174 2017-03-17 Tom de Vries <tom@codesourcery.com>
19176 * gcov-dump.c (print_usage): Print bug_report_url.
19178 2017-03-17 Richard Biener <rguenther@suse.de>
19180 PR middle-end/80050
19181 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19182 (parser::peek): Likewise.
19184 2017-03-17 Richard Biener <rguenther@suse.de>
19186 PR tree-optimization/80048
19187 * sese.c (free_sese_info): Properly release rename_map and
19188 copied_bb_map elements.
19190 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19192 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19193 Add linked-list forward and backlinks. Insert on
19194 construction, remove on destruction.
19195 (class pass_store_merging): Add m_stores_head field.
19196 (pass_store_merging::terminate_and_process_all_chains):
19197 Iterate over m_stores_head list.
19198 (pass_store_merging::terminate_all_aliasing_chains):
19200 (pass_store_merging::execute): Check for debug stmts first.
19201 Push new chains onto the m_stores_head stack.
19203 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19206 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19207 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19208 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19210 2017-03-16 Jeff Law <law@redhat.com>
19212 PR tree-optimization/71437
19213 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19214 member function. Implementation moved into after_dom_children
19215 member function and into the threader's thread_outgoing_edges
19217 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19218 some code into new thread_outgoing_edges.
19219 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19220 definition. Simplify marker handling (do it here). Assume we always
19221 have the available expression and the const/copies tables.
19222 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19224 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19225 * tree-vrp.c (equiv_stack): No longer file scoped.
19226 (vrp_dom_walker): New class.
19227 (vrp_dom_walker::before_dom_children): New member function.
19228 (vrp_dom_walker::after_dom_children): Likewise.
19229 (identify_jump_threads): Setup domwalker. Use it rather than
19230 walking edges in a random order by hand. Simplify setup/finalization.
19231 (finalize_jump_threads): Remove.
19232 (vrp_finalize): Do not call identify_jump_threads here.
19233 (execute_vrp): Do it here instead and call thread_through_all_blocks
19236 PR tree-optimization/71437
19237 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19239 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19241 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19242 (dom_opt_dom_walker::thread_across_edge): Remove
19243 handle_dominating_asserts argument. All callers changed.
19244 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19245 changes. Remove calls to lhs_of_dominating_assert. Other
19246 uses of handle_dominating_asserts turn into unconditional code
19247 (simplify_control_stmt_condition_1): Likewise.
19248 (simplify_control_stmt_condition): Likewise.
19249 (thread_through_normal_block, thread_across_edge): Likewise.
19250 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19251 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19252 object if it is not an SSA_NAME.
19253 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19254 before calling into the VRP specific simplifiers.
19255 (identify_jump_threads): Remove handle_dominating_asserts
19258 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19261 * tree-diagnostic.c (default_tree_printer): No longer static.
19262 * tree-diagnostic.h (default_tree_printer): New prototype.
19264 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19266 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19267 Change ins into fmov.
19269 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19271 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19272 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19273 Use h_con constraint for operand 1.
19274 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19275 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19277 2017-03-15 Jeff Law <law@redhat.com>
19279 PR tree-optimization/71437
19280 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19281 (record_temporary_equivalences): Use it.
19283 PR tree-optimization/71437
19284 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19285 tree-ssa-scopedtables.
19286 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19287 (record_conditions, record_cond, vuse_eq): Likewise.
19288 (record_edge_info): Adjust to API tweak of record_conditions.
19289 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19290 (record_temporary_equivalences, optimize_stmt): Likewise.
19291 (eliminate_redundant_computations): Likewise.
19292 (record_equivalences_from_stmt): Likewise.
19293 * tree-ssa-scopedtables.c: Include options.h and params.h.
19294 (vuse_eq): New function, moved from tree-ssa-dom.c
19295 (build_and_record_new_cond): Likewise.
19296 (record_conditions): Likewise. Accept vector of conditions rather
19297 than edge_equivalence structure for first argument.
19298 for the first argument.
19299 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19300 from tree-ssa-dom.c.
19301 (avail_exprs_stack::record_cond): Likewise.
19302 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19303 from tree-ssa-dom.c.
19304 (avail_exprs_stack): Add new member functions lookup_avail_expr
19306 (record_conditions): Declare.
19308 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19311 * lra-constraints.c (process_alt_operands): Increase reject for
19312 reloading an input/output operand.
19314 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19317 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19318 insns to convert from signed/unsigned char/short to IEEE 128-bit
19320 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19322 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19325 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19326 subreg of inner mode for values already in registers.
19328 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19330 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19331 iteration reg is used after the loop.
19333 2017-03-14 Martin Sebor <msebor@redhat.com>
19335 PR tree-optimization/79800
19336 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19337 precision in negative-positive range.
19338 (format_floating): Call non-const overload with adjusted precision.
19340 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19343 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19346 2017-03-14 Martin Sebor <msebor@redhat.com>
19348 PR middle-end/80020
19349 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19350 * builtins.def (aligned_alloc): Use it.
19353 * Makefile.in (GTFILES): Add calls.c.
19354 * calls.c: Include "gt-calls.h".
19356 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19358 PR rtl-optimization/79728
19359 * regs.h (struct target_regs): New field
19360 x_contains_allocatable_regs_of_mode.
19361 (contains_allocatable_regs_of_mode): New macro.
19362 * reginfo.c (init_reg_sets_1): Initialize it, and change
19363 contains_reg_of_mode so it includes global regs as well.
19364 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19365 rather than contains_regs_of_mode.
19367 2017-03-14 Martin Liska <mliska@suse.cz>
19369 * doc/invoke.texi: Document options that can't be combined with
19370 -fcheck-pointer-bounds.
19372 2017-03-14 Martin Liska <mliska@suse.cz>
19374 PR middle-end/79831
19375 * doc/invoke.texi (-Wchkp): Document the option.
19377 2017-03-14 Martin Liska <mliska@suse.cz>
19379 * Makefile.in: Install gcov-dump.
19381 2017-03-14 Martin Liska <mliska@suse.cz>
19383 * multiple_target.c (expand_target_clones): Bail out for
19384 an invalid attribute.
19386 2017-03-14 Richard Biener <rguenther@suse.de>
19388 * alias.c (struct alias_set_entry): Pack properly.
19389 * cfgloop.h (struct loop): Likewise.
19390 * cse.c (struct set): Likewise.
19391 * ipa-utils.c (struct searchc_env): Likewise.
19392 * loop-invariant.c (struct invariant): Likewise.
19393 * lra-remat.c (struct cand): Likewise.
19394 * recog.c (struct change_t): Likewise.
19395 * rtl.h (struct address_info): Likewise.
19396 * symbol-summary.h (function_summary): Likewise.
19397 * tree-loop-distribution.c (struct partition): Likewise.
19398 * tree-object-size.c (struct object_size_info): Likewise.
19399 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19400 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19401 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19402 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19403 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19404 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19405 (struct _stmt_vec_info): Likewise.
19407 2017-03-14 Martin Liska <mliska@suse.cz>
19410 * multiple_target.c (create_dispatcher_calls): Check that
19411 a target can create a function dispatcher.
19413 2017-03-14 Martin Liska <mliska@suse.cz>
19416 * multiple_target.c (expand_target_clones): Drop local.local
19417 flag for default implementation.
19419 2017-03-14 Richard Biener <rguenther@suse.de>
19421 PR tree-optimization/80030
19422 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19424 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19426 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19427 gcc_fallthrough() instead of __attribute__((fallthrough));
19429 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19431 * doc/gcc.texi: Remove "up" link to (DIR).
19432 * doc/gccint.texi: Ditto.
19434 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19436 * doc/install.texi (Specific) <avr>: Remove reference to
19439 2017-03-13 Jeff Law <law@redhat.com>
19441 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19442 attribute rather than comments.
19444 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19445 match_scratch operand is highest.
19447 2017-03-13 Martin Liska <mliska@suse.cz>
19449 PR middle-end/78339
19450 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19451 is a CHKP clone, use original declaration.
19453 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19455 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19456 (arc_conditional_register_usage): Use a different allocation order
19457 when optimizing for size.
19458 * common/config/arc/arc-common.c (arc_option_optimization_table):
19459 Section anchors default on when optimizing for size.
19461 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19463 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19465 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19467 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19468 * config/arc/arc.md (cpu_facility): Add cd variant.
19469 (*movqi_insn): Add code density variant.
19470 (*movhi_insn): Likewise.
19471 (*movqi_insn): Likewise.
19472 (*addsi3_mixed): Likewise.
19473 (subsi3_insn): Likewise.
19475 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19477 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19479 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19481 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19482 expressions with MINUS and UNARY ops.
19484 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19487 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19489 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19490 between vec_select and vector argument.
19491 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19492 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19493 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19494 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19495 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19496 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19498 2017-03-13 Richard Biener <rguenther@suse.de>
19501 * params.def (vect-max-peeling-for-alignment): Fix typo.
19503 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19505 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19506 issue that only occurred with binutils below 2.18.
19508 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19510 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19511 refer to binutils 2.11/2.12 minimum.
19513 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19515 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19516 ftp.kernel.org and simplify binutils requirement.
19518 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19520 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19522 (Optimize Options): Ditto. Also remove redundancy.
19524 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19526 PR translation/79848
19527 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19529 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19530 to G_ to avoid double translation.
19532 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19534 PR translation/79923
19535 * auto-profile.c (get_combined_location): Convert leading
19536 character of diagnostics to lower case and remove trailing period.
19537 (read_profile): Likewise for various diagnostics.
19538 * config/arm/arm.c (arm_option_override): Remove trailing period
19539 from various diagnostics.
19540 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19541 (msp430_expand_delay_cycles): Likewise.
19543 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19546 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19547 full command-line argument, rather than just "str".
19548 (aarch64_validate_march): Likewise.
19549 (aarch64_validate_mtune): Likewise.
19551 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19553 PR rtl-optimization/78911
19554 * lra-assigns.c (must_not_spill_p): New function.
19555 (spill_for): Use it.
19557 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19559 PR tree-optimization/79981
19560 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19561 ATOMIC_COMPARE_EXCHANGE ifn result.
19562 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19563 IFN_ATOMIC_COMPARE_EXCHANGE.
19565 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19568 * opts.c (parse_sanitizer_options): Add missing question mark to
19569 "did you mean" message.
19571 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19573 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19575 (VMULEUH_UNS): Likewise.
19576 (VMULOUB_UNS): Likewise.
19577 (VMULOUH_UNS): Likewise.
19578 * config/rs6000/rs6000.c (builtin_function_type): Remove
19579 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19581 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19584 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19585 x with result of extra_parsing_for_operand_code_0.
19586 (function_reader::extra_parsing_for_operand_code_0): Convert
19587 return type from void to rtx, returning x. When reading
19588 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19589 larger size containing struct block_symbol.
19591 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19594 -mfloat128-hardware without -m64.
19596 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19599 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19600 entries to the case statement that marks unsigned arguments to
19601 overloaded functions.
19603 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19606 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19608 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19611 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19612 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19614 2017-03-10 Martin Liska <mliska@suse.cz>
19618 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19619 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19622 2017-03-10 Olivier Hainque <hainque@adacore.com>
19624 * tree-switch-conversion (array_value_type): Start by resetting
19625 candidate type to it's main variant.
19627 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19629 PR rtl-optimization/79909
19630 * combine.c (try_combine): Use simplify_replace_rtx on individual
19631 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19632 of the whole CALL_INSN_FUNCTION_USAGE.
19634 PR tree-optimization/79972
19635 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19636 get_range_info on SSA_NAMEs. Formatting fixes.
19638 2017-03-10 Richard Biener <rguenther@suse.de>
19639 Jakub Jelinek <jakub@redhat.com>
19641 PR tree-optimization/77975
19642 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19643 edge to be constant.
19644 (get_val_for): For constant x return it. Formatting fix.
19645 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19646 would use the same bases as the current one.
19648 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19650 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19651 instead of vec_select for V1TImode.
19652 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19654 (VSX_LE_128): Add V1TI to this mode iterator.
19655 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19656 (*vsx_le_perm_store_<mode>): Likewise.
19657 (pre-reload splitter for VSX stores): Likewise.
19658 (post-reload splitter for VSX stores): Likewise.
19659 (*vsx_xxpermdi2_le_<mode>): Likewise.
19660 (*vsx_lxvd2x2_le_<mode>): Likewise.
19661 (*vsx_stxvd2x2_le_<mode>): Likewise.
19663 2017-03-09 Michael Eager <eager@eagercon.com>
19665 Correct failures with --enable-checking=yes,rtl.
19667 * config/microblaze/microblaze.c (microblaze_expand_shift):
19668 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19669 test for const0_rtx.
19670 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19671 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19673 2017-03-09 Richard Biener <rguenther@suse.de>
19675 PR tree-optimization/79977
19676 * graphite-scop-detection.c (scop_detection::merge_sese):
19677 Handle the case of extra exits to blocks dominating the entry.
19679 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19681 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19684 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19686 PR rtl-optimization/79949
19687 * lra-constraints.c (process_alt_operands): Check memory when
19688 trying to predict a cycle. Print about the overall increase.
19690 2017-03-09 Richard Biener <rguenther@suse.de>
19692 PR middle-end/79971
19693 * gimple-expr.c (useless_type_conversion_p): Preserve
19694 TYPE_SATURATING for fixed-point types.
19696 2017-03-09 Richard Biener <rguenther@suse.de>
19699 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19700 alignment of BLKmode params.
19702 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19705 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19706 (VALL_NO_V2Q): Likewise.
19708 * config/aarch64/aarch64-simd.md
19709 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19711 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19712 VALL_NO_V2Q mode iterator.
19713 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
19715 2017-03-09 Martin Liska <mliska@suse.cz>
19717 PR tree-optimization/79631
19718 * tree-chkp-opt.c (chkp_is_constant_addr): Call
19719 tree_int_cst_sign_bit just for INTEGER constants.
19721 2017-03-09 Martin Liska <mliska@suse.cz>
19725 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
19728 2017-03-09 Marek Polacek <polacek@redhat.com>
19731 * tree.c (inchash::add_expr): Handle TREE_VEC.
19733 2017-03-09 Martin Liska <mliska@suse.cz>
19736 (chkp_narrow_size_and_offset): New function.
19737 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
19738 (void chkp_parse_bit_field_ref): New function.
19739 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
19740 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
19742 2017-03-09 Martin Liska <mliska@suse.cz>
19745 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
19746 (chkp_find_bounds_1): Remove gcc_unreachable.
19748 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19751 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19752 BUILT_IN_SYNC*, determine the access type from the size suffix and
19753 always build a MEM_REF with that type. Handle forgotten
19754 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19757 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19758 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19759 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19760 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19761 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19762 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19763 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19764 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19765 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19766 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19767 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19768 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19769 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19770 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19771 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19772 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19773 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19774 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19775 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19776 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19777 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19778 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19779 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19780 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19781 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19782 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19783 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19784 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19785 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19786 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19787 definitions outside of __OPTIMIZE__ guarded section.
19790 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19791 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19792 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19793 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19796 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19798 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19799 ("vfenez<mode>"): Add missing constraints.
19801 2017-03-08 Martin Sebor <msebor@redhat.com>
19804 * config/nds32/nds32.c (nds32_option_override):
19805 Fix misspelled diagnostic.
19807 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19810 * gimplify.c (gimplify_omp_for): Replace index var in outer
19811 taskloop statement with an artificial variable and add
19812 OMP_CLAUSE_PRIVATE clause for it.
19814 2017-03-08 Richard Biener <rguenther@suse.de>
19816 PR tree-optimization/79955
19817 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19818 for accesses that are completely outside of the variable.
19820 2017-03-08 Andrew Haley <aph@redhat.com>
19822 PR tree-optimization/79943
19823 * tree-ssa-loop-split.c (compute_new_first_bound): When
19824 calculating the new upper bound, (END-BEG) should be added, not
19827 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19829 * config/avr/avr.md (setmemhi): Make sure match_dup
19830 operand number comes before match_scratch.
19832 2017-03-08 Richard Biener <rguenther@suse.de>
19834 PR tree-optimization/79920
19835 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19836 with ncopies == 1 to ...
19837 (vect_transform_slp_perm_load): ... here. Properly compute
19838 all element loads by iterating VF times over the group. Do
19839 not handle ncopies (computed in a broken way) in
19840 vect_create_mask_and_perm.
19842 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19845 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19846 is a uniform vector, use uniform_vector_p return value instead of
19847 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19849 2017-03-07 Marek Polacek <polacek@redhat.com>
19851 PR middle-end/79809
19852 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19853 (alloca_call_type): Likewise.
19855 2017-03-07 Martin Liska <mliska@suse.cz>
19857 * gcov.c (process_args): Put comment to correct location.
19859 2017-03-07 Martin Liska <mliska@suse.cz>
19861 PR middle-end/68270
19862 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19863 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19864 (chkp_narrow_bounds_for_field): Likewise.
19865 (chkp_parse_array_and_component_ref): Pass one more argument to
19868 2017-03-07 Richard Biener <rguenther@suse.de>
19870 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19873 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19875 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19876 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19878 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19881 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19882 to end of description.
19883 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19885 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19887 PR rtl-optimization/79901
19888 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19890 (*avx512f_<code><mode>3<mask_name>): ... this.
19891 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19892 iterator instead of VI8_AVX2_AVX512BW.
19894 PR rtl-optimization/79901
19895 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19896 min/max expander, expand it using expand_vec_cond_expr.
19899 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19902 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19905 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19906 to void * for PCH reasons.
19907 * dwarf2out.c (output_loc_operands, output_die): Cast
19908 v.val_vec.array to unsigned char *.
19910 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19913 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19916 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19918 PR rtl-optimization/79571
19919 * lra-constraints.c (process_alt_operands): Calculate static
19920 reject and subtract it from overall when only addresses will be
19923 2017-03-06 Julia Koval <julia.koval@intel.com>
19926 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
19927 incoming stack boundary to 128 for 64-bit targets.
19929 2017-03-06 Richard Biener <rguenther@suse.de>
19931 PR tree-optimization/79894
19932 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
19933 to NULL after folding it.
19935 2017-03-06 Richard Biener <rguenther@suse.de>
19937 PR tree-optimization/79824
19938 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
19939 check disabling peeling for gaps.
19941 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
19943 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19944 attributes): Document gettimeofday.
19946 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19948 * config/s390/s390.c (s390_option_override_internal): Set
19949 PARAM_MIN_VECT_LOOP_BOUND
19951 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19953 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
19954 * config/s390/s390.md: Likewise.
19956 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19959 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
19960 (<avx2_avx512>_perm<mode>): Rename to ...
19961 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
19963 (<avx512>_perm<mode>_mask): Rename to ...
19964 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
19966 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
19967 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
19968 instead of VI8F_256_512.
19969 (avx512f_perm<mode>): New define_expand.
19970 (avx512f_perm<mode>_mask): Likewise.
19971 (avx512f_perm<mode>_1<mask_name>): New define_insn.
19972 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
19974 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19976 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
19977 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
19979 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
19981 2017-03-06 Martin Liska <mliska@suse.cz>
19984 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
19985 when having a SSA NAME w/o VAR_DECL assigned to it.
19987 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19989 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
19990 msa_dpsub_<su>_d): Fix MODE for vec_select.
19992 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19994 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
19996 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
19998 2017-03-06 Richard Biener <rguenther@suse.de>
20000 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20001 * plugin.c (register_plugin_info): Likewise.
20002 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20004 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20006 * config/i386/sse.md (sse_storehps, sse_storelps,
20007 avx_<castmode><avxsizesuffix>_<castmode>,
20008 avx512f_<castmode><avxsizesuffix>_<castmode>,
20009 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20010 in condition that at least one operand is not a MEM.
20012 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20014 PR middle-end/79805
20015 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20016 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20018 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20019 gimple_call_nothrow_p flag based on whether original builtin can throw.
20020 If it can, emit following stmts on the fallthrough edge.
20021 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20022 don't create new bb if inserting just debug stmts on the edge, try to
20023 insert them on the fallthru bb or just reset debug stmts.
20025 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20028 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20029 restore recog_data (including the operand rtxes inside it) around
20030 the call to get_insn_template.
