1 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6 when the two are known to be equal.
8 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9 Alan Hayward <alan.hayward@arm.com>
10 David Sherwood <david.sherwood@arm.com>
12 * simplify-rtx.c (simplify_const_unary_operation): Use
13 is_a <scalar_int_mode> instead of checking for a nonzero
14 precision. Forcibly convert op_mode to a scalar_int_mode
15 in that case. More clearly differentiate the operand and
16 result modes and use the former when deciding what the value
17 of a count-bits operation should be. Use is_int_mode instead
18 of checking for a MODE_INT. Remove redundant check for whether
19 this mode has a zero precision.
21 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
22 Alan Hayward <alan.hayward@arm.com>
23 David Sherwood <david.sherwood@arm.com>
25 * optabs.c (widen_leading): Change the type of the mode argument
26 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
27 (widen_bswap): Likewise.
28 (expand_parity): Likewise.
29 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
30 (expand_ffs): Likewise.
31 (epand_unop): Check for scalar integer modes before calling the
34 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
35 Alan Hayward <alan.hayward@arm.com>
36 David Sherwood <david.sherwood@arm.com>
38 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
40 (expand_expr_real_1): Update call accordingly.
42 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
43 Alan Hayward <alan.hayward@arm.com>
44 David Sherwood <david.sherwood@arm.com>
46 * expmed.c (store_bit_field_using_insv): Add op0_mode and
47 value_mode arguments. Use scalar_int_mode internally.
48 (store_bit_field_1): Rename the new integer mode from imode
49 to op0_mode and use it instead of GET_MODE (op0). Update calls
50 to store_split_bit_field, store_bit_field_using_insv and
51 store_fixed_bit_field.
52 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
53 Use scalar_int_mode internally. Use a bit count rather than a mode
54 when calculating the largest bit size for get_best_mode.
55 Update calls to store_split_bit_field and store_fixed_bit_field_1.
56 (store_fixed_bit_field_1): Add mode and value_mode arguments.
57 Remove assertion that OP0 has a scalar integer mode.
58 (store_split_bit_field): Add op0_mode and value_mode arguments.
59 Update calls to extract_fixed_bit_field.
60 (extract_bit_field_using_extv): Add an op0_mode argument.
61 Use scalar_int_mode internally.
62 (extract_bit_field_1): Rename the new integer mode from imode to
63 op0_mode and use it instead of GET_MODE (op0). Update calls to
64 extract_split_bit_field, extract_bit_field_using_extv and
65 extract_fixed_bit_field.
66 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
67 to extract_split_bit_field and extract_fixed_bit_field_1.
68 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
69 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
71 (extract_split_bit_field): Add an op0_mode argument. Update call
72 to extract_fixed_bit_field.
74 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
75 Alan Hayward <alan.hayward@arm.com>
76 David Sherwood <david.sherwood@arm.com>
78 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
79 * explow.c (hard_function_value): Likewise.
80 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
81 convert_to_mode call outside the loop.
82 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
83 for the mode iterator. Require the mode specified by max_pieces
85 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
87 (copy_blkmode_to_reg): Likewise.
88 (set_storage_via_setmem): Likewise.
89 * optabs.c (prepare_cmp_insn): Likewise.
90 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
91 * stor-layout.c (finish_bitfield_representative): Likewise.
93 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
94 Alan Hayward <alan.hayward@arm.com>
95 David Sherwood <david.sherwood@arm.com>
97 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
98 * expr.c (convert_move): Use them.
99 (convert_modes): Likewise.
100 (store_expr_with_bounds): Likewise.
102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
103 Alan Hayward <alan.hayward@arm.com>
104 David Sherwood <david.sherwood@arm.com>
106 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
107 parameter for the mode of "x". Remove the "known_x", "known_mode"
108 and "known_ret" arguments. Change the type of the mode argument
110 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
111 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
112 (reg_num_sign_bit_copies_for_combine): Likewise.
113 * rtlanal.c (nonzero_bits1): Likewise.
114 (num_sign_bit_copies1): Likewise.
115 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
116 (reg_num_sign_bit_copies_general): Likewise.
117 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
118 (reg_nonzero_bits_general): Likewise.
120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
121 Alan Hayward <alan.hayward@arm.com>
122 David Sherwood <david.sherwood@arm.com>
124 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
125 than in subroutines. Return 1 for non-integer modes.
126 (cached_num_sign_bit_copies): Change the type of the mode parameter
128 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
129 classes. Handle CONST_INT_P first and then check whether X also
130 has a scalar integer mode. Check the same thing for inner registers
131 of a SUBREG and for values that are being extended or truncated.
133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
134 Alan Hayward <alan.hayward@arm.com>
135 David Sherwood <david.sherwood@arm.com>
137 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
138 in subroutines. Return the mode mask for non-integer modes.
139 (cached_nonzero_bits): Change the type of the mode parameter
141 (nonzero_bits1): Likewise. Remove early exit for other mode
142 classes. Handle CONST_INT_P first and then check whether X
143 also has a scalar integer mode.
145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
146 Alan Hayward <alan.hayward@arm.com>
147 David Sherwood <david.sherwood@arm.com>
149 * combine.c (extract_left_shift): Add a mode argument and update
151 (make_compound_operation_int): Change the type of the mode parameter
152 to scalar_int_mode and update the call to extract_left_shift.
154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
155 Alan Hayward <alan.hayward@arm.com>
156 David Sherwood <david.sherwood@arm.com>
158 * combine.c (simplify_and_const_int): Change the type of the mode
159 parameter to scalar_int_mode.
160 (simplify_and_const_int_1): Likewise. Update recursive call.
162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * combine.c (simplify_compare_const): Check that the mode is a
167 scalar_int_mode (rather than VOIDmode) before testing its
169 (simplify_comparison): Move COMPARISON_P handling out of the
170 loop and restrict the latter part of the loop to scalar_int_modes.
171 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
172 and when considering SUBREG_REGs. Use is_int_mode instead of
173 checking GET_MODE_CLASS against MODE_INT.
175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
176 Alan Hayward <alan.hayward@arm.com>
177 David Sherwood <david.sherwood@arm.com>
179 * combine.c (try_widen_shift_mode): Move check for equal modes to...
180 (simplify_shift_const_1): ...here. Use scalar_int_mode for
181 shift_unit_mode and for modes involved in scalar shifts.
183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
184 Alan Hayward <alan.hayward@arm.com>
185 David Sherwood <david.sherwood@arm.com>
187 * combine.c (force_int_to_mode): New function, split out from...
188 (force_to_mode): ...here. Keep xmode up-to-date and use it
189 instead of GET_MODE (x).
191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
192 Alan Hayward <alan.hayward@arm.com>
193 David Sherwood <david.sherwood@arm.com>
195 * optabs-query.h (extraction_insn::struct_mode): Change type to
196 opt_scalar_int_mode and update comment.
197 (extraction_insn::field_mode): Change type to scalar_int_mode.
198 (extraction_insn::pos_mode): Likewise.
199 * combine.c (make_extraction): Update accordingly.
200 * optabs-query.c (get_traditional_extraction_insn): Likewise.
201 (get_optab_extraction_insn): Likewise.
202 * recog.c (simplify_while_replacing): Likewise.
203 * expmed.c (narrow_bit_field_mem): Change the type of the mode
204 parameter to opt_scalar_int_mode.
206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
207 Alan Hayward <alan.hayward@arm.com>
208 David Sherwood <david.sherwood@arm.com>
210 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
211 to a scalar_int_mode instead of a machine_mode.
212 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
213 (get_best_mode): Return a boolean and use a pointer argument to store
214 the selected mode. Replace the limit mode parameter with a bit limit.
215 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
216 for the values returned by bit_field_mode_iterator::next_mode.
217 (store_bit_field): Update call to get_best_mode.
218 (store_fixed_bit_field): Likewise.
219 (extract_fixed_bit_field): Likewise.
220 * expr.c (optimize_bitfield_assignment_op): Likewise.
221 * fold-const.c (optimize_bit_field_compare): Likewise.
222 (fold_truth_andor_1): Likewise.
223 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
224 Update for new type of m_mode.
225 (get_best_mode): As above.
227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
228 Alan Hayward <alan.hayward@arm.com>
229 David Sherwood <david.sherwood@arm.com>
231 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
232 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
233 (store_bit_field): Check is_a <scalar_int_mode> before calling
234 strict_volatile_bitfield_p.
235 (extract_bit_field): Likewise.
237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
238 Alan Hayward <alan.hayward@arm.com>
239 David Sherwood <david.sherwood@arm.com>
241 * target.def (cstore_mode): Return a scalar_int_mode.
242 * doc/tm.texi: Regenerate.
243 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
244 * targhooks.h (default_cstore_mode): Likewise.
245 * targhooks.c (default_cstore_mode): Likewise, using a forced
247 * expmed.c (emit_cstore): Expect the target of the cstore to be
250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
251 Alan Hayward <alan.hayward@arm.com>
252 David Sherwood <david.sherwood@arm.com>
254 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
255 * builtins.c (expand_builtin_signbit): Use it.
256 * cfgexpand.c (expand_debug_expr): Likewise.
257 * dojump.c (do_jump): Likewise.
258 (do_compare_and_jump): Likewise.
259 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
260 * expmed.c (make_tree): Likewise.
261 * expr.c (expand_expr_real_2): Likewise.
262 (expand_expr_real_1): Likewise.
263 (try_casesi): Likewise.
264 * fold-const-call.c (fold_const_call_ss): Likewise.
265 * fold-const.c (unextend): Likewise.
266 (extract_muldiv_1): Likewise.
267 (fold_single_bit_test): Likewise.
268 (native_encode_int): Likewise.
269 (native_encode_string): Likewise.
270 (native_interpret_int): Likewise.
271 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
272 * internal-fn.c (expand_addsub_overflow): Likewise.
273 (expand_neg_overflow): Likewise.
274 (expand_mul_overflow): Likewise.
275 (expand_arith_overflow): Likewise.
276 * match.pd: Likewise.
277 * stor-layout.c (layout_type): Likewise.
278 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
279 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
280 * tree-ssanames.c (get_range_info): Likewise.
281 * tree-switch-conversion.c (array_value_type) Likewise.
282 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
283 (vect_recog_divmod_pattern): Likewise.
284 (vect_recog_mixed_size_cond_pattern): Likewise.
285 * tree-vrp.c (extract_range_basic): Likewise.
286 (simplify_float_conversion_using_ranges): Likewise.
287 * tree.c (int_fits_type_p): Likewise.
288 * ubsan.c (instrument_bool_enum_load): Likewise.
289 * varasm.c (mergeable_string_section): Likewise.
290 (narrowing_initializer_constant_valid_p): Likewise.
291 (output_constant): Likewise.
293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
294 Alan Hayward <alan.hayward@arm.com>
295 David Sherwood <david.sherwood@arm.com>
297 * machmode.h (NARROWEST_INT_MODE): New macro.
298 * expr.c (alignment_for_piecewise_move): Use it instead of
299 GET_CLASS_NARROWEST_MODE (MODE_INT).
300 (push_block): Likewise.
301 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
303 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
306 Alan Hayward <alan.hayward@arm.com>
307 David Sherwood <david.sherwood@arm.com>
309 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
311 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
312 * optabs.c (expand_unop): Likewise.
314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
315 Alan Hayward <alan.hayward@arm.com>
316 David Sherwood <david.sherwood@arm.com>
318 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
319 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
320 (popcount_loc_descriptor): Likewise.
321 (bswap_loc_descriptor): Likewise.
322 (rotate_loc_descriptor): Likewise.
323 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
324 calling the functions above.
326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
327 Alan Hayward <alan.hayward@arm.com>
328 David Sherwood <david.sherwood@arm.com>
330 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
332 (try_combine): Likewise.
333 (simplify_if_then_else): Likewise.
334 * cse.c (cse_insn): Likewise.
335 * dwarf2out.c (mem_loc_descriptor): Likewise.
336 * emit-rtl.c (gen_lowpart_common): Likewise.
337 * simplify-rtx.c (simplify_truncation): Likewise.
338 (simplify_binary_operation_1): Likewise.
339 (simplify_const_relational_operation): Likewise.
340 (simplify_ternary_operation): Likewise.
341 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
344 Alan Hayward <alan.hayward@arm.com>
345 David Sherwood <david.sherwood@arm.com>
347 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
348 * reload.c (push_reload): Likewise.
349 (find_reloads): Likewise.
351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
352 Alan Hayward <alan.hayward@arm.com>
353 David Sherwood <david.sherwood@arm.com>
355 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
356 (make_compound_operation_int): Likewise.
357 (change_zero_ext): Likewise.
358 * expr.c (convert_move): Likewise.
359 (convert_modes): Likewise.
360 * fwprop.c (forward_propagate_subreg): Likewise.
361 * loop-iv.c (get_biv_step_1): Likewise.
362 * optabs.c (widen_operand): Likewise.
363 * postreload.c (move2add_valid_value_p): Likewise.
364 * recog.c (simplify_while_replacing): Likewise.
365 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
366 (simplify_binary_operation_1): Likewise. Remove redundant
369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
370 Alan Hayward <alan.hayward@arm.com>
371 David Sherwood <david.sherwood@arm.com>
373 * combine.c (combine_simplify_rtx): Add checks for
374 is_a <scalar_int_mode>.
375 (simplify_if_then_else): Likewise.
376 (make_field_assignment): Likewise.
377 (simplify_comparison): Likewise.
378 * ifcvt.c (noce_try_bitop): Likewise.
379 * loop-invariant.c (canonicalize_address_mult): Likewise.
380 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
383 Alan Hayward <alan.hayward@arm.com>
384 David Sherwood <david.sherwood@arm.com>
386 * simplify-rtx.c (simplify_binary_operation_1): Use
387 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
390 Alan Hayward <alan.hayward@arm.com>
391 David Sherwood <david.sherwood@arm.com>
393 * wide-int.h (int_traits<unsigned char>) New class.
394 (int_traits<unsigned short>) Likewise.
395 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
396 Use GET_MODE_UNIT_PRECISION and remove redundant test for
398 * combine.c (set_nonzero_bits_and_sign_copies): Use
399 is_a <scalar_int_mode>.
400 (find_split_point): Likewise.
401 (combine_simplify_rtx): Likewise.
402 (simplify_logical): Likewise.
403 (expand_compound_operation): Likewise.
404 (expand_field_assignment): Likewise.
405 (make_compound_operation): Likewise.
406 (extended_count): Likewise.
407 (change_zero_ext): Likewise.
408 (simplify_comparison): Likewise.
409 * dwarf2out.c (scompare_loc_descriptor): Likewise.
410 (ucompare_loc_descriptor): Likewise.
411 (minmax_loc_descriptor): Likewise.
412 (mem_loc_descriptor): Likewise.
413 (loc_descriptor): Likewise.
414 * expmed.c (init_expmed_one_mode): Likewise.
415 * lra-constraints.c (lra_constraint_offset): Likewise.
416 * optabs.c (prepare_libcall_arg): Likewise.
417 * postreload.c (move2add_note_store): Likewise.
418 * reload.c (operands_match_p): Likewise.
419 * rtl.h (load_extend_op): Likewise.
420 * rtlhooks.c (gen_lowpart_general): Likewise.
421 * simplify-rtx.c (simplify_truncation): Likewise.
422 (simplify_unary_operation_1): Likewise.
423 (simplify_binary_operation_1): Likewise.
424 (simplify_const_binary_operation): Likewise.
425 (simplify_const_relational_operation): Likewise.
426 (simplify_subreg): Likewise.
427 * stor-layout.c (bitwise_mode_for_mode): Likewise.
428 * var-tracking.c (adjust_mems): Likewise.
429 (prepare_call_arguments): Likewise.
431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
432 Alan Hayward <alan.hayward@arm.com>
433 David Sherwood <david.sherwood@arm.com>
435 * machmode.h (is_int_mode): New fuction.
436 * combine.c (find_split_point): Use it.
437 (combine_simplify_rtx): Likewise.
438 (simplify_if_then_else): Likewise.
439 (simplify_set): Likewise.
440 (simplify_shift_const_1): Likewise.
441 (simplify_comparison): Likewise.
442 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
443 * cse.c (notreg_cost): Likewise.
444 (cse_insn): Likewise.
445 * cselib.c (cselib_lookup_1): Likewise.
446 * dojump.c (do_jump_1): Likewise.
447 (do_compare_rtx_and_jump): Likewise.
448 * dse.c (get_call_args): Likewise.
449 * dwarf2out.c (rtl_for_decl_init): Likewise.
450 (native_encode_initializer): Likewise.
451 * expmed.c (emit_store_flag_1): Likewise.
452 (emit_store_flag): Likewise.
453 * expr.c (convert_modes): Likewise.
454 (store_field): Likewise.
455 (expand_expr_real_1): Likewise.
456 * fold-const.c (fold_read_from_constant_string): Likewise.
457 * gimple-ssa-sprintf.c (get_format_string): Likewise.
458 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
459 * optabs.c (expand_binop): Likewise.
460 (expand_unop): Likewise.
461 (expand_abs_nojump): Likewise.
462 (expand_one_cmpl_abs_nojump): Likewise.
463 * simplify-rtx.c (mode_signbit_p): Likewise.
464 (val_signbit_p): Likewise.
465 (val_signbit_known_set_p): Likewise.
466 (val_signbit_known_clear_p): Likewise.
467 (simplify_relational_operation_1): Likewise.
468 * tree.c (vector_type_mode): Likewise.
470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
471 Alan Hayward <alan.hayward@arm.com>
472 David Sherwood <david.sherwood@arm.com>
474 * machmode.h (smallest_mode_for_size): Fix formatting.
475 (smallest_int_mode_for_size): New function.
476 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
477 instead of smallest_mode_for_size.
478 * combine.c (make_extraction): Likewise.
479 * config/arc/arc.c (arc_expand_movmem): Likewise.
480 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
481 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
482 * config/s390/s390.c (s390_expand_insv): Likewise.
483 * config/sparc/sparc.c (assign_int_registers): Likewise.
484 * config/spu/spu.c (spu_function_value): Likewise.
485 (spu_function_arg): Likewise.
486 * coverage.c (get_gcov_type): Likewise.
487 (get_gcov_unsigned_t): Likewise.
488 * dse.c (find_shift_sequence): Likewise.
489 * expmed.c (store_bit_field_1): Likewise.
490 * expr.c (convert_move): Likewise.
491 (store_field): Likewise.
492 * internal-fn.c (expand_arith_overflow): Likewise.
493 * optabs-query.c (get_best_extraction_insn): Likewise.
494 * optabs.c (expand_twoval_binop_libfunc): Likewise.
495 * stor-layout.c (layout_type): Likewise.
496 (initialize_sizetypes): Likewise.
497 * targhooks.c (default_get_mask_mode): Likewise.
498 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
501 Alan Hayward <alan.hayward@arm.com>
502 David Sherwood <david.sherwood@arm.com>
504 * machmode.h (opt_mode::else_blk): New function.
505 (int_mode_for_mode): Declare.
506 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
507 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
509 * cfgexpand.c (expand_debug_expr): Likewise.
510 * combine.c (gen_lowpart_or_truncate): Likewise.
511 (gen_lowpart_for_combine): Likewise.
512 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
513 * config/avr/avr.c (avr_to_int_mode): Likewise.
514 (avr_out_plus_1): Likewise.
515 (avr_out_plus): Likewise.
516 (avr_out_round): Likewise.
517 * config/i386/i386.c (ix86_split_to_parts): Likewise.
518 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
519 (s390_expand_vcond): Likewise.
520 * config/spu/spu.c (spu_split_immediate): Likewise.
521 (spu_expand_mov): Likewise.
522 * dse.c (get_stored_val): Likewise.
523 * expmed.c (store_bit_field_1): Likewise.
524 (convert_extracted_bit_field): Use int_mode_for_mode instead of
526 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
527 (extract_low_bits): Likewise.
528 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
529 handling rather than repeating the check.
530 (emit_group_store): Likewise.
531 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
532 * optabs.c (expand_absneg_bit): Likewise.
533 (expand_copysign_absneg): Likewise.
534 (expand_copysign_bit): Likewise.
535 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
536 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
537 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
538 * var-tracking.c (prepare_call_arguments): Likewise.
539 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
540 int_mode_for_mode instead of mode_for_size.
541 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
544 Alan Hayward <alan.hayward@arm.com>
545 David Sherwood <david.sherwood@arm.com>
547 * machmode.h (int_mode_for_size): New function.
548 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
549 instead of mode_for_size.
550 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
552 * combine.c (expand_field_assignment): Use int_mode_for_size
553 instead of mode_for_size.
554 (make_extraction): Likewise.
555 (simplify_shift_const_1): Likewise.
556 (simplify_comparison): Likewise.
557 * dojump.c (do_jump): Likewise.
558 * dwarf2out.c (mem_loc_descriptor): Likewise.
559 * emit-rtl.c (init_derived_machine_modes): Likewise.
560 * expmed.c (flip_storage_order): Likewise.
561 (convert_extracted_bit_field): Likewise.
562 * expr.c (copy_blkmode_from_reg): Likewise.
563 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
564 * internal-fn.c (expand_mul_overflow): Likewise.
565 * lower-subreg.c (simple_move): Likewise.
566 * optabs-libfuncs.c (init_optabs): Likewise.
567 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
568 * tree.c (vector_type_mode): Likewise.
569 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
570 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
571 * tree-vect-generic.c (expand_vector_parallel): Likewise.
572 * tree-vect-stmts.c (vectorizable_load): Likewise.
573 (vectorizable_store): Likewise.
575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
576 Alan Hayward <alan.hayward@arm.com>
577 David Sherwood <david.sherwood@arm.com>
579 * coretypes.h (pod_mode): New type.
580 (scalar_int_mode_pod): New typedef.
581 * machmode.h (pod_mode): New class.
582 (int_n_data_t::m): Change type to scalar_int_mode_pod.
583 * genmodes.c (emit_mode_int_n): Update accordingly.
584 * lower-subreg.h (target_lower_subreg): Change type to
586 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
590 Alan Hayward <alan.hayward@arm.com>
591 David Sherwood <david.sherwood@arm.com>
593 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
594 machine_mode to scalar_int_mode.
595 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
596 (rs6000_option_override_internal): Remove cast to int.
597 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
598 machine_mode to scalar_int_mode.
599 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
600 (rs6000_option_override_internal): Remove cast to int.
601 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
602 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
604 * config/s390/s390.c (s390_expand_builtin): Likewise.
605 * coretypes.h (scalar_int_mode): New type.
606 (opt_scalar_int_mode): New typedef.
607 * machmode.h (scalar_int_mode): New class.
608 (scalar_int_mode::includes_p): New function.
609 (byte_mode): Change type to scalar_int_mode.
610 (word_mode): Likewise.
611 (ptr_mode): Likewise.
612 * emit-rtl.c (byte_mode): Likewise.
613 (word_mode): Likewise.
614 (ptr_mode): Likewise.
615 (init_derived_machine_modes): Update accordingly.
616 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
617 and MODE_PARTIAL_INT.
618 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
622 Alan Hayward <alan.hayward@arm.com>
623 David Sherwood <david.sherwood@arm.com>
625 * target.def (libgcc_floating_mode_supported_p): Take a
627 * doc/tm.texi: Regenerate.
628 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
630 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
631 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
635 Alan Hayward <alan.hayward@arm.com>
636 David Sherwood <david.sherwood@arm.com>
638 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
639 * doc/tm.texi: Regenerate.
640 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
641 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
642 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
643 * targhooks.h (default_floatn_mode): Likewise.
644 * targhooks.c (default_floatn_mode): Likewise.
645 * tree.c (build_common_tree_nodes): Update accordingly.
647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
648 Alan Hayward <alan.hayward@arm.com>
649 David Sherwood <david.sherwood@arm.com>
651 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
652 (mode_iterator::iterate_p): Likewise.
653 (mode_iterator::get_wider): Likewise.
654 * expr.c (init_expr_target): Use opt_scalar_float_mode.
656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
657 Alan Hayward <alan.hayward@arm.com>
658 David Sherwood <david.sherwood@arm.com>
660 * coretypes.h (opt_scalar_float_mode): New typedef.
661 * machmode.h (float_mode_for_size): New function.
662 * emit-rtl.c (double_mode): Delete.
663 (init_emit_once): Use float_mode_for_size.
664 * stor-layout.c (layout_type): Likewise.
665 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
668 Alan Hayward <alan.hayward@arm.com>
669 David Sherwood <david.sherwood@arm.com>
671 * output.h (assemble_real): Take a scalar_float_mode.
672 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
673 * config/arm/arm.md (consttable_4): Likewise.
674 (consttable_8): Likewise.
675 (consttable_16): Likewise.
676 * config/mips/mips.md (consttable_float): Likewise.
677 * config/s390/s390.c (s390_output_pool_entry): Likewise.
678 * varasm.c (assemble_real): Take a scalar_float_mode.
679 (output_constant_pool_2): Update accordingly.
680 (output_constant): Likewise.
682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
683 Alan Hayward <alan.hayward@arm.com>
684 David Sherwood <david.sherwood@arm.com>
686 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
687 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
688 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
689 (native_encode_real): Likewise.
690 (native_interpret_real): Likewise.
691 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
692 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
695 Alan Hayward <alan.hayward@arm.com>
696 David Sherwood <david.sherwood@arm.com>
698 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
699 <scalar_float_mode>. Simplify.
700 (gen_extend_conv_libfunc): Likewise.
702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
703 Alan Hayward <alan.hayward@arm.com>
704 David Sherwood <david.sherwood@arm.com>
706 * coretypes.h (scalar_float_mode): New type.
707 * machmode.h (mode_traits::from_int): Use machine_mode if
708 USE_ENUM_MODES is defined.
709 (is_a): New function.
711 (dyn_cast): Likewise.
712 (scalar_float_mode): New class.
713 (scalar_float_mode::includes_p): New function.
714 (is_float_mode): Likewise.
715 * gdbhooks.py (MachineModePrinter): New class.
716 (build_pretty_printer): Use it for scalar_float_mode.
717 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
718 (format_helper::format_helper): Turn into a template.
719 * genmodes.c (get_mode_class): New function.
720 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
721 or machine_mode if none.
722 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
723 as_a <scalar_float_mode>.
724 * dwarf2out.c (mem_loc_descriptor): Likewise.
725 (insert_float): Likewise.
726 (add_const_value_attribute): Likewise.
727 * simplify-rtx.c (simplify_immed_subreg): Likewise.
728 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
729 (expand_unop): Update accordingly.
730 (expand_abs_nojump): Likewise.
731 (expand_copysign_absneg): Take a scalar_float_mode.
732 (expand_copysign_bit): Likewise.
733 (expand_copysign): Update accordingly.
735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
736 Alan Hayward <alan.hayward@arm.com>
737 David Sherwood <david.sherwood@arm.com>
739 * coretypes.h (opt_mode): New class.
740 * machmode.h (opt_mode): Likewise.
741 (opt_mode::else_void): New function.
742 (opt_mode::require): Likewise.
743 (opt_mode::exists): Likewise.
744 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
745 (GET_MODE_2XWIDER_MODE): Likewise.
746 (mode_iterator::get_wider): Update accordingly.
747 (mode_iterator::get_2xwider): Likewise.
748 (mode_iterator::get_known_wider): Likewise, turning into a template.
749 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
750 forcing a wider mode to exist.
751 * config/cr16/cr16.h (LONG_REG_P): Likewise.
752 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
753 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
754 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
755 * lower-subreg.c (init_lower_subreg): Likewise.
756 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
757 on the final iteration.
758 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
759 a wider mode exists before asking for a move pattern.
760 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
761 forcing a wider mode to exist.
762 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
763 returning false if no such mode exists.
764 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
765 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
766 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
767 Avoid checking for a MODE_INT if we already know the mode is not a
769 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
770 forcing a wider mode to exist.
771 (expmed_mult_highpart_optab): Likewise.
772 (expmed_mult_highpart): Likewise.
773 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
775 * lto-streamer-in.c (lto_input_mode_table): Likewise.
776 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
777 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
778 * internal-fn.c (expand_mul_overflow): Update use of
779 GET_MODE_2XWIDER_MODE.
780 * omp-low.c (omp_clause_aligned_alignment): Likewise.
781 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
783 (convert_plusminus_to_widen): Likewise.
784 * tree-switch-conversion.c (array_value_type): Likewise.
785 * var-tracking.c (emit_note_insn_var_location): Likewise.
786 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
787 Return false inside rather than outside the loop if no wider mode
789 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
790 and GET_MODE_2XWIDER_MODE
791 (can_compare_p): Use else_void.
792 * gdbhooks.py (OptMachineModePrinter): New class.
793 (build_pretty_printer): Use it for opt_mode.
795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
796 Alan Hayward <alan.hayward@arm.com>
797 David Sherwood <david.sherwood@arm.com>
799 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
800 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
803 Alan Hayward <alan.hayward@arm.com>
804 David Sherwood <david.sherwood@arm.com>
806 * machmode.h (mode_traits): New structure.
807 (get_narrowest_mode): New function.
808 (mode_iterator::start): Likewise.
809 (mode_iterator::iterate_p): Likewise.
810 (mode_iterator::get_wider): Likewise.
811 (mode_iterator::get_known_wider): Likewise.
812 (mode_iterator::get_2xwider): Likewise.
813 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
814 (FOR_EACH_MODE): Likewise.
815 (FOR_EACH_MODE_FROM): Likewise.
816 (FOR_EACH_MODE_UNTIL): Likewise.
817 (FOR_EACH_WIDER_MODE): Likewise.
818 (FOR_EACH_2XWIDER_MODE): Likewise.
819 * builtins.c (expand_builtin_strlen): Use new mode iterators.
820 * combine.c (simplify_comparison): Likewise
821 * config/i386/i386.c (type_natural_mode): Likewise.
822 * cse.c (cse_insn): Likewise.
823 * dse.c (find_shift_sequence): Likewise.
824 * emit-rtl.c (init_derived_machine_modes): Likewise.
825 (init_emit_once): Likewise.
826 * explow.c (hard_function_value): Likewise.
827 * expmed.c (extract_fixed_bit_field_1): Likewise.
828 (extract_bit_field_1): Likewise.
829 (expand_divmod): Likewise.
830 (emit_store_flag_1): Likewise.
831 * expr.c (init_expr_target): Likewise.
832 (convert_move): Likewise.
833 (alignment_for_piecewise_move): Likewise.
834 (widest_int_mode_for_size): Likewise.
835 (emit_block_move_via_movmem): Likewise.
836 (copy_blkmode_to_reg): Likewise.
837 (set_storage_via_setmem): Likewise.
838 (compress_float_constant): Likewise.
839 * omp-low.c (omp_clause_aligned_alignment): Likewise.
840 * optabs-query.c (get_best_extraction_insn): Likewise.
841 * optabs.c (expand_binop): Likewise.
842 (expand_twoval_unop): Likewise.
843 (expand_twoval_binop): Likewise.
844 (widen_leading): Likewise.
845 (widen_bswap): Likewise.
846 (expand_parity): Likewise.
847 (expand_unop): Likewise.
848 (prepare_cmp_insn): Likewise.
849 (prepare_float_lib_cmp): Likewise.
850 (expand_float): Likewise.
851 (expand_fix): Likewise.
852 (expand_sfix_optab): Likewise.
853 * postreload.c (move2add_use_add2_insn): Likewise.
854 * reg-stack.c (reg_to_stack): Likewise.
855 * reginfo.c (choose_hard_reg_mode): Likewise.
856 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
857 * stor-layout.c (mode_for_size): Likewise.
858 (smallest_mode_for_size): Likewise.
859 (mode_for_vector): Likewise.
860 (finish_bitfield_representative): Likewise.
861 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
862 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
863 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
864 * var-tracking.c (prepare_call_arguments): Likewise.
866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
867 Alan Hayward <alan.hayward@arm.com>
868 David Sherwood <david.sherwood@arm.com>
870 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
871 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
872 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
873 * machmode.h (mode_size): Move earlier in file.
874 (mode_precision): Likewise.
875 (mode_inner): Likewise.
876 (mode_nunits): Likewise.
877 (mode_unit_size): Likewise.
878 (unit_unit_precision): Likewise.
879 (mode_wider): Likewise.
880 (mode_2xwider): Likewise.
881 (machine_mode): New class.
882 (mode_to_bytes): New function.
883 (mode_to_bits): Likewise.
884 (mode_to_precision): Likewise.
885 (mode_to_inner): Likewise.
886 (mode_to_unit_size): Likewise.
887 (mode_to_unit_precision): Likewise.
888 (mode_to_nunits): Likewise.
889 (GET_MODE_SIZE): Use mode_to_bytes.
890 (GET_MODE_BITSIZE): Use mode_to_bits.
891 (GET_MODE_PRECISION): Use mode_to_precision.
892 (GET_MODE_INNER): Use mode_to_inner.
893 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
894 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
895 (GET_MODE_NUNITS): Use mode_to_nunits.
896 * system.h (ALWAYS_INLINE): New macro.
897 * config/powerpcspe/powerpcspe-c.c
898 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
899 int for arg1_mode and arg2_mode.
901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
902 Alan Hayward <alan.hayward@arm.com>
903 David Sherwood <david.sherwood@arm.com>
905 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
906 Prefix mode names with E_ in case statements.
907 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
908 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
909 (aarch64_split_simd_move): Likewise.
910 (aarch64_gen_storewb_pair): Likewise.
911 (aarch64_gen_loadwb_pair): Likewise.
912 (aarch64_gen_store_pair): Likewise.
913 (aarch64_gen_load_pair): Likewise.
914 (aarch64_get_condition_code_1): Likewise.
915 (aarch64_constant_pool_reload_icode): Likewise.
916 (get_rsqrte_type): Likewise.
917 (get_rsqrts_type): Likewise.
918 (get_recpe_type): Likewise.
919 (get_recps_type): Likewise.
920 (aarch64_gimplify_va_arg_expr): Likewise.
921 (aarch64_simd_container_mode): Likewise.
922 (aarch64_emit_load_exclusive): Likewise.
923 (aarch64_emit_store_exclusive): Likewise.
924 (aarch64_expand_compare_and_swap): Likewise.
925 (aarch64_gen_atomic_cas): Likewise.
926 (aarch64_emit_bic): Likewise.
927 (aarch64_emit_atomic_swap): Likewise.
928 (aarch64_emit_atomic_load_op): Likewise.
929 (aarch64_evpc_trn): Likewise.
930 (aarch64_evpc_uzp): Likewise.
931 (aarch64_evpc_zip): Likewise.
932 (aarch64_evpc_ext): Likewise.
933 (aarch64_evpc_rev): Likewise.
934 (aarch64_evpc_dup): Likewise.
935 (aarch64_gen_ccmp_first): Likewise.
936 (aarch64_gen_ccmp_next): Likewise.
937 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
938 (alpha_emit_xfloating_libcall): Likewise.
939 (emit_insxl): Likewise.
940 (alpha_arg_type): Likewise.
941 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
942 (arc_preferred_simd_mode): Likewise.
943 (arc_secondary_reload): Likewise.
944 (get_arc_condition_code): Likewise.
945 (arc_print_operand): Likewise.
946 (arc_legitimate_constant_p): Likewise.
947 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
948 * config/arc/arc.md (casesi_load): Likewise.
949 (casesi_compact_jump): Likewise.
950 * config/arc/predicates.md (proper_comparison_operator): Likewise.
951 (cc_use_register): Likewise.
952 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
953 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
954 (arm_init_iwmmxt_builtins): Likewise.
955 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
956 (neon_expand_vector_init): Likewise.
957 (arm_attr_length_move_neon): Likewise.
958 (maybe_get_arm_condition_code): Likewise.
959 (arm_emit_vector_const): Likewise.
960 (arm_preferred_simd_mode): Likewise.
961 (arm_output_iwmmxt_tinsr): Likewise.
962 (thumb1_output_casesi): Likewise.
963 (thumb2_output_casesi): Likewise.
964 (arm_emit_load_exclusive): Likewise.
965 (arm_emit_store_exclusive): Likewise.
966 (arm_expand_compare_and_swap): Likewise.
967 (arm_evpc_neon_vuzp): Likewise.
968 (arm_evpc_neon_vzip): Likewise.
969 (arm_evpc_neon_vrev): Likewise.
970 (arm_evpc_neon_vtrn): Likewise.
971 (arm_evpc_neon_vext): Likewise.
972 (arm_validize_comparison): Likewise.
973 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
974 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
975 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
976 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
977 (c6x_preferred_simd_mode): Likewise.
978 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
979 (epiphany_rtx_costs): Likewise.
980 * config/epiphany/predicates.md (proper_comparison_operator):
982 * config/frv/frv.c (condexec_memory_operand): Likewise.
983 (frv_emit_move): Likewise.
984 (output_move_single): Likewise.
985 (output_condmove_single): Likewise.
986 (frv_hard_regno_mode_ok): Likewise.
987 (frv_matching_accg_mode): Likewise.
988 * config/h8300/h8300.c (split_adds_subs): Likewise.
989 (h8300_rtx_costs): Likewise.
990 (h8300_print_operand): Likewise.
991 (compute_mov_length): Likewise.
992 (output_logical_op): Likewise.
993 (compute_logical_op_length): Likewise.
994 (compute_logical_op_cc): Likewise.
995 (h8300_shift_needs_scratch_p): Likewise.
996 (output_a_shift): Likewise.
997 (compute_a_shift_length): Likewise.
998 (compute_a_shift_cc): Likewise.
999 (expand_a_rotate): Likewise.
1000 (output_a_rotate): Likewise.
1001 * config/i386/i386.c (classify_argument): Likewise.
1002 (function_arg_advance_32): Likewise.
1003 (function_arg_32): Likewise.
1004 (function_arg_64): Likewise.
1005 (function_value_64): Likewise.
1006 (ix86_gimplify_va_arg): Likewise.
1007 (ix86_legitimate_constant_p): Likewise.
1008 (put_condition_code): Likewise.
1009 (split_double_mode): Likewise.
1010 (ix86_avx256_split_vector_move_misalign): Likewise.
1011 (ix86_expand_vector_logical_operator): Likewise.
1012 (ix86_split_idivmod): Likewise.
1013 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1014 (ix86_build_const_vector): Likewise.
1015 (ix86_build_signbit_mask): Likewise.
1016 (ix86_match_ccmode): Likewise.
1017 (ix86_cc_modes_compatible): Likewise.
1018 (ix86_expand_branch): Likewise.
1019 (ix86_expand_sse_cmp): Likewise.
1020 (ix86_expand_sse_movcc): Likewise.
1021 (ix86_expand_int_sse_cmp): Likewise.
1022 (ix86_expand_vec_perm_vpermi2): Likewise.
1023 (ix86_expand_vec_perm): Likewise.
1024 (ix86_expand_sse_unpack): Likewise.
1025 (ix86_expand_int_addcc): Likewise.
1026 (ix86_split_to_parts): Likewise.
1027 (ix86_vectorize_builtin_gather): Likewise.
1028 (ix86_vectorize_builtin_scatter): Likewise.
1029 (avx_vpermilp_parallel): Likewise.
1030 (inline_memory_move_cost): Likewise.
1031 (ix86_tieable_integer_mode_p): Likewise.
1032 (x86_maybe_negate_const_int): Likewise.