20032 2017-03-03 Martin Sebor <msebor@redhat.com>
20034 PR tree-optimization/79699
20035 * context.c (context::~context): Free MPFR caches to avoid
20036 a memory leak on program exit.
20038 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20040 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20041 Use wide_int::ulow () instead of .elt (0).
20043 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20045 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20046 (*pushxf): Limit oF constraint to 32bit targets and add oC
20047 constraint for 64bit targets.
20048 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20049 (*pushdf): Change rmF constraint to rmC.
20051 2017-03-03 Martin Liska <mliska@suse.cz>
20053 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20054 Remove unused variable.
20056 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20059 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20060 is a memory operand, increase num_memory.
20061 (ix86_expand_args_builtin): Likewise.
20063 2017-03-03 Jan Hubicka <jh@suse.cz>
20066 * ipa-devirt.c (maybe_record_node): Properly handle
20067 __cxa_pure_virtual visibility.
20069 2017-03-03 Martin Liska <mliska@suse.cz>
20071 PR tree-optimization/79803
20072 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20074 (pass_loop_prefetch::execute): Disabled optimization if an
20075 assumption about L1 cache size is not met.
20077 2017-03-03 Martin Liska <mliska@suse.cz>
20079 PR rtl-optimization/79574
20080 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20081 (hash_scan_set): Likewise.
20082 (dump_hash_table): Likewise.
20083 (hoist_code): Likewise.
20085 2017-03-03 Richard Biener <rguenther@suse.de>
20087 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20088 in place of uhigh ().
20089 (fixed_convert_from_real): Likewise.
20091 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20094 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20096 2017-03-03 Richard Biener <rguenther@suse.de>
20098 PR middle-end/79818
20099 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20100 TYPE_OVERFLOW_UNDEFINED check.
20102 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20104 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20106 (vector_ae_<mode>_p): Likewise.
20107 (vector_nez_<mode>_p): Likewise.
20108 (vector_ne_v2di_p): Likewise.
20109 (vector_ae_v2di_p): Likewise.
20110 (vector_ne_<mode>_p): Likewise.
20111 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20113 (vsx_tsqrt<mode>2_fe): Likewise.
20115 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20118 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20120 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20122 PR rtl-optimization/79780
20123 * cprop.c (one_cprop_pass): When second and further conditional trap
20124 in a single basic block is turned into an unconditional trap, turn it
20125 into a deleted note to avoid RTL verification failures.
20127 2017-03-02 Richard Biener <rguenther@suse.de>
20129 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20131 2017-03-02 Richard Biener <rguenther@suse.de>
20133 PR tree-optimization/79345
20135 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20136 param and abort the walk, returning -1 if it is hit.
20137 (walk_aliased_vdefs): Take a limit param and pass it on.
20138 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20139 defaulting to 0 and return a signed int.
20140 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20141 (check_defs): New helper.
20142 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20143 about uninitialized memory.
20144 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20145 bogus uninitialized warning.
20146 (fixed_convert_from_real): Likewise.
20148 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20150 PR tree-optimization/66768
20151 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20152 iv_use if base object can't be determined.
20154 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20156 PR tree-optimization/79345
20157 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20158 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20159 (get_pattern_stats): Initialize it.
20160 * genemit.c (gen_expand): Verify match_scratch numbers come after
20161 match_operand/match_dup numbers.
20162 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20163 match_scratch numbers.
20164 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20166 * config/s390/s390.md (trunctdsd2): Likewise.
20168 2017-03-02 Richard Biener <rguenther@suse.de>
20170 * wide-int.h (wide_int_storage::operator=): Implement in terms
20173 2017-03-02 Richard Biener <rguenther@suse.de>
20175 PR tree-optimization/79777
20176 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20177 the to insert expression to sth existing.
20179 2017-03-01 Martin Sebor <msebor@redhat.com>
20181 PR middle-end/79692
20182 * gimple-ssa-sprintf.c
20183 (directive::known_width_and_precision): New function.
20184 (format_integer): Use it.
20185 (get_mpfr_format_length): Consider the full range of precision
20186 when computing %g output with the # flag. Set the likely byte
20187 count to 3 rather than 1 when precision is indeterminate.
20188 (format_floating): Correct the lower bound of precision.
20190 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20192 * doc/invoke.texi: Document default code model for 64-bit Linux.
20194 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20197 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20198 udiv rather than div since input pattern is unsigned.
20200 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20202 * config/i386/i386.c (print_reg): Warn for values of
20203 unsupported size in integer register.
20205 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20208 * config/rs6000/predicates.md (current_file_function_operand): Do
20209 not allow self calls to be local if the function is replaceable.
20211 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20214 * config/rs6000/altivec.h (vec_ctz and others): Change the
20215 preprocessor macro that controls conditional compilation from
20216 _ARCH_PWR9 to __POWER9_VECTOR__.
20217 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20218 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20219 control (instead of _ARCH_PWR9 control) so that template
20220 definition uses power9-specific function.
20221 (vec_any_eq): Likewise.
20222 (vec_all_ne): Change macro definition to use a power9-specific
20223 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20224 _ARCH_PWR9 control).
20225 (vec_any_eq) Likewise.
20226 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20227 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20228 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20229 support for xvcmpnedp instruction.
20230 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20231 macro expansion so that Power9 implementation of vec_all_ne does
20232 not use the AltiVec predicate framework.
20233 (VCMPNEH_P): Likewise.
20234 (VCMPNEW_P): Likewise.
20235 (VCMPNED_P): Likewise.
20236 (VCMPNEFP_P): Likewise.
20237 (VCMPNEDP_P): Likewise.
20238 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20239 implementation of vec_any_eq to not use AltiVec predicate
20241 (VCMPAEH_P): Likewise.
20242 (VCMPAEW_P): Likewise.
20243 (VCMPAED_P): Likewise.
20244 (VCMPAEFP_P): Likewise.
20245 (VCMPAEDP_P): Likewise.
20246 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20247 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20248 not use the AltiVec predicate framework.
20249 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20250 of vec_any_eq to not use AltiVec predicate framework.
20251 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20252 support for predefined __POWER9_VECTOR__ macro to indicate that
20253 Power9 instruction selection is enabled.
20254 (altivec_overloaded_builtins): Remove extraneous
20255 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20256 function argument types RS6000_BTI_bool_V16QI and
20257 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20258 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20259 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20260 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20261 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20262 Power9 for implementations of vec_cmpne. Change the signature for
20263 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20264 (representing vec_all_ne) to remove the previously described first
20265 argument of type RS6000_BTI_INTSI, as this was an artifact of
20266 reliance on the AltiVec predicate framework, which is no longer
20267 used in the implementation of these functions. Add
20268 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20269 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20270 since, unlike the AltiVec predicate framework implementation, we
20271 do not share function descriptors between vec_alle and vec_anyeq.
20272 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20273 set of modes that receive special treatment even when
20274 TARGET_P9_VECTOR is true. The special treatment emits code that
20275 does not depend on Power9 instructions.
20276 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20277 define_expand to not rely on AltiVec predicate framework.
20278 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20280 (vector_ne_v2di_p): Change this define_expand to not rely on
20281 AltiVec predicate framework.
20282 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20284 (vector_ne_<mode>_p): Change this define_expand to not rely on
20285 AltiVec predicate framework.
20286 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20288 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20289 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20290 define_insn pattern.
20291 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20292 define_insn pattern because the xvcmpne<VSs>. instruction is not
20294 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20295 instruction is not supported.
20297 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20299 * config/nvptx/nvptx.c: Include intl.h.
20301 2017-03-01 Martin Jambor <mjambor@suse.cz>
20304 * ipa-prop.h (ipa_bits): Removed field known.
20305 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20306 to pointers. Adjusted their comments to warn about their sharing.
20307 (ipcp_transformation_summary): Change bits to a vector of pointers.
20308 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20309 (ipa_get_ipa_bits_for_value): Declare.
20310 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20311 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20312 (ipa_bits_hash_table): Likewise.
20313 (ipa_vr_ggc_hash_traits): Likewise.
20314 (ipa_vr_hash_table): Likewise.
20315 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20316 being pointers and vr_known being removed.
20317 (ipa_set_jf_unknown): Likewise.
20318 (ipa_get_ipa_bits_for_value): New function.
20319 (ipa_set_jfunc_bits): Likewise.
20320 (ipa_get_value_range): New overloaded functions.
20321 (ipa_set_jfunc_vr): Likewise.
20322 (ipa_compute_jump_functions_for_edge): Use the above functions to
20323 construct bits and vr parts of jump functions.
20324 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20325 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20327 (ipcp_grow_transformations_if_necessary): Also allocate
20328 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20330 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20331 them. Fix too long lines.
20332 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20333 vr_known being removed.
20334 (ipa_read_jump_function): Use new setter functions to construct bits
20335 and vr parts of jump functions or set them to NULL.
20336 (write_ipcp_transformation_info): Adjust for bits being pointers.
20337 (read_ipcp_transformation_info): Likewise.
20338 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20340 Include gt-ipa-prop.h.
20341 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20343 (ipcp_store_bits_results): Likewise.
20344 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20345 Do not write to existing jump functions but use a temporary instead.
20347 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20350 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20351 attempt to use its first operand as BIT_FIELD_REF base.
20353 2017-03-01 Richard Biener <rguenther@suse.de>
20355 PR middle-end/79721
20356 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20357 interpolating formula in wrapping arithmetic.
20358 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20360 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20362 PR tree-optimization/79734
20363 * tree-vect-generic.c (expand_vector_condition): Optimize
20364 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20365 Handle VEC_COND_EXPR where comparison has different inner width from
20366 type's inner width.
20368 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20370 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20371 markup, and similar issues. Remove @opindex entries for things
20372 that aren't options. Add missing -mmpy-option entries.
20374 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20376 PR tree-optimization/79737
20377 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20378 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20379 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20380 instead of byte_size. Formatting fix.
20381 (shift_bytes_in_array_right): Formatting fix.
20383 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20386 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20387 condition on optimize for the leaf function test.
20389 2017-02-28 Martin Liska <mliska@suse.cz>
20392 * read-rtl-function.c (function_reader::handle_unknown_directive):
20393 Bail out when one uses -flto.
20395 2017-02-28 Martin Liska <mliska@suse.cz>
20397 * common.opt: Replace space with tabular for options of <number>
20399 * config/i386/i386.opt: Show <number> value for
20400 -mlarge-data-threshold.
20401 * opts.c (print_filtered_help): Do not display number in hexadecimal
20404 2017-02-28 Martin Liska <mliska@suse.cz>
20406 * common.opt: Fix --help=option -Q for options which are of
20409 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20411 * config/i386/i386.c (print_reg): Error out for values
20412 of 8-bit size in invalid integer register.
20414 2017-02-28 Martin Sebor <msebor@redhat.com>
20416 PR tree-optimization/79691
20417 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20419 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20422 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20423 gcc_unreachable with output_operand_lossage.
20425 2017-02-28 Richard Biener <rguenther@suse.de>
20427 PR tree-optimization/79740
20428 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20430 (visit_nary_op): Insert the nary into the hashtable if we
20431 pattern-matched sth.
20432 * tree-ssa-pre.c (eliminate_insert): Robustify.
20434 2017-02-28 Richard Biener <rguenther@suse.de>
20436 PR middle-end/79731
20437 * fold-const.c (decode_field_reference): Reject out-of-bound
20440 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20442 * config/i386/i386.c: Include intl.h.
20443 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20444 instead of just cond ? "..." : "...".
20445 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20446 * coverage.c (read_counts_file): Likewise.
20447 * omp-offload.c: Include intl.h.
20448 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20449 of just cond ? "..." : "...".
20450 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20451 of just cond ? "..." : "...".
20453 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20456 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20458 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20459 'tune for' CPU name.
20460 * config/arm/arm-cpu-data.h: Regenerated.
20462 2017-02-28 Richard Biener <rguenther@suse.de>
20464 PR tree-optimization/79732
20465 * tree-inline.c (expand_call_inline): Do not shadow var.
20467 2017-02-28 Richard Biener <rguenther@suse.de>
20469 PR tree-optimization/79723
20470 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20471 address-space properly.
20473 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20475 * doc/optinfo.texi (Optimization groups): Fix option used for
20477 * doc/invoke.texi (-fopt-info): Document "omp".
20478 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20479 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20480 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20481 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20482 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20484 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20486 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20487 OPTGROUP_OMP with "omp".
20489 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20492 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20493 for arithmetic shift of unsigned V2DI.
20495 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20497 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20498 arc/linux.h headers.
20499 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20500 (LINK_SPEC): Likewise.
20501 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20502 (EXTRA_SPECS): Likewise.
20503 (STARTFILE_SPEC): Likewise.
20504 (ENDFILE_SPEC): Likewise.
20505 (LIB_SPEC): Likewise.
20506 (TARGET_SDATA_DEFAULT): Likewise.
20507 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20508 (MULTILIB_DEFAULTS): Likewise.
20509 (DWARF2_UNWIND_INFO): Likewise.
20510 * config/arc/big.h: New file.
20511 * config/arc/elf.h: Likewise.
20512 * config/arc/linux.h: Likewise.
20513 * config/arc/t-uClibc: Remove.
20515 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20517 PR tree-optimization/77536
20518 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20519 (tree_transform_and_unroll_loop): Use above function to compute the
20520 estimated niter of unrolled loop and use it when scaling profile.
20521 Also use count info rather than frequency if it's non-zero.
20522 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20523 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20524 (vect_transform_loop): Call above function.
20526 2017-02-27 Richard Biener <rguenther@suse.de>
20528 PR tree-optimization/45397
20529 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20530 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20531 (visit_nary_op): Add pattern matching for CSEing sign-changed
20532 or truncated operations with wider ones.
20534 2017-02-27 Richard Biener <rguenther@suse.de>
20536 PR tree-optimization/79690
20537 * tree-vect-stmts.c (vectorizable_store): Use vector type
20538 built from the DR with address-space.
20540 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20542 * doc/invoke.texi (Optimize Options): Refine the description
20543 of asan-use-after-return.
20545 2017-02-25 Alan Modra <amodra@gmail.com>
20547 PR rtl-optimization/79584
20548 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20549 base, not ad->base_term, the reg within base. Remove assertion
20550 that ad->base == ad->base_term. Replace gen_int_mode using
20551 bogus mode with const0_rtx.
20553 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20555 PR middle-end/79396
20556 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20557 FMA_EXPR like tcc_binary or tcc_unary.
20559 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20562 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20564 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20565 (output_loc_operands): Handle DW_OP_call_ref and
20566 DW_OP_GNU_variable_value.
20567 (struct variable_value_struct): New type.
20568 (struct variable_value_hasher): Likewise.
20569 (variable_value_hash): New variable.
20570 (string_types): Remove.
20571 (copy_loc_descr): New function.
20572 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20573 (prepend_loc_descr_to_each): New function.
20574 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20575 instead of add_loc_descr_to_each if the first argument is single
20576 location list and the second has multiple.
20577 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20578 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20579 when looking for variable value which doesn't have other location info.
20580 (loc_list_from_tree): Formatting fix.
20581 (gen_array_type_die): Simplify DW_AT_string_length handling.
20582 (adjust_string_types): Remove.
20583 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20584 string_types. Call resolve_variable_values.
20585 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20586 (resolve_addr_in_expr): Likewise. Add A argument.
20587 (copy_deref_exprloc): Remove deref argument. Adjust for the
20588 original expression being DW_OP_GNU_variable_value with optionally
20589 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20590 optionally after it.
20591 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20592 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20593 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20594 (variable_value_hasher::hash, variable_value_hasher::equal): New
20596 (resolve_variable_value_in_expr, resolve_variable_value,
20597 resolve_variable_values, note_variable_value_in_expr,
20598 note_variable_value): New functions.
20599 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20602 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20605 * opts.h (handle_generated_option): Add GENERATED_P argument.
20606 * opts-common.c (handle_option): Adjust function comment.
20607 (handle_generated_option): Add GENERATED_P argument, pass it to
20609 (control_warning_option): Pass false to handle_generated_option
20611 * opts.c (maybe_default_option): Pass true to handle_generated_option
20613 * optc-gen.awk: Likewise.
20615 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20617 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20618 a REG, look at the REG it is a SUBREG of.
20619 (splitter for cmpeqsi_t): Ditto.
20621 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20623 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20624 the special USEs with the pattern of the insn, not the insn itself.
20626 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20629 * doc/invoke.texi: Document -mload-store-pairs.
20631 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20632 Sandra Loosemore <sandra@codesourcery.com>
20634 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20635 argument isn't a CONST_INT.
20636 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20637 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20638 (nios2_validate_compare): Bypass alternate compare logic if *op2
20639 is not a CONST_INT.
20640 (ldstwm_operation_p): Return false if first_base is not a REG or
20641 if first_offset is not a CONST_INT.
20643 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20645 * config/cris/cris.md: Use correct operand in a define_peephole2.
20647 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20649 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20651 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20653 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20654 this_insn if it is an INSN or JUMP_INSN.
20655 (force_offsettable): Look at base, not at addr.
20656 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20657 on things that aren't necessarily CONST_INTs.
20659 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20661 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20662 -mfpmath=sse is the default also for x86-32 targets with SSE2
20663 instruction set when @option{-ffast-math} is enabled
20665 2017-02-24 Jeff Law <law@redhat.com>
20667 PR rtl-optimizatoin/79286
20668 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20671 2017-02-24 Richard Biener <rguenther@suse.de>
20673 PR tree-optimization/79389
20674 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20677 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20679 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20680 function comment to reflect reality.
20681 (loop_exits_before_overflow): Fix typo in function description.
20683 2017-02-24 Richard Biener <rguenther@suse.de>
20685 PR tree-optimization/79389
20686 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20687 properly that a threading opportunity exists. Detect conditional
20688 copy/constant propagation opportunities.
20690 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20692 * config/visium/visium.md (type): Add trap.
20693 (b): New mode attribute.
20694 (*btst): Rename into...
20695 (*btst<mode>): ...this and adjust.
20696 (*cbranchsi4_btst_insn): Rename into...
20697 (*cbranch<mode>4_btst_insn): ...this and adjust.
20698 (trap): New define_insn.
20700 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20702 PR tree-optimization/79389
20703 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20704 (noce_reversed_cond_code): New function.
20705 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20706 reversed_comparison_code. Formatting fix.
20707 (noce_try_store_flag): Test rev_cond != NULL in addition to
20708 reversed_comparison_code.
20709 (noce_try_store_flag_constants): Likewise.
20710 (noce_try_store_flag_mask): Likewise.
20711 (noce_try_addcc): Use rev_cond if non-NULL instead of
20712 reversed_comparison_code.
20713 (noce_try_cmove_arith): Likewise. Formatting fixes.
20714 (noce_try_minmax, noce_try_abs): Clear rev_cond.
20715 (noce_find_if_block): Initialize rev_cond.
20716 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
20717 instead of false as last argument never attempt to reverse it
20720 2017-02-23 Bin Cheng <bin.cheng@arm.com>
20722 PR tree-optimization/79663
20723 * tree-predcom.c (combine_chains): Process refs in reverse order
20724 only for ZERO length chains, and add explaining comment.
20726 2017-02-23 Jeff Law <law@redhat.com>
20728 PR tree-optimization/79578
20729 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
20730 in call to operand_equal_p.
20732 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
20735 * config/i386/cpuid.h: Fix another undefined behavior.
20737 2017-02-23 Richard Biener <rguenther@suse.de>
20739 PR tree-optimization/79683
20740 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
20741 vector types for data-refs.
20743 2017-02-23 Martin Liska <mliska@suse.cz>
20745 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
20747 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20749 PR middle-end/79665
20750 * internal-fn.c (get_range_pos_neg): Moved to ...
20751 * tree.c (get_range_pos_neg): ... here. No longer static.
20752 * tree.h (get_range_pos_neg): New prototype.
20753 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20754 are known to be in between 0 and signed maximum inclusive, try to
20755 expand both unsigned and signed divmod and use the cheaper one from
20758 2017-02-22 Jeff Law <law@redhat.com>
20760 PR tree-optimization/79578
20761 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20762 to compare base operands.
20764 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20767 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20768 gpc_reg_operand instead of fpr_reg_operand.
20770 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20772 * config/mips/mips.c (mips_return_in_memory): Force FP
20773 vector types to be returned in memory for o32 ABI.
20775 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20777 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20778 instead of DW_TAG_member for static data member declarations and don't
20779 set no_linkage_name for static inline data members.
20780 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20783 2017-02-22 Martin Liska <mliska@suse.cz>
20785 * doc/invoke.texi: Replace inequality signs with square brackets
20788 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20790 PR tree-optimization/68644
20791 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20793 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20796 * lra-constraints.c (simplify_operand_subreg): Handle
20797 WORD_REGISTER_OPERATIONS targets.
20799 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20802 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20803 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20804 elimination by swapping fld*.
20806 2017-02-22 Richard Biener <rguenther@suse.de>
20808 PR tree-optimization/79673
20809 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20810 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20811 irrelevant address-space qualifiers and avoiding a
20812 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20814 2017-02-22 Richard Biener <rguenther@suse.de>
20816 PR tree-optimization/79666
20817 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20818 to not symbolically negate if that may introduce undefined
20821 2017-02-22 Martin Liska <mliska@suse.cz>
20824 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20825 * data-streamer-out.c (streamer_write_gcov_count_stream):
20827 * value-prof.c (stream_out_histogram_value): Make assert more
20828 precise based on type of counter.
20830 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20833 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20834 Use nonimmediate_operand instead of memory_operand for operand 1.
20835 (float-extend standard_x87sse_constant_load splitter): Ditto.
20837 2017-02-21 Jeff Law <law@redhat.com>
20839 PR tree-optimization/79621
20840 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20841 blocks with edges to themselves.
20843 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20846 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20847 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20848 Use gimple_call_builtin_p.
20851 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20852 on temporarily removed DEBUG_INSNs.
20854 PR tree-optimization/79649
20855 * tree-loop-distribution.c (classify_partition): Give up on
20856 non-generic address space loads/stores.
20858 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20860 * doc/loop.texi (Loop manipulation): Remove nonexistent
20861 tree_ssa_loop_version from the documentation.
20862 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20864 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20867 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20868 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20869 * config/rs6000/rs6000.c: Include except.h.
20870 (rs6000_expand_split_stack_prologue): Call
20871 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20873 2017-02-21 Martin Jambor <mjambor@suse.cz>
20876 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20877 have been analyzed.
20879 2017-02-21 Martin Jambor <mjambor@suse.cz>
20881 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20882 for backward compatibility only.
20883 * doc/invoke.texi (Option Summary): Remove all references to
20884 -fipa-cp-alignment.
20886 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20890 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20892 * lra-constraints.c (curr_insn_transform): Handle
20893 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20895 2017-02-21 Martin Liska <mliska@suse.cz>
20897 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20899 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20901 PR translation/79638
20902 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20904 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20907 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20908 (arm_function_ok_for_sibcall): Return false for an indirect call by
20909 descriptor if all the argument registers are used.
20910 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20911 alignment of the function.
20913 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20915 PR tree-optimization/61441
20916 * simplify-rtx.c (simplify_const_unary_operation): For
20917 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20918 the sNaN unmodified.
20920 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20922 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20923 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20924 instead of SYSTEM_HEADER_DIR.
20926 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
20927 Martin Liška <mliska@suse.cz>
20929 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
20930 Fix typos and grammar, use active voice, and clarify.
20932 2017-02-20 Marek Polacek <polacek@redhat.com>
20934 PR middle-end/79537
20935 * gimplify.c (gimplify_expr): Handle unused *&&L;.
20938 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
20940 2017-02-20 Jakub Jelinek <jakub@redhat.com>
20943 * config/i386/i386.c (ix86_expand_builtin): Handle
20944 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
20945 ix86_builtins_isa[fcode].isa as a requirement of those
20946 flags and any other flag in the bitmask.
20947 (ix86_init_mmx_sse_builtins): Use 0 instead of
20948 ~OPTION_MASK_ISA_64BIT as mask.
20949 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
20950 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
20951 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
20952 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
20954 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20957 * lra-constraints.c (split_reg): Check requested split mode
20958 is supported by the register.
20960 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20962 * lra-constraints.c (simplify_operand_subreg): Remove early
20965 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20968 * lra-constraints.c (curr_insn_transform): Tighten condition
20969 for converting SUBREG reloads from OP_OUT to OP_INOUT.
20971 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20974 * lra-constraints.c (curr_insn_transform): Handle
20975 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20977 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
20980 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
20982 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
20984 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
20987 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
20990 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20992 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
20995 2017-02-18 Jakub Jelinek <jakub@redhat.com>
20997 * final.c (last_columnnum, override_columnnum): New variables.
20998 (final_start_function): Set last_columnnum, pass it to begin_prologue
20999 hook and pass 0 to dwarf2out_begin_prologue.
21000 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21001 to source_line debug hook.
21002 (notice_source_line): Compute last_columnnum and for debug_column_info
21003 return true on column changes.
21004 * debug.h (struct gcc_debug_hooks): Add column argument to
21005 source_line and begin_prologue hooks.
21006 (debug_nothing_int_charstar_int_bool): Remove prototype.
21007 (debug_nothing_int_int_charstar,
21008 debug_nothing_int_int_charstar_int_bool): New prototypes.
21009 (dwarf2out_begin_prologue): Add column argument.
21010 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21011 begin_prologue hooks.
21012 (debug_nothing_int_charstar_int_bool): Remove.
21013 (debug_nothing_int_int_charstar,
21014 debug_nothing_int_int_charstar_int_bool): New functions.
21015 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21016 through to dwarf2out_source_line.
21017 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21018 (dwarf2out_source_line): Add column argument, emit it if requested.
21019 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21021 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21022 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21023 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21024 through to dwarf2out_begin_prologue.
21025 (vmsdbgout_source_line): Add column argument, pass it through to
21026 dwarf2out_source_line.
21027 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21028 dbxout_source_line caller.
21029 (dbxout_source_line): Add column argument.
21031 * common.opt (gno-column-info, gcolumn-info): New options.
21032 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21033 (check_die): Also test for multiple DW_AT_decl_column attributes.
21034 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21035 DW_AT_decl_column if requested.
21036 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21038 (gen_variable_die): Likewise.
21039 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21040 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21043 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21044 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21045 (ix86_handle_option): Handle OPT_m3dnowa.
21046 * doc/invoke.texi (-m3dnowa): Document.
21047 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21048 -m3dnowa instead of -m3dnow -march=athlon.
21051 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21052 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21054 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21057 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21058 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21059 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21060 generator for vsx_xxpermdi_<mode>_be.
21061 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21062 force big-endian semantics.
21063 (vsx_xxpermdi_<mode>_be): New define_expand with same
21064 implementation as previous version of vsx_xxpermdi_<mode>.
21066 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21068 PR tree-optimization/79327
21069 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21070 variable, its initialization and use.
21072 2017-02-17 Julia Koval <julia.koval@intel.com>
21074 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21075 (OPTION_MASK_ISA_PKU_UNSET): New.
21076 (ix86_handle_option): Handle -mrdpid.
21077 * config/i386/cpuid.h (bit_RDPID): New.
21078 * config/i386/driver-i386.c (host_detect_local_cpu):
21079 Detect RDPID feature.
21080 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21081 * config/i386/i386-c.c (ix86_target_macros_internal):
21083 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21084 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21085 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21086 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21087 * config/i386/i386.md (define_insn "rdpid"): New.
21088 * config/i386/i386.opt Add -mrdpid.
21089 * config/i386/immintrin.h (_rdpid_u32): New.
21091 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21093 PR rtl-optimization/79541
21094 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21095 instead of transforming it into USE.
21097 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21099 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21100 If HONOR_SNANS (SFmode) force the input to a register.
21101 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21102 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21103 an frsp or similar insn.
21105 2017-02-17 Martin Liska <mliska@suse.cz>
21107 PR rtl-optimization/79577
21108 * params.def (selsched-max-sched-times): Increase minimum to 1.
21110 2017-02-17 Martin Liska <mliska@suse.cz>
21112 PR rtl-optimization/79574
21113 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21115 2017-02-17 Martin Liska <mliska@suse.cz>
21117 PR tree-optimization/79529
21118 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21119 ssa_defined_default_def_p to handle cases which are implicitly
21121 * tree-ssa.c (ssa_defined_default_def_p): New function.
21122 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21123 which are implicitly defined.
21124 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21126 2017-02-17 Richard Biener <rguenther@suse.de>
21128 PR middle-end/79576
21129 * params.def (max-ssa-name-query-depth): Limit to 10.
21131 2017-02-17 Richard Biener <rguenther@suse.de>
21133 PR tree-optimization/79552
21134 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21137 2017-02-17 Richard Biener <rguenther@suse.de>
21140 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21142 2017-02-17 Marek Polacek <polacek@redhat.com>
21144 PR middle-end/79536
21145 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21146 (fold_negate_expr): New wrapper.
21148 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21150 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21151 Correct terminology and de-emphasize pre-standard behavior.
21153 2017-02-16 Alan Modra <amodra@gmail.com>
21155 PR rtl-optimization/79286
21156 * ira.c (def_dominates_uses): New function.
21157 (update_equiv_regs): Don't create an equivalence for insns that
21158 may trap where the register def does not dominate the use.
21160 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21162 PR rtl-optimization/78127
21163 * lra.c (lra): Call lra_eliminate before finish the loop after
21166 2017-02-16 Richard Biener <rguenther@suse.de>
21168 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21170 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21171 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21172 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21173 (isl_val_int_from_wi): New function.
21174 (extract_affine_gmp): Rename to ...
21175 (extract_affine_wi): ... this, take a widest_int.
21176 (extract_affine_int): Just wrap extract_affine_wi.
21177 (add_param_constraints): Use isl_val_int_from_wi.
21178 (add_loop_constraints): Likewise, and extract_affine_wi.
21180 2017-02-15 Jeff Law <law@redhat.com>
21182 PR middle-end/79521
21183 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21184 ira_init_register_move_cost_if_necessary.
21186 2017-02-15 Martin Sebor <msebor@redhat.com>
21188 PR middle-end/32003
21189 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21190 removed in a prior commit.
21192 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21194 PR tree-optimization/79347
21195 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21196 counters during peeling.
21198 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21200 * Makefile.in (site.exp): Remove "set ISLVER".
21202 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21205 * real.c (real_from_integer): Call real_convert even for decimal.
21207 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21210 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21212 2017-02-14 Andrew Pinski <apinski@cavium.com>
21214 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21215 cores and change the partno/implementer to be correct.
21216 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21217 the 'B" as the implementer.
21218 * config/aarch64/aarch64-tune.md: Regenerate.
21220 2017-02-14 Carl Love <cel@us.ibm.com>
21222 * config/rs6000/rs6000.c: Add case statement entry to make the
21223 xvcvuxdsp built-in argument unsigned.
21224 * config/rs6000/vsx.md: Fix the source and return operand types so they
21225 match the instruction definitions from the ISA document. Fix typo
21226 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21229 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21232 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21233 member early_clobber_alts.
21234 * lra-lives.c (reg_early_clobber_p): New.
21235 (process_bb_lives): Use it.
21236 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21237 (debug_operand_data): Initialize early_clobber_alts.
21238 (setup_operand_alternative): Set up early_clobber_alts.
21239 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21240 alternatives to new_insn_reg.
21241 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21243 (lra_update_insn_regno_info): Pass the new arg.
21245 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21247 PR middle-end/79505
21248 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21249 (new_oacc_loop_raw): Don't clear already cleared fields.
21252 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21253 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21254 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21256 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21259 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21261 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21264 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21265 the extra instruction to the right place to store 128-bit constant
21268 2017-02-14 Martin Sebor <msebor@redhat.com>
21270 PR middle-end/79448
21271 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21272 warning for strings of unknown length.
21274 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21276 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21278 2017-02-14 Jeff Law <law@redhat.com>
21281 * ira-costs.c (scan_one_insn): Initialize register move costs
21282 for pseudos seen in USE/CLOBBER insns.
21284 PR tree-optimization/79095
21285 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21286 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21287 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21288 if the operands are known to be not equal, then the resulting range
21290 (intersect_ranges): If the new range is ~[0,0] and the old range is
21291 wide, then prefer ~[0,0].
21292 * tree-vrp.c (overflow_comparison_p_1): New function.
21293 (overflow_comparison_p): New function.
21294 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21295 if NAME is used in an overflow test.
21296 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21297 overflow check that can be expressed as an equality test, then adjust
21298 ops to be that equality test.
21300 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21302 * config/s390/s390-builtin-types.def: Remove flags argument.
21303 * config/s390/s390.c (s390_init_builtins): Likewise.
21305 2017-02-14 Martin Liska <mliska@suse.cz>
21307 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21308 vector. Fix trailing white spaces.
21310 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21312 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21315 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21317 PR rtl-optimization/68664
21318 * config/arm/arm.c (arm_sched_can_speculate_insn):
21319 New function. Declare prototype.
21320 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21322 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21324 PR rtl-optimization/68664
21325 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21327 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21329 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21331 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21332 max skip bytes for function, loop and jump.
21334 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21336 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21337 ABS_EXPR for gimple dump.
21339 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21342 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21344 PR tree-optimization/79408
21345 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21346 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21347 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21348 also if rhs1 is INTEGER_CST.
21350 2017-02-14 Richard Biener <rguenther@suse.de>
21352 PR middle-end/79432
21353 * tree-into-ssa.c (insert_phi_nodes): When the function can
21354 have abnormal edges rewrite SSA names with broken use-def
21355 dominance out of SSA and register them for PHI insertion.
21357 2017-02-13 Martin Sebor <msebor@redhat.com>
21359 PR middle-end/79496
21360 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21361 clearing info.nowrite flag when snprintf size argument is a range.
21363 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21365 * cprop.c (cprop_jump): Add missing space in string literal.
21366 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21367 (get_constraint_for_component_ref): Likewise.
21368 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21369 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21370 * lra-constraints.c (process_alt_operands): Likewise.
21371 * ipa-inline.c (inline_small_functions): Likewise.
21372 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21373 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21374 * trans-mem.c (diagnose_tm_1_op): Likewise.
21375 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21376 (grid_parallel_clauses_gridifiable): Likewise.
21378 * config/nvptx/mkoffload.c (process): Add space in between
21381 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21382 "MOD4_SSE_REGS" and "ALL_REGS".
21384 * spellcheck.c (test_data): Add , in between "foo" and "food".
21386 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21389 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21390 boundary crossing check and subsequent code generation agree.
21392 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21394 * config/aarch64/aarch64.c (has_memory_op): Delete.
21395 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21398 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21400 PR rtl-optimization/79388
21401 PR rtl-optimization/79450
21402 * combine.c (distribute_notes): When removing TEM_INSN for which
21403 corresponding dest has last value recorded, invalidate that last
21406 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21408 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21409 of explicit '@'. Add missing assembly comment marker on branch costs
21412 2017-02-13 Nathan Sidwell <nathan@acm.org>
21414 * gengtype-lex.l (<in_struct>): Add '/'.
21416 2017-02-13 Martin Liska <mliska@suse.cz>
21419 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21421 2017-02-13 Richard Biener <rguenther@suse.de>
21423 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21425 * configure: Re-generate.
21426 * config.in: Likewise.
21427 * graphite-dependences.c: Simplify as if
21428 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21429 * graphite-isl-ast-to-gimple.c: Likewise.
21430 * graphite-optimize-isl.c: Likewise.
21431 * graphite-poly.c: Likewise.
21432 * graphite-sese-to-poly.c: Likewise.
21433 * graphite.h: Likewise.
21434 * toplev.c: Include isl/version.h and use isl_version () for
21435 printing the ISL version.
21436 * doc/install.texi: Update ISL requirement.
21438 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21440 * doc/standards.texi (Standards): Update reference to
21443 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21445 * doc/extend.texi (Named Address Spaces): sourceware.org now
21447 * doc/install.texi (Binaries): Ditto.
21450 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21452 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21453 standard terminology "stringize"/"stringizing" throughout.
21454 * doc/cppinternals.texi: Likewise.
21456 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21458 * doc/extend.texi: Fix some spelling mistakes and typos.
21459 * doc/invoke.texi: Likewise.
21461 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21464 * params.def (inline-min-speedup) Change from 10 to 8.
21466 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21468 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21471 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21474 * ipa-inline-analysis.c (get_minimal_bb): New function.
21475 (record_modified): Use it.
21476 (remap_edge_change_prob): Handle also ancestor functions.
21478 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21480 * doc/contrib.texi (Contributors): Remove broken link into
21481 the Mauve CVS repository.
21483 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21485 PR middle-end/79454
21486 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21487 result computation whenever lhs doesn't have vector mode, not
21488 just when it has BLKmode.
21490 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21492 * doc/makefile.texi (profiledbootstrap): Refer to the
21493 installation instructions only in textual form.
21495 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21498 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21500 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21502 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21503 (Specific): Update mingw-w64 reference.
21505 (Specific): Remove broken link to Renesas RX processor.
21507 2017-02-10 Richard Biener <rguenther@suse.de>
21509 * toplev.c (process_options): Do not mention obsolete graphite
21510 options when printing sorry message about missing graphite support.
21511 Mention -floop-nest-optimize.
21513 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21515 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21516 (vtst_p16): Likewise.
21517 (vtstq_p8): Likewise.
21518 (vtstq_p16): Likewise.
21520 (vtstq_p64): Likewise.
21521 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21522 (vset_lane_p64): New.
21523 (vsetq_lane_p64): New.
21525 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21527 PR tree-optimization/79411
21528 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21529 stmt operands are SSA_NAMEs used in abnormal phis.
21530 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21533 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21536 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21539 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21541 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21543 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21545 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21549 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21550 not arbitrary TREE_CONSTANT.
21553 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21554 "omp declare target link" attribute unless is_global_var.
21555 * omp-offload.c (find_link_var_op): Likewise.
21557 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21558 Chung-Lin Tang <cltang@codesourcery.com>
21560 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21562 (gimplify_adjust_omp_clauses): Don't delete TILE.
21563 (gimplify_omp_for): Deal with TILE.
21564 * internal-fn.c (expand_GOACC_TILE): New function.
21565 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21567 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21568 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21570 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21571 avoid DIV for outermost collapse var.
21572 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21573 Remove out of date comments, fix whitespace.
21574 * omp-general.c (omp_extract_for_data): Deal with tiling.
21575 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21576 adjust OLF_DIM_BASE value.
21577 (struct omp_for_data): Add tiling field.
21578 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21579 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21580 for auto loops. Remove default auto determining, moved to
21581 oacc_loop_fixed_partitions.
21582 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21583 stmts, add e_mask field.
21584 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21585 (oacc_thread_numbers): Use oacc_dim_call.
21586 (oacc_xform_tile): New.
21587 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21588 (finish_oacc_loop): Adjust for ifns vector.
21589 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21590 add case for GOACC_TILE fns.
21591 (oacc_loop_xform_loop): Delete.
21592 (oacc_loop_process): Iterate over call list directly, and add
21593 handling for GOACC_TILE fns.
21594 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21596 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21597 vector partitioning to outer loops. Assign 2 partitions to loops
21598 when available. Add TILE handling.
21599 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21600 (execite_oacc_device_lower): Process GOACC_TILE fns,
21601 ignore unknown specs.
21602 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21603 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21604 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21606 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21608 * configure.ac (ACX_BUGURL): Update.
21609 * configure: Regenerate.
21611 2017-02-09 Richard Biener <rguenther@suse.de>
21613 PR tree-optimization/69823
21614 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21615 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21617 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21619 * config/arc/arc-c.def: Add __NPS400__ definition.
21620 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21621 (TARGET_NPS400): Define.
21623 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21625 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21627 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21628 pointer, arch_info.
21629 (arc_cpu_types): Fill the arch_info field with a pointer into the
21630 arc_arch_types table.
21631 (arc_selected_cpu): Declare.
21632 * config/arc/arc.c (arc_selected_cpu): Make global.
21633 (arc_selected_arch): Delete.
21634 (arc_base_cpu): Delete.
21635 (arc_override_options): Remove references to deleted variables,
21636 update access to arch information.
21637 (ARC_OPT): Update access to arch information.
21638 (ARC_OPTX): Likewise.
21639 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21640 (TARGET_ARC600): Update access to arch information.