1033 (ix86_expand_vector_init_duplicate): Likewise.
1034 (ix86_expand_vector_init_one_nonzero): Likewise.
1035 (ix86_expand_vector_init_one_var): Likewise.
1036 (ix86_expand_vector_init_concat): Likewise.
1037 (ix86_expand_vector_init_interleave): Likewise.
1038 (ix86_expand_vector_init_general): Likewise.
1039 (ix86_expand_vector_set): Likewise.
1040 (ix86_expand_vector_extract): Likewise.
1041 (emit_reduc_half): Likewise.
1042 (ix86_emit_i387_round): Likewise.
1043 (ix86_mangle_type): Likewise.
1044 (ix86_expand_round_sse4): Likewise.
1045 (expand_vec_perm_blend): Likewise.
1046 (canonicalize_vector_int_perm): Likewise.
1047 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1048 (expand_vec_perm_1): Likewise.
1049 (expand_vec_perm_interleave3): Likewise.
1050 (expand_vec_perm_even_odd_pack): Likewise.
1051 (expand_vec_perm_even_odd_1): Likewise.
1052 (expand_vec_perm_broadcast_1): Likewise.
1053 (ix86_vectorize_vec_perm_const_ok): Likewise.
1054 (ix86_expand_vecop_qihi): Likewise.
1055 (ix86_expand_mul_widen_hilo): Likewise.
1056 (ix86_expand_sse2_abs): Likewise.
1057 (ix86_expand_pextr): Likewise.
1058 (ix86_expand_pinsr): Likewise.
1059 (ix86_preferred_simd_mode): Likewise.
1060 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1061 * config/i386/sse.md (*andnot<mode>3): Likewise.
1062 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1063 (*<code><mode>3): Likewise.
1064 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1065 (ia64_expand_atomic_op): Likewise.
1066 (ia64_arg_type): Likewise.
1067 (ia64_mode_to_int): Likewise.
1068 (ia64_scalar_mode_supported_p): Likewise.
1069 (ia64_vector_mode_supported_p): Likewise.
1070 (expand_vec_perm_broadcast): Likewise.
1071 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1072 (iq2000_function_arg_advance): Likewise.
1073 (iq2000_function_arg): Likewise.
1074 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1075 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1076 (m68k_libcall_value): Likewise.
1077 (m68k_function_value): Likewise.
1078 (sched_attr_op_type): Likewise.
1079 * config/mcore/mcore.c (mcore_output_move): Likewise.
1080 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1082 (microblaze_function_arg): Likewise.
1083 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1084 (mips_print_operand): Likewise.
1085 (mips_mode_ok_for_mov_fmt_p): Likewise.
1086 (mips_vector_mode_supported_p): Likewise.
1087 (mips_preferred_simd_mode): Likewise.
1088 (mips_expand_vpc_loongson_even_odd): Likewise.
1089 (mips_expand_vec_unpack): Likewise.
1090 (mips_expand_vi_broadcast): Likewise.
1091 (mips_expand_vector_init): Likewise.
1092 (mips_expand_vec_reduc): Likewise.
1093 (mips_expand_msa_cmp): Likewise.
1094 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1095 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1096 (cc_flags_for_mode): Likewise.
1097 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1098 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1099 (nds32_output_casesi_pc_relative): Likewise.
1100 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1101 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1102 (nvptx_gen_unpack): Likewise.
1103 (nvptx_gen_pack): Likewise.
1104 (nvptx_gen_shuffle): Likewise.
1105 (nvptx_gen_wcast): Likewise.
1106 (nvptx_preferred_simd_mode): Likewise.
1107 * config/pa/pa.c (pa_secondary_reload): Likewise.
1108 * config/pa/predicates.md (base14_operand): Likewise.
1109 * config/powerpcspe/powerpcspe-c.c
1110 (altivec_resolve_overloaded_builtin): Likewise.
1111 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1113 (rs6000_preferred_simd_mode): Likewise.
1114 (output_vec_const_move): Likewise.
1115 (rs6000_expand_vector_extract): Likewise.
1116 (rs6000_split_vec_extract_var): Likewise.
1117 (reg_offset_addressing_ok_p): Likewise.
1118 (rs6000_legitimate_offset_address_p): Likewise.
1119 (rs6000_legitimize_address): Likewise.
1120 (rs6000_emit_set_const): Likewise.
1121 (rs6000_const_vec): Likewise.
1122 (rs6000_emit_move): Likewise.
1123 (spe_build_register_parallel): Likewise.
1124 (rs6000_darwin64_record_arg_recurse): Likewise.
1125 (swap_selector_for_mode): Likewise.
1126 (spe_init_builtins): Likewise.
1127 (paired_init_builtins): Likewise.
1128 (altivec_init_builtins): Likewise.
1129 (do_load_for_compare): Likewise.
1130 (rs6000_generate_compare): Likewise.
1131 (rs6000_expand_float128_convert): Likewise.
1132 (emit_load_locked): Likewise.
1133 (emit_store_conditional): Likewise.
1134 (rs6000_output_function_epilogue): Likewise.
1135 (rs6000_handle_altivec_attribute): Likewise.
1136 (rs6000_function_value): Likewise.
1137 (emit_fusion_gpr_load): Likewise.
1138 (emit_fusion_p9_load): Likewise.
1139 (emit_fusion_p9_store): Likewise.
1140 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1141 (fusion_gpr_mem_load): Likewise.
1142 (fusion_addis_mem_combo_load): Likewise.
1143 (fusion_addis_mem_combo_store): Likewise.
1144 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1145 (fusion_gpr_mem_load): Likewise.
1146 (fusion_addis_mem_combo_load): Likewise.
1147 (fusion_addis_mem_combo_store): Likewise.
1148 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1150 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1151 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1152 (rs6000_preferred_simd_mode): Likewise.
1153 (output_vec_const_move): Likewise.
1154 (rs6000_expand_vector_extract): Likewise.
1155 (rs6000_split_vec_extract_var): Likewise.
1156 (reg_offset_addressing_ok_p): Likewise.
1157 (rs6000_legitimate_offset_address_p): Likewise.
1158 (rs6000_legitimize_address): Likewise.
1159 (rs6000_emit_set_const): Likewise.
1160 (rs6000_const_vec): Likewise.
1161 (rs6000_emit_move): Likewise.
1162 (rs6000_darwin64_record_arg_recurse): Likewise.
1163 (swap_selector_for_mode): Likewise.
1164 (paired_init_builtins): Likewise.
1165 (altivec_init_builtins): Likewise.
1166 (rs6000_expand_float128_convert): Likewise.
1167 (emit_load_locked): Likewise.
1168 (emit_store_conditional): Likewise.
1169 (rs6000_output_function_epilogue): Likewise.
1170 (rs6000_handle_altivec_attribute): Likewise.
1171 (rs6000_function_value): Likewise.
1172 (emit_fusion_gpr_load): Likewise.
1173 (emit_fusion_p9_load): Likewise.
1174 (emit_fusion_p9_store): Likewise.
1175 * config/rx/rx.c (rx_gen_move_template): Likewise.
1176 (flags_from_mode): Likewise.
1177 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1178 (s390_slb_comparison): Likewise.
1179 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1180 (s390_vector_mode_supported_p): Likewise.
1181 (s390_cc_modes_compatible): Likewise.
1182 (s390_match_ccmode_set): Likewise.
1183 (s390_canonicalize_comparison): Likewise.
1184 (s390_emit_compare_and_swap): Likewise.
1185 (s390_branch_condition_mask): Likewise.
1186 (s390_rtx_costs): Likewise.
1187 (s390_secondary_reload): Likewise.
1188 (__SECONDARY_RELOAD_CASE): Likewise.
1189 (s390_expand_cs): Likewise.
1190 (s390_preferred_simd_mode): Likewise.
1191 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1192 * config/sh/sh.c (sh_print_operand): Likewise.
1193 (dump_table): Likewise.
1194 (sh_secondary_reload): Likewise.
1195 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1196 * config/sh/sh.md (casesi_worker_1): Likewise.
1197 (casesi_worker_2): Likewise.
1198 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1199 (fcc_comparison_operator): Likewise.
1200 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1201 (emit_soft_tfmode_cvt): Likewise.
1202 (sparc_preferred_simd_mode): Likewise.
1203 (output_cbranch): Likewise.
1204 (sparc_print_operand): Likewise.
1205 (sparc_expand_vec_perm_bmask): Likewise.
1206 (vector_init_bshuffle): Likewise.
1207 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1208 (spu_vector_mode_supported_p): Likewise.
1209 (spu_expand_insv): Likewise.
1210 (spu_emit_branch_or_set): Likewise.
1211 (spu_handle_vector_attribute): Likewise.
1212 (spu_builtin_splats): Likewise.
1213 (spu_builtin_extract): Likewise.
1214 (spu_builtin_promote): Likewise.
1215 (spu_expand_sign_extend): Likewise.
1216 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1217 (tilegx_simd_int): Likewise.
1218 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1219 (tilepro_simd_int): Likewise.
1220 * config/v850/v850.c (const_double_split): Likewise.
1221 (v850_print_operand): Likewise.
1222 (ep_memory_offset): Likewise.
1223 * config/vax/vax.c (vax_rtx_costs): Likewise.
1224 (vax_output_int_move): Likewise.
1225 (vax_output_int_add): Likewise.
1226 (vax_output_int_subtract): Likewise.
1227 * config/visium/predicates.md (visium_branch_operator): Likewise.
1228 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1229 (visium_print_operand_address): Likewise.
1230 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1231 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1232 (xtensa_expand_conditional_branch): Likewise.
1233 (xtensa_copy_incoming_a7): Likewise.
1234 (xtensa_output_literal): Likewise.
1235 * dfp.c (decimal_real_maxval): Likewise.
1236 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1243 (mode_nunits_inline): Likewise.
1244 (mode_inner_inline): Likewise.
1245 (mode_unit_size_inline): Likewise.
1246 (mode_unit_precision_inline): Likewise.
1247 (emit_insn_modes_h): Likewise. Also emit a #define of the
1249 (emit_mode_wider): Add an E_ prefix to mode names.
1250 (emit_mode_complex): Likewise.
1251 (emit_mode_inner): Likewise.
1252 (emit_mode_adjustments): Likewise.
1253 (emit_mode_int_n): Likewise.
1254 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1255 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1256 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1257 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1258 (CRC32_BUILTIN, ENTRY): Likewise.
1259 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1260 (aarch64_pop_regs): Likewise.
1261 (aarch64_process_components): Likewise.
1262 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1263 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1264 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1265 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1266 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1267 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1269 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1270 (pcmpistr): Likewise.
1271 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1272 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1273 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1275 * config/rl78/rl78.c (mduc_regs): Likewise.
1276 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1277 (htm_expand_builtin): Likewise.
1278 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1279 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1280 * config/xtensa/xtensa.c (print_operand): Likewise.
1281 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1282 (NUM_MODE_VECTOR_INT): Likewise.
1283 * genoutput.c (null_operand): Likewise.
1284 (output_operand_data): Likewise.
1285 * genrecog.c (print_parameter_value): Likewise.
1286 * lra.c (debug_operand_data): Likewise.
1288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1290 * dwarf2out.c (scompare_loc_descriptor_wide)
1291 (scompare_loc_descriptor_narrow): New functions, split out from...
1292 (scompare_loc_descriptor): ...here.
1293 * expmed.c (emit_store_flag_int): New function, split out from...
1294 (emit_store_flag): ...here.
1296 2017-08-30 Richard Biener <rguenther@suse.de>
1298 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1299 (dwarf2out_early_finish): Move setting of AT_pubnames from
1300 early debug output to early finish.
1302 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1304 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1305 and -mdata-region to the assembler.
1307 2017-08-30 Richard Biener <rguenther@suse.de>
1309 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1311 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1312 (dwarf2out_early_global_decl): Only generate a DIE for the
1313 abstract origin if it doesn't already exist or is a declaration DIE.
1314 (resolve_addr): Do not add the linkage name twice when
1315 generating a stub DIE for the DW_TAG_GNU_call_site target.
1317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1319 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1320 Use machine_mode rather than int for arg1_mode.
1322 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1325 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1326 that the second argument of the built-in functions to unpack
1327 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1328 switch statement instead a lot of if statements.
1329 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1330 Allow 64-bit values to be in Altivec registers as well as
1331 traditional floating point registers.
1332 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1334 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1336 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1337 MAX_REGS_PER_ADDRESS == 1.
1339 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1341 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1342 * config/i386/i386.c (ix86_option_override_internal): Simplify
1343 setting of opts->x_flag_entry.
1345 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1346 Jakub Jelinek <jakub@redhat.com>
1347 Richard Biener <rguenther@suse.de>
1349 PR tree-optimization/81503
1350 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1351 folded constant fits in the target type; reorder tests for clarity.
1353 2017-08-29 Martin Liska <mliska@suse.cz>
1355 * passes.def: Include pass_lower_switch.
1356 * stmt.c (dump_case_nodes): Remove and move to
1357 tree-switch-conversion.
1358 (case_values_threshold): Likewise.
1359 (expand_switch_as_decision_tree_p): Likewise.
1360 (emit_case_decision_tree): Likewise.
1361 (expand_case): Likewise.
1362 (balance_case_nodes): Likewise.
1363 (node_has_low_bound): Likewise.
1364 (node_has_high_bound): Likewise.
1365 (node_is_bounded): Likewise.
1366 (emit_case_nodes): Likewise.
1367 (struct simple_case_node): New struct.
1368 (add_case_node): Remove.
1369 (emit_case_dispatch_table): Use vector instead of case_list.
1370 (reset_out_edges_aux): Remove.
1371 (compute_cases_per_edge): Likewise.
1372 (expand_case): Build list of simple_case_node.
1373 (expand_sjlj_dispatch_table): Use it.
1374 * tree-switch-conversion.c (struct case_node): Moved from
1375 stmt.c and adjusted.
1376 (emit_case_nodes): Likewise.
1377 (node_has_low_bound): Likewise.
1378 (node_has_high_bound): Likewise.
1379 (node_is_bounded): Likewise.
1380 (case_values_threshold): Likewise.
1381 (reset_out_edges_aux): Likewise.
1382 (compute_cases_per_edge): Likewise.
1383 (add_case_node): Likewise.
1384 (dump_case_nodes): Likewise.
1385 (balance_case_nodes): Likewise.
1386 (expand_switch_as_decision_tree_p): Likewise.
1387 (emit_jump): Likewise.
1388 (emit_case_decision_tree): Likewise.
1389 (try_switch_expansion): Likewise.
1390 (do_jump_if_equal): Likewise.
1391 (emit_cmp_and_jump_insns): Likewise.
1392 (fix_phi_operands_for_edge): New function.
1393 (record_phi_operand_mapping): Likewise.
1394 (class pass_lower_switch): New pass.
1395 (pass_lower_switch::execute): New function.
1396 (make_pass_lower_switch): Likewise.
1397 (conditional_probability):
1398 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1399 * tree-pass.h: Add make_pass_lower_switch.
1401 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1404 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1407 2017-08-29 Richard Biener <rguenther@suse.de>
1409 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1410 we do not add a DW_AT_inline attribute twice.
1411 (gen_subprogram_die): Remove code setting DW_AT_inline on
1412 DECL_ABSTRACT_P nodes.
1414 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1416 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1417 calls to internal functions.
1418 (gimplify_modify_expr): Likewise.
1419 * tree-call-cdce.c (use_internal_fn): Likewise.
1420 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1421 (convert_to_divmod): Set the nothrow flag.
1422 * tree-if-conv.c (predicate_mem_writes): Likewise.
1423 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1424 (vectorizable_call): Likewise.
1425 (vectorizable_store): Likewise.
1426 (vectorizable_load): Likewise.
1427 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1428 (vect_recog_mask_conversion_pattern): Likewise.
1430 2017-08-29 Martin Liska <mliska@suse.cz>
1433 * gcc.c (driver_handle_option): Add new argument.
1434 * opts-common.c (handle_option): Pass
1435 target_option_override_hook.
1436 * opts-global.c (lang_handle_option): Add new option.
1437 (set_default_handlers): Add new argument.
1438 (decode_options): Likewise.
1439 * opts.c (target_handle_option): Likewise.
1440 (common_handle_option): Call target_option_override_hook.
1441 * opts.h (struct cl_option_handler_func): Add hook for
1442 target option override.
1443 (struct cl_option_handlers): Likewise.
1444 (set_default_handlers): Add new argument.
1445 (decode_options): Likewise.
1446 (common_handle_option): Likewise.
1447 (target_handle_option): Likewise.
1448 * toplev.c (toplev::main): Pass targetm.target_option.override
1451 2017-08-29 Richard Biener <rguenther@suse.de>
1452 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1454 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1455 life to the active subtree.
1457 2017-08-28 Jeff Law <law@redhat.com>
1459 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1460 derive_equivalences.
1461 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1463 (edge_info::derive_equivalences): New private member function
1465 * tree-ssa-dom.c (class edge_info): Changed from a struct
1466 to a class. Add ctor/dtor, methods and data members.
1467 (edge_info::edge_info): Renamed from allocate_edge_info.
1468 Initialize additional members.
1469 (edge_info::~edge_info): New.
1470 (free_dom_edge_info): Delete the edge info.
1471 (record_edge_info): Use new class & associated member functions.
1472 Tighten forms for testing for edge equivalences.
1473 (record_temporary_equivalences): Iterate over the simple
1474 equivalences rather than assuming there's only one per edge.
1475 (cprop_into_successor_phis): Iterate over the simple
1476 equivalences rather than assuming there's only one per edge.
1477 (optimize_stmt): Use operand_equal_p rather than pointer
1478 equality for mini-DSE code.
1480 2017-08-28 Nathan Sidwell <nathan@acm.org>
1482 * gcc.c (execute): Fold SIGPIPE handling into switch
1483 statement. Adjust internal error message.
1485 2017-08-28 Richard Biener <rguenther@suse.de>
1488 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1489 Do nothing for removed DIEs.
1491 2017-08-28 Richard Biener <rguenther@suse.de>
1493 PR tree-optimization/81977
1494 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1497 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1500 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1501 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1502 using targetm.gen_mem_thread_fence.
1504 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1507 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1508 predicate to register_operand. Reorder operands.
1509 (*btr<mode>): Ditto.
1510 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1511 (*btr<mode>_mask): Ditto.
1513 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1515 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1516 * config/rs6000/xmmintrin.h: New file.
1517 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1519 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1522 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1523 parameter and_insn and return it.
1524 (recombine_lvx_pattern): Insert a copy to ensure availability of
1525 the base register of the copied masking operation at the point of
1526 the instruction replacement.
1527 (recombine_stvx_pattern): Likewise.
1529 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1531 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1532 undocumented switches.
1533 (-mpower9-dform-vector): Likewise.
1534 (-mpower9-dform): Likewise.
1535 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1536 comments to delete references to -mpower9-dform* switches.
1537 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1538 Delete reference to -mpower9-dform* switches, test for
1539 -mpower9-vector instead.
1540 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1541 (OTHER_P9_VECTOR_MASKS): Likewise.
1542 (POWERPC_MASKS): Likewise.
1543 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1544 tests against -mpower9-dform* to -mpower9-vector. Delete code
1545 that checked for -mpower9-dform* consistancy with other options.
1546 Add test for -mpower9-misc to enable other power9 switches.
1547 (rs6000_init_hard_regno_mode_ok): Likewise.
1548 (rs6000_option_override_internal): Likewise.
1549 (rs6000_emit_prologue): Likewise.
1550 (rs6000_emit_epilogue): Likewise.
1551 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1552 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1553 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1555 (emit_fusion_p9_store): Likewise.
1556 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1557 resetting these macros if the assembler does not support ISA 3.0
1559 (TARGET_P9_DFORM_VECTOR): Likewise.
1560 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1561 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1563 2017-08-25 Alan Modra <amodra@gmail.com>
1565 PR rtl-optimization/81747
1566 * cse.c (cse_extended_basic_block): Don't attempt to record
1567 equivalences for degenerate conditional branches that branch
1568 to their fall-through.
1570 2017-08-24 Martin Sebor <msebor@redhat.com>
1573 * gimple-fold.c (size_must_be_zero_p): New function.
1574 (gimple_fold_builtin_memory_op): Call it.
1576 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1578 * config/rs6000/mm_malloc.h: New file.
1580 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1582 PR tree-optimization/81913
1583 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1584 analysis when either IVs in condition can wrap.
1586 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1588 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1589 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1591 2017-08-24 Richard Biener <rguenther@suse.de>
1594 * targhooks.c (default_target_can_inline_p): Properly
1595 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1596 is present and always compare.
1597 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1598 infer -mfpmath=sse from TARGET_SSE_P.
1599 (ix86_can_inline_p): Properly use target_option_default_node when
1600 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1602 2017-08-24 Richard Biener <rguenther@suse.de>
1605 * dwarf2out.c (output_die): Handle flag_generate_offload like
1607 (output_comp_unit): Likewise.
1608 (gen_array_type_die): Likewise.
1609 (dwarf2out_early_finish): Likewise.
1610 (note_variable_value_in_expr): Likewise.
1611 (dwarf2out_finish): Likewise. Adjust assert.
1612 * cgraphunit.c (symbol_table::compile): Move setting of
1613 flag_generate_offload earlier ...
1614 (symbol_table::finalize_compilation_unit): ... here, before
1615 early debug finalization.
1617 2017-08-24 Richard Biener <rguenther@suse.de>
1619 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1620 and ipa-fnsummary.h.
1621 (ix86_can_inline_p): When ix86_fpmath flags do not match
1622 check whether the callee uses FP math at all.
1624 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1627 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1628 instead of TYPE_PRECISION.
1630 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1631 Alan Hayward <alan.hayward@arm.com>
1632 David Sherwood <david.sherwood@arm.com>
1634 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1636 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1638 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1640 (ix86_frame::stack_realign_allocate): New field.
1641 (struct machine_frame_state): Modify comments.
1642 (machine_frame_state::sp_realigned_fp_end): New field.
1643 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1645 (sp_valid_at): Add assertion to assure no attempt to access invalid
1646 offset of a realigned stack.
1647 (fp_valid_at): Likewise.
1648 (choose_baseaddr): Modify comments.
1649 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1650 ix86_expand_prologue.
1651 (ix86_expand_prologue): Modify stack realignment and allocation.
1652 (ix86_expand_epilogue): Modify comments.
1653 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1654 avx2_runtime, avx512f, and avx512f_runtime.
1656 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1658 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1659 (mstackrealign): Do not init to -1.
1660 * config/i386/i386.c (ix86_option_override_internal):
1661 Check opts_set, not opts when setting default value of
1662 opts->x_ix86_force_align_arg_pointer.
1664 2017-08-23 Richard Biener <rguenther@suse.de>
1666 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1667 lang_hooks.decl_printable_name.
1668 * print-rtl-function.c (print_rtx_function): Likewise.
1669 * tree-pretty-print.c (dump_function_header): Likewise.
1671 2017-08-23 Richard Biener <rguenther@suse.de>
1674 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1675 -g0 at compile-time.
1677 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1680 * doc/sourcebuild.texi (Other hardware attributes):
1683 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1686 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1687 Fix single-use check.
1689 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1691 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1693 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1695 * doc/install.texi: Modify to add more details on running selected
1698 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1700 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1701 is combined with -mabi=ms.
1702 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1705 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1707 PR tree-optimization/81488
1708 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1709 and cached_basis fields.
1710 (MAX_SPREAD): New constant.
1711 (alloc_cand_and_find_basis): Initialize new fields.
1712 (clear_visited): New function.
1713 (create_phi_basis_1): Rename from create_phi_basis, set visited
1714 and cached_basis fields.
1715 (create_phi_basis): New wrapper function.
1716 (phi_add_costs_1): Rename from phi_add_costs, add spread
1717 parameter, set visited field, short-circuit when limits reached.
1718 (phi_add_costs): New wrapper function.
1719 (record_phi_increments_1): Rename from record_phi_increments, set
1721 (record_phi_increments): New wrapper function.
1722 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1723 (phi_incr_cost): New wrapper function.
1724 (all_phi_incrs_profitable_1): Rename from
1725 all_phi_incrs_profitable, set visited field.
1726 (all_phi_incrs_profitable): New wrapper function.
1728 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1729 Alan Hayward <alan.hayward@arm.com>
1730 David Sherwood <david.sherwood@arm.com>
1732 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1733 that takes the outer and inner modes.
1734 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1735 comparison as the canonical test for a paradoxical subreg.
1736 * combine.c (simplify_set): Use paradoxical_subreg_p.
1737 (make_extraction): Likewise.
1738 (force_to_mode): Likewise.
1739 (rtx_equal_for_field_assignment_p): Likewise.
1740 (gen_lowpart_for_combine): Likewise.
1741 (simplify_comparison): Likewise.
1742 * cse.c (equiv_constant): Likewise.
1743 * expmed.c (store_bit_field_1): Likewise.
1744 * final.c (alter_subreg): Likewise.
1745 * fwprop.c (propagate_rtx): Likewise.
1746 (forward_propagate_subreg): Likewise.
1747 * ira-conflicts.c (ira_build_conflicts): Likewise.
1748 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1749 * lra-constraints.c (curr_insn_transform): Likewise.
1750 (split_reg): Likewise.
1751 * lra-eliminations.c (move_plus_up): Likewise.
1752 (lra_eliminate_regs_1): Likewise.
1753 * recog.c (general_operand): Likewise.
1754 * ree.c (combine_reaching_defs): Likewise.
1755 * reload.c (push_reload): Likewise.
1756 (find_reloads): Likewise.
1757 * reload1.c (elimination_effects): Likewise.
1758 (compute_reload_subreg_offset): Likewise.
1759 (choose_reload_regs): Likewise.
1760 * rtlanal.c (subreg_lsb_1): Likewise.
1761 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1762 (simplify_subreg): Likewise.
1763 * var-tracking.c (track_loc_p): Likewise.
1764 * emit-rtl.c (byte_lowpart_offset): Likewise.
1765 (paradoxical_subreg_p): Delete out-of-line definition.
1767 2017-08-22 Jeff Law <law@redhat.com>
1769 PR tree-optimization/81741
1770 PR tree-optimization/71947
1771 * tree-ssa-dom.c: Include tree-inline.h.
1772 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1773 equivalences if one is more expensive to compute than the other.
1774 * tree-ssa-scopedtables.h (class const_or_copies): Make
1775 record_const_or_copy_raw method private.
1776 (class avail_exprs_stack): New method simplify_binary_operation.
1777 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1778 avail_exprs_stack::simplify_binary_operation as needed.
1779 (avail_exprs_stack::simplify_binary_operation): New function.
1781 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
1783 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
1784 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
1785 (DOT_SYMBOLS): Likewise.
1786 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
1787 (RELOCATABLE_NEEDS_FIXUP): Likewise.
1788 (RS6000_ABI_NAME): Likewise.
1789 (TARGET_CMODEL): Likewise.
1790 (TOC_SECTION_ASM_OP): Likewise.
1791 (SET_CMODEL): New macro.
1792 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
1794 2017-08-22 Richard Biener <rguenther@suse.de>
1796 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
1797 to allow for free-lang-data replacements similar to verify_type_variant.
1799 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * config/aarch64/aarch64.md (casesi): Use DImode rather than
1804 VOIDmode for the LABEL_REF.
1806 2017-08-22 Richard Biener <rguenther@suse.de>
1808 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
1810 2017-08-22 Richard Biener <rguenther@suse.de>
1812 * common.opt (feliminate-dwarf2-dups): Ignore.
1813 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
1814 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
1815 same_die_p_wrap, compute_section_prefix,
1816 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
1817 (comdat_symbol_id, comdat_symbol_number): Likewise.
1818 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
1820 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
1821 (output_die): Mark unreachable path unreachable.
1822 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
1823 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
1824 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
1825 (dwarf2out_early_finish): Likewise.
1827 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
1829 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
1831 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
1834 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
1835 not VAR_P. Filter attribute warnings with OPT_Wattributes.
1836 (avr_attribute_table) <io, io_low, address>: Initialize
1837 .decl_required with true.
1839 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
1841 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
1842 undocumented debugging options.
1843 (-mvsx-scalar-double): Likewise.
1844 (-mallow-df-permute): Likewise.
1845 (-mvectorize-builtins): Likewise.
1846 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1847 (rs6000_builtin_vectorized_function): Likewise.
1848 (rs6000_builtin_md_vectorized_function): Likewise.
1849 (rs6000_opt_vars): Likewise.
1851 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
1854 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
1855 (*btrq_imm): Rename from *btrq.
1856 (*btcq_imm): Rename from *btcq.
1857 (btsc): New code attribute.
1858 (*<btsc><mode>): New insn pattern.
1859 (*btr<mode>): Ditto.
1860 (*<btsc><mode>_mask): New insn_and_split pattern.
1861 (*btr<mode>_mask): Ditto.
1863 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1864 Alan Hayward <alan.hayward@arm.com>
1865 David Sherwood <david.sherwood@arm.com>
1867 * function.c (pad_below): Simplify padding calculation.
1869 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1870 Alan Hayward <alan.hayward@arm.com>
1871 David Sherwood <david.sherwood@arm.com>
1873 * target.def (function_prologue): Remove frame size argument.
1874 (function_epilogue): Likewise.
1875 * doc/tm.texi: Regenerate.
1876 * final.c (final_start_function): Update call to function_prologue.
1877 (final_end_function): Update call to function_epilogue.
1878 (default_function_pro_epilogue): Remove frame size argument.
1879 * output.h (default_function_pro_epilogue): Likewise.
1880 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
1881 (arm_output_function_prologue): Likewise.
1882 * config/frv/frv.c (frv_function_prologue): Likewise.
1883 (frv_function_epilogue): Likewise.
1884 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
1885 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
1886 (ia64_output_function_epilogue): Likewise.
1887 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
1888 (m32r_output_function_epilogue): Likewise.
1889 * config/microblaze/microblaze.c (microblaze_function_prologue)
1890 (microblaze_function_epilogue): Likewise.
1891 * config/mips/mips.c (mips_output_function_prologue): Likewise.
1892 (mips_output_function_epilogue): Likewise.
1893 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
1894 (mmix_target_asm_function_epilogue): Likewise.
1895 * config/msp430/msp430.c (msp430_start_function): Likewise.
1896 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
1897 (nds32_asm_function_epilogue): Likewise.
1898 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
1899 * config/pa/pa.c (pa_output_function_prologue): Likewise.
1900 (pa_output_function_epilogue): Likewise.
1901 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
1902 (rs6000_output_function_epilogue): Likewise.
1903 * config/rl78/rl78.c (rl78_start_function): Likewise.
1904 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
1905 (rs6000_output_function_epilogue): Likewise.
1906 * config/rx/rx.c (rx_output_function_prologue): Likewise.
1907 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
1908 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
1909 (sparc_asm_function_epilogue): Likewise.
1911 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1913 * tree.h (type_has_mode_precision_p): New function.
1914 * convert.c (convert_to_integer_1): Use it.
1915 * expr.c (expand_expr_real_2): Likewise.
1916 (expand_expr_real_1): Likewise.
1917 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
1918 * match.pd: Likewise.
1919 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
1920 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
1921 * tree-tailcall.c (process_assignment): Likewise.
1922 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1923 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
1924 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
1925 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1926 (vectorizable_assignment): Likewise.
1927 (vectorizable_shift): Likewise.
1928 (vectorizable_operation): Likewise.
1929 * tree-vrp.c (register_edge_assert_for_2): Likewise.
1931 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
1933 * match.pd: Add pow (C, x) simplification.
1935 2017-08-21 Richard Biener <rguenther@suse.de>
1937 PR tree-optimization/81900
1938 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
1939 for blocks with abnormal predecessors.
1940 (compute_antic): Do not set visited flag prematurely.
1942 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
1945 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
1947 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1949 * stor-layout.h (vector_type_mode): Move to...
1950 * tree.h (vector_type_mode): ...here.
1951 * stor-layout.c (vector_type_mode): Move to...
1952 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
1954 2017-08-21 Richard Biener <rguenther@suse.de>
1956 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
1957 register_external_die hooks.
1958 (debug_false_tree_charstarstar_uhwistar): Declare.
1959 (debug_nothing_tree_charstar_uhwi): Likewise.
1960 * debug.c (do_nothing_debug_hooks): Adjust.
1961 (debug_false_tree_charstarstar_uhwistar): New do nothing.
1962 (debug_nothing_tree_charstar_uhwi): Likewise.
1963 * dbxout.c (dbx_debug_hooks): Adjust.
1964 (xcoff_debug_hooks): Likewise.
1965 * sdbout.c (sdb_debug_hooks): Likewise.
1966 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1967 * dwarf2out.c (macinfo_label_base): New global.
1968 (dwarf2out_register_external_die): New function for the
1969 register_external_die hook.
1970 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
1971 (dwarf2_debug_hooks): Use them.
1972 (dwarf2_lineno_debug_hooks): Adjust.
1973 (struct die_struct): Add with_offset flag.
1974 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
1975 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
1976 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
1977 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
1978 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
1979 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
1980 defining section names for the early LTO debug variants.
1981 (reset_indirect_string): New helper.
1982 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
1983 (print_dw_val): Add support for offsetted symbol references.
1984 (get_ultimate_context): Split out from is_cxx.
1985 (is_cxx): Use get_ultimate_context.
1986 (is_fortran): Add decl overload.
1987 (compute_comp_unit_symbol): Split out worker from
1988 compute_section_prefix.
1989 (compute_section_prefix): Call compute_comp_unit_symbol and
1990 set comdat_type_p here.
1991 (output_die): Skip DIE symbol output for the LTO added one.
1992 Handle DIE symbol references with offset.
1993 (output_comp_unit): Guard section name mangling properly.
1994 For LTO debug sections emit a symbol at the section beginning
1995 which we use to refer to its DIEs.
1996 (add_abstract_origin_attribute): For DIEs registered via
1997 dwarf2out_register_external_die directly refer to the early
1998 DIE rather than indirectly through the shadow one we created.
1999 Remove obsolete call to dwarf2out_abstract_function for
2000 non-function/block origins.
2001 (gen_array_type_die): When generating early LTO debug do
2002 not emit DW_AT_string_length.
2003 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2004 late when in LTO. As suggested place a gcc_unreachable for
2005 the DECL_ABSTRACT_P case.
2006 (gen_subprogram_die): Avoid another specification DIE
2007 for early built declarations/definitions for the late LTO case.
2008 (gen_variable_die): Add type references for late duplicated VLA dies
2010 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2011 we have the abstract instance already.
2012 (process_scope_var): Adjust decl DIE contexts in LTO which
2013 first puts them in limbo.
2014 (gen_decl_die): Do not generate type DIEs late apart from
2015 types for VLAs or for decls we do not yet have a DIE. Do not
2016 call dwarf2out_abstract_function late.
2017 (dwarf2out_early_global_decl): Make sure to create DIEs
2018 for abstract instances of a decl first.
2019 (dwarf2out_late_global_decl): Adjust comment.
2020 (output_macinfo_op): With multiple macro sections use
2021 macinfo_label_base to distinguish labels.
2022 (output_macinfo): Likewise. Update macinfo_label_base.
2023 Pass in the line info label.
2024 (note_variable_value_in_expr): When generating LTO resolve
2025 all variable values here by generating DIEs as needed.
2026 (init_sections_and_labels): Add early LTO debug flag parameter
2027 and generate different sections and names if set. Add generation
2028 counter for the labels so we can have multiple of them.
2029 (reset_dies): Helper to allow DIEs to be output multiple times.
2030 (dwarf2out_finish): When outputting DIEs to the fat part of an
2031 LTO object first reset DIEs.
2032 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2033 (modified_type_die): Check for decl_ultimate_origin being self
2035 (gen_type_die_with_usage): Likewise.
2036 (gen_typedef_die): Allow decl_ultimate_origin being self.
2037 (set_decl_abstract_flags): Remove.
2038 (set_block_abstract_flags): Likewise.
2039 (dwarf2out_abstract_function): Treat the early generated DIEs
2040 as the abstract copy and only add DW_AT_inline and
2041 DW_AT_artificial here and call set_decl_origin_self.
2042 If the DIE has an abstract origin don't do anything.
2043 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2044 if we have none yet (Go fails to build one, PR78628).
2045 (variably_modified_type_p): Prevent endless recursion for Ada
2046 cyclic pointer types.
2047 * lto-streamer-in.c: Include debug.h.
2048 (dref_queue): New global.
2049 (lto_read_tree_1): Stream in DIE references.
2050 (lto_input_tree): Register DIE references.
2051 (input_function): Stream DECL_DEBUG_ARGS.
2052 * lto-streamer-out.c: Include debug.h.
2053 (lto_write_tree_1): Output DIE references.
2054 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2055 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2056 (output_function): Stream DECL_DEBUG_ARGS.
2057 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2058 Stream DECL_ABSTRACT_ORIGIN.
2059 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2060 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2061 DECL_CONTEXT for file-scope decls.
2062 * lto-streamer.h (struct dref_entry): Declare.
2063 (dref_queue): Likewise.
2064 * cfgexpand.c (pass_expand::execute): Do not call the
2065 outlining_inline_function hook here.
2066 * lto-wrapper.c (debug_obj): New global.
2067 (tool_cleanup): Unlink it if required.
2068 (debug_objcopy): New function.
2069 (run_gcc): Handle early debug sections in the IL files by
2070 extracting them to separate files, partially linkin them and
2071 feeding the result back as result to the linker.
2072 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2073 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2074 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2075 sections into a separate segment.
2076 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2078 (darwin_asm_dwarf_section): Likewise.
2079 (darwin_asm_output_dwarf_offset): Likewise.
2080 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2082 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2083 Alan Hayward <alan.hayward@arm.com>
2084 David Sherwood <david.sherwood@arm.com>
2086 * read-md.h (md_reader::record_potential_iterator_use): Replace
2087 pointer argument with an rtx and an index.
2088 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2089 (apply_mode_iterator): Likewise.
2090 (apply_code_iterator): Likewise.
2091 (apply_int_iterator): Likewise.
2092 (apply_subst_iterator): Likewise.
2093 (record_iterator_use): Likewise.
2094 (record_attribute_use): Likewise.
2095 (md_reader::record_potential_iterator_use): Likewise. Update calls
2096 to record_iterator_use and apply_iterator.
2097 (iterator_use): Replace ptr with x and index.
2098 (attribute_use): Likewise.
2099 (apply_attribute_uses): Update calls to apply_iterator.
2100 (apply_iterators): Likewise. Update initialization of iterator_use.
2101 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2102 and record_potential_iterator_use.
2103 (rtx_reader::read_rtx_operand): Likewise.
2105 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2106 Alan Hayward <alan.hayward@arm.com>
2107 David Sherwood <david.sherwood@arm.com>
2109 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2112 2017-08-21 Richard Biener <rguenther@suse.de>
2115 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2116 at struct end conservatively when comparing common bases.
2118 2017-08-21 Richard Biener <rguenther@suse.de>
2120 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2121 (mem_ref_in_stmt): Remove.
2122 (determine_max_movement): Use ref index to get at the reference.
2123 (invariantness_dom_walker::before_dom_children): Deal with
2124 lim data already initialized.
2125 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2127 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2129 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2130 (TARGET_ISA_ROUND): Ditto.
2131 (TARGET_ROUND): Ditto.