21641 (TARGET_ARC601): Likewise.
21642 (TARGET_ARC700): Likewise.
21643 (TARGET_EM): Likewise.
21644 (TARGET_HS): Likewise.
21645 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21648 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21651 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21652 condition/operands for integer GE/LE/GEU/LEU operations.
21654 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21656 PR translation/79397
21657 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21660 2017-02-08 Martin Jambor <mjambor@suse.cz>
21663 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21664 whether allocation happened.
21665 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21666 nothing was allocated.
21668 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21670 PR tree-optimization/79408
21671 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21672 constant, but SSA_NAME with a known integer range, use the minimum
21673 of that range instead of op1 to determine if modulo can be replaced
21674 with its first operand.
21676 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21678 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21680 2017-02-08 Richard Biener <rguenther@suse.de>
21682 PR tree-optimization/71824
21683 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21684 Check all loops contained in the merged region.
21686 2017-02-07 Andrew Pinski <apinski@cavium.com>
21688 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21690 2017-02-07 Andrew Pinski <apinski@cavium.com>
21692 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21693 (thunderxt88): Likewise.
21694 (thunderxt81): Disable LSE and change v8.1 to v8.
21695 (thunderxt83): Likewise.
21697 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21698 Richard Biener <rguenther@suse.de>
21700 PR middle-end/79399
21701 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21702 type from int to size_t.
21703 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21705 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21707 PR rtl-optimization/79386
21708 * cprop.c (bypass_conditional_jumps): Initialize
21709 bypass_last_basic_block already before splitting bbs after
21710 unconditional traps...
21711 (bypass_conditional_jumps): ... rather than here.
21714 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
21715 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
21716 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
21717 fix -masm=intel patterns.
21719 2017-02-07 Richard Biener <rguenther@suse.de>
21721 PR tree-optimization/79256
21722 PR middle-end/79278
21723 * builtins.c (get_object_alignment_2): Use min_align_of_type
21724 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
21725 and ADJUST_FIELD_ALIGN.
21727 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
21729 * doc/tm.texi: Regenerate.
21730 * stor-layout.c (layout_decl): Adjust.
21731 (update_alignment_for_field): Likewise.
21732 (place_field): Likewise.
21733 (min_align_of_type): Likewise.
21734 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
21735 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
21736 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
21737 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
21738 * config/frv/frv.c (frv_adjust_field_align): Likewise.
21739 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
21740 * config/i386/i386.c (x86_field_alignment): Likewise.
21741 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
21742 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
21743 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
21744 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
21745 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
21746 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21750 2017-01-30 Richard Biener <rguenther@suse.de>
21752 PR tree-optimization/79256
21753 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21754 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21757 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21759 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21760 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21761 builtins to SImode and emit a zero-extend, if necessary.
21763 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21765 * docs/invoke.texi (RISC-V Options): Alphabetize.
21767 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21769 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21772 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21774 * config/riscv/riscv.c: New file.
21775 * common/config/riscv/riscv-common.c: Likewise.
21776 * config.gcc: Likewise.
21777 * config/riscv/constraints.md: Likewise.
21778 * config/riscv/elf.h: Likewise.
21779 * config/riscv/generic.md: Likewise.
21780 * config/riscv/linux.h: Likewise.
21781 * config/riscv/multilib-generator: Likewise.
21782 * config/riscv/peephole.md: Likewise.
21783 * config/riscv/pic.md: Likewise.
21784 * config/riscv/predicates.md: Likewise.
21785 * config/riscv/riscv-builtins.c: Likewise.
21786 * config/riscv/riscv-c.c: Likewise.
21787 * config/riscv/riscv-ftypes.def: Likewise.
21788 * config/riscv/riscv-modes.def: Likewise.
21789 * config/riscv/riscv-opts.h: Likewise.
21790 * config/riscv/riscv-protos.h: Likewise.
21791 * config/riscv/riscv.h: Likewise.
21792 * config/riscv/riscv.md: Likewise.
21793 * config/riscv/riscv.opt: Likewise.
21794 * config/riscv/sync.md: Likewise.
21795 * config/riscv/t-elf-multilib: Likewise.
21796 * config/riscv/t-linux: Likewise.
21797 * config/riscv/t-linux-multilib: Likewise.
21798 * config/riscv/t-riscv: Likewise.
21799 * configure.ac: Likewise.
21800 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21801 Waterman as RISC-V maintainers.
21802 * doc/install.texi: Add RISC-V entries.
21803 * doc/invoke.texi: Add RISC-V options section.
21804 * doc/md.texi: Add RISC-V constraints section.
21805 * configure: Regenerated.
21807 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21810 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21811 false values to be constant vectors with all 0 or all 1 bits set.
21812 (vcondu<mode><mode>): Likewise.
21813 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21815 (fpmask_comparison_operator): Update comment.
21816 (vecint_comparison_operator): New predicate.
21817 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21818 vector conditionals when the true and false values are constant
21819 vectors with all 0 bits or all 1 bits set.
21821 2017-02-06 Martin Sebor <msebor@redhat.com>
21823 PR tree-optimization/79376
21824 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21826 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21828 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21829 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21830 to simplify split condition.
21832 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21834 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21835 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21838 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21840 PR rtl-optimization/68664
21841 * target.def (can_speculate_insn): New hook.
21842 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21843 * doc/tm.texi: Regenerate.
21844 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21845 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21846 (rs6000_sched_can_speculate_insn): New function.
21848 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21850 PR tree-optimization/79284
21851 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21852 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21853 vectorizable_mask_load_store, vectorizable_operation,
21854 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21855 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21856 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21857 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21858 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21859 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21860 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21861 is_gimple_assign (stmt). Replace another such test with
21862 is_gimple_assign (stmt).
21864 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21867 * config/avr/avr.c (rtl-iter.h): Include it.
21868 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21869 (avr_legitimate_combined_insn): ...and implementation.
21871 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21873 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21874 * config/s390/s390.c (s390_const_operand_ok)
21875 (s390_canonicalize_comparison, s390_extract_part)
21876 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21877 (s390_contiguous_bitmask_p, s390_rtx_costs)
21878 (legitimize_pic_address): Likewise.
21879 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21880 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21881 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21882 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21883 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21885 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21887 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21888 REGNO($0) == REGNO($1).
21890 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21892 * config/s390/linux.h(SIZE_TYPE): Add comment.
21894 2017-02-06 Julian Brown <julian@codesourcery.com>
21895 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21896 Virendra Pathak <virendra.pathak@broadcom.com>
21898 * config/aarch64/aarch64-cores.def: Change the scheduler
21900 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21901 * config/aarch64/thunderx2t99.md: New file.
21903 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21905 * doc/standards.texi (Go Language): Update link to language
21908 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21910 * tree-eh.c (lower_resx): Sanitize profile.
21911 (cleanup_empty_eh_move_lp): Likewise.
21913 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21916 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21918 * cfgloopmanip.h (loop_version): Update prototype.
21919 * modulo-sched.c (sms_schedule): Update call of loop_version.
21920 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21921 * tree-parloops.c (gen_parallel_loop): Likewise.
21922 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
21923 * tree-ssa-loop-split.c (split_loop): Likewise.
21924 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
21925 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21927 2017-02-05 Martin Liska <mliska@suse.cz>
21930 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
21932 (print_operand_address): Initialize a struct to zero.
21934 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21936 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
21937 garbage collector only in textual form.
21939 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21941 * doc/extend.texi (x86 specific memory model extensions for
21942 transactional memory): Simplify a phrase.
21944 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
21947 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
21948 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
21949 (atomic_storedi_1): Likewise.
21951 2017-02-04 Jakub Jelinek <jakub@redhat.com>
21953 PR tree-optimization/79338
21954 * tree-parloops.c (gather_scalar_reductions): Don't call
21955 vect_analyze_loop_form for loop->inner before destroying loop's
21958 2017-02-03 Martin Sebor <msebor@redhat.com>
21960 PR tree-optimization/79327
21961 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
21962 when precision has resulted in leading zeros.
21963 (format_integer): Adjust the likely counter to assume an unknown
21964 argument that may be zero is non-zero.
21966 2017-02-03 Jason Merrill <jason@redhat.com>
21969 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
21970 avoid copying non-taken branch.
21972 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21974 PR tree-optimization/79340
21975 * tree-vect-loop.c (vectorizable_reduction): Release
21976 vec_defs elements after safe_splicing them into other vectors.
21979 PR tree-optimization/79327
21980 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
21981 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
21983 (format_integer): Use wide_int_to_tree instead of build_int_cst
21984 + to_?hwi. If argmin is NULL, just set argmin and argmax to
21985 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
21986 of shortest and longest sequence.
21988 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
21990 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
21991 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
21993 2017-02-03 Walter Lee <walt@tilera.com>
21996 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
21997 after initial stackframe link reg save.
21998 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22000 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22003 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22004 wu for stxssp alternative.
22006 2017-02-03 Martin Sebor <msebor@redhat.com>
22008 PR tree-optimization/79352
22009 * gimple-fold.c (get_range_strlen): Add argument.
22010 (get_range_strlen): Change return type to bool.
22011 (get_maxval_strlen): Pass in a dummy argument.
22012 * gimple-fold.h (get_range_strlen): Change return type to bool.
22013 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22014 * tree.h (array_at_struct_end_p): Add argument.
22015 * tree.c (array_at_struct_end_p): Handle it.
22017 2017-02-03 Martin Liska <mliska@suse.cz>
22020 * multiple_target.c (create_dispatcher_calls): Redirect edge
22021 from a caller of a dispatcher.
22022 (expand_target_clones): Make the clones local.
22023 (ipa_target_clone): Do both target clones and resolvers.
22024 (ipa_dispatcher_calls): Remove the pass.
22025 (pass_dispatcher_calls::gate): Likewise.
22026 (make_pass_dispatcher_calls): Likewise.
22027 * passes.def (pass_target_clone): Put as very first IPA early
22030 2017-02-03 Martin Liska <mliska@suse.cz>
22032 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22033 in case of a function with ifunc attribute.
22035 2017-02-03 Martin Liska <mliska@suse.cz>
22037 * cgraph.c (cgraph_node::dump): Dump function version info.
22038 * symtab.c (symtab_node::dump_base): Add missing new line.
22040 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22042 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22043 (ifcombine_ifandif): Use it.
22045 2017-02-03 Martin Liska <mliska@suse.cz>
22047 * doc/invoke.texi: Document default value for
22048 use-after-scope-direct-emission-threshold.
22050 2017-02-03 Martin Liska <mliska@suse.cz>
22052 PR tree-optimization/79339
22053 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22054 (format_floating): Likewise.
22056 2017-02-03 Martin Liska <mliska@suse.cz>
22059 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22061 (ipa_node_params_t::remove): Likewise.
22062 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22063 initialization from removed ipa_node_params_t::insert.
22064 (ipa_node_params::~ipa_node_params): Move from removed
22065 ipa_node_params_t::release.
22066 * symbol-summary.h (symbol_summary::m_released): New member.
22067 Do not release a summary twice. Do not allow to call finalizer
22068 for types of a summary that live in GGC memory.
22070 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22072 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22075 2017-02-02 Martin Sebor <msebor@redhat.com>
22077 PR middle-end/79275
22078 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22079 (format_string): Tighten up the range of output for non-constant
22080 strings and correct the expected range for wide non-constant strings.
22082 2017-02-02 Martin Sebor <msebor@redhat.com>
22084 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22086 PR middle-end/32003
22087 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22089 (-fdump-tree-@var): Add to index and document how to come up
22090 with pass-specific option and dump file names.
22091 (-fdump-passes): Clarify where to look for output.
22093 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22095 PR middle-end/77445
22096 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22097 statistics of the analyzed path; allow threading for speed when
22098 any of BBs along the path are optimized for speed.
22100 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22102 PR middle-end/78468
22103 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22104 settings of the virtual registers.
22107 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22109 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22110 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22113 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22115 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22116 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22118 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22120 * config/s390/s390.md: Add missing comments with the expanded
22122 * config/s390/vector.md: Likewise.
22123 * config/s390/vx-builtins.md: Likewise.
22125 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22128 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22129 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22130 conditions on a single line.
22132 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22134 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22135 __S390_VX__ to __VX__.
22137 2017-02-01 Andrew Pinski <apinski@cavium.com>
22139 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22140 stmt_info to record_stmt_cost.
22141 (vect_get_known_peeling_cost): Pass stmt_info if known to
22143 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22144 cpu_vector_cost field into
22145 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22146 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22147 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22148 splitting of scalar_stmt_cost and vec_stmt_cost.
22149 (thunderx_vector_cost): Likewise.
22150 (cortexa57_vector_cost): LIkewise.
22151 (exynosm1_vector_cost): Likewise.
22152 (xgene1_vector_cost): Likewise.
22153 (thunderx2t99_vector_cost): Improve after the splitting of the two
22155 (aarch64_builtin_vectorization_cost): Update for the splitting of
22156 scalar_stmt_cost and vec_stmt_cost.
22158 2017-02-01 Torvald Riegel <triegel@redhat.com>
22159 Richard Henderson <rth@redhat.com>
22161 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22162 conditional on existance of a fast atomic load.
22163 * optabs-query.c (can_atomic_load_p): New function.
22164 * optabs-query.h (can_atomic_load_p): Declare it.
22165 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22166 no fast atomic load is available for the particular size of access.
22167 (expand_atomic_compare_and_swap): Likewise.
22168 (expand_atomic_load): Likewise.
22169 (expand_atomic_store): Likewise.
22170 (expand_atomic_fetch_op): Likewise.
22171 * testsuite/lib/target-supports.exp
22172 (check_effective_target_sync_int_128): Remove x86 because it provides
22173 no fast atomic load.
22174 (check_effective_target_sync_int_128_runtime): Likewise.
22176 2017-02-01 Richard Biener <rguenther@suse.de>
22178 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22179 (graphite_transform_loops): Provide opt-info for optimized nests.
22180 * tree-parloop.c (parallelize_loops): Provide opt-info for
22181 parallelized loops.
22183 2017-02-01 Richard Biener <rguenther@suse.de>
22185 PR middle-end/79315
22186 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22187 was not set before.
22189 2017-02-01 Richard Biener <rguenther@suse.de>
22191 PR tree-optimization/71824
22192 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22193 Verify the loops are valid in the merged SESE region.
22194 (scop_detection::can_represent_loop_1): Check analyzing the
22195 evolution of the number of iterations in the region succeeds.
22197 2017-01-31 Ian Lance Taylor <iant@golang.org>
22199 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22200 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22202 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22204 PR preprocessor/79210
22205 * input.c (get_substring_ranges_for_loc): Replace line_width
22206 assertion with error-handling.
22208 2017-01-31 Richard Biener <rguenther@suse.de>
22210 PR tree-optimization/77318
22211 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22212 (create_pw_aff_from_tree): Take loop parameter.
22213 (add_condition_to_pbb): Pass loop of the condition to
22214 create_pw_aff_from_tree.
22216 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22218 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22219 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22221 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22225 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22227 (convert_int_to_float128): Likewise.
22228 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22229 (convert_int_to_float128): Likewise.
22230 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22231 (UNSPEC_IEEE128_CONVERT): Likewise.
22232 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22233 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22234 Use local variables for IBM extended format.
22235 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22236 (fix_trunc<mode>si2_fprs): Likewise.
22237 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22238 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22239 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22240 to know that we can now have integers of all sizes in vector
22242 (fix<uns>_<mode>di2_hw): Likewise.
22243 (float<uns>_<mode>si2_hw): Likewise.
22244 (fix_<mode>si2_hw): Likewise.
22245 (fixuns_<mode>si2_hw): Likewise.
22246 (float<uns>_<mode>di2_hw): Likewise.
22247 (float_<mode>di2_hw): Likewise.
22248 (float_<mode>si2_hw): Likewise.
22249 (floatuns_<mode>di2_hw): Likewise.
22250 (floatuns_<mode>si2_hw): Likewise.
22251 (xscvqp<su>wz_<mode>): Delete, no longer used.
22252 (xscvqp<su>dz_<mode>): Likewise.
22253 (xscv<su>dqp_<mode>): Likewise.
22254 (ieee128_mfvsrd_64bit): Likewise.
22255 (ieee128_mfvsrd_32bit): Likewise.
22256 (ieee128_mfvsrwz): Likewise.
22257 (ieee128_mtvsrw): Likewise.
22258 (ieee128_mtvsrd_64bit): Likewise.
22259 (ieee128_mtvsrd_32bit): Likewise.
22261 2017-01-31 Martin Liska <mliska@suse.cz>
22264 * ipa-prop.c (ipa_free_all_node_params): Call release method
22265 instead of ~sumbol_summary to not to trigger double times
22268 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22270 PR tree-optimization/71691
22271 * bitmap.h (class auto_bitmap): New.
22272 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22273 is_maybe_undefined instead of ssa_undefined_value_p.
22275 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22277 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22278 __S390_ARCH_LEVEL__ to __ARCH__.
22280 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22282 PR tree-optimization/79267
22283 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22284 if should_remove_lhs_p is true.
22286 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22289 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22290 (add_alignment_attribute): New.
22291 (base_type_die): Add alignment attribute.
22292 (subrange_type_die): Likewise.
22293 (modified_type_die): Likewise.
22294 (gen_array_type_die): Likewise.
22295 (gen_descr_array_type_die: Likewise.
22296 (gen_enumeration_type_die): Likewise.
22297 (gen_subprogram_die): Likewise.
22298 (gen_variable_die): Likewise.
22299 (gen_field_die): Likewise.
22300 (gen_ptr_to_mbr_type_die): Likewise.
22301 (gen_struct_or_union_type_die): Likewise.
22302 (gen_subroutine_type_die): Likewise.
22303 (gen_typedef_die): Likewise.
22304 (base_type_cmp): Compare alignment attribute.
22306 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22309 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22310 (setb_unsigned) New pattern for setb with CCUNS.
22311 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22312 subfc./subfe sequence to avoid overflow problems. Generate a
22313 shorter sequence with cmpld/setb for power9.
22314 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22315 for generating subfc. instruction.
22316 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22317 now uses this instruction.
22319 2017-01-30 Ian Lance Taylor <iant@google.com>
22322 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22323 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22325 2017-01-30 Martin Sebor <msebor@redhat.com>
22327 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22328 Move constant to the right of a relational operator.
22329 (get_mpfr_format_length, format_character, format_string): Ditto.
22330 (should_warn_p, maybe_warn): Same.
22332 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22334 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22337 * asan.c (get_translation_unit_decl): Remove function.
22338 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22340 2017-01-30 Martin Liska <mliska@suse.cz>
22342 PR gcov-profile/79259
22343 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22344 -fprofile-generate.
22346 2017-01-30 Martin Liska <mliska@suse.cz>
22349 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22350 Initialize variables with NULL value.
22352 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22355 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22357 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22359 2017-01-30 Richard Biener <rguenther@suse.de>
22361 PR tree-optimization/79276
22362 * tree-vrp.c (process_assert_insertions): Properly adjust common
22363 when removing a duplicate.
22365 * gcc.dg/torture/pr79276.c: New testcase.
22367 2017-01-30 Richard Biener <rguenther@suse.de>
22369 PR tree-optimization/79256
22370 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22371 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22373 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22375 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22378 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22379 ("*r<noxa>sbg_<mode>_sll_bitmask")
22380 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22381 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22382 Use contiguous_bitmask_nowrap_operand.
22384 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22387 * config/rs6000/altivec.h (vec_xl): Revise #define.
22388 (vec_xst): Likewise.
22390 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22392 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22394 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22396 PR rtl-optimization/79194
22397 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22398 traps before call to bypass_conditional_jumps.
22400 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22402 PR tree-optimization/71374
22403 * lra-constraints.c (check_conflict_input_operands): New.
22404 (match_reload): Use it.
22406 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22409 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22410 account to calculate conflict_set.
22412 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22414 PR rtl-optimization/78559
22415 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22416 other_insn in combine.
22418 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22420 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22421 uint16_type_node for BT_UINT16.
22423 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22425 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22426 "RTL Tests" to menu.
22427 (GIMPLE Tests): New node.
22428 (RTL Tests): New node.
22430 2017-01-27 Richard Biener <rguenther@suse.de>
22432 PR tree-optimization/79245
22433 * tree-loop-distribution.c (distribute_loop): Apply cost
22434 modeling also to detected patterns.