2132 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2133 * config/i386/i386.md: Ditto.
2134 * config/i386/sse.md: Ditto.
2135 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2136 with OPTION_MASK_ISA_SSE4_1.
2138 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2141 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2142 __builtin_ia32_lzcnt_u16.
2144 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2147 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2148 (rs6000_set_current_function): Rewrite function to use it.
2150 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2153 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2154 and TYPE_WARN_IF_NOT_ALIGN.
2155 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2156 (handle_warn_if_not_align): New.
2157 (place_union_field): Call handle_warn_if_not_align.
2158 (place_field): Call handle_warn_if_not_align.
2159 Copy TYPE_WARN_IF_NOT_ALIGN.
2160 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2161 (layout_type): Likewise.
2162 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2164 (tree_decl_common): Add warn_if_not_align.
2165 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2166 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2167 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2168 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2169 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2170 * doc/extend.texi: Document warn_if_not_aligned attribute.
2171 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2173 2017-08-17 Martin Liska <mliska@suse.cz>
2176 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2177 (get_data_dependence): Use it as pointer type.
2178 (distribute_loop): Likewise.
2180 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2182 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2183 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2184 (p8_vmrgow_<mode>_direct): New define_insn.
2185 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2186 handle endianness for vmrgew and vmrgow permute patterns.
2188 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2190 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2191 * config/rs6000/rs6000-cpus.def: Remove comment.
2192 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2193 (POWERPC_MASKS): Likewise.
2194 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2195 use of TARGET_VSX_TIMODE.
2196 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2197 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2198 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2199 (rs6000_option_override_internal): Remove dead code.
2200 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2201 (rs6000_legitimize_reload_address): Likewise.
2202 (rs6000_legitimate_address_p): Likewise.
2203 (rs6000_opt_masks): Delete "vsx-timode".
2204 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2205 from function comment.
2206 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2207 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2208 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2210 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2211 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2212 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2215 (define_peephole2): Likewise.
2217 2017-08-17 Martin Sebor <msebor@redhat.com>
2220 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2221 past the end of an array.
2222 (test_pp_format): Add test cases.
2224 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2226 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2227 missing ECF_NOTHROW flags.
2229 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2232 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2233 operands residing in integer registers.
2234 (*vsx_le_perm_load_<mode>): Likewise.
2235 (*vsx_le_perm_store_<mode>): Likewise.
2236 (define_peephole2): Add peepholes to optimize the above.
2238 2017-08-17 Marek Polacek <polacek@redhat.com>
2241 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2242 to mimic what shorten_compare did. Change the return type to bool.
2243 (fold_cond_expr_with_comparison): Update call to
2244 operand_equal_for_comparison_p.
2245 (fold_ternary_loc): Likewise.
2247 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2249 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2251 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2253 2017-08-17 Richard Biener <rguenther@suse.de>
2255 * tree-ssa-structalias.c (solve_graph): When propagating
2256 to successors update the graphs succ edges and avoid duplicate work.
2258 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2261 * config/i386/i386.c (ix86_option_override_internal): Save target
2262 specific options after ix86_stack_protector_guard_reg was changed.
2264 2017-08-17 Richard Biener <rguenther@suse.de>
2266 PR tree-optimization/81827
2267 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2268 (new_var_info): Initialize it conservatively.
2269 (get_call_vi): Mark register vars.
2270 (new_scalar_tmp_constraint_exp): Likewise.
2271 (handle_rhs_call): Likewise.
2272 (handle_const_call): Likewise.
2273 (create_function_info_for): Likewise.
2274 (solve_constraints): Sort varinfos to separate register from
2275 non-register vars to pack points-to solution bitmaps during
2278 2017-08-17 Marek Polacek <polacek@redhat.com>
2280 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2282 2017-08-17 Richard Biener <rguenther@suse.de>
2284 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2285 to true when overflow is undefined and we saturated the result.
2287 2017-08-17 Alan Modra <amodra@gmail.com>
2290 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2291 Don't use store multiple if only one reg needs saving.
2292 (interesting_frame_related_regno): New function.
2293 (rs6000_frame_related): Don't emit frame info for regs that
2295 (rs6000_emit_epilogue): Likewise.
2297 2017-08-16 Nathan Sidwell <nathan@acm.org>
2299 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2300 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2301 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2302 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2303 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2304 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2305 (verify_type): Adjust for TYPE_BINFO move.
2306 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2307 process TYPE_BINFO directly.
2308 (hash_tree): Likewise.
2309 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2311 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2314 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2316 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2318 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2321 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2322 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2323 Add ix86_binary_operator_ok to insn constraint.
2324 (*iordi_1_bts): Ditto.
2325 (*xordi_1_btc): Ditto.
2326 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2327 Update corresponding peephole2 pattern.
2331 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2333 PR tree-optimization/81832
2334 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2335 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2337 2017-08-16 Marek Polacek <polacek@redhat.com>
2340 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2341 regarding size_zero_node.
2343 2017-08-16 Martin Liska <mliska@suse.cz>
2346 * config.gcc: Respect previously set extra_objs in case
2349 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2351 PR tree-optimization/81835
2352 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2353 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2354 not depend on the phi.
2356 2017-08-16 Alan Modra <amodra@gmail.com>
2358 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2361 2017-08-16 Alan Modra <amodra@gmail.com>
2363 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2364 (save_reg_p): ..into this. Update all callers.
2365 (first_reg_to_save): Simplify.
2367 2017-08-16 Alan Modra <amodra@gmail.com>
2369 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2372 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2376 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2377 constant count if that count is less than 32.
2379 2017-08-15 Nathan Sidwell <nathan@acm.org>
2381 * gcc.c (execute): Emit friendlier message if inferior is killed
2382 by an external cause.
2384 2017-08-15 Richard Biener <rguenther@suse.de>
2386 PR tree-optimization/81790
2387 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2388 CONSTRUCTORs from simplifying and VN.
2390 2017-08-14 Martin Sebor <msebor@redhat.com>
2392 * builtin-attrs.def: Add comments.
2394 2017-08-14 Martin Sebor <msebor@redhat.com>
2397 * doc/extend.texi (attribute nonstring): Document new attribute.
2399 2017-08-14 Martin Sebor <msebor@redhat.com>
2402 * tree-diagnostic.c (default_tree_printer): Handle %G.
2403 * gimple-pretty-print.h (percent_G_format): Declare new function.
2404 * gimple-pretty-print.c (percent_G_format): Define.
2405 * tree-pretty-print.c (percent_K_format): Add argument.
2407 2017-08-14 Martin Sebor <msebor@redhat.com>
2409 PR translation/79998
2410 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2411 Remove a stray space.
2413 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2416 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2417 (*iordi_1_bts): Ditto.
2418 (*xordi_1_btc): Ditto.
2420 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2423 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2424 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2426 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2427 quoted strings, and make more translator-friendly.
2428 (darwin_rs6000_override_options): Likewise.
2429 (rs6000_option_override_internal): Likewise.
2430 (rs6000_return_in_memory): Fix overlong line.
2431 (init_cmulative_args): Use quoted strings, and make more
2432 translator-friendly.
2433 (rs6000_pass_by_reference): Fix overlong line.
2434 (def_builtin): Use quoted strings.
2435 (altivec_expand_predicate_builtin): Use quoted strings, and make
2436 more translator-friendly.
2437 (htm_expand_builtin): Use quoted strings.
2438 (cpu_expand_builtin): Use quoted strings, and make more
2439 translator-friendly.
2440 (altivec_expand_builtin): Likewise.
2441 (paired_expand_predicate_builtin): Likewise.
2442 (rs6000_invalid_builtin): Likewise.
2443 (builtin_function_type): Use quoted strings.
2444 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2445 more translator-friendly.
2446 (rs6000_trampoline_init): Likewise.
2447 (rs6000_handle_altivec_attribute): Likewise.
2448 (rs6000_inner_target_options): Use quoted strings.
2449 (rs6000_disable_incompatible_switches): Likewise.
2450 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2451 strings, and make more translator-friendly.
2452 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2454 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2456 PR tree-optimization/81799
2457 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2458 cond_expr to simple gimple operand.
2460 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2463 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2465 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2469 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2470 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2471 TARGET_GASISR_PROLOGUES.
2472 * config/avr/avr.c (avr_option_override): Same.
2473 (avr_pass_pre_proep::execute): Same.
2475 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2478 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2479 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2481 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2483 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2484 define_insn_and_split. Split to a memory load from 0 in
2485 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2486 using PTR mode iterator.
2487 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2488 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2489 (*add_tp_<mode>): Redefine as define_insn_and_split.
2490 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2491 address space. Merge with *add_tp_x32 using PTR mode iterator.
2492 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2493 Split to an add with a memory load from 0 in
2494 DEFAULT_TLS_SEG_REG address space.
2496 2017-08-12 Andrew Pinski <apinski@cavium.com>
2498 * config/aarch64/aarch64-option-extensions.def (rdma):
2499 Fix feature string to what Linux prints out in /proc/cpuinfo.
2501 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2504 * dwarf2out.c (modified_type_die): For C typedef types that have
2505 an ultimate origin, process the ultimate origin instead of the
2507 (gen_typedef_die): Assert that input DECLs have no ultimate
2509 (gen_type_die_with_usage): For typedef variants that have an
2510 ultimate origin, just call gen_decl_die on the original DECL.
2511 (process_scope_var): Avoid creating DIEs for local typedefs and
2512 concrete static variables.
2514 2017-08-12 Alan Modra <amodra@gmail.com>
2518 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2519 match gnu-user.h startfile.
2520 (ENDFILE_LINUX_SPEC): Similarly.
2522 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2525 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2527 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2529 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2530 * config/aarch64/aarch64.md (mov<mode>): Change.
2531 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2532 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2533 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2535 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2537 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2538 for storage order barriers.
2540 2017-08-11 Martin Liska <mliska@suse.cz>
2543 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2544 variables of void type.
2546 2017-08-11 Martin Liska <mliska@suse.cz>
2548 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2549 TARGET_SUPPORTS_ALIASES.
2550 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2551 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2552 (optimize_weakref): Likewise.
2553 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2554 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2555 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2557 2017-08-11 Martin Liska <mliska@suse.cz>
2560 * config/i386/i386.c (make_resolver_func): Do complete
2561 refactoring of the function.
2563 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2566 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2567 * config/i386/i386.c (ix86_stack_protect_guard): Use
2568 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2569 * doc/invoke.texi (x86 Options): Document
2570 -mstack-protector-guard-symbol= option.
2572 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2574 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2575 * config/i386/i386.c (ix86_split_stack_guard): New function.
2576 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2577 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2578 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2579 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2580 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2581 (split_stack_space_check): Call ix86_split_stack_guard.
2583 2017-08-10 Martin Sebor <msebor@redhat.com>
2585 * print-tree.c (print_node): Print location using the established
2587 Replace spaces with colons.
2588 (debug_raw, debug): Ditto.
2590 2017-08-10 Martin Sebor <msebor@redhat.com>
2593 * pretty-print.c (pp_format): Correct the handling of %s precision.
2595 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2598 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2600 (ix86_finalize_stack_frame_flags): This. Also clear
2601 frame_pointer_needed if -fno-omit-frame-pointer is used without
2603 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2604 with ix86_finalize_stack_frame_flags.
2605 (ix86_expand_epilogue): Likewise.
2606 (ix86_expand_split_stack_prologue): Likewise.
2607 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2609 2017-08-10 Martin Liska <mliska@suse.cz>
2612 * c-attribs.c (handle_target_attribute):
2613 Report warning for an empty string argument of target attribute.
2615 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2618 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2620 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2621 or DECL_NONLOCAL labels.
2622 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2623 or DECL_NONLOCAL labels here.
2625 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2627 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2628 to indicate when early gimple folding has been disabled.
2629 (rs6000_gimple_fold_builtin): Add debug content.
2630 (rs6000_invalid_builtin): Fix whitespace.
2631 (rs6000_expand_builtin): Fix whitespace.
2632 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2634 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2637 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2638 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2640 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2642 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2643 (qdf24xx): Likewise.
2644 * config/aarch64/aarch64.md: Include falkor.md.
2645 * config/aarch64/falkor.md: New.
2647 2017-08-09 Marek Polacek <polacek@redhat.com>
2650 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2651 * diagnostic.c (emit_diagnostic): Add a comment.
2652 (emit_diagnostic_valist): New function.
2654 2017-08-09 Marek Polacek <polacek@redhat.com>
2657 * input.c (make_location): New overload.
2658 * input.h (make_location): Declare.
2660 2017-08-08 Alan Modra <amodra@gmail.com>
2661 H.J. Lu <hongjiu.lu@intel.com>
2664 * gcc.c (NO_PIE_SPEC): Delete.
2665 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2667 (LINK_PIE_SPEC): ..to here.
2668 (LINK_COMMAND_SPEC): Support -no-pie.
2669 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2670 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2672 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2673 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2674 (ENDFILE_CRTEND_SPEC): Similarly.
2676 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2679 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2680 (mstack-protector-guard-offset=): Ditto.
2681 * config/i386/i386.c (ix86_option_override): Handle
2682 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2684 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2685 ix86_stack_protect_guard_offset variables.
2686 (TARGET_STACK_PROTECT_GUARD): Always define.
2687 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2688 and -mstack-protector-guard-offset= options.
2690 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2692 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2693 boundary case for the last candidate.
2695 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2697 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2698 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2700 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2703 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2704 * config/aarch64/aarch64-builtins.c
2705 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2706 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2707 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2709 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2710 Andrew Pinski <pinskia@gmail.com>
2713 * internal-fn.def (XORSIGN): New.
2714 * optabs.def (xorsign_optab): New.
2715 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2716 (convert_expand_mult_copysign): New.
2717 (pass_optimize_widening_mul::execute): Call
2718 convert_expand_mult_copysign.
2720 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2722 PR tree-optimization/81354
2723 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2724 Insert on edges rather than explicitly creating landing pads.
2725 (analyze_candidates_and_replace): Commit edge inserts.
2727 2017-08-08 Richard Biener <rguenther@suse.de>
2730 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2731 (expand_simple_operations): Also look through ADDR_EXPRs with
2732 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2734 2017-08-08 Richard Biener <rguenther@suse.de>
2736 PR tree-optimization/81723
2737 * tree-vect-slp.c (struct bst_traits): New hash traits.
2738 (bst_fail): New global.
2739 (vect_build_slp_tree_2): New worker, split out from ...
2740 (vect_build_slp_tree): ... this now wrapping it with using
2741 bst_fail set to cache SLP tree build fails. Properly handle
2743 (vect_analyze_slp_instance): Allocate and free bst_fail.
2745 2017-08-08 Martin Liska <mliska@suse.cz>
2748 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2749 LABEL_DECLs that can be from a different function.
2751 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2753 PR tree-optimization/81744
2754 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2755 loop's number of iterations.
2757 2017-08-08 Martin Liska <mliska@suse.cz>
2759 * asan.c: Include header files.
2760 * attribs.c (build_decl_attribute_variant): New function moved
2762 (build_type_attribute_qual_variant): Likewise.
2763 (cmp_attrib_identifiers): Likewise.
2764 (simple_cst_list_equal): Likewise.
2765 (omp_declare_simd_clauses_equal): Likewise.
2766 (attribute_value_equal): Likewise.
2767 (comp_type_attributes): Likewise.
2768 (build_type_attribute_variant): Likewise.
2769 (lookup_ident_attribute): Likewise.
2770 (remove_attribute): Likewise.
2771 (merge_attributes): Likewise.
2772 (merge_type_attributes): Likewise.
2773 (merge_decl_attributes): Likewise.
2774 (merge_dllimport_decl_attributes): Likewise.
2775 (handle_dll_attribute): Likewise.
2776 (attribute_list_equal): Likewise.
2777 (attribute_list_contained): Likewise.
2778 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2779 (lookup_attribute_by_prefix): Likewise.
2780 * bb-reorder.c: Include header files.
2781 * builtins.c: Likewise.
2782 * calls.c: Likewise.
2783 * cfgexpand.c: Likewise.
2784 * cgraph.c: Likewise.
2785 * cgraphunit.c: Likewise.
2786 * convert.c: Likewise.
2787 * dwarf2out.c: Likewise.
2788 * final.c: Likewise.
2789 * fold-const.c: Likewise.
2790 * function.c: Likewise.
2791 * gimple-expr.c: Likewise.
2792 * gimple-fold.c: Likewise.
2793 * gimple-pretty-print.c: Likewise.
2794 * gimple.c: Likewise.
2795 * gimplify.c: Likewise.
2796 * hsa-common.c: Likewise.
2797 * hsa-gen.c: Likewise.
2798 * internal-fn.c: Likewise.
2799 * ipa-chkp.c: Likewise.
2800 * ipa-cp.c: Likewise.
2801 * ipa-devirt.c: Likewise.
2802 * ipa-fnsummary.c: Likewise.
2803 * ipa-inline.c: Likewise.
2804 * ipa-visibility.c: Likewise.
2806 * lto-cgraph.c: Likewise.
2807 * omp-expand.c: Likewise.
2808 * omp-general.c: Likewise.
2809 * omp-low.c: Likewise.
2810 * omp-offload.c: Likewise.
2811 * omp-simd-clone.c: Likewise.
2812 * opts-global.c: Likewise.
2813 * passes.c: Likewise.
2814 * predict.c: Likewise.
2815 * sancov.c: Likewise.
2816 * sanopt.c: Likewise.
2817 * symtab.c: Likewise.
2818 * toplev.c: Likewise.
2819 * trans-mem.c: Likewise.
2820 * tree-chkp.c: Likewise.
2821 * tree-eh.c: Likewise.
2822 * tree-into-ssa.c: Likewise.
2823 * tree-object-size.c: Likewise.
2824 * tree-parloops.c: Likewise.
2825 * tree-profile.c: Likewise.
2826 * tree-ssa-ccp.c: Likewise.
2827 * tree-ssa-live.c: Likewise.
2828 * tree-ssa-loop.c: Likewise.
2829 * tree-ssa-sccvn.c: Likewise.
2830 * tree-ssa-structalias.c: Likewise.
2831 * tree-ssa.c: Likewise.
2832 * tree-streamer-in.c: Likewise.
2833 * tree-vectorizer.c: Likewise.
2834 * tree-vrp.c: Likewise.
2836 * ubsan.c: Likewise.
2837 * varasm.c: Likewise.
2838 * varpool.c: Likewise.
2839 * tree.c: Remove functions moved to attribs.[ch].
2841 * config/aarch64/aarch64.c: Add attrs.h header file.
2842 * config/alpha/alpha.c: Likewise.
2843 * config/arc/arc.c: Likewise.
2844 * config/arm/arm.c: Likewise.
2845 * config/avr/avr.c: Likewise.
2846 * config/bfin/bfin.c: Likewise.
2847 * config/c6x/c6x.c: Likewise.
2848 * config/cr16/cr16.c: Likewise.
2849 * config/cris/cris.c: Likewise.
2850 * config/darwin.c: Likewise.
2851 * config/epiphany/epiphany.c: Likewise.
2852 * config/fr30/fr30.c: Likewise.
2853 * config/frv/frv.c: Likewise.
2854 * config/ft32/ft32.c: Likewise.
2855 * config/h8300/h8300.c: Likewise.
2856 * config/i386/winnt.c: Likewise.
2857 * config/ia64/ia64.c: Likewise.
2858 * config/iq2000/iq2000.c: Likewise.
2859 * config/lm32/lm32.c: Likewise.
2860 * config/m32c/m32c.c: Likewise.
2861 * config/m32r/m32r.c: Likewise.
2862 * config/m68k/m68k.c: Likewise.
2863 * config/mcore/mcore.c: Likewise.
2864 * config/microblaze/microblaze.c: Likewise.
2865 * config/mips/mips.c: Likewise.
2866 * config/mmix/mmix.c: Likewise.
2867 * config/mn10300/mn10300.c: Likewise.
2868 * config/moxie/moxie.c: Likewise.
2869 * config/msp430/msp430.c: Likewise.
2870 * config/nds32/nds32-isr.c: Likewise.
2871 * config/nds32/nds32.c: Likewise.
2872 * config/nios2/nios2.c: Likewise.
2873 * config/nvptx/nvptx.c: Likewise.
2874 * config/pa/pa.c: Likewise.
2875 * config/pdp11/pdp11.c: Likewise.
2876 * config/powerpcspe/powerpcspe.c: Likewise.
2877 * config/riscv/riscv.c: Likewise.
2878 * config/rl78/rl78.c: Likewise.
2879 * config/rx/rx.c: Likewise.
2880 * config/s390/s390.c: Likewise.
2881 * config/sh/sh.c: Likewise.
2882 * config/sol2.c: Likewise.
2883 * config/sparc/sparc.c: Likewise.
2884 * config/spu/spu.c: Likewise.
2885 * config/stormy16/stormy16.c: Likewise.
2886 * config/tilegx/tilegx.c: Likewise.
2887 * config/tilepro/tilepro.c: Likewise.
2888 * config/v850/v850.c: Likewise.
2889 * config/vax/vax.c: Likewise.
2890 * config/visium/visium.c: Likewise.
2891 * config/xtensa/xtensa.c: Likewise.
2893 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2896 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
2897 constraints since the -mupper-regs-* switches have been
2899 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
2900 into a vector from a double word element that was extracted from
2901 another vector, and eliminate extra XXPERMDI instructions.
2902 (vsx_concat_<mode>_2): Likewise.
2903 (vsx_concat_<mode>_3): Likewise.
2904 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
2905 concat to allow optimizing inserts from previous extracts.
2907 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
2909 * config/i386/i386.c (ix86_stack_protect_guard): Generate
2910 memory reference to a SSP offset in TLS address space.
2911 (ix86_print_operand) <case '@'>: Remove.
2912 (ix86_print_operand_punct_valid_p): Remove '@' code.
2913 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
2915 (stack_tls_protect_set_<mode>): Remove.
2916 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
2917 (stack_tls_protect_test_<mode>): Remove.
2918 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
2920 2017-08-07 Olivier Hainque <hainque@adacore.com>
2923 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
2925 2017-08-07 Douglas Rupp <rupp@adacore.com>
2927 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
2928 variable was referenced as multidir in command.
2930 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2933 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
2936 2017-08-07 Martin Liska <mliska@suse.cz>
2938 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
2939 * config/rl78/rl78.c: Add include of attribs.h.
2940 * config/sh/sh.c: Likewise.
2941 * config/v850/v850.c: Likewise.
2943 2017-08-07 Tom de Vries <tom@codesourcery.com>
2946 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
2948 2017-08-07 Martin Liska <mliska@suse.cz>
2950 * config/mips/mips.c: Include attribs.h.
2952 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
2955 * doc/invoke.texi: Document change in behvaior for -Ofast for
2958 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
2960 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
2962 (aarch64_pop_regs): Likewise.
2963 (aarch64_gen_load_pair): Likewise.
2964 (aarch64_save_callee_saves): Likewise.
2965 (aarch64_restore_callee_saves): Likewise.
2967 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2969 * config/i386/i386.c: Revert the last change.
2971 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2974 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2976 (ix86_finalize_stack_frame_flags): This. Also clear
2977 frame_pointer_needed if -fno-omit-frame-pointer is used without
2979 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2980 with ix86_finalize_stack_frame_flags.
2981 (ix86_expand_epilogue): Likewise.
2982 (ix86_expand_split_stack_prologue): Likewise.
2984 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2987 * config/i386/i386.c (get_builtin_code_for_version): Set priority
2988 to P_AES for Westmere.
2990 2017-08-07 Jonathan Yong <10walls@gmail.com>
2992 * config/i386/mingw.opt (fset-stack-executable): Removed.
2993 * config/i386/cygming.opt (fset-stack-executable): Moved
2995 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
2997 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
2999 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3001 2017-08-07 Marek Polacek <polacek@redhat.com>
3004 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3006 2017-08-07 Martin Liska <mliska@suse.cz>
3008 * attribs.h (canonicalize_attr_name): New function.
3009 (cmp_attribs): Move from c-format.c and adjusted.
3010 (is_attribute_p): Moved from tree.h.
3011 * tree-inline.c: Add new includes.
3012 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3013 (private_is_attribute_p): Remove.
3014 (private_lookup_attribute): Likewise.
3015 (private_lookup_attribute_by_prefix): Simplify.
3016 (remove_attribute): Use is_attribute_p.
3017 * tree.h: Remove removed declarations.
3019 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3022 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3023 instead of computing it in the function. Formatting fix.
3024 (expand_case): Don't rely on default_edge being the first edge,
3025 clear it if removing it, pass default_edge to
3026 emit_case_dispatch_table.
3027 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3030 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3032 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3033 insn in the function, emit NOP after the insn.
3035 2017-08-06 Tom de Vries <tom@codesourcery.com>
3037 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3040 2017-08-06 Tom de Vries <tom@codesourcery.com>
3042 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3045 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3047 PR tree-optimization/57371
3048 * match.pd: New pattern.
3050 2017-08-04 Marek Polacek <polacek@redhat.com>
3053 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3054 perform the computation in offset_int.
3056 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3058 PR tree-optimization/81136
3059 * tree-vectorizer.h: Include tree-hash-traits.h.
3060 (vec_base_alignments): New typedef.
3061 (vec_info): Add a base_alignments field.
3062 (vect_record_base_alignments): Declare.
3063 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3065 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3066 (create_data_ref): Add an is_conditional_in_stmt argument.
3067 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3068 the is_conditional_in_stmt field.
3069 (data_ref_loc): Add an is_conditional_in_stmt field.
3070 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3071 (find_data_references_in_stmt): Update call to create_data_ref.
3072 (graphite_find_data_references_in_stmt): Likewise.
3073 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3074 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3075 (vect_record_base_alignment): New function.
3076 (vect_record_base_alignments): Likewise.
3077 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3078 for nested statements even if we fail to compute a misalignment.
3079 Use pooled base alignments for unconditional references.
3080 (vect_find_same_alignment_drs): Compare base addresses instead
3082 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3083 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3085 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3087 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3088 Add an explicit name for the enum. Use auto_vec for slp_instances
3090 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3092 (_bb_vec_info): Add a constructor and destructor.
3093 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3094 (destroy_loop_vec_info): Delete.
3095 (vect_destroy_datarefs): Likewise.
3096 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3097 (vec_info::vec_info): New function.
3098 (vec_info::~vec_info): Likewise.
3099 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3100 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3101 destroy_loop_vec_info.
3102 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3103 (_loop_vec_info::_loop_vec_info): ...this.
3104 (destroy_loop_vec_info): Replace with...
3105 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3106 the stmt_vec_infos. Leave handling of vec_info information to its
3107 destructor. Remove explicit vector releases.
3108 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3109 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3110 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3111 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3112 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3113 (destroy_bb_vec_info): Replace with...
3114 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3115 information to its destructor.
3116 (vect_slp_analyze_bb_1): Use new and delete instead of
3117 new_bb_vec_info and destroy_bb_vec_info.
3118 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3121 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3123 * tree-data-ref.h (subscript): Add access_fn field.
3124 (data_dependence_relation): Add could_be_independent_p.
3125 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3126 (same_access_functions): Move to tree-data-ref.c.
3127 * tree-data-ref.c (ref_contains_union_access_p): New function.
3128 (access_fn_component_p): Likewise.
3129 (access_fn_components_comparable_p): Likewise.
3130 (dr_analyze_indices): Add a reference to access_fn_component_p.
3131 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3133 (constant_access_functions): Likewise.
3134 (add_other_self_distances): Likewise.
3135 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3136 (initialize_data_dependence_relation): Use XCNEW and remove
3137 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3138 of access functions that have the same type. Allow the
3139 subsequence to end with different bases in some circumstances.
3140 Record the chosen access functions in SUB_ACCESS_FN.
3141 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3142 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3143 (subscript_dependence_tester_1): Likewise dra and drb.
3144 (build_classic_dist_vector): Update calls accordingly.
3145 (subscript_dependence_tester): Likewise.
3146 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3147 DDR_COULD_BE_INDEPENDENT_P.
3148 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3149 comp_alias_ddrs instead of may_alias_ddrs.
3150 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3152 (vect_analyze_data_ref_dependence): Use it if
3153 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3154 distance vectors if that fails.
3155 (dependence_distance_ge_vf): New function.
3156 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3157 LOOP_VINFO_MAY_ALIAS_DDRS.
3159 2017-08-04 Richard Biener <rguenther@suse.de>
3162 * fold-const.c (fold_binary_loc): Properly restrict
3163 minus_var0 && minus_var1 case when associating undefined overflow
3166 2017-08-04 Tom de Vries <tom@codesourcery.com>
3168 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3170 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3172 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3173 Don't start diagnostic messages with a capital letter.
3174 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3176 (rs6000_invalid_builtin): Likewise.
3177 (rs6000_trampoline_init): Likewise.
3179 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3182 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3183 after setting changeable df flags.
3185 2017-08-03 Richard Biener <rguenther@suse.de>
3187 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3188 up if the use is in USE - X.
3190 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3192 * toplev.c (dumpfile.h): New include.
3193 (internal_error_reentered): New static function. Use it...
3194 (internal_error_function): ...here to handle reentered internal_error.
3196 2017-08-03 Richard Biener <rguenther@suse.de>
3199 * fold-const.c (split_tree): Add minus_var and minus_con
3200 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3201 here but always use minus_*.
3202 (associate_trees): Assert we never associate with MINUS_EXPR
3203 and NULL first operand. Do not recurse for PLUS_EXPR operands
3204 when associating as MINUS_EXPR either.
3205 (fold_binary_loc): Track minus_var and minus_con.
3207 2017-08-03 Tom de Vries <tom@codesourcery.com>
3210 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3211 ACCEL_COMPILER, apply finish_options on
3212 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3214 2017-08-03 Tom de Vries <tom@codesourcery.com>
3217 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3218 function_entry_patch_area_size > 0.
3220 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3223 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3224 instead of 10??LU, perform unit multiplication in wide_int,
3225 don't change alloc_object_size_limit if the limit is larger
3228 PR tree-optimization/81655
3229 PR tree-optimization/81588
3230 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3231 the case when ranges[i].low and high are 1 for unsigned type with
3235 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3236 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3238 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3240 * tree-vrp.h: Add include guard.
3242 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3245 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3246 (ud2): New insn pattern.
3247 * config/i386/i386.c (ix86_expand_epilogue):
3248 For naked functions, generate ud2 instead of trap insn.
3250 2017-08-02 Marek Polacek <polacek@redhat.com>
3253 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3255 2017-08-02 Tom de Vries <tom@codesourcery.com>
3256 Cesar Philippidis <cesar@codesourcery.com>
3258 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3259 Add missing edge probabilities.
3261 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3263 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3266 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3269 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3270 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3271 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3272 prologue_seq sequences - if any.
3274 2017-08-02 Richard Biener <rguenther@suse.de>
3276 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3277 via vectors if supported, integer extracts via punning if supported
3278 or otherwise vector extracts.
3280 2017-08-02 Richard Biener <rguenther@suse.de>
3282 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3284 (bitmap_insert_into_set): ... this.
3286 2017-08-02 Richard Biener <rguenther@suse.de>
3288 PR tree-optimization/81633
3290 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3292 PR tree-optimization/71752
3293 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3295 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3297 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3298 (machine_function::call_ms2sysv_pad_out): Remove field.
3299 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3300 (ix86_compute_frame_layout): Likewise.
3302 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3305 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3306 attribute with interrupt attribute.
3308 2017-08-01 Andrew Pinski <apinski@cavium.com>
3310 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3311 BIT_INSERT_EXPR's operand 1
3312 to see if the types precision matches.
3314 2017-08-01 Martin Liska <mliska@suse.cz>
3317 * builtins.c (expand_builtin_memcpy_args): Remove.
3318 (expand_builtin_memcpy): Call newly added function
3319 expand_builtin_memory_copy_args.
3320 (expand_builtin_memcpy_with_bounds): Likewise.
3321 (expand_builtin_mempcpy): Remove last argument.
3322 (expand_builtin_mempcpy_with_bounds): Likewise.
3323 (expand_builtin_memory_copy_args): New function created from
3324 expand_builtin_mempcpy_args with small modifications.
3325 (expand_builtin_mempcpy_args): Remove.
3326 (expand_builtin_stpcpy): Remove unused argument.
3327 (expand_builtin): Likewise.
3328 (expand_builtin_with_bounds): Likewise.
3330 2017-08-01 Martin Liska <mliska@suse.cz>
3333 Make mempcpy more optimal (PR middle-end/70140).
3335 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3339 __builtin_vec_cmpne verify both arguments are compatible vectors
3340 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3341 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3342 move computation of aligned to after checking the argument types.
3346 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3347 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3350 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3351 Doug Rupp <rupp@adacore.com>
3352 Olivier Hainque <hainque@adacore.com>
3354 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3355 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3357 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3358 for TARGET_OS_CPP_BUILTIN.
3359 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3360 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3362 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3363 passing required abi options to the assembler for EABI configurations.
3364 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3365 of .text.hot and .text.unlikely sections for kernel modules when
3366 using ARM style exceptions.
3367 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3368 options. Add EXTRA_CC1_SPEC.
3369 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3371 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3373 (ARM_TARGET2_DWARF_FORMAT): Define.
3374 * config/arm/t-vxworks: Adjust multilib control to removal of the
3375 Diab command line options.
3377 2017-08-01 Martin Liska <mliska@suse.cz>
3379 PR gcov-profile/81561
3380 * gcov.c (unblock): Make unblocking safe as we need to preserve
3381 index correspondence of blocks and block_lists.
3383 2017-08-01 Richard Biener <rguenther@suse.de>
3385 PR tree-optimization/81181
3386 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3387 (compute_antic): ... end of iteration here.
3389 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3391 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3392 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3393 (ftree-slp-vectorize): Likewise.
3394 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3395 can no longer be set independent of flag_tree_loop_vectorize.
3396 * omp-general.c (emp_max_vf): Likewise.
3397 * opts.c (enable_fdo_optimizations): Remove references to
3398 flag_tree_vectorize, these are now implicit.
3399 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3400 and leave it for the options machinery.
3402 2017-08-01 Martin Liska <mliska@suse.cz>
3405 * builtins.c (expand_builtin_memcpy_args): Remove.
3406 (expand_builtin_memcpy): Call newly added function
3407 expand_builtin_memory_copy_args.
3408 (expand_builtin_memcpy_with_bounds): Likewise.
3409 (expand_builtin_mempcpy): Remove last argument.
3410 (expand_builtin_mempcpy_with_bounds): Likewise.
3411 (expand_builtin_memory_copy_args): New function created from
3412 expand_builtin_mempcpy_args with small modifications.
3413 (expand_builtin_mempcpy_args): Remove.
3414 (expand_builtin_stpcpy): Remove unused argument.
3415 (expand_builtin): Likewise.
3416 (expand_builtin_with_bounds): Likewise.
3418 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3421 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3422 print "ds:" only for immediates in generic address space.
3424 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3427 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3428 (ix86_function_ok_for_sibcall): Return false for naked functions.
3430 2017-08-01 Richard Biener <rguenther@suse.de>
3432 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3433 (compute_antic): Seed worklist with exit block predecessors.
3434 * cfganal.c (dfs_find_deadend): For a cycle return the source
3435 of the edge closing it.
3437 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3439 * config/aarch64/aarch64.c
3440 (aarch64_can_const_movi_rtx_p): Move 0 check.
3442 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3444 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3445 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3447 * match.pd: Ditto in address comparison pattern.
3449 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3451 PR tree-optimization/81627
3452 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3453 closed ssa form for store-store chain.
3455 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3457 PR tree-optimization/81620
3458 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3459 for store-store chain.
3461 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3463 PR tree-optimization/81588
3464 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3465 ranges[i].in_p, invert comparison code ccode. For >/>=,
3466 swap rhs1 and rhs2 and comparison code unconditionally,
3467 for </<= don't do that. Don't swap rhs1/rhs2 again if
3468 ranges[i].in_p, instead invert comparison code ccode if
3469 opcode or oe->rank is BIT_IOR_EXPR.
3472 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3473 a direct optab to conversion optab.
3474 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3475 with GET_MODE_INNER as last argument instead of optab_handler.
3476 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3477 vector extraction if possible and optab is available.
3478 * expr.c (store_constructor): Use convert_optab_handler instead
3479 of optab_handler. Use vector initialization from smaller
3480 vectors if possible and optab is available.
3481 * tree-vect-stmts.c (vectorizable_load): Likewise.
3482 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3484 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3485 of vec_init from half-sized vectors with the same element mode.
3486 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3487 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3488 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3489 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3490 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3491 after mode in gen_vec_extract* calls.
3492 (vec_extract<mode>): Renamed to ...
3493 (vec_extract<mode><ssescalarmodelower>): ... this.
3494 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3495 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3496 element mode after mode in gen_vec_init* calls.
3497 (VEC_INIT_HALF_MODE): New mode iterator.
3498 (vec_init<mode>): Renamed to ...
3499 (vec_init<mode><ssescalarmodelower>): ... this.
3500 (vec_init<mode><ssehalfvecmodelower>): New expander.
3501 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3502 (vec_extractv2sfsf): ... this.
3503 (vec_initv2sf): Renamed to ...
3504 (vec_initv2sfsf): ... this.
3505 (vec_extractv2si): Renamed to ...
3506 (vec_extractv2sisi): ... this.
3507 (vec_initv2si): Renamed to ...
3508 (vec_initv2sisi): ... this.
3509 (vec_extractv4hi): Renamed to ...
3510 (vec_extractv4hihi): ... this.
3511 (vec_initv4hi): Renamed to ...
3512 (vec_initv4hihi): ... this.
3513 (vec_extractv8qi): Renamed to ...
3514 (vec_extractv8qiqi): ... this.
3515 (vec_initv8qi): Renamed to ...
3516 (vec_initv8qiqi): ... this.
3517 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3518 (vec_init<mode>): Renamed to ...
3519 (vec_init<mode><VEC_base_l>): ... this.
3520 (vec_extract<mode>): Renamed to ...
3521 (vec_extract<mode><VEC_base_l>): ... this.
3522 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3523 (vec_initv2sfsf): ... this.
3524 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3525 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3526 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3527 element mode after mode in gen_vec_init* calls.
3528 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3529 (vec_init<mode><Vel>): ... this.
3530 (vec_extract<mode>): Renamed to ...
3531 (vec_extract<mode><Vel>): ... this.
3532 * config/aarch64/iterators.md (Vel): New mode attribute.
3533 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3534 Add element mode after mode in gen_vec_extract* calls.
3535 * config/s390/vector.md (non_vec_l): New mode attribute.
3536 (vec_extract<mode>): Renamed to ...
3537 (vec_extract<mode><non_vec_l>): ... this.
3538 (vec_init<mode>): Renamed to ...
3539 (vec_init<mode><non_vec_l>): ... this.
3540 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3541 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3543 * config/arm/iterators.md (V_elem_l): New mode attribute.
3544 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3545 (vec_extract<mode><V_elem_l>): ... this.
3546 (vec_extractv2di): Renamed to ...
3547 (vec_extractv2didi): ... this.
3548 (vec_init<mode>): Renamed to ...
3549 (vec_init<mode><V_elem_l>): ... this.
3550 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3551 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3552 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3553 Add element mode after gen_vec_extract* calls.