22436 2017-01-27 Richard Biener <rguenther@suse.de>
22438 PR tree-optimization/71433
22439 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22440 (compare_assert_loc): New function.
22441 (process_assert_insertions): Sort and optimize assert locations
22442 to remove duplicates and push down identical assertions on
22443 edges to their destination block.
22445 2017-01-27 Richard Biener <rguenther@suse.de>
22447 PR tree-optimization/79244
22448 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22449 out SSA names even if abnormal.
22451 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22453 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22454 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22455 instead of MPFR_RNDN.
22457 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22460 * arm.c (arm_option_override): Don't call build_target_option_node
22461 until after doing all option overrides.
22462 (arm_valid_target_attribute_tree): Likewise.
22464 2017-01-27 Martin Liska <mliska@suse.cz>
22466 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22467 for {cd}tors and C++ {cd}tors.
22469 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22471 * config/s390/s390.md ("*setmem_long_and")
22472 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22474 2017-01-26 Martin Sebor <msebor@redhat.com>
22476 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22479 2017-01-26 Martin Sebor <msebor@redhat.com>
22481 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22482 HAVE_DFmode before using XFmode or DFmode.
22483 (parse_directive): Avoid using the z length modifier to avoid
22484 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22486 PR middle-end/78703
22487 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22488 to accept adjustment as an array.
22489 (get_int_range): New function.
22490 (struct directive): Make width and prec arrays.
22491 (directive::set_width, directive::set_precision): Call get_int_range.
22492 (format_integer, format_floating): Handle width and precision ranges.
22493 (format_string, parse_directive): Same.
22495 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22498 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22499 comdat_type_p set, just clone them, but keep the children in the
22503 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22504 which have direct callers with -fvar-tracking-assignments enabled
22506 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22507 inside of type units.
22509 2017-01-26 Martin Sebor <msebor@redhat.com>
22511 PR middle-end/78703
22512 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22514 (struct format_result): Replace number_chars, number_chars_min,
22515 and number_chars_max with a single member of struct result_range.
22517 (format_result::operator+=): Adjust.
22518 (struct fmtresult): Remove bounded. Handle likely and unlikely
22520 (fmtresult::adjust_for_width_or_precision): New function.
22521 (fmtresult:type_max_digits): New function.
22522 (bytes_remaining): Handle likely and unlikely counters.
22523 (min_bytes_remaining): Remove.
22524 (format_percent): Simplify.
22525 (format_integer, format_floating): Set likely and unlikely counters.
22526 (get_string_length, format_character, format_string): Same.
22527 (format_plain, should_warn_p): New function.
22528 (maybe_warn): Call should_warn_p. Update diagnostic messages
22529 and handle those for all directives, including plain strings.
22530 (format_directive): Handle likely and unlikely counters.
22531 Remove unnecessary quoting from diagnostics. Add an informational
22533 (add_bytes): Remove.
22534 (pass_sprintf_length::compute_format_length): Simplify.
22535 (try_substitute_return_value): Handle likely and unlikely counters.
22537 2017-01-26 Carl Love <cel@us.ibm.com>
22539 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22540 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22542 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22545 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22546 endianess for subregs into account.
22547 * lra-constraints.c (lra_constraints): Do risky transformations
22548 always on the first iteration.
22549 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22550 last_call_used_reg_set.
22551 (process_bb_lives): Define and use last_call_used_reg_set.
22552 * lra.c (lra): Always continue after lra_constraints on the first
22555 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22557 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22559 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22561 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22563 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22564 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22565 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22566 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22567 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22568 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22569 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22570 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22571 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22573 2017-01-26 Marek Polacek <polacek@redhat.com>
22576 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22577 for the third operand.
22579 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22581 PR middle-end/79236
22582 * omp-low.c (struct omp_context): Add simt_stmt field.
22583 (scan_omp_for): Return omp_context *.
22584 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22585 context to the _simt_ SIMD stmt.
22586 (lower_omp_for): For combined SIMD with sibling _simt_
22587 SIMD, make sure to use the same decls in _looptemp_
22588 clauses as in the sibling.
22590 2017-01-26 David Sherwood <david.sherwood@arm.com>
22592 PR middle-end/79212
22593 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22596 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22599 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22600 emit fld b; fld a; if possible.
22602 * brig-builtins.def: Update copyright years.
22603 * config/arm/arm_acle_builtins.def: Update copyright years.
22605 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22608 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22609 constraint instead of o for the stxsd instruction.
22611 2017-01-25 Carl Love <cel@us.ibm.com>
22613 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22614 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22616 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22618 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22620 2017-01-25 Richard Biener <rguenther@suse.de>
22622 PR tree-optimization/69264
22623 * target.def (vector_alignment_reachable): Improve documentation.
22624 * doc/tm.texi: Regenerate.
22625 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22627 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22628 earlier changes with respect to TYPE_USER_ALIGN.
22629 (vector_alignment_reachable_p): Likewise. Improve dumping.
22631 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22634 * config/arm/arm.md (xordi3): Force constant operand into a register
22637 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22639 * doc/invoke.texi (-fstore-merging): Correct default optimization
22640 levels at which it is enabled.
22641 (-O): Move -fstore-merging from list to...
22644 2017-01-25 Richard Biener <rguenther@suse.de>
22647 * omp-expand.c: Include debug.h.
22648 (expand_omp_taskreg): Make sure to generate early debug before
22649 outlining anything from a function.
22650 (expand_omp_target): Likewise.
22651 (grid_expand_target_grid_body): Likewise.
22653 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22656 * asan.c (get_translation_unit_decl): New function.
22657 (asan_add_global): Extract modules file name from globals
22658 TRANSLATION_UNIT_DECL name.
22660 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22663 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22664 for long calls with APCS frame and VFP.
22666 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22668 * cfg.c (original_copy_tables_initialized_p): New function.
22669 * cfg.h (original_copy_tables_initialized_p): New decl.
22670 * cfgrtl.c (relink_block_chain): Guard the call to
22671 free_original_copy_tables with a call to
22672 original_copy_tables_initialized_p.
22673 * cgraph.h (symtab_node::native_rtl_p): New decl.
22674 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22675 (symtab_node::needed_p): Don't assert for early assembly output
22676 for __RTL functions.
22677 (cgraph_node::finalize_function): Set "force_output" for __RTL
22679 (cgraph_node::analyze): Bail out early for __RTL functions.
22680 (analyze_functions): Update assertion to support __RTL functions.
22681 (cgraph_node::expand): Bail out early for __RTL functions.
22682 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22684 * function.h (struct function): Update comment for field
22686 * gimple-expr.c: Include "tree-pass.h".
22687 (gimple_has_body_p): Return false for __RTL functions.
22688 * Makefile.in (OBJS): Add run-rtl-passes.o.
22689 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22691 (gcc::pass_manager::get_clean_slate): New accessor.
22692 * passes.c: Include "insn-addr.h".
22693 (should_skip_pass_p): Add logging. Update logic for running
22694 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22695 property-provider override so it is only done for gimple passes.
22697 (skip_pass): New function.
22698 (execute_one_pass): Call skip_pass when skipping passes.
22699 * read-md.c (md_reader::read_char): Support filtering
22700 the input to a subset of line numbers.
22701 (md_reader::md_reader): Initialize fields
22702 m_first_line and m_last_line.
22703 (md_reader::read_file_fragment): New function.
22704 * read-md.h (md_reader::read_file_fragment): New decl.
22705 (md_reader::m_first_line): New field.
22706 (md_reader::m_last_line): New field.
22707 * read-rtl-function.c (function_reader::create_function): Only
22708 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22709 curr_properties. Set DECL_INITIAL to a dummy block.
22710 (read_rtl_function_body_from_file_range): New function.
22711 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22713 * run-rtl-passes.c: New file.
22714 * run-rtl-passes.h: New file.
22716 2017-01-24 Jeff Law <law@redhat.com>
22718 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
22721 2017-01-24 Bin Cheng <bin.cheng@arm.com>
22723 PR tree-optimization/79159
22724 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
22725 (record_nonwrapping_iv): Improve boundary using above function if no
22726 value range information.
22728 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
22729 Martin Jambor <mjambor@suse.cz>
22731 * brig-builtins.def: New file.
22732 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
22733 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
22734 (DEF_HSAIL_SAT_BUILTIN): Likewise.
22735 (DEF_HSAIL_INTR_BUILTIN): Likewise.
22736 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
22737 * builtin-types.def (BT_INT8): New.
22738 (BT_INT16): Likewise.
22739 (BT_UINT8): Likewise.
22740 (BT_UINT16): Likewise.
22741 (BT_FN_ULONG): Likewise.
22742 (BT_FN_UINT_INT): Likewise.
22743 (BT_FN_UINT_ULONG): Likewise.
22744 (BT_FN_UINT_LONG): Likewise.
22745 (BT_FN_UINT_PTR): Likewise.
22746 (BT_FN_ULONG_PTR): Likewise.
22747 (BT_FN_INT8_FLOAT): Likewise.
22748 (BT_FN_INT16_FLOAT): Likewise.
22749 (BT_FN_UINT32_FLOAT): Likewise.
22750 (BT_FN_UINT16_FLOAT): Likewise.
22751 (BT_FN_UINT8_FLOAT): Likewise.
22752 (BT_FN_UINT64_FLOAT): Likewise.
22753 (BT_FN_UINT16_UINT32): Likewise.
22754 (BT_FN_UINT32_UINT16): Likewise.
22755 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22756 (BT_FN_INT_PTR_INT): Likewise.
22757 (BT_FN_UINT_PTR_UINT): Likewise.
22758 (BT_FN_LONG_PTR_LONG): Likewise.
22759 (BT_FN_ULONG_PTR_ULONG): Likewise.
22760 (BT_FN_VOID_UINT64_UINT64): Likewise.
22761 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22762 (BT_FN_INT8_INT8_INT8): Likewise.
22763 (BT_FN_INT16_INT16_INT16): Likewise.
22764 (BT_FN_INT_INT_INT): Likewise.
22765 (BT_FN_UINT_FLOAT_UINT): Likewise.
22766 (BT_FN_FLOAT_UINT_UINT): Likewise.
22767 (BT_FN_ULONG_UINT_UINT): Likewise.
22768 (BT_FN_ULONG_UINT_PTR): Likewise.
22769 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22770 (BT_FN_UINT_UINT_UINT): Likewise.
22771 (BT_FN_VOID_UINT_PTR): Likewise.
22772 (BT_FN_UINT_UINT_PTR: Likewise.
22773 (BT_FN_UINT32_UINT64_PTR): Likewise.
22774 (BT_FN_INT_INT_UINT_UINT): Likewise.
22775 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22776 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22777 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22778 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22779 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22780 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22781 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22782 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22783 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22784 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22785 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22786 * doc/frontends.texi: List BRIG FE.
22787 * doc/install.texi (Testing): Add BRIG tesring requirements.
22788 * doc/invoke.texi (Overall Options): Mention BRIG.
22789 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22791 2017-01-24 Richard Biener <rguenther@suse.de>
22793 PR translation/79208
22794 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22796 2017-01-24 Martin Jambor <mjambor@suse.cz>
22799 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22800 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22801 and known_contexts.
22803 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22805 PR middle-end/79123
22806 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22807 casts from signed to unsigned really don't have a range.
22809 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22811 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22812 GMP_RNDx for compatiblity.
22814 2017-01-24 Martin Liska <mliska@suse.cz>
22817 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22818 that would prevent us to call alloca with -1 as argument.
22820 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22822 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22823 Avoid -Wformat-security warning.
22825 2017-01-23 Andrew Pinski <apinski@cavium.com>
22827 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22830 2017-01-23 Martin Sebor <msebor@redhat.com>
22832 PR middle-end/78703
22833 * gimple-ssa-sprintf.c (warn_level): New global.
22834 (format_integer): Use it here and throughout the rest of the file.
22835 Use the same switch to compute sign as base.
22836 (maybe_warn): New function.
22837 (format_directive): Factor out warnings into maybe_warn.
22838 Add debugging output. Use warn_level.
22839 (add_bytes): Use warn_level.
22840 (pass_sprintf_length::compute_format_length): Add debugging output.
22841 (try_substitute_return_value): Same.
22842 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22844 PR middle-end/78703
22845 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22846 (struct fmtresult, format_integer, format_floating): Adjust.
22847 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22848 (get_string_length, format_string,format_directive): Same.
22849 (pass_sprintf_length::compute_format_length): Same.
22850 (try_substitute_return_value): Simplify slightly.
22852 PR middle-end/78703
22853 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22854 (fmtresult::operator+=): Outlined.
22855 (struct fmtresult): Add ctors.
22856 (struct conversion_spec): Rename...
22857 (struct directive): ...to this. Add and remove data members.
22858 (directive::set_width, directive::set_precision): New functions.
22859 (format_percent): Use fmtresult ctor.
22860 (get_width_and_precision): Remove.
22861 (format_integer): Make naming changes. Avoid computing width and
22863 (format_floating): Same. Adjust indentation.
22864 (format_character, format_none): New functions.
22865 (format_string): Moved character handling to format_character.
22866 (format_directive): Remove arguments, change return type.
22867 (parse_directive): New function.
22868 (pass_sprintf_length::compute_format_length): Move directive
22869 parsing to parse_directive.
22871 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22873 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22874 (assign_assembler_name_if_needed): ... this.
22875 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22876 (assign_assembler_name_if_needed): ... this.
22877 (free_lang_data_in_cgraph): Adjust callers.
22878 * cgraphunit.c (cgraph_node::analyze): Likewise.
22879 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22882 2017-01-23 Richard Biener <rguenther@suse.de>
22884 PR tree-optimization/79088
22885 PR tree-optimization/79188
22886 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22887 resetting loop bounds after last path deletion. Reset loop
22888 bounds of the target loop, make code match the comments.
22889 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22890 Make sure loops need no fixups.
22892 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22894 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22895 exponent support with double type for first argument.
22896 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22897 type returned by __builtin_vec_extract_sig,
22898 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22899 functions from "vector int" to "vector unsigned int" or from
22900 "vector long long int" to "vector unsigned long long int".
22901 Changed type returned by __builtin_vec_extract_exp,
22902 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22903 functions from "vector int" to "vector unsigned int" or from
22904 "vector long long int" to "vector unsigned long long int".
22905 Changed return type of __builtin_vec_test_data_class,
22906 __builtin_vec_test_data_class_sp, and
22907 __builtin_vec_test_data_class_dp from "vector int" to
22908 "vector bool int" or from "vector long long int" to "vector bool
22909 long long int" and changed second argument type from "unsigned
22910 int" to "int". Added new overloaded function forms "vector float
22911 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22912 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22913 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22914 double, vector unsigned long long int)" and "vector double
22915 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22916 long int)". Changed return type of
22917 __builtin_scalar_test_data_class and
22918 __builtin_scalar_test_data_class_sp and
22919 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22920 int" and changed second argument from "unsigned int" to "int".
22921 Changed type returned by __builtin_scalar_test_neg,
22922 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
22923 from "int" to "bool int". Added new overloaded function form
22924 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
22925 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
22926 exponent double-precision with floating point first argument.
22927 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
22928 documentation of scalar_test_data_class, scalar_test_neg,
22929 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
22930 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
22931 vec_test_data_class built-in functions to reflect refinements in
22932 their type signatures.
22934 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
22936 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
22938 (aarch64_elf_asm_destructor): Likewise.
22940 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
22942 PR rtl-optimization/78634
22943 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
22944 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
22945 * ifcvt.c (noce_try_cmove): Add missing cost check.
22947 PR rtl-optimization/71724
22948 * combine.c (if_then_else_cond): Look for situations where it is
22949 beneficial to undo the work of one of the recursive calls.
22951 2017-01-23 Bin Cheng <bin.cheng@arm.com>
22953 PR tree-optimization/70754
22954 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
22955 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
22956 combined stmt before it if not NULL.
22957 (combine_chains): Process refs reversely and compute dominance point
22960 2017-01-23 Martin Liska <mliska@suse.cz>
22962 PR tree-optimization/79196
22963 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
22964 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
22966 (strlen_optimize_stmt): Call the renamed function.
22968 2017-01-23 Michael Matz <matz@suse.de>
22970 PR tree-optimization/78384
22971 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
22973 2017-01-23 Richard Biener <rguenther@suse.de>
22975 PR tree-optimization/79186
22976 * tree-vrp.c (register_new_assert_for): Make sure we've seen
22977 both incoming edges before moving an assert.
22979 2017-01-23 Martin Jambor <mjambor@suse.cz>
22981 * ipa-prop.c (load_from_param_1): Removed.
22982 (load_from_unmodified_param): Bits from load_from_param_1 put back
22984 (load_from_param): Removed.
22985 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
22986 with stmt. Reverted back to use of load_from_unmodified_param.
22988 2017-01-23 Martin Jambor <mjambor@suse.cz>
22991 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
22992 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
22993 field a pointer to garbage collected vector, mark lattices and
22994 ipcp_orig_node with GTY((skip)).
22995 (ipa_get_param_count): Adjust to descriptors being a pointer.
22996 (ipa_get_param): Likewise.
22997 (ipa_get_type): Likewise.
22998 (ipa_get_param_move_cost): Likewise.
22999 (ipa_set_param_used): Likewise.
23000 (ipa_get_controlled_uses): Likewise.
23001 (ipa_set_controlled_uses): Likewise.
23002 (ipa_is_param_used): Likewise.
23003 (ipa_node_params_t): Move into garbage collector. New methods insert
23005 (ipa_node_params_sum): Annotate wth GTY(()).
23006 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23008 (ipa_load_from_parm_agg): Adjust declaration.
23009 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23010 * ipa-profile.c (ipa_profile): Likewise.
23011 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23012 (ipa_populate_param_decls): Make descriptors parameter garbage
23014 (ipa_dump_param): Adjust to descriptors being a pointer.
23015 (ipa_alloc_node_params): Likewise.
23016 (ipa_initialize_node_params): Likewise.
23017 (load_from_param_1): Make descriptors parameter garbage collected.
23018 (load_from_unmodified_param): Likewise.
23019 (load_from_param): Likewise.
23020 (ipa_load_from_parm_agg): Likewise.
23021 (ipa_node_params::~ipa_node_params): Removed.
23022 (ipa_free_all_node_params): Remove call to delete operator.
23023 (ipa_node_params_t::insert): New.
23024 (ipa_node_params_t::remove): Likewise.
23025 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23026 copy known_csts and known_contexts vectors.
23027 (ipa_read_node_info): Adjust to descriptors being a pointer.
23028 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23030 (ipcp_transform_function): Make descriptors variable garbage
23033 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23035 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23036 * config/i386/avx512dqintrin.h: Ditto.
23037 * config/i386/avx512fintrin.h: Ditto.
23038 * config/i386/i386.c: Handle new builtins.
23039 * config/i386/i386-builtin.def: Add new builtins.
23040 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23041 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23043 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23044 Martin Liska <mliska@suse.cz>
23046 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23047 * asan.c (asan_expand_poison_ifn): Support stores and use
23048 appropriate ASAN report function.
23049 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23050 * internal-fn.def (ASAN_POISON_USE): Declare.
23051 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23052 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23053 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23054 ASAN_POISON calls w/o LHS.
23055 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23056 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23057 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23058 * gimplify.c (asan_poison_variables): Add attribute
23059 use_after_scope_memory to variables that really needs to live
23061 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23062 having the attribute.
23064 2017-01-23 Martin Liska <mliska@suse.cz>
23066 * asan.c (create_asan_shadow_var): New function.
23067 (asan_expand_poison_ifn): Likewise.
23068 * asan.h (asan_expand_poison_ifn): New declaration.
23069 * internal-fn.c (expand_ASAN_POISON): Likewise.
23070 * internal-fn.def (ASAN_POISON): New builtin.
23071 * sanopt.c (pass_sanopt::execute): Expand
23072 asan_expand_poison_ifn.
23073 * tree-inline.c (copy_decl_for_dup_finish): Make function
23075 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23076 * tree-ssa.c (is_asan_mark_p): New function.
23077 (execute_update_addresses_taken): Rewrite local variables
23078 (identified just by use-after-scope as addressable) into SSA.