3554 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3555 (vec_init<mode><unitmode>): ... this.
3556 (vec_extract<mode>): Renamed to ...
3557 (vec_extract<mode><unitmode>): ... this.
3558 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3559 (vec_init<mode><unitmode>): ... this.
3560 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3561 (vec_initv2sfsf): ... this.
3562 (vec_extractv2sf): Renamed to ...
3563 (vec_extractv2sfsf): ... this.
3564 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3565 Add element mode after gen_vec_extract* calls.
3566 * config/mips/mips.md (unitmode): New mode iterator.
3567 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3568 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3569 * config/spu/spu.md (inner_l): New mode attribute.
3570 (vec_init<mode>): Renamed to ...
3571 (vec_init<mode><inner_l>): ... this.
3572 (vec_extract<mode>): Renamed to ...
3573 (vec_extract<mode><inner_l>): ... this.
3574 * config/sparc/sparc.md (veltmode): New mode iterator.
3575 (vec_init<VMALL:mode>): Renamed to ...
3576 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3577 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3578 (vec_initv2sisi): ... this.
3579 (vec_initv2sf): Renamed to ...
3580 (vec_initv2sfsf): ... this.
3581 (vec_extractv2sf): Renamed to ...
3582 (vec_extractv2sfsf): ... this.
3583 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3584 (vec_init<mode>): Renamed to ...
3585 (vec_init<mode><VEC_base_l>): ... this.
3586 (vec_extract<mode>): Renamed to ...
3587 (vec_extract<mode><VEC_base_l>): ... this.
3588 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3589 (vec_initv2sfsf): ... this.
3590 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3591 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3592 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3593 gen_vec_init* calls.
3595 2017-08-01 Richard Biener <rguenther@suse.de>
3597 PR tree-optimization/81297
3598 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3599 TREE_OVERFLOW from INTEGER_CSTs.
3601 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3603 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3605 2017-07-31 Carl Love <cel@us.ibm.com>
3607 * config/rs6000/rs6000-c: Add support for built-in functions
3608 vector signed char vec_xl_be (signed long long, signed char *);
3609 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3610 vector signed int vec_xl_be (signed long long, signed int *);
3611 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3612 vector signed long long vec_xl_be (signed long long, signed long long *);
3613 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3614 vector signed short vec_xl_be (signed long long, signed short *);
3615 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3616 vector double vec_xl_be (signed long long, double *);
3617 vector float vec_xl_be (signed long long, float *);
3618 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3619 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3620 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3622 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3623 (altivec_expand_builtin): Add switch statement to call
3624 altivec_expand_xl_be for each builtin.
3625 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3626 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3627 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3628 __builtin_vsx_le_be_v16qi.
3629 * doc/extend.texi: Update the built-in documentation file for the
3630 new built-in functions.
3632 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3635 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3637 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3639 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3641 * config.gcc: Add z14.
3642 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3643 CPU model numbers for z13s and z14.
3644 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3646 * config/s390/s390-opts.h (enum processor_type): Rename
3647 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3648 * config/s390/s390.c (processor_table): Add field for CPU name to
3649 be passed to Binutils.
3650 (s390_asm_output_machine_for_arch): Use the new field in
3651 processor_table for Binutils.
3652 (s390_expand_builtin): Replace arch12 with z14.
3653 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3654 (s390_get_sched_attrmask): Likewise.
3655 (s390_get_unit_mask): Likewise.
3656 * config/s390/s390.opt: Add z14 to processor_type enum.
3658 2017-07-31 Martin Jambor <mjambor@suse.cz>
3661 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3662 regardless of optimization level.
3664 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3665 Martin Liska <mliska@suse.cz>
3667 * predict.def: Remove old comment and adjust probability.
3668 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3671 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3674 * config/i386/i386.c (ix86_function_naked): New function.
3675 (ix86_can_use_return_insn_p): Return false for naked functions.
3676 (ix86_expand_prologue): Skip prologue for naked functions.
3677 (ix86_expand_epilogue): Skip epilogue for naked functions
3678 and emit trap instruction.
3679 (ix86_warn_func_return): New function.
3680 (ix86_attribute_table): Add "naked" attribute specification.
3681 (TARGET_WARN_FUNC_RETURN): Define.
3682 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3684 2017-07-31 Martin Liska <mliska@suse.cz>
3686 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3687 (dump_gimple_bb_header): Always dump BB info.
3688 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3690 2017-07-31 Martin Liska <mliska@suse.cz>
3693 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3694 also with current_function_decl non-null equality.
3696 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3699 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3700 change type to the element type, instead add eltype variable and
3701 use it where we are interested in the element type.
3703 PR tree-optimization/81603
3704 * ipa-polymorphic-call.c
3705 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3706 offset arithmetic in offset_int, bail out if the resulting bit offset
3707 doesn't fit into shwi.
3709 2017-07-31 Martin Liska <mliska@suse.cz>
3711 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3712 (gimplify_save_expr): Fix comment.
3714 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3717 * config/i386/i386.c (ix86_function_arg): Update arguments for
3719 (ix86_compute_frame_layout): Set the initial stack offset to
3720 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3721 INCOMING_FRAME_SP_OFFSET.
3722 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3723 stack before exception handler returns.
3724 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3725 the 'ERROR_CODE' for exception handler.
3727 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3729 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3730 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3731 (ASM_OUTPUT_REG_POP): Ditto.
3732 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3733 instead of asm_fprintf to output pure string.
3735 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3737 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3738 to imported_module_or_decl hook.
3739 (debug_nothing_tree_tree_tree_bool): Remove.
3740 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3741 * debug.c (do_nothing_debug_hooks): Use
3742 debug_nothing_tree_tree_tree_bool_bool instead of
3743 debug_nothing_tree_tree_tree_bool.
3744 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3745 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3746 * sdbout.c (sdb_debug_hooks): Likewise.
3747 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3748 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3750 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3751 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3752 attribute, don't add anything.
3754 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3756 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3757 (fold_build2_stat_loc): Likewise.
3758 (fold_build3_stat_loc): Likewise.
3759 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3760 (fold_build1_loc): Remove macro.
3761 (fold_build2_loc): Likewise.
3762 (fold_build3_loc): Likewise.
3764 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3766 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3767 (gimple_build_debug_bind_source_stat): Likewise.
3768 * gimple.h (gimple_build_debug_bind): Remove macro.
3769 (gimple_build_debug_bind_source): Likewise.
3771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3773 * bitmap.c (bitmap_alloc): Adjust.
3774 (bitmap_gc_alloc): Likewise.
3775 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3779 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3780 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3781 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
3782 (bitmap_gc_alloc_stat): Likewise.
3783 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
3785 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3787 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
3788 * rtl.h (shallow_copy_rtx): Remove macro.
3790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3792 * emit-rtl.c (gen_raw_REG): Adjust.
3793 * gengenrtl.c (gendef): Likewise.
3794 * rtl.c (rtx_alloc_stat): Remove _stat from name.
3795 * rtl.h (rtx_alloc): Remove macro.
3797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3799 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
3800 (build_tree_list_stat): Likewise.
3801 * tree.h (build_tree_list): Remove macro.
3802 (build_tree_list_vec): Likewise.
3804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3806 * tree.c (make_vector_stat): Remove _stat from name.
3807 (build_vector_stat): Likewise.
3808 * tree.h (make_vector_stat): Remove macro.
3809 (build_vector_stat): Likewise.
3811 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3813 * tree.h (build_var_debug_value): Remove prototype.
3815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3817 * tree.c (tree_cons_stat): Remove _stat from name.
3818 * tree.h (tree_cons): Remove macro.
3820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3822 * tree.c (build_vl_exp_stat): Remove _stat from name.
3823 * tree.h (build_vl_exp): Remove macro.
3825 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3827 * tree.c (build_decl_stat): Remove _stat from name.
3828 * tree.h (build_decl): Remove macro.
3830 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3832 * gimple.c (gimple_build_with_ops_stat): Adjust.
3833 (gimple_alloc_stat): Remove _stat from name.
3834 * gimple.h (gimple_alloc): Remove macro.
3836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3838 * tree.c (make_tree_vec_stat): Remove _stat from name.
3839 (grow_tree_vec_stat): Likewise.
3840 * tree.h (make_tree_vec_stat): Adjust prototype.
3841 (grow_tree_vec_stat): Likewise.
3842 (make_tree_vec): Remove macro.
3843 (grow_tree_vec): Likewise.
3845 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3847 * fold-const.c (fold_build1_stat_loc): Adjust.
3848 (fold_build2_stat_loc): Likewise.
3849 (fold_build3_stat_loc): Likewise.
3850 * tree.c (build0_stat): Remove _stat from name.
3851 (build1_stat): Likewise.
3852 (build2_stat): Likewise.
3853 (build3_stat): Likewise.
3854 (build4_stat): Likewise.
3855 (build5_stat): Likewise.
3856 * tree.h (build1_loc): Remove macro, and rename _stat function
3858 (build2_loc): Likewise.
3859 (build3_loc): Likewise.
3860 (build4_loc): Likewise.
3861 (build5_loc): Likewise.
3863 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3865 * tree.c (make_int_cst_stat): Remove _stat from name.
3866 * tree.h (make_int_cst_stat): Adjust prototype.
3867 (make_int_cst): Remove macro.
3869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3871 * tree.c (make_tre_binfo_stat): Remove _stat from name.
3872 * tree.h (make_tree_binfo_stat): Adjust prototype.
3873 (make_tree_binfo): Remove.
3875 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3877 * tree.c (copy_node_stat): Rename to copy_node.
3878 (build_distinct_type_copy): Adjust.
3879 * tree.h (copy_node_stat): Adjust prototype.
3880 (copy_node): Remove macro.
3882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3884 * tree.c (make_node_stat): rename to make_node.
3885 (build_tree_list_stat): Adjust.
3886 (build0_stat): Likewise.
3887 (build2_stat): Likewise.
3888 (build3_stat): Likewise.
3889 (build4_stat): Likewise.
3890 (build5_stat): Likewise.
3891 (build_decl_stat): Likewise.
3892 * tree.h (make_node_stat): Adjust prototype.
3893 (make_node): remove macro.
3895 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
3897 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
3898 (PPC_FEATURE2_SCV): Likewise.
3899 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
3901 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3903 * config/aarch64/aarch64.c
3904 (aarch64_internal_mov_immediate): Add new special pattern.
3905 * config/aarch64/aarch64.md (*movdi_aarch64):
3906 Add reg/32bit const mov case.
3908 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3909 Richard Sandiford <richard.sandiford@linaro.org>
3911 * config/aarch64/aarch64.md (mov<mode>): Generalize.
3912 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3913 Add integer and movi cases.
3914 (movi-split-hf-df-sf split, fp16): New.
3915 (enabled): Added TARGET_FP_F16INST.
3916 * config/aarch64/iterators.md (GPF_HF): New.
3917 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
3919 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3921 * config/aarch64/aarch64.c
3922 (aarch64_simd_container_mode): Add prototype.
3923 (aarch64_expand_mov_immediate): Add HI support.
3924 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
3925 (aarch64_can_const_movi_rtx_p): New.
3926 (aarch64_preferred_reload_class):
3927 Remove restrictions of using FP registers for certain SIMD operations.
3928 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
3929 (aarch64_valid_floating_const): Add integer move validation.
3930 (aarch64_simd_imm_scalar_p): Remove.
3931 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
3932 (aarch64_legitimate_constant_p): Expand list of supported cases.
3933 * config/aarch64/aarch64-protos.h
3934 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
3935 (aarch64_reinterpret_float_as_int): New.
3936 (aarch64_simd_imm_scalar_p): Remove.
3937 * config/aarch64/constraints.md (Uvi): New.
3938 (Dd): Split into Ds and new Dd.
3939 * config/aarch64/aarch64.md (*movsi_aarch64):
3941 (*movdi_aarch64): Add SIMD mov case.
3943 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3945 * tree-predcom.c: (struct chain): Handle store-store chain in which
3946 stores for elimination only store loop invariant values.
3947 (execute_pred_commoning_chain): Ditto.
3948 (prepare_initializers_chain_store_elim): Ditto.
3949 (prepare_finalizers): Ditto.
3950 (is_inv_store_elimination_chain): New function.
3951 (initialize_root_vars_store_elim_1): New function.
3953 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3955 * tree-predcom.c: Revise general description of the pass.
3956 (enum chain_type): New enum type for store elimination.
3957 (struct chain): New field supporting store elimination.
3958 (struct component): Ditto.
3959 (dump_chain): Dump store-stores chain.
3960 (release_chain): Release resources.
3961 (split_data_refs_to_components): Compute and create component
3962 contains only stores for elimination.
3963 (get_chain_last_ref_at): New function.
3964 (make_invariant_chain): Initialization.
3965 (make_rooted_chain): Specify chain type in parameter and record it.
3966 (add_looparound_copies): Skip for store-stores chain.
3967 (determine_roots_comp): Compute type of chain and pass it to
3969 (initialize_root_vars_store_elim_2): New function.
3970 (finalize_eliminated_stores): New function.
3971 (remove_stmt): Handle store for elimination.
3972 (execute_pred_commoning_chain): Execute predictive commoning on
3974 (determine_unroll_factor): Skip unroll for store-stores chain.
3975 (prepare_initializers_chain_store_elim): New function.
3976 (prepare_initializers_chain): Hanlde store-store chain.
3977 (prepare_finalizers_chain, prepare_finalizers): New function.
3978 (tree_predictive_commoning_loop): Return integer value indicating
3979 if loop is unrolled or lcssa form is corrupted.
3980 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
3982 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3984 * tree-predcom.c (initialize_root): Delete.
3985 (execute_pred_commoning_chain): Initialize root vars and replace
3986 reference of non-combined chain directly, rather than call above
3989 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3991 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
3992 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
3994 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3996 * tree-predcom.c (struct chain): New field init_seq.
3997 (release_chain): Release init_seq.
3998 (prepare_initializers_chain): Record intialization stmts in above
4000 (insert_init_seqs): New function.
4001 (tree_predictive_commoning_loop): Call insert_init_seqs.
4003 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4005 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4007 2017-07-28 Richard Biener <rguenther@suse.de>
4009 * match.pd: Remove superfluous :c.
4010 * genmatch.c (simplify::id): Add member.
4011 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4013 (current_id): New global.
4014 (dt_node::parent): Move from ...
4015 (dt_operand::parent): ... here. Add for_id member.
4016 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4017 (decision_tree::find_node): Relax order requirement when
4018 merging DT_TRUE nodes to ones inbetween the current simplify
4019 and the one we try to merge with. Add diagnostic whenever
4020 we need to enforce pattern order by not merging.
4021 (decision_tree::insert): Set current_id.
4022 (decision_tree::print_node): Dump parent node and for_id.
4023 (parser::last_id): Add member.
4024 (parser::push_simplify): Assign unique id.
4025 (parser::parser): Initialize last_id.
4027 2017-07-28 Martin Liska <mliska@suse.cz>
4030 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4031 gimple_build_debug_bind.
4033 2017-07-28 Richard Biener <rguenther@suse.de>
4035 PR tree-optimization/81502
4036 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4038 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4040 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4041 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4042 for BIT_FIELD_REF args.
4043 * fold-const.c (make_bit_field_ref): Likewise.
4044 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4046 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4049 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4050 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4051 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4052 Or it into SANITIZER_UNDEFINED.
4053 * ubsan.c: Include gimple-fold.h and varasm.h.
4054 (ubsan_expand_ptr_ifn): New function.
4055 (instrument_pointer_overflow): New function.
4056 (maybe_instrument_pointer_overflow): New function.
4057 (instrument_object_size): Formatting fix.
4058 (pass_ubsan::execute): Call instrument_pointer_overflow
4059 and maybe_instrument_pointer_overflow.
4060 * internal-fn.c (expand_UBSAN_PTR): New function.
4061 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4062 * sanitizer.def (__ubsan_handle_pointer_overflow,
4063 __ubsan_handle_pointer_overflow_abort): New builtins.
4064 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4065 * internal-fn.def (UBSAN_PTR): New internal function.
4066 * opts.c (sanitizer_opts): Add pointer-overflow.
4067 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4068 * fold-const.c (build_range_check): Compute pointer range check in
4069 integral type if pointer arithmetics would be needed. Formatting
4072 2017-07-28 Martin Liska <mliska@suse.cz>
4075 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4076 parameters that are of a variable-length.
4078 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4080 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4082 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4083 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4084 (CRT_CALL_STATIC_FUNCTION): Likewise.
4085 (ASM_DEFAULT_SPEC): New define.
4086 (ASM_SPEC32): Likewise.
4087 (ASM_SPEC64): Likewise.
4088 (ASM_SPEC_COMMON): Likewise.
4089 (ASM_SPEC): Likewise.
4090 (INVALID_64BIT): Likewise.
4091 (LINK_OS_DEFAULT_SPEC): Likewise.
4092 (LINK_OS_SPEC32): Likewise.
4093 (LINK_OS_SPEC64): Likewise.
4094 (POWERPC_LINUX): Likewise.
4095 (PTRDIFF_TYPE): Likewise.
4096 (RESTORE_FP_PREFIX): Likewise.
4097 (RESTORE_FP_SUFFIX): Likewise.
4098 (SAVE_FP_PREFIX): Likewise.
4099 (SAVE_FP_SUFFIX): Likewise.
4100 (SIZE_TYPE): Likewise.
4101 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4102 (TARGET_64BIT): Likewise.
4103 (TARGET_64BIT): Likewise.
4104 (TARGET_AIX): Likewise.
4105 (WCHAR_TYPE_SIZE): Likewise.
4106 (WCHAR_TYPE): Undefine.
4107 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4108 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4109 (CPP_OS_RTEMS_SPEC): Delete.
4110 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4111 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4113 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4115 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4117 PR tree-optimization/81578
4118 * tree-parloops.c (build_new_reduction): Bail out if
4119 reduction_code isn't one of the standard OpenMP reductions.
4120 Move the details printing after that decision.
4122 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4124 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4125 related to reload_in_progress.
4126 (splat_input_operand): Likewise.
4127 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4129 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4131 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4132 (TARGET_INSTANTIATE_DECLS): Likewise.
4133 (legitimate_indexed_address_p): Delete reload_in_progress code.
4134 (rs6000_debug_legitimate_address_p): Likewise.
4135 (rs6000_eliminate_indexed_memrefs): Likewise.
4136 (rs6000_emit_le_vsx_store): Likewise.
4137 (rs6000_emit_move_si_sf_subreg): Likewise.
4138 (rs6000_emit_move): Likewise.
4139 (register_to_reg_type): Likewise.
4140 (rs6000_pre_atomic_barrier): Likewise.
4141 (rs6000_machopic_legitimize_pic_address): Likewise.
4142 (rs6000_allocate_stack_temp): Likewise.
4143 (rs6000_address_for_fpconvert): Likewise.
4144 (rs6000_address_for_altivec): Likewise.
4145 (rs6000_secondary_memory_needed_rtx): Delete function.
4146 (rs6000_check_sdmode): Likewise.
4147 (rs6000_alloc_sdmode_stack_slot): Likewise.
4148 (rs6000_instantiate_decls): Likewise.
4149 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4150 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4151 Delete reload_in_progress.
4152 (*vec_reload_and_plus_<mptrsize>): Likewise.
4153 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4154 (vsx_div_v2di): Likewise.
4155 (vsx_udiv_v2di): Likewise.
4157 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4159 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4160 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4161 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4162 (rs6000_debug_reg_global): Delete print of LRA status.
4163 (rs6000_option_override_internal): Delete dead LRA related code.
4164 (rs6000_lra_p): Delete function.
4165 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4167 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4169 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4170 * config/riscv/rtems.h: New file.