23080 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23082 * doc/install.texi (Specific): opensource.apple.com uses https
23083 now. Remove trailing slash.
23085 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23087 * README.Portability: Remove note on an Irix compatibility issue.
23089 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23091 * gcov.c (INCLUDE_ALGORITHM): Define.
23092 (INCLUDE_VECTOR): Define.
23093 No longer include <vector> and <algorithm> directly.
23095 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23097 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23099 * doc/invoke.texi (Code Gen Options): Ditto.
23101 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23104 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23106 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23108 rtl-optimization/79125
23109 * cprop.c (local_cprop_pass): Handle cases where we make an
23110 unconditional trap.
23112 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23116 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23117 read from, for big endian.
23119 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23121 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23122 register pauth builtins for LP64 only.
23124 2017-01-20 Marek Polacek <polacek@redhat.com>
23127 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23130 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23132 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23134 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23135 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23136 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23138 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23141 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23142 in big-endian mode when they are not a single duplicated value.
23144 2017-01-20 Richard Biener <rguenther@suse.de>
23146 * BASE-VER: Bump to 7.0.1.
23148 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23150 * omp-low.c (omplow_simd_context): New struct. Use it...
23151 (lower_rec_simd_input_clauses): ...here and...
23152 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23153 references to idx, lane, max_vf, is_simt.
23155 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23157 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23160 2017-01-20 Martin Jambor <mjambor@suse.cz>
23162 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23163 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23165 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23166 (GTFILES): Rename hsa.c to hsa-common.c.
23167 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23168 * hsa-dump.c: Likewise.
23169 * hsa-gen.c: Likewise.
23170 * hsa-regalloc.c: Likewise.
23171 * ipa-hsa.c: Likewise.
23172 * omp-expand.c: Likewise.
23173 * omp-low.c: Likewise.
23174 * toplev.c: Likewise.
23176 2017-01-20 Marek Polacek <polacek@redhat.com>
23179 * doc/invoke.texi: Document -Wduplicated-branches.
23180 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23181 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23182 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23183 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23184 return 0 only when not OEP_LEXICOGRAPHIC.
23185 (fold_build_cleanup_point_expr): Use the expression
23186 location when building CLEANUP_POINT_EXPR.
23187 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23188 * tree.c (add_expr): Handle error_mark_node.
23190 2017-01-20 Martin Liska <mliska@suse.cz>
23193 * tree-profile.c (init_ic_make_global_vars): Do not call
23195 (gimple_init_gcov_profiler): Likewise.
23197 2017-01-20 Martin Liska <mliska@suse.cz>
23200 * cgraph.h (maybe_create_reference): Remove argument and
23202 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23204 * ipa-cp.c (create_specialized_node): Likewise.
23205 * symtab.c (symtab_node::maybe_create_reference): Handle
23206 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23208 2017-01-20 Martin Liska <mliska@suse.cz>
23210 * read-rtl-function.c (function_reader::create_function): Use
23211 build_decl instread of build_decl_stat.
23213 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23215 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23216 * config/i386/avx512dqintrin.h: Ditto.
23217 * config/i386/avx512fintrin.h: Ditto.
23218 * config/i386/i386-builtin-types.def: Add new types.
23219 * config/i386/i386.c: Handle new types.
23220 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23221 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23222 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23223 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23224 (__builtin_ia32_kshiftridi): New.
23225 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23227 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23231 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23232 define to rs6000_init_stack_protect_guard.
23233 (rs6000_init_stack_protect_guard): New function.
23235 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23236 Yunqiang Su <yunqiang.su@imgtec.com>
23238 * config.gcc (supported_defaults): Add madd4.
23239 (with_madd4): Add validation.
23240 (all_defaults): Add madd4.
23241 * config/mips/mips.opt (mmadd4): New option.
23242 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23244 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23246 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23247 (ISA_HAS_FUSED_MADD4): Likewise.
23248 * doc/invoke.texi (-mmadd4): Document the new option.
23249 * doc/install.texi (--with-madd4): Document the new option.
23251 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23253 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23254 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23255 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23256 (aarch64_init_pauth_hint_builtins): New.
23257 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23258 (aarch64_expand_builtin): Expand new builtins.
23260 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23262 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23263 * combine-stack-adj.c (no_unhandled_cfa): Handle
23264 REG_CFA_TOGGLE_RA_MANGLE.
23265 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23266 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23267 info for return address signing.
23268 (aarch64_expand_epilogue): Likewise.
23270 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23272 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23273 * config/aarch64/aarch64-protos.h
23274 (aarch64_return_address_signing_enabled): New declaration.
23275 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23277 (aarch64_expand_prologue): Sign return address before it's pushed onto
23279 (aarch64_expand_epilogue): Authenticate return address fetched from
23281 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23282 (aarch64_attributes): New function attributes for "sign-return-address".
23283 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23284 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23285 ("*do_return"): Generate combined instructions according to key index.
23286 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23287 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23289 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23290 * config/aarch64/aarch64.opt (msign-return-address=): New.
23291 * doc/extend.texi (AArch64 Function Attributes): Documents
23292 "sign-return-address=".
23293 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23295 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23297 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23298 overall option summary.
23300 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23302 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23303 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23304 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23305 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23307 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23309 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23310 -mpower9-minmax by default for -mcpu=power9.
23311 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23312 128-bit floating point.
23314 2017-01-20 Alan Modra <amodra@gmail.com>
23316 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23317 optimizing for size.
23319 2017-01-20 Alan Modra <amodra@gmail.com>
23322 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23323 for strcmp and strncmp from corresponding builtin decl.
23325 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23327 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23328 instead of i386/rtems-64.h.
23329 * config/i386/rtems-64.h: Remove.
23331 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23335 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23337 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23339 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23341 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23342 Change int to HOST_WIDE_INT.
23343 * config/aarch64/aarch64-protos.h
23344 (aarch64_simd_gen_const_vector_dup): Likewise.
23345 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23347 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23349 * langhooks-def.h (lhd_type_for_size): New decl.
23350 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23351 * langhooks.c (lhd_type_for_size): New function, taken from
23354 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23356 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23357 define_bypass for CR latency.
23358 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23359 (power9-alu2): Add define_bypass for CR latency.
23361 (power9-mul): Update insn latency.
23362 (power9-mul-compare): Update insn latency, bypass latency and remove
23365 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23367 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23369 * config/aarch64/aarch64.md
23370 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23371 aarch64_nopcrelative_literal_loads.
23372 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23374 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23376 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23377 TARGET_LOONGSON_3A.
23378 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23380 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23383 * config.gcc (supported_defaults): Add lxc1-sxc1.
23384 (with_lxc1_sxc1): Add validation.
23385 (all_defaults): Add lxc1-sxc1.
23386 * config/mips/mips.opt (mlxc1-sxc1): New option.
23387 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23389 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23390 __mips_no_lxc1_sxc1.
23391 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23392 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23393 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23395 2017-01-19 Richard Biener <rguenther@suse.de>
23397 PR tree-optimization/72488
23398 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23399 sure to restore SSA info.
23400 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23402 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23404 PR rtl-optimization/79121
23405 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23406 of the inner type when shifting an extended value.
23408 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23411 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23412 interposable aliases.
23414 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23417 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23418 Use the evmergelohi instruction.
23419 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23420 (mov_sitf_e500_subreg8_2_be): Likewise.
23421 (mov_sitf_e500_subreg12_2_le): Likewise.
23422 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23423 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23424 (mov_sitf_e500_subreg8_2_le): Likewise.
23425 (mov_sitf_e500_subreg12_2_be): Likewise.
23427 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23429 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23430 attribute from vecsimple to vecperm.
23431 (altivec_vbpermq2): Likewise.
23433 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23436 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23438 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23439 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23440 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23441 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23442 case where N arg is SIZE_MAX.
23443 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23444 (cmpstrsi): Add pattern.
23446 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23448 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23449 __builtin_vec_revb builtins.
23450 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23451 built-in functions to support generation of the ISA 3.0 XXBR<x>
23452 vector byte reverse instructions.
23453 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23454 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23455 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23456 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23457 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23458 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23459 (P9V_BUILTIN_VEC_REVB): Likewise.
23460 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23461 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23462 (p9_xxbrq_v16qi): Likewise.
23463 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23464 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23465 (p9_xxbrh_v8hi): Likewise.
23466 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23467 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23468 vec_revb built-in functions.
23470 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23472 PR rtl-optimization/78952
23473 * config/i386/i386.md (any_extract): New code iterator.
23474 (*insvqi_2): Use any_extract for source operand.
23475 (*insvqi_3): Use any_shiftrt for source operand.
23477 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23479 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23481 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23483 2017-01-18 Matthias Klose <doko@ubuntu.com>
23485 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23487 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23489 * config/rs6000/altivec.h (vec_bperm): Change #define.
23490 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23491 (altivec_vbpermq2): New define_insn.
23492 (altivec_vbpermd): Likewise.
23493 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23494 function interface.
23495 (VBPERMD): Likewise.
23496 (VBPERM): New polymorphic function interface.
23497 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23498 Add entries for P9V_BUILTIN_VEC_VBPERM.
23499 * doc/extend.texi: Add interfaces for vec_bperm.
23501 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23503 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23504 first letter of error messages.
23505 (s390_resolve_overloaded_builtin): Likewise.
23506 * config/s390/s390.c (s390_expand_builtin): Likewise.
23507 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23508 (s390_valid_target_attribute_inner_p): Likewise.
23509 * config/s390/s390.md ("tabort"): Likewise.
23511 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23513 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23514 (ISA_AVOID_DIV_HILO): New macro.
23515 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23516 (ISA_HAS_DDIV): Likewise.
23518 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23520 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23522 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23524 * doc/invoke.texi (fabi-version): Spelling fix.
23526 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23529 * doc/invoke.texi (fabi-version): Mention mangling fix for
23532 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23535 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23537 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23540 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23543 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23545 (stack_protect_test): Ditto.
23546 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23547 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23549 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23550 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23551 -mstack-protector-guard-offset=.
23552 (RS/6000 and PowerPC Options): Ditto.
23554 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23556 * config/i386/i386.h (MASK_CLASS_P): New define.
23557 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23558 there are no registers from different register sets also when
23559 mask registers are used. Update function comment.
23560 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23561 to (*k/*r) and (*k/*km) alternatives.
23563 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23565 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23566 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23567 (EH_RETURN_HANDLER_RTX): New define.
23568 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23569 Force frame pointer in EH return functions.
23570 (aarch64_expand_epilogue): Add barrier for eh_return.
23571 (aarch64_final_eh_return_addr): Remove.
23572 (aarch64_eh_return_handler_rtx): New function.
23573 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23575 (aarch64_eh_return_handler_rtx): New prototype.
23577 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23579 * config/rs6000/altivec.h (vec_rlmi): New #define.
23580 (vec_vrlnm): Likewise.
23581 (vec_rlnm): Likewise.
23582 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23583 (UNSPEC_VRLNM): Likewise.
23584 (VIlong): New mode iterator.
23585 (altivec_vrl<VI_char>mi): New define_insn.
23586 (altivec_vrl<VI_char>nm): Likewise.
23587 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23589 (VRLDNM): Likewise.
23590 (RLNM): New polymorphic function entry.
23591 (VRLWMI): New monomorphic function entry.
23592 (VRLDMI): Likewise.
23593 (RLMI): New polymorphic function entry.
23594 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23595 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23596 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23599 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23602 * dwarf2out.c (field_byte_offset): Restore the
23603 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23604 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23605 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23608 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23611 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23613 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23616 * dwarf2out.c (add_data_member_location_attribute): For constant
23617 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23618 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23619 DW_AT_byte_size attributes.
23621 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23623 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23624 after forcing to constant memory when the code model is medium.
23626 2017-01-17 Julia Koval <julia.koval@intel.com>
23629 * config/i386/avx512fintrin.h
23630 (_mm512_i32gather_ps): Change __addr type to void const*.
23631 (_mm512_mask_i32gather_ps): Ditto.
23632 (_mm512_i32gather_pd): Ditto.
23633 (_mm512_mask_i32gather_pd): Ditto.
23634 (_mm512_i64gather_ps): Ditto.
23635 (_mm512_mask_i64gather_ps): Ditto.
23636 (_mm512_i64gather_pd): Ditto.
23637 (_mm512_mask_i64gather_pd): Ditto.
23638 (_mm512_i32gather_epi32): Ditto.
23639 (_mm512_mask_i32gather_epi32): Ditto.
23640 (_mm512_i32gather_epi64): Ditto.
23641 (_mm512_mask_i32gather_epi64): Ditto.
23642 (_mm512_i64gather_epi32): Ditto.
23643 (_mm512_mask_i64gather_epi32): Ditto.
23644 (_mm512_i64gather_epi64): Ditto.
23645 (_mm512_mask_i64gather_epi64): Ditto.
23646 (_mm512_i32scatter_ps): Change __addr type to void*.
23647 (_mm512_mask_i32scatter_ps): Ditto.
23648 (_mm512_i32scatter_pd): Ditto.
23649 (_mm512_mask_i32scatter_pd): Ditto.
23650 (_mm512_i64scatter_ps): Ditto.
23651 (_mm512_mask_i64scatter_ps): Ditto.
23652 (_mm512_i64scatter_pd): Ditto.
23653 (_mm512_mask_i64scatter_pd): Ditto.
23654 (_mm512_i32scatter_epi32): Ditto.
23655 (_mm512_mask_i32scatter_epi32): Ditto.
23656 (_mm512_i32scatter_epi64): Ditto.
23657 (_mm512_mask_i32scatter_epi64): Ditto.
23658 (_mm512_i64scatter_epi32): Ditto.
23659 (_mm512_mask_i64scatter_epi32): Ditto.
23660 (_mm512_i64scatter_epi64): Ditto.
23661 (_mm512_mask_i64scatter_epi64): Ditto.
23662 * config/i386/avx512pfintrin.h
23663 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23664 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23665 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23666 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23667 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23668 (_mm512_prefetch_i32scatter_ps): Ditto.
23669 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23670 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23671 (_mm512_prefetch_i64scatter_pd): Ditto.
23672 (_mm512_prefetch_i64scatter_ps): Ditto.
23673 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23674 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23675 * config/i386/avx512vlintrin.h
23676 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23677 (_mm_mmask_i32gather_ps): Ditto.
23678 (_mm256_mmask_i32gather_pd): Ditto.
23679 (_mm_mmask_i32gather_pd): Ditto.
23680 (_mm256_mmask_i64gather_ps): Ditto.
23681 (_mm_mmask_i64gather_ps): Ditto.
23682 (_mm256_mmask_i64gather_pd): Ditto.
23683 (_mm_mmask_i64gather_pd): Ditto.
23684 (_mm256_mmask_i32gather_epi32): Ditto.
23685 (_mm_mmask_i32gather_epi32): Ditto.
23686 (_mm256_mmask_i32gather_epi64): Ditto.
23687 (_mm_mmask_i32gather_epi64): Ditto.
23688 (_mm256_mmask_i64gather_epi32): Ditto.
23689 (_mm_mmask_i64gather_epi32): Ditto.
23690 (_mm256_mmask_i64gather_epi64): Ditto.
23691 (_mm_mmask_i64gather_epi64): Ditto.
23692 (_mm256_i32scatter_ps): Change __addr type to void*.
23693 (_mm256_mask_i32scatter_ps): Ditto.
23694 (_mm_i32scatter_ps): Ditto.
23695 (_mm_mask_i32scatter_ps): Ditto.
23696 (_mm256_i32scatter_pd): Ditto.
23697 (_mm256_mask_i32scatter_pd): Ditto.
23698 (_mm_i32scatter_pd): Ditto.
23699 (_mm_mask_i32scatter_pd): Ditto.
23700 (_mm256_i64scatter_ps): Ditto.
23701 (_mm256_mask_i64scatter_ps): Ditto.
23702 (_mm_i64scatter_ps): Ditto.
23703 (_mm_mask_i64scatter_ps): Ditto.
23704 (_mm256_i64scatter_pd): Ditto.
23705 (_mm256_mask_i64scatter_pd): Ditto.
23706 (_mm_i64scatter_pd): Ditto.
23707 (_mm_mask_i64scatter_pd): Ditto.
23708 (_mm256_i32scatter_epi32): Ditto.
23709 (_mm256_mask_i32scatter_epi32): Ditto.
23710 (_mm_i32scatter_epi32): Ditto.
23711 (_mm_mask_i32scatter_epi32): Ditto.
23712 (_mm256_i32scatter_epi64): Ditto.
23713 (_mm256_mask_i32scatter_epi64): Ditto.
23714 (_mm_i32scatter_epi64): Ditto.
23715 (_mm_mask_i32scatter_epi64): Ditto.
23716 (_mm256_i64scatter_epi32): Ditto.
23717 (_mm256_mask_i64scatter_epi32): Ditto.
23718 (_mm_i64scatter_epi32): Ditto.
23719 (_mm_mask_i64scatter_epi32): Ditto.
23720 (_mm256_i64scatter_epi64): Ditto.
23721 (_mm256_mask_i64scatter_epi64): Ditto.
23722 (_mm_i64scatter_epi64): Ditto.
23723 (_mm_mask_i64scatter_epi64): Ditto.
23724 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
23725 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
23726 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
23727 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
23728 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
23729 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
23730 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
23731 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
23732 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
23733 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
23734 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
23735 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
23736 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
23737 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
23738 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
23739 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
23740 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
23741 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
23742 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
23743 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
23744 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
23745 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
23746 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
23747 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23748 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23749 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23750 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23751 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23752 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23753 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23754 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23755 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23756 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23757 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23758 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23759 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23760 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23761 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23762 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23763 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23764 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23765 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23766 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23767 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23768 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23769 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23770 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23771 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23772 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23773 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23774 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23775 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23776 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23777 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23778 definitions accordingly.
23780 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23781 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23784 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23787 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23790 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23791 hard regs for inner regno.
23793 2017-01-17 Martin Liska <mliska@suse.cz>
23796 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23797 assumption and add comment.
23799 2017-01-17 Nathan Sidwell <nathan@acm.org>
23801 * ipa-visibility.c (localize_node): New function, broken out of ...
23802 (function_and_variable_visibility): ... here. Call it.
23804 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23806 PR middle-end/77445
23807 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23808 correctly set frequency of oudgoing edge.
23809 (duplicate_thread_path): Fix profile updating.
23811 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23814 * configure.ac: Add GCC_BASE_VER.
23815 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23816 version from BASE-VER file.
23817 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23818 (gcc.o): Depend on $(BASEVER).
23819 * common.opt (dumpfullversion): New option.
23820 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23821 * doc/invoke.texi: Document -dumpfullversion.
23822 * doc/install.texi: Document --with-gcc-major-version-only.
23823 * configure: Regenerated.
23825 2017-01-17 Richard Biener <rguenther@suse.de>
23827 PR tree-optimization/71433
23828 * tree-vrp.c (register_new_assert_for): Merge same asserts
23829 on all incoming edges.
23830 (process_assert_insertions_for): Handle insertions at the
23833 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23835 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23836 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23838 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23841 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23844 2017-01-17 Alan Modra <amodra@gmail.com>
23847 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23848 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23849 symbolic stack limit when pic.
23851 2017-01-16 Martin Sebor <msebor@redhat.com>
23853 PR tree-optimization/78608
23854 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23856 2017-01-16 Jeff Law <law@redhat.com>
23859 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23860 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23861 for several include directories that may be relative to sysroot.
23862 * config/i386/x-mingw32 (gplus_includedir): Define.
23863 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23864 (native_system_includedir): Likewise.
23865 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23866 override if TARGET_SYSTEM_ROOT is defined.
23867 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23869 PR tree-optimization/79090
23870 PR tree-optimization/33562
23871 PR tree-optimization/61912
23872 PR tree-optimization/77485
23873 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23874 and computed trims into the dump file.
23876 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23878 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23880 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23883 * gimplify.c (gimplify_init_constructor): If want_value and
23884 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23888 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23889 sequence. Formatting fixes.
23890 (doloop_optimize): Formatting fixes.
23893 * gcc.c (debug_level_greater_than_spec_func): New function.
23894 (static_spec_functions): Add debug-level-gt spec function.
23895 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23897 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23898 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23899 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23900 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23901 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23902 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23904 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23906 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23907 QImode fixups to general and mask registers only.
23909 2017-01-16 Carl Love <cel@us.ibm.com>
23911 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23912 for built-in functions
23913 vector signed char vec_nabs (vector signed char)
23914 vector signed short vec_nabs (vector signed short)
23915 vector signed int vec_nabs (vector signed int)
23916 vector signed long long vec_nabs (vector signed long long)
23917 vector float vec_nabs (vector float)
23918 vector double vec_nabs (vector double)
23919 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23921 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23922 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
23923 * doc/extend.texi: Update the documentation file for the new built-in
23926 2017-01-16 Martin Sebor <msebor@redhat.com>
23928 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
23931 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23933 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
23934 UNSPEC_VSX__XXSPLTD to require special splat handling.
23936 2017-01-16 David Malcolm <dmalcolm@redhat.com>
23939 * system.h: Poison strndup.
23941 2017-01-16 Alan Modra <amodra@gmail.com>
23944 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
23947 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
23949 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
23951 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
23953 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
23954 call recog here. Assert that INSN_CODE (insn) is non-negative.
23956 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23959 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
23961 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
23962 in the currently scheduled RTL fragment.
23964 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23966 PR rtl-optimization/78751
23967 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
23970 2017-01-14 Jeff Law <law@redhat.com>
23972 PR tree-optimization/79090
23973 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
23974 variable length stores.
23975 (compute_trims): Delete dead assignment to *trim_tail.
23976 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
23979 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
23981 PR rtl-optimization/78626
23982 PR rtl-optimization/78727
23983 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
23984 of a block, and split such blocks after everything else is finished.
23986 2017-01-14 Alan Modra <amodra@gmail.com>
23989 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
23990 target legitimate_combined_insn.
23991 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
23992 (rs6000_legitimate_combined_insn): New function.
23993 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
23995 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
23996 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
23997 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
23999 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24001 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24003 2017-01-13 Jeff Law <law@redhat.com>
24005 PR tree-optimization/33562
24006 PR tree-optimization/61912
24007 PR tree-optimization/77485
24008 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24010 (delete_dead_assignment): Likewise.
24011 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24012 statement to delete_dead_call and delete_dead_assignment.
24014 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24017 * substring-locations.c (format_warning_va): Strengthen case 1 so
24018 that both endpoints of the substring must be within the format
24019 range for just the substring to be printed.
24021 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24023 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24024 * config/i386/i386.c (ix86_target_string): Add missing options
24025 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24026 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24027 flags_other and ix86_target_other to flags2_other. Display unknown
24029 (ix86_valid_target_attribute_inner_p): Add missing options and
24030 reorder options by implied ISAs, as in ix86_target_string.
24032 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24034 * hash-table.h (hash_table::too_empty_p): New function.
24035 (hash_table::expand): Use it.
24036 (hash_table::traverse): Likewise.
24037 (hash_table::empty_slot): Use sizeof (value_type) instead of
24038 sizeof (PTR) to convert bytes to elements. Shrink the table
24039 if the current size is excessive for the current number of
24042 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24044 * ira-costs.c (record_reg_classes): Break from the inner loop
24045 early once alt_fail is known to be true. Update outer loop
24046 handling accordingly.
24048 2017-01-13 Jeff Law <law@redhat.com>
24050 * tree-ssa-dse.c (decrement_count): New function.
24051 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24052 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24053 when we know the partially dead statement is a mem* function.
24055 PR tree-optimization/61912
24056 PR tree-optimization/77485
24057 * tree-ssa-dse.c: Include expr.h.
24058 (maybe_trim_constructor_store): New function.
24059 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24061 PR tree-optimization/33562
24062 PR tree-optimization/61912
24063 PR tree-optimization/77485
24064 * doc/invoke.texi: Document new dse-max-object-size param.
24065 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24066 * tree-ssa-dse.c: Include params.h.
24067 (dse_store_status): New enum.
24068 (initialize_ao_ref_for_dse): New, partially extracted from
24070 (valid_ao_ref_for_dse, normalize_ref): New.
24071 (setup_live_bytes_from_ref, compute_trims): Likewise.
24072 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24073 (maybe_trim_partially_dead_store): Likewise.
24074 (maybe_trim_complex_store): Likewise.
24075 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24076 Track what bytes live from the original store. Return tri-state
24077 for dead, partially dead or live.
24078 (dse_dom_walker): Add constructor, destructor and new private members.
24079 (delete_dead_call, delete_dead_assignment): New extracted from
24081 (dse_optimize_stmt): Make a member of dse_dom_walker.
24082 Use initialize_ao_ref_for_dse.
24084 PR tree-optimization/33562
24085 PR tree-optimization/61912
24086 PR tree-optimization/77485
24087 * sbitmap.h (bitmap_count_bits): Prototype.
24088 (bitmap_clear_range, bitmap_set_range): Likewise.
24089 * sbitmap.c (bitmap_clear_range): New function.
24090 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24092 2017-01-13 Martin Liska <mliska@suse.cz>
24095 * function.c (set_cfun): Add new argument force.
24096 * function.h (set_cfun): Likewise.
24097 * ipa-inline-transform.c (inline_call): Use the function when
24098 strict alising from is dropped for function we inline to.
24100 2017-01-13 Richard Biener <rguenther@suse.de>
24102 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24103 for dumping GIMPLE INTEGER_CSTs.
24105 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24107 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24108 to 201112L since C++17.
24110 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24113 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24114 if -fsanitize=kernel-address is present.
24116 2017-01-13 Richard Biener <rguenther@suse.de>
24118 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24119 as _Literal ( type ) number in case usual suffixes do not
24120 preserve all information.
24122 2017-01-13 Richard Biener <rguenther@suse.de>
24124 PR tree-optimization/77283
24125 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24126 and ssa-iterators.h.
24127 (is_feasible_trace): Implement a cost model based on joiner
24130 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24133 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24134 char or short to __float128/_Float128 directly.
24136 2017-01-12 Martin Sebor <msebor@redhat.com>
24138 to -Wformat-overflow.
24139 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24140 (min_bytes_remaining): Same.
24141 (get_string_length): Same.
24142 (format_string): Same.
24143 (format_directive): Same.
24145 (pass_sprintf_length::handle_gimple_call): Same.
24147 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24149 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24150 info.nowrite calls with no lhs that can't throw. Return bool
24151 whether gsi_remove has been called or not.
24152 (pass_sprintf_length::handle_gimple_call): Return bool whether
24153 try_substitute_return_value called gsi_remove. Formatting fix.
24154 (pass_sprintf_length::execute): Don't use gsi_remove if
24155 handle_gimple_call returned true.
24158 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24159 be removed due to side-effects, don't remove following barrier nor
24160 turn the successor edge into fallthru edge.
24162 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24165 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24166 element-reversing loads and stores as not swappable.
24168 2017-01-12 Nathan Sidwell <nathan@acm.org>
24169 Nicolai Stange <nicstange@gmail.com>
24171 * combine.c (try_combine): Don't ignore result of overlap checking
24172 loop. Combine overlap & asm check into single loop.
24174 2017-01-12 Richard Biener <rguenther@suse.de>
24176 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24177 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24179 2017-01-12 Richard Biener <rguenther@suse.de>
24181 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24182 and TS_TARGET_OPTION directly derive from TS_BASE.
24183 * tree-core.h (tree_optimization_option): Derive from tree_base.
24184 (tree_target_option): Likewise.
24186 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24188 * config/i386/i386.c (memory_address_length): Increase len
24189 only when rip_relative_addr_p returns false.
24191 2017-01-11 Julia Koval <julia.koval@intel.com>
24193 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24194 (OPTION_MASK_ISA_SGX_SET): New.
24195 (ix86_handle_option): Handle OPT_msgx.
24196 * config.gcc: Added sgxintrin.h.
24197 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24198 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24199 * config/i386/i386.c (ix86_target_string): Add -msgx.
24201 (ix86_option_override_internal): Handle new options.
24202 (ix86_valid_target_attribute_inner_p): Add sgx.
24203 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24204 * config/i386/i386.opt: Add msgx.
24205 * config/i386/sgxintrin.h: New file.
24206 * config/i386/x86intrin.h: Add sgxintrin.h.
24208 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24211 * fold-const.c (maybe_nonzero_address): Return 1 for function
24213 (tree_single_nonzero_warnv_p): Don't handle function local objects
24217 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24220 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24223 * opts.c: Include "spellcheck.h"
24224 (struct string_fragment): New struct.
24225 (struct edit_distance_traits<const string_fragment &>): New
24227 (get_closest_sanitizer_option): New function.
24228 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24230 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24232 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24234 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24235 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24236 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24237 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24238 for initial die_offset if dwarf_split_debug_info.
24239 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24240 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24242 (output_skeleton_debug_sections): Formatting fix. Use
24243 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24244 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24246 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24248 * config/arm/cortex-a53.md: Add bypasses for
24249 cortex_a53_r2f_cvt.
24250 (cortex_a53_r2f): Only use for transfers.
24251 (cortex_a53_f2r): Likewise.
24252 (cortex_a53_r2f_cvt): Add reservation for conversions.
24253 (cortex_a53_f2r_cvt): Likewise.
24255 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24257 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24258 to all inlined functions, change static to extern.
24260 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24263 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24265 (arm_assemble_integer): Likewise.
24267 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24269 * config.gcc: Use new awk script to check CPU, FPU and architecture
24270 parameters for --with-... options.
24271 * config/arm/parsecpu.awk: New file
24272 * config/arm/arm-cpus.in: New file.
24273 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24275 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24277 * config/arm/t-arm: Update dependency rules.
24278 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24279 of processing .def files.
24280 * config/arm/genopt.sh: Deleted.
24281 * config/arm/gentune.sh: Deleted.
24282 * config/arm/arm-cores.def: Deleted.
24283 * config/arm/arm-arches.def: Deleted.
24284 * config/arm/arm-fpus.def: Deleted.
24285 * config/arm/arm-tune.md: Regenerated.
24286 * config/arm/arm-tables.opt: Regenerated.
24287 * config/arm/arm-cpu.h: New generated file.
24288 * config/arm/arm-cpu-data.h: New generated file.
24289 * config/arm/arm-cpu-cdata.h: New generated file.
24291 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24294 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24296 (input_varpool_node): Unpack dynamically_initialized bit.
24298 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24300 PR rtl-optimization/79032
24301 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24302 the alignment of the adjusted memory reference against that of MODE,
24303 instead of the alignment of the original memory reference.
24305 2017-01-11 Martin Jambor <mjambor@suse.cz>
24307 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24309 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24310 decorated functions.
24312 2017-01-11 Richard Biener <rguenther@suse.de>
24314 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24315 set range/nonnull info for PHI results. Do not set it on
24316 stmts marked for removal.
24318 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24320 * expr.c (store_field): In the bitfield case, fetch the return value
24321 from the registers before applying a single big-endian adjustment.
24322 Always do a final load for a BLKmode value not larger than a word.
24324 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24327 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24328 that we correctly handle column numbers greater than
24329 LINE_MAP_MAX_COLUMN_NUMBER.
24331 2017-01-10 Martin Sebor <msebor@redhat.com>
24333 PR middle-end/78245
24334 * gimple-ssa-sprintf.c (get_destination_size): Call
24335 {init,fini}object_sizes.
24336 * tree-object-size.c (addr_object_size): Adjust.
24337 (pass_through_call): Adjust.
24338 (pass_object_sizes::execute): Adjust.
24339 * tree-object-size.h (fini_object_sizes): Declare.
24341 2017-01-10 Martin Sebor <msebor@redhat.com>
24343 PR tree-optimization/78775
24344 * builtins.c (get_size_range): Move...
24345 * calls.c: ...to here.
24346 (alloc_max_size): Accept zero argument.
24347 (operand_signed_p): Remove.
24348 (maybe_warn_alloc_args_overflow): Call get_size_range.
24349 * calls.h (get_size_range): Declare.
24351 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24353 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24354 from TI's devices.csv file as of September 2016.
24355 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24357 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24359 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24360 * doc/invoke.texi: Likewise.
24361 * doc/md.texi: Likewise.
24362 * doc/objc.texi: Likewise.
24364 2017-01-10 Joshua Conner <joshconner@google.com>
24366 * config/arm/fuchsia-elf.h: New file.
24367 * config/fuchsia.h: New file.
24368 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24369 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24371 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24373 2016-01-10 Richard Biener <rguenther@suse.de>
24375 PR tree-optimization/79034
24376 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24377 Propagate out degenerate PHIs in the joiner.
24379 2017-01-10 Martin Liska <mliska@suse.cz>
24381 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24382 (sort_congruence_classes_by_decl_uid): Likewise.
24383 (sort_congruence_class_groups_by_decl_uid): Likewise.
24384 (sem_item_optimizer::merge_classes): Sort class, groups in these
24385 classes and members in the groups by DECL_UID of declarations.
24386 This would make merge operations stable.
24388 2017-01-10 Martin Liska <mliska@suse.cz>
24390 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24391 usage of m_classes_vec.
24392 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24393 (sem_item_optimizer::get_group_by_hash): Likewise.
24394 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24395 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24396 (sem_item_optimizer::verify_classes): Likewise.
24397 (sem_item_optimizer::process_cong_reduction): Likewise.
24398 (sem_item_optimizer::dump_cong_classes): Likewise.
24399 (sem_item_optimizer::merge_classes): Likewise.
24400 * ipa-icf.h (congruence_class_hash): Rename from
24401 congruence_class_group_hash. Remove declaration of m_classes_vec.
24403 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24405 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24406 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24407 * config.gcc: Add avx512vpopcntdqintrin.h.
24408 * config/i386/avx512vpopcntdqintrin.h: New.
24409 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24410 * config/i386/i386-builtin-types.def: Add new types.
24411 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24412 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24413 __builtin_ia32_vpopcountq_v8di_mask): New.
24414 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24415 __AVX512VPOPCNTDQ__.
24416 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24417 (PTA_AVX512VPOPCNTDQ): Define.
24418 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24419 TARGET_AVX512VPOPCNTDQ_P): Define.
24420 * config/i386/i386.opt: Add mavx512vpopcntdq.
24421 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24422 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24424 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24426 PR middle-end/77484
24427 * predict.def (PRED_CALL): Set to 67.
24429 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24431 * expr.c (store_field): In the bitfield case, if the value comes from
24432 a function call and is of an aggregate type returned in registers, do
24433 not modify the field mode; extract the value in all cases if the mode
24434 is BLKmode and the size is not larger than a word.
24436 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24439 * config/i386/cpuid.h: Fix undefined behavior.
24441 2017-01-04 Jeff Law <law@redhat.com>
24443 PR tree-optimization/79007
24444 PR tree-optimization/67955
24445 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24446 conservative for pt.null when flag_non_call_exceptions is on.
24448 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24450 PR translation/79019
24451 PR translation/79020
24452 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24453 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24454 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24456 * config/avr/avr.opt (maccumulate-args): Likewise.
24457 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24458 * common.opt (freport-bug): Likewise.
24459 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24460 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24461 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24462 translatable string.
24463 * config/i386/i386.c (function_value_32): Likewise.
24464 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24465 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24467 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24468 * common/config/msp430/msp430-common.c (msp430_handle_option):
24470 * symtab.c (symtab_node::verify_base): Likewise.
24471 * opts.c (set_debug_level): Likewise.
24472 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24474 missing whitespace to translatable strings.
24475 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24476 * config/sh/superh.h: Likewise.
24477 * config/i386/xopintrin.h: Likewise.
24478 * config/i386/znver1.md: Likewise.
24479 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24480 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24481 * double-int.h (struct double_int): Likewise.
24482 * double-int.c (div_and_round_double): Likewise.
24483 * wide-int.cc: Likewise.
24484 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24485 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24486 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24487 (crossjumps_occurred): ... this.
24488 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24491 PR tree-optimization/78899
24492 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24493 returning bool return struct loop *, NULL for failure and the new
24495 (versionable_outer_loop_p): Don't version outer loop if it has
24496 dont_vectorized bit set.
24497 (tree_if_conversion): When versioning outer loop, ensure
24498 tree_if_conversion is performed also on the inner loop of the
24499 non-vectorizable outer loop copy.
24500 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24501 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24502 prevent vectorization of it.
24503 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24504 the outer loop vectorization of the non-scalar version is attempted
24505 before vectorization of the inner loop in scalar version. If
24506 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24507 vectorization of its inner loop.
24508 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24509 has 2 inner loops, rename also on edges from bb whose single pred
24510 is outer_loop->header. Fix typo in function comment.
24512 2017-01-09 Martin Sebor <msebor@redhat.com>
24515 * asan.c (asan_emit_stack_protection): Increase local buffer size
24516 to avoid snprintf truncation warning.
24518 2017-01-09 Andrew Pinski <apinski@cavium.com>
24520 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24521 to reference thunderx2t99 for the tuning structure
24522 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24524 (thunderx2t99_extra_costs): This.
24525 * config/aarch64/aarch64-tune.md: Regenerate.
24526 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24527 (vulcan_addrcost_table): This.
24528 (vulcan_regmove_cost): Rename to ...
24529 (thunderx2t99_regmove_cost): This.
24530 (vulcan_vector_cost): Rename to ...
24531 (thunderx2t99_vector_cost): this.
24532 (vulcan_branch_cost): Rename to ...
24533 (thunderx2t99_branch_cost): This.
24534 (vulcan_tunings): Rename to ...
24535 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24536 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24538 2017-01-09 Martin Jambor <mjambor@suse.cz>
24542 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24543 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24544 (propagate_vr_accross_jump_function): Use the above function for all
24545 value range computations for pass-through jump functions and type
24546 converasion from explicit value range values.
24547 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24548 parameters from TYPE_ARG_TYPES.
24549 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24550 (ipa_write_node_info): Stream type of the actual argument.
24551 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24553 2017-01-09 Martin Liska <mliska@suse.cz>
24556 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24557 (lookup_compiler): Do not show error message with have_E.
24559 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24561 PR tree-optimization/78938
24562 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24563 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24564 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24565 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24568 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24570 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24573 2017-01-09 Richard Biener <rguenther@suse.de>
24575 PR tree-optimization/78997
24576 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24577 name condition properly.
24579 2017-01-09 Richard Biener <rguenther@suse.de>
24582 * dwarf2out.c (is_cxx): New overload with context.
24583 (is_naming_typedef_decl): Use it.
24585 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24587 * invoke.texi (Option Summary): Correct spacing in option lists
24588 and add line breaks to fix over-long lines.
24590 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24592 PR middle-end/17660
24594 * extend.texi (Common Variable Attributes): Add xref to GCC
24595 Internals manual to explain mode attribute keywords.
24597 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24600 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24601 and Preprocessor Options.
24602 (Options for Linking): Document -pthread here....
24603 (RS/6000 and PowerPC Options): ...not here.
24604 (Solaris 2 Options): ...or here.
24605 * doc/cppopts.texi: Document -pthread.
24607 2017-01-08 Martin Sebor <msebor@redhat.com>
24609 PR middle-end/77708
24610 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24611 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24612 New member functions.
24613 (format_directive): Used them.
24615 (pass_sprintf_length::handle_gimple_call): Same.
24616 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24617 to avoid truncation for any argument.
24618 (extract_affine_mul): Same.
24619 * tree.c (get_file_function_name): Same.
24621 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24623 PR middle-end/77484
24624 * predict.def (PRED_INDIR_CALL): Set to 86.
24626 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24628 PR preprocessor/54124
24629 * doc/cppopts.texi: Reformat -d subtable to list the full name
24630 of the options. Add cross-reference to the docs for the general
24631 compiler -d options.
24632 * doc/invoke.texi (Developer Options): Add cross-reference to the
24633 preprocessor-specific -d option documentation.
24635 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24637 PR preprocessor/13498
24638 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24639 redudant material, and reflect new command-line options.
24640 (System Headers): Likewise.
24642 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24644 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24645 -isystem, and -idirafter. Copy-edit.
24646 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24647 default for -ftrack-macro-expansion. Delete obsolete and
24648 badly-formatted implementation details about -fdebug-cpp output.
24649 * doc/cppwarnopts.texi: Copy-edit.
24651 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24654 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24655 that the transition from a max line width >= 1<<10 to narrower
24656 lines works correctly.
24658 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24660 * doc/options.texi (PerFunction): New.
24661 * opt-functions.awk (switch_flags): Map both Optimization and
24662 PerFunction to CL_OPTIMIZATION.
24663 * opth-gen.awk: Test for PerFunction flag along with
24665 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24666 it only when the latter is present. Skip those that don't in
24667 the hash function generator.
24668 * common.opt (fvar-tracking): Mark as PerFunction instead of
24670 (fvar-tracking-assignments): Likewise.
24671 (fvar-tracking-assignments-toggle): Likewise.
24672 (fvar-tracking-uninit): Likewise.
24674 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24676 PR translation/79018
24677 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24680 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24683 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24684 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24685 TARGET_LONG_JUMP_TABLE_OFFSETS.
24686 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24687 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24688 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24689 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24690 * config/m68k/m68k.md (tablejump expander): Likewise.
24691 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24692 TARGET_LONG_JUMP_TABLE_OFFSETS.
24693 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24694 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24696 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24697 David Holsgrove <david.holsgrove@xilinx.com>
24699 * common/config/microblaze/microblaze-common.c
24700 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24701 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24703 * config/microblaze/microblaze.c (microblaze_must_save_register)
24704 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24706 (microblaze_eh_return): New function.
24707 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24708 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24709 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24710 * config/microblaze/microblaze.md (eh_return): New pattern.
24712 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24714 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
24715 GCC_DIAGNOSTIC_STRINGIFY): Define.
24717 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
24719 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24721 * config/arm/arm.md (<mcrr>): New.
24723 * config/arm/arm.c (arm_arch5te): New.
24724 (arm_option_override): Set arm_arch5te.
24725 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
24727 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
24728 (arm_mcrr_qualifiers): ... this. New.
24729 (MRRC_QUALIFIERS): Define to...
24730 (arm_mrrc_qualifiers): ... this. New.
24731 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
24733 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
24734 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
24735 (MRRCI, mrrc, MRRC): New.
24736 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
24737 VUNSPEC_MRRC2): New.
24739 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24741 * config/arm/arm.md (<mcr>): New.
24743 * config/arm/arm.c (arm_coproc_builtin_available): Add
24744 support for mcr, mrc, mcr2 and mrc2.
24745 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
24746 (arm_mcr_qualifiers): ... this. New.
24747 (MRC_QUALIFIERS): Define to ...
24748 (arm_mrc_qualifiers): ... this. New.
24749 (MCR_QUALIFIERS): Define to ...
24750 (arm_mcr_qualifiers): ... this. New.
24751 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24753 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24754 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24755 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24756 VUNSPEC_MRC2): New.
24758 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24760 * config/arm/arm.md (*ldc): New.
24764 * config/arm/arm.c (arm_coproc_builtin_available): Add
24765 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24766 (arm_coproc_ldc_stc_legitimate_address): New.
24767 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24768 'qualifier_const_pointer'.
24769 (LDC_QUALIFIERS): Define to...
24770 (arm_ldc_qualifiers): ... this. New.
24771 (STC_QUALIFIERS): Define to...
24772 (arm_stc_qualifiers): ... this. New.
24773 * config/arm/arm-protos.h
24774 (arm_coproc_ldc_stc_legitimate_address): New.
24775 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24776 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24777 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24778 stc2, stcl, stc2l): New.
24779 * config/arm/constraints.md (Uz): New.
24780 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24781 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24782 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24783 VUNSPEC_STC2L): New.
24785 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24787 * config/arm/arm.md (<cdp>): New.
24788 * config/arm/arm.c (neon_const_bounds): Rename this ...
24789 (arm_const_bounds): ... this.
24790 (arm_coproc_builtin_available): New.
24791 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24792 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24793 (CDP_QUALIFIERS): Define to...
24794 (arm_cdp_qualifiers): ... this. New.
24796 (arm_expand_builtin_args): Add case for 6 arguments.
24797 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24798 (arm_const_bounds): ... this.
24799 (arm_coproc_builtin_available): New.
24800 * config/arm/arm_acle.h (__arm_cdp): New.
24802 * config/arm/arm_acle_builtins.def (cdp): New.
24804 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24805 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24806 'arm_const_bounds'.
24807 * config/arm/types.md (coproc): New.
24808 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24809 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24810 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24811 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24813 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24815 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24816 (UBINOP_QUALIFIERS): New.
24818 (acle_builtin_data): New. Change comment.
24819 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24820 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24821 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24822 arm_acle_builtins.def.
24823 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24824 (arm_init_acle_builtins): New.
24825 (CRC32_BUILTIN): Remove.
24826 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24827 crc32cb, crc32ch and crc32cw.
24828 (arm_init_crc32_builtins): Remove.
24829 (arm_init_builtins): Use arm_init_acle_builtins rather
24830 than arm_init_crc32_builtins.
24831 (arm_expand_acle_builtin): New.
24832 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24833 * config/arm/arm_acle_builtins.def: New.
24835 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24837 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24838 (arm_builtin_datum): ... this.
24839 (arm_init_neon_builtin): Rename to ...
24840 (arm_init_builtin): ... this. Add a new parameters PREFIX
24841 and USE_SIG_IN_NAME.
24842 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24843 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24844 'arm_builtin_datum'.
24845 (arm_init_vfp_builtins): Likewise.
24846 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24847 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24848 (arm_expand_neon_args): Rename to ...
24849 (arm_expand_builtin_args): ... this. Rename builtin_arg
24850 enum values and differentiate between ARG_BUILTIN_MEMORY
24851 and ARG_BUILTIN_NEON_MEMORY.
24852 (arm_expand_neon_builtin_1): Rename to ...
24853 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24854 values, arm_expand_builtin_args and add bool parameter NEON.
24855 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24856 (arm_expand_vfp_builtin): Likewise.
24857 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24859 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24861 PR middle-end/77484
24862 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24863 * predict.c (tree_estimate_probability_bb): Reverse direction of
24864 polymorphic call predictor.
24866 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24868 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24869 (determine_pass_name_match): ...this new function and...
24870 (should_skip_pass_p): ...this new function.
24872 2017-01-06 Nathan Sidwell <nathan@acm.org>
24874 * ipa-visibility.c (function_and_variable_visibility): Reformat
24875 comments and long lines. Remove extrneous if.
24876 * symtab.c (symtab_node::make_decl_local): Fix code format.
24877 (symtab_node::set_section_for_node): Fix comment typo.
24879 2017-01-06 Martin Liska <mliska@suse.cz>
24882 * lra-constraints.c: Rename invariant to lra_invariant.
24883 * predict.c (set_even_probabilities): Initialize e to NULL.
24885 2017-01-05 Martin Sebor <msebor@redhat.com>
24887 PR tree-optimization/78910
24888 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24889 (format_integer): Correct off-by-one error in the handling
24890 of precision with negative numbers in signed conversions..
24892 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24894 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24896 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24898 PR tree-optimization/71016
24899 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24900 factor_out_conditional_conversion. Formatting fix.
24901 (factor_out_conditional_conversion): Add cond_stmt argument.
24902 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24903 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24906 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24908 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24909 read-rtl-function.o, and selftest-rtl.o.
24910 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24911 (selftest::aarch64_test_loading_full_dump): New function.
24912 (selftest::aarch64_run_selftests): New function.
24913 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24914 selftest::aarch64_run_selftests.
24915 * config/i386/i386.c
24916 (selftest::ix86_test_loading_dump_fragment_1): New function.
24917 (selftest::ix86_test_loading_call_insn): New function.
24918 (selftest::ix86_test_loading_full_dump): New function.
24919 (selftest::ix86_test_loading_unspec): New function.
24920 (selftest::ix86_run_selftests): Call the new functions.
24921 * emit-rtl.c (maybe_set_max_label_num): New function.
24922 * emit-rtl.h (maybe_set_max_label_num): New decl.
24923 * function.c (instantiate_decls): Guard call to
24924 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
24925 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
24927 * gensupport.c (gen_reader::gen_reader): Pass "false"
24928 for new "compact" param of rtx_reader.
24929 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
24930 rather than an empty string for NULL strings.
24931 * read-md.c: Potentially include config.h rather than bconfig.h.
24932 Wrap include of errors.h with #ifdef GENERATOR_FILE.
24933 (have_error): New global, copied from errors.c.
24934 (md_reader::read_name): Rename to...
24935 (md_reader::read_name_1): ...this, adding "out_loc" param,
24936 and converting "missing name or number" to returning false, rather
24938 (md_reader::read_name): Reimplement in terms of read_name_1.
24939 (md_reader::read_name_or_nil): New function.
24940 (md_reader::read_string): Handle "(nil)" by returning NULL.
24941 (md_reader::md_reader): Add new param "compact".
24942 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
24943 (md_reader::read_file): New method.
24944 * read-md.h (md_reader::md_reader): Add new param "compact".
24945 (md_reader::read_file): New method.
24946 (md_reader::is_compact): New accessor.
24947 (md_reader::read_name): Convert return type from void to file_location.
24948 (md_reader::read_name_or_nil): New decl.
24949 (md_reader::read_name_1): New decl.
24950 (md_reader::m_compact): New field.
24951 (noop_reader::noop_reader): Pass "false" for new "compact" param
24953 (rtx_reader::rtx_reader): Add new "compact" param.
24954 (rtx_reader::read_rtx_operand): Make virtual and convert return
24955 type from void to rtx.
24956 (rtx_reader::read_until): New decl.
24957 (rtx_reader::handle_any_trailing_information): New virtual function.
24958 (rtx_reader::postprocess): New virtual function.
24959 (rtx_reader::finalize_string): New virtual function.
24960 (rtx_reader::m_in_call_function_usage): New field.
24961 (rtx_reader::m_reuse_rtx_by_id): New field.
24962 * read-rtl-function.c: New file.
24963 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
24964 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
24965 (selftest::verify_three_block_rtl_cfg): New decl.
24966 * read-rtl-function.h: New file.
24967 * read-rtl.c: Potentially include config.h rather than bconfig.h.
24968 For host, include function.h, memmodel.h, and emit-rtl.h.
24969 (one_time_initialization): New function.
24970 (struct compact_insn_name): New struct.
24971 (compact_insn_names): New array.
24972 (find_code): Handle insn codes in compact dumps.
24973 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
24974 (bind_subst_iter_and_attr): Likewise.
24975 (add_condition_to_string): Likewise.
24976 (add_condition_to_rtx): Likewise.
24977 (apply_attribute_uses): Likewise.
24978 (add_current_iterators): Likewise.
24979 (apply_iterators): Likewise.
24980 (initialize_iterators): Guard usage of apply_subst_iterator with
24981 #ifdef GENERATOR_FILE.
24982 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
24983 (md_reader::read_mapping): Likewise.
24984 (add_define_attr_for_define_subst): Likewise.
24985 (add_define_subst_attr): Likewise.
24986 (read_subst_mapping): Likewise.
24987 (check_code_iterator): Likewise.
24988 (rtx_reader::read_rtx): Likewise. Move one-time initialization
24990 (one_time_initialization): New function.
24991 (rtx_reader::read_until): New method.
24992 (read_flags): New function.
24993 (parse_reg_note_name): New function.
24994 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
24995 Handle reuse_rtx ids.
24996 Wrap iterator lookup within #ifdef GENERATOR_FILE.
24997 Add parsing support for RTL dumps, mirroring the special-cases in
24998 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
24999 values, and calling handle_any_trailing_information.
25000 (rtx_reader::read_rtx_operand): Convert return type from void
25001 to rtx, returning return_rtx. Handle case 'e'. Call
25002 finalize_string on XSTR and XTMPL fields.
25003 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25004 "(nil)" values were omitted. Call the postprocess vfunc on the
25006 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25007 class ctor. Initialize m_in_call_function_usage. Call
25008 one_time_initialization.
25009 * rtl-tests.c (selftest::test_uncond_jump): Call
25010 set_new_first_and_last_insn.
25011 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25012 * selftest-rtl.c: New file.
25013 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25014 (selftest::get_insn_by_uid): New decl.
25015 * selftest-run-tests.c (selftest::run_tests): Call
25016 read_rtl_function_c_tests.
25017 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25018 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25021 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25023 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25024 operands in a special way. Assert that pos+len <= mode precision.
25026 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25028 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25029 3 argument Alias with unlimited for the negative form.
25030 (fno-vect-cost-model): Removed.
25032 2017-01-05 Martin Liska <mliska@suse.cz>
25034 * hsa-gen.c (gen_hsa_divmod): New function.
25035 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25037 2017-01-05 Martin Liska <mliska@suse.cz>
25040 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25043 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25045 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25046 small constant length operands.
25048 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25050 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25051 between loop iterations.
25053 2017-01-05 Martin Liska <mliska@suse.cz>
25056 * gimplify.c (gimplify_decl_expr): Compare to
25057 asan_poisoned_variables instread of checking flags.
25058 (gimplify_target_expr): Likewise.
25059 (gimplify_expr): Likewise.
25060 (gimplify_function_tree): Conditionally initialize
25061 asan_poisoned_variables.
25063 2017-01-04 Jeff Law <law@redhat.com>
25065 PR tree-optimizatin/78812
25066 * rtl.h (contains_mem_rtx_p): Prototype.
25067 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25068 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25069 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25070 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25071 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25073 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25075 * input.c (assert_char_at_range): Default-initialize actual_range.
25077 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25079 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25080 to match the caller.
25082 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25084 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25085 insns after final jump in test to emit dummy move.
25087 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25089 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25090 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25092 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25094 * multiple_target.c (create_dispatcher_calls): Init e_next.
25095 * tree-ssa-loop-split.c (split_loop): Init border.
25096 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25099 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25104 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25105 (altivec_register_operand): Do not return true if the operand
25106 contains a SUBREG mixing SImode and SFmode.
25107 (vsx_register_operand): Likewise.
25108 (vsx_reg_sfsubreg_ok): New predicate.
25109 (vfloat_operand): Do not return true if the operand contains a
25110 SUBREG mixing SImode and SFmode.
25111 (vint_operand): Likewise.
25112 (vlogical_operand): Likewise.
25113 (gpc_reg_operand): Likewise.
25114 (int_reg_operand): Likewise.
25115 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25116 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25117 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25119 (rs6000_emit_move_si_sf_subreg): New helper function.
25120 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25121 fixup SUBREGs involving SImode and SFmode.
25122 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25123 numbers for the new peephole2 optimization.
25124 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25125 the GLIBC math library that do AND/IOR/XOR operations on single
25126 precision floating point.
25127 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25128 target macros to say whether we need to avoid SUBREGs mixing
25130 (TARGET_ALLOW_SF_SUBREG): Likewise.
25131 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25132 (UNSPEC_SI_FROM_SF): Likewise.
25133 (iorxor): Change spacing.
25134 (and_ior_xor): New iterator for AND, IOR, and XOR.
25135 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25136 (movdi_from_sf_zero_ext): Likewise.
25137 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25138 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25139 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25140 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25141 (fms<mode>4): Likewise.
25142 (fnma<mode>4): Likewise.
25143 (fnms<mode>4): Likewise.
25144 (nfma<mode>4): Likewise.
25145 (nfms<mode>4): Likewise.
25147 2017-01-04 Marek Polacek <polacek@redhat.com>
25150 * doc/invoke.texi: Document -Wpointer-compare.
25152 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25154 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25157 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25158 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25161 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25164 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25165 documentation of the powerpc_popcntb_ok attribute.
25166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25167 code to issue warning messages if a requested CPU configuration is
25168 not supported by the binary (assembler and loader) toolchain.
25169 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25170 made to define a built-in function that has been disabled.
25171 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25172 made to define a built-in function that has been disabled.
25173 (altivec_init_builtins): Add comment explaining why definition
25174 of the DST built-in functions is not preceded by an assertion
25175 check. Add assertions to prevent ICE if attempts are made to
25176 define an altivec predicate or an abs* built-in function that has
25178 (htm_init_builtins): Add comment explaining why definition of the
25179 htm built-in functions is not preceded by an assertion check.
25181 2017-01-04 Jeff Law <law@redhat.com>
25183 PR tree-optimizatin/67955
25184 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25185 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25186 the points-to solution does not include pt_null. Use DECL_PT_UID
25189 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25191 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25192 Use gen_int_mode instead of gen_lopwart for const_int operands.
25194 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25196 PR tree-optimization/71563
25197 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25198 out of range value - has low bits known to be zero.
25200 2017-01-04 Alan Modra <amodra@gmail.com>
25202 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25203 * configure: Regenerate.
25204 * config.in: Regenerate.
25206 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25209 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25210 a substring of the message, but strcmp with the whole message. Ifdef
25211 ENABLE_NLS, translate the message first using dgettext.
25213 2017-01-03 Jeff Law <law@redhat.com>
25215 PR tree-optimizatin/78856
25216 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25217 (mark_threaded_blocks): Remove code to truncate thread paths that
25218 cross multiple loop headers. Instead invalidate the cached loop
25219 iteration information and handle case of a thread path walking
25220 into an irreducible region.
25222 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25225 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25226 assertions. Add support for doing the signbit if the IEEE 128-bit
25227 floating point value is in a GPR.
25228 * config/rs6000/rs6000.md (Fsignbit): Delete.
25229 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25230 Update the length attribute if the value is in a GPR.
25231 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25232 the sign or zero extension instruction, since the value is always 0/1.
25233 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25236 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25237 extracting SImode to a GPR register so that we can generate a
25238 store, limit the vector to be in a traditional Altivec register
25239 for the vextuwrx instruction.
25241 2017-01-03 Ian Lance Taylor <iant@google.com>
25243 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25245 2017-01-03 Martin Sebor <msebor@redhat.com>
25247 PR tree-optimization/78696
25248 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25249 precision. Use MPFR for %f for greater fidelity. Correct handling
25251 (pass_sprintf_length::compute_format_length): Set width and precision
25252 specified by asrerisk to void_node for vararg functions.
25253 (try_substitute_return_value): Adjust dump output.
25255 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25257 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25259 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25261 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25262 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25263 -mlra/-mno-lra was passed to the compiler.
25265 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25267 PR rtl-optimization/65618
25268 * emit-rtl.c (try_split): Move initialization of "before" and
25269 "after" to just before the call to emit_insn_after_setloc.
25271 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25273 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25275 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25277 * dwarf2out.c (gen_enumeration_type_die): When
25278 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25280 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25282 PR tree-optimization/78965
25283 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25284 Change first argument from const call_info & to call_info &. For %n
25285 set info.nowrite to false.
25287 PR middle-end/78901
25288 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25289 possibly throwing calls.
25291 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25292 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25293 and fns handling, rather than in a separate case SSA_NAME.
25295 2017-01-02 Jeff Law <law@redhat.com>
25297 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25298 fixes for first_period and second_period variables.
25300 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25303 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25304 (*insvqi_1): New insn pattern.
25305 (*insvqi_1_mem_rex64): Ditto.
25306 (*insvqi_2): Ditto.
25307 (*insvqi_3): Rename from *insvqi.
25309 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25311 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25313 * doc/cfg.texi (Edges): Remove reference to Java.
25314 (Maintaining the CFG): Ditto.
25316 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25318 PR middle-end/77674
25319 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25320 transparent aliases.
25322 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25324 PR middle-end/77484
25325 * predict.def (PRED_CALL): Update hitrate.
25326 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25327 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25328 into direct/indirect/polymorphic variants.
25330 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25332 Update copyright years.
25334 * gcc.c (process_command): Update copyright notice dates.
25335 * gcov-dump.c (print_version): Ditto.
25336 * gcov.c (print_version): Ditto.
25337 * gcov-tool.c (print_version): Ditto.
25338 * gengtype.c (create_file): Ditto.
25339 * doc/cpp.texi: Bump @copying's copyright year.
25340 * doc/cppinternals.texi: Ditto.
25341 * doc/gcc.texi: Ditto.
25342 * doc/gccint.texi: Ditto.
25343 * doc/gcov.texi: Ditto.
25344 * doc/install.texi: Ditto.
25345 * doc/invoke.texi: Ditto.
25347 Copyright (C) 2017 Free Software Foundation, Inc.
25349 Copying and distribution of this file, with or without modification,
25350 are permitted in any medium without royalty provided the copyright
25351 notice and this notice are preserved.