4172 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4173 Sudakshina Das <sudi.das@arm.com>
4175 * config/aarch64/aarch64.md
4176 (define_split for and<mode>3nr_compare): Move
4177 non aarch64_logical_operand to a register.
4178 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4179 register immediate operand to a register.
4180 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4182 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4185 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4187 2017-07-27 Richard Biener <rguenther@suse.de>
4189 PR tree-optimization/81573
4190 PR tree-optimization/81494
4191 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4192 multi defuse cycle case.
4194 2017-07-27 Richard Biener <rguenther@suse.de>
4196 PR tree-optimization/81571
4197 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4200 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4202 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4203 earlier and only if MASK_FPU is set. Adjust formatting.
4205 2017-07-27 Martin Liska <mliska@suse.cz>
4207 * opt-functions.awk: Add validation of value of Init.
4208 * optc-gen.awk: Pass new argument.
4210 2017-07-27 Martin Liska <mliska@suse.cz>
4212 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4213 Fix wrong condition.
4215 2017-07-27 Martin Liska <mliska@suse.cz>
4217 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4218 BBs and edges seen by autoFDO.
4220 2017-07-27 Richard Biener <rguenther@suse.de>
4222 PR tree-optimization/81502
4223 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4224 with incompatible but same sized type.
4225 (execute_update_addresses_taken): Likewise.
4227 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4229 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4230 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4232 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4235 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4236 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4237 Change s_operand to memory_operand.
4239 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4241 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4242 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4243 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4244 Emit instructions rather than returning an expression. Handle TFmode
4245 and KFmode by casting to TImode.
4246 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4247 (rs6000_emit_le_vsx_store): Likewise.
4248 * config/rs6000/vsx.md (VSX_TI): New iterator.
4249 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4250 (*vsx_le_undo_permute_<mode>): Likewise.
4251 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4252 emit the split sequence.
4253 (*vsx_le_perm_store_<mode>): Likewise.
4255 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4257 PR tree-optimization/81555
4258 PR tree-optimization/81556
4259 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4260 if true, force CHANGED for the recursive invocation.
4261 (reassociate_bb): Remember original length of ops array, pass
4262 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4264 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4265 attributes for noipa attribute. For naked attribute use
4266 lookup_attribute first before lookup_attribute_spec.
4267 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4269 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4270 for functions with noipa attribute.
4271 (cgraph_externally_visible_p): Return true for functions with noipa
4273 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4274 for functions with noipa attribute.
4275 * doc/extend.texi: Document noipa function attribute.
4276 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4277 also for functions with noipa attribute.
4278 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4280 2017-07-26 Andrew Pinski <apinski@cavium.com>
4282 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4283 vec_unalign_load_cost and vec_unalign_store_cost.
4285 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4287 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4288 -mvsx-small-integer option.
4289 (ISA_3_0_MASKS_IEEE): Likewise.
4290 (OTHER_VSX_VECTOR_MASKS): Likewise.
4291 (POWERPC_MASKS): Likewise.
4292 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4293 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4294 code, only testing for DImode being allowed in non-VSX floating
4296 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4297 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4299 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4300 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4301 TARGET_P8_VECTOR test.
4302 (rs6000_secondary_reload_simple_move): Likewise.
4303 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4304 since TARGET_P9_VECTOR was already tested.
4305 (rs6000_opt_masks): Remove -mvsx-small-integer.
4306 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4307 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4309 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4310 test for TARGET_VEXTRACTUB was used, and that uses
4312 (p9 extract splitter): Likewise.
4313 (vsx_extract_<mode>_di_p9): Likewise.
4314 (vsx_extract_<mode>_store_p9): Likewise.
4315 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4316 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4317 the elimination of TARGET_VSX_SMALL_INTEGER.
4318 (vsx_extract_<mode>_p8): Likewise.
4319 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4320 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4321 (vsx_set_<mode>_p9): Likewise.
4322 (vsx_set_v4sf_p9): Likewise.
4323 (vsx_set_v4sf_p9_zero): Likewise.
4324 (vsx_insert_extract_v4sf_p9): Likewise.
4325 (vsx_insert_extract_v4sf_p9_2): Likewise.
4326 * config/rs6000/rs6000.md (sign extend splitter): Change
4327 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4328 (floatsi<mode>2_lfiwax_mem): Likewise.
4329 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4330 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4331 since a test for TARGET_P9_VECTOR was used.
4332 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4333 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4334 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4335 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4336 TARGET_P8_VECTOR test.
4337 (fix_trunc<mode>si2_stfiwx): Likewise.
4338 (fix_trunc<mode>si2_internal): Likewise.
4339 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4340 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4342 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4343 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4344 TARGET_P8_VECTOR test.
4345 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4346 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4347 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4349 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4350 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4351 since a test for TARGET_P9_VECTOR was used.
4352 (splitter for loading small constants): Likewise.
4354 2017-07-26 Andrew Pinski <apinski@cavium.com>
4356 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4359 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4362 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4363 (fp_valid_at): Likewise.
4365 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4367 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4368 (qdf24xx_addrcost_table): Likewise.
4369 (cortexa57_tunings): Update to use generic_branch_cost.
4370 (cortexa72_tunings): Likewise.
4371 (cortexa73_tunings): Likewise.
4372 (qdf24xx_tunings): Likewise.
4374 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4376 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4377 (thunderx2t99_branch_cost): Likewise.
4378 (cortexa35_tunings): Update to use generic_branch_cost.
4379 (cortexa53_tunings): Likewise.
4380 (cortexa57_tunings): Likewise.
4381 (cortexa72_tunings): Likewise.
4382 (cortexa73_tunings): Likewise.
4383 (thunderx2t99_tunings): Likewise.
4385 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4387 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4388 (sparc_option_override): Honour MASK_FSMULD.
4389 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4390 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4391 * config/sparc/sparc.opt (mfsmuld): New option.
4392 * doc/invoke.texi (mfsmuld): Document option.
4394 2017-07-26 Marek Polacek <polacek@redhat.com>
4397 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4399 2017-07-26 Richard Biener <rguenther@suse.de>
4401 * gimple-match-head.c (do_valueize): Return OP if valueize
4403 (get_def): New helper to get at the def stmt of a SSA name
4405 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4406 do_valueize to get at the def stmt.
4407 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4409 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4412 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4414 2017-07-26 Martin Liska <mliska@suse.cz>
4417 * function.c (expand_function_start): Make expansion of
4418 nonlocal_goto_save_area after parm_birth_insn.
4420 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4422 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4423 from all CPU target flags enable members.
4425 2017-07-26 Richard Biener <rguenther@suse.de>
4427 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4428 (decision_tree::gen): Make 'type' const.
4429 (write_predicate): Likewise.
4431 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4433 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4434 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4435 (rs6000_option_override_internal): Likewise.
4436 (rs6000_expand_vector_set): Likewise.
4437 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4438 (TARGET_UPPER_REGS_SF): Likewise.
4439 (TARGET_UPPER_REGS_DI): Likewise.
4440 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4441 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4442 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4443 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4444 (Splitters for DI constants in Altivec registers): Likewise.
4445 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4446 (vsx_set_v4sf_p9): Likewise.
4447 (vsx_set_v4sf_p9_zero): Likewise.
4448 (vsx_insert_extract_v4sf_p9): Likewise.
4449 (vsx_insert_extract_v4sf_p9_2): Likewise.
4451 2017-07-25 Carl Love <cel@us.ibm.com>
4453 * doc/extend.texi: Update the built-in documentation file for the
4454 existing built-in functions
4455 vector signed char vec_cnttz (vector signed char);
4456 vector unsigned char vec_cnttz (vector unsigned char);
4457 vector signed short vec_cnttz (vector signed short);
4458 vector unsigned short vec_cnttz (vector unsigned short);
4459 vector signed int vec_cnttz (vector signed int);
4460 vector unsigned int vec_cnttz (vector unsigned int);
4461 vector signed long long vec_cnttz (vector signed long long);
4462 vector unsigned long long vec_cnttz (vector unsigned long long);
4464 2017-07-25 Andrew Pinski <apinski@cavium.com>
4466 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4467 accesses where the use is for the first operand of a BIT_INSERT.
4469 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4472 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4473 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4475 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4477 * config/i386/gstabs.h: Delete.
4478 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4480 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4482 * config/i386/i386.c (ix86_decompose_address): Do not check for
4483 register RTX when looking at index_reg or base_reg.
4484 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4486 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4488 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4489 to update EH info here.
4491 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4493 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4495 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4497 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4499 2017-07-25 Torsten Duwe <duwe@suse.de>
4501 * common.opt: Introduce -fpatchable-function-entry
4502 command line option, and its variables function_entry_patch_area_size
4503 and function_entry_patch_area_start.
4504 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4505 including a two-value parser.
4506 * target.def (print_patchable_function_entry): New target hook.
4507 * targhooks.h (default_print_patchable_function_entry): New function.
4508 * targhooks.c (default_print_patchable_function_entry): Likewise.
4509 * toplev.c (process_options): Switch off IPA-RA if
4510 patchable function entries are being generated.
4511 * varasm.c (assemble_start_function): Look at the
4512 patchable-function-entry command line switch and current
4513 function attributes and maybe generate NOP instructions by
4514 calling the print_patchable_function_entry hook.
4515 * doc/extend.texi: Document patchable_function_entry attribute.
4516 * doc/invoke.texi: Document -fpatchable_function_entry
4517 command line option.
4518 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4520 * doc/tm.texi: Re-generate.
4522 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4525 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4526 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4528 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4530 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4532 2017-07-25 Richard Biener <rguenther@suse.de>
4534 PR tree-optimization/81455
4535 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4536 not walk in cycles when looking for guards.
4538 2017-07-25 Richard Biener <rguenther@suse.de>
4540 PR tree-optimization/81529
4541 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4542 when optimizing backedge uses.
4544 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4546 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4548 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4549 to dl_section_ref. On AIX, append an expression to subtract
4550 the size of the section length to dl_section_ref.
4552 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4554 * configure.ac: If any of the config.* scripts fail, exit 1.
4555 * configure: Regenerate.
4557 2017-07-25 Richard Biener <rguenther@suse.de>
4560 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4561 of immediate uses, be more verbose on errors.
4563 2017-07-25 Richard Biener <rguenther@suse.de>
4565 PR tree-optimization/81510
4566 * tree-vect-loop.c (vect_is_simple_reduction): When the
4567 reduction stmt is not inside the loop bail out.
4569 2017-07-25 Richard Biener <rguenther@suse.de>
4571 PR tree-optimization/81303
4572 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4573 profitability check against LOOP_VINFO_NITERSM1.
4575 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4577 * domwalk.c (cmp_bb_postorder): Simplify.
4578 (sort_bbs_postorder): New function. Use it...
4579 (dom_walker::walk): ...here to optimize common cases.
4581 2017-07-25 Martin Liska <mliska@suse.cz>
4584 * ipa-visibility.c (function_and_variable_visibility): Make the
4585 redirection just on target that supports aliasing.
4586 Fix GNU coding style.
4588 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4591 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4593 * config/arm/rtems.h: Likewise.
4594 * config/bfin/rtems.h: Likewise.
4595 * config/i386/rtemself.h: Likewise.
4596 * config/lm32/rtems.h: Likewise.
4597 * config/m32c/rtems.h: Likewise.
4598 * config/m68k/rtemself.h: Likewise.
4599 * config/microblaze/rtems.h: Likewise.
4600 * config/mips/rtems.h: Likewise.
4601 * config/moxie/rtems.h: Likewise.
4602 * config/nios2/rtems.h: Likewise.
4603 * config/powerpcspe/rtems.h: Likewise.
4604 * config/rs6000/rtems.h: Likewise.
4605 * config/rtems.h: Likewise.
4606 * config/sh/rtems.h: Likewise.
4607 * config/sh/rtemself.h: Likewise.
4608 * config/sparc/rtemself.h: Likewise.
4610 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4613 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4614 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4615 * tree-ssa-structalias.c (alias_get_name): Same.
4617 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4620 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4621 instructions if no du chain is found.
4623 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4625 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4627 2017-07-25 Richard Biener <rguenther@suse.de>
4630 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4633 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4635 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4637 (ISA_2_7_MASKS_SERVER): Likewise.
4638 (ISA_3_0_MASKS_IEEE): Likewise.
4639 (OTHER_P8_VECTOR_MASKS): Likewise.
4640 (OTHER_VSX_VECTOR_MASKS): Likewise.
4641 (POWERPC_MASKS): Likewise.
4642 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4643 duplicate list of options.
4644 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4645 explicit -mupper-regs options.
4646 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4647 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4648 alias for -mupper-regs-df.
4649 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4650 (rs6000_init_hard_regno_mode_ok): Likewise.
4651 (rs6000_option_override_internal): Likewise.
4652 (rs6000_opt_masks): Likewise.
4653 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4654 options in terms of whether -mvsx or -mpower8-vector was used.
4655 (TARGET_UPPER_REGS_DI): Likewise.
4656 (TARGET_UPPER_REGS_SF): Likewise.
4657 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4658 -mupper-regs-* options.
4660 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4662 * passes.c (emergency_dump_function): Print some empty lines and a
4663 header before the RTL dump.
4665 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4667 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4669 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4672 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4673 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4675 2017-07-24 Carl Love <cel@us.ibm.com>
4677 * config/rs6000/rs6000-c.c: Add support for built-in functions
4678 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4679 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4680 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4681 vec_extract_fp_from_shortl): Add defines for the two builtins.
4682 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4683 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4685 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4686 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4687 * doc/extend.texi: Update the built-in documentation file for the
4688 new built-in function.
4690 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4693 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4695 * doc/generic.texi: Likewise.
4696 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4697 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4699 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4701 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4702 (aarch64_mls_elt_merge<mode>): Likewise.
4704 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4706 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4707 having __cxa_atexit.
4709 2017-07-23 Michael Collison <michael.collison@arm.com>
4711 * config/arm/arm.c (arm_option_override): Deprecate
4712 use of -mstructure-size-boundary.
4713 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4714 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4716 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4719 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4720 Reduce cost estimate for direct moves.
4722 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4725 * config/i386/i386.c (ix86_option_override_internal): Disable
4726 BMI, BMI2 and TBM instructions for -m16.
4728 2017-07-21 Carl Love <cel@us.ibm.com>
4730 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4731 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4732 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4733 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4734 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4735 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4736 VMULOSW): New enum "unspec" values.
4737 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4738 altivec_vmulosw): New patterns.
4739 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4740 VMULOSW): Add definitions.
4742 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4744 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4745 (qdf24xx): Likewise.
4746 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4747 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4748 (AARCH64_FL_V8_1): Renumber.
4749 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4750 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4751 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4752 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4753 rdma to feature modifiers list.
4755 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4758 * ipa-visibility (function_and_variable_visibility): Convert
4759 recursive PLT call to direct call if appropriate.
4761 2017-07-21 Andrew Pinski <apinski@cavium.com>
4763 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4764 operand 1 to see if the types precision matches.
4765 * fold-const.c (operand_equal_p): Likewise.
4767 2017-07-21 Richard Biener <rguenther@suse.de>
4769 PR tree-optimization/81303
4770 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4771 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4772 (vect_peeling_hash_get_lowest_cost): Adjust.
4773 (vect_enhance_data_refs_alignment): Likewise. Use
4774 vect_get_peeling_costs_all_drs to compute the penalty for no
4775 peeling to match up costs.
4777 2017-07-21 Richard Biener <rguenther@suse.de>
4779 PR tree-optimization/81500
4780 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4781 we didn't identify a reduction path.
4783 2017-07-21 Tom de Vries <tom@codesourcery.com>
4784 Cesar Philippidis <cesar@codesourcery.com>
4786 PR gcov-profile/81442
4787 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
4790 2017-07-21 Tom de Vries <tom@codesourcery.com>
4793 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
4795 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
4796 nvptx_override_options_after_change.
4798 2017-07-21 Ulrich Drepper <drepper@redhat.com>
4800 * dwarf2out.c (output_file_names): Avoid double testing for
4803 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
4805 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
4807 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4809 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
4811 (try_crossjump_to_edge): Do not punt on partitioned functions.
4813 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4815 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4816 Put all BBs reachable only via paths crossing cold region to cold
4818 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
4820 2016-07-21 Richard Biener <rguenther@suse.de>
4822 PR tree-optimization/81303
4823 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
4824 into account prologue and epilogue iterations when raising
4825 min_profitable_iters to sth at least covering one vector iteration.
4827 2017-07-21 Tamar Christina <tamar.christina@arm.com>
4829 * config/arm/arm.c (arm_test_cpu_arch_dat):
4832 2017-07-20 Nathan Sidwell <nathan@acm.org>
4834 Remove TYPE_METHODS.
4835 * tree.h (TYPE_METHODS): Delete.
4836 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
4837 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
4838 (dbxout_type_methods): Scan TYPE_FIELDS.
4839 (dbxout_type): Don't check TYPE_METHODS here.
4840 * function.c (use_register_for_decl): Always ignore register for
4841 class types when not optimizing.
4842 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
4843 * tree.c (free_lang_data_in_type): Stitch out member functions and
4844 templates from TYPE_FIELDS.
4845 (build_distinct_type_copy, verify_type_variant,
4846 verify_type): Member fns are on TYPE_FIELDS.
4847 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
4848 * tree-pretty-print.c (dump_generic_node): Likewise.
4850 2017-07-20 Jakub Jelinek <jakub@redhat.com>
4853 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
4854 V2TImode and V4TImode.
4855 (ix86_expand_vector_extract): Likewise.
4856 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
4857 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
4858 (ssescalarmode): Handle V4TImode and V2TImode.
4859 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
4860 (*vec_extractv2ti, *vec_extractv4ti): New insns.
4861 (VEXTRACTI128_MODE): New mode iterator.
4862 (splitter for *vec_extractv?ti first element): New.
4863 (VEC_INIT_MODE): New mode iterator.
4864 (vec_init<mode>): Consolidate 3 expanders into one using
4865 VEC_INIT_MODE mode iterator.
4867 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4869 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
4870 non_spilled_static_chain_regno_p.
4872 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4874 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
4876 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
4878 * bb-reorder.c (connect_traces): Allow copying of blocks within
4881 2017-07-20 Richard Biener <rguenther@suse.de>
4883 * gimple.h (gimple_phi_result): Add gphi * overload.
4884 (gimple_phi_result_ptr): Likewise.
4885 (gimple_phi_arg): Likewise. Adjust index assert to only
4886 allow actual argument accesses rather than all slots available
4888 (gimple_phi_arg_def): Add gphi * overload.
4889 * tree-phinodes.c (make_phi_node): Initialize only actual
4891 (resize_phi_node): Clear memory not covered by old node,
4892 do not initialize excess argument slots.
4893 (reserve_phi_args_for_new_edge): Initialize new argument slot
4896 2017-07-20 Bin Cheng <bin.cheng@arm.com>
4898 PR tree-optimization/81388
4900 2016-07-21 Bin Cheng <bin.cheng@arm.com>
4902 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
4903 by removing computation of may_be_zero.
4905 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4906 Tom de Vries <tom@codesourcery.com>
4909 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
4910 when gimple level profile disagrees with what RTL expander did.
4912 2017-07-20 Richard Biener <rguenther@suse.de>
4914 PR tree-optimization/61171
4915 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
4916 (vect_analyze_stmt): Add slp instance parameter.
4917 (vectorizable_reduction): Likewise.
4918 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
4919 (vect_is_simple_reduction): Deal with chains not detected
4920 as SLP reduction chain, specifically not properly associated
4921 chains containing a mix of plus/minus.
4922 (get_reduction_op): Remove.
4923 (get_initial_defs_for_reduction): Simplify, pass in whether
4924 this is a reduction chain, pass in the SLP node for the PHIs.
4925 (vect_create_epilog_for_reduction): Get the SLP instance as
4927 (vectorizable_reduction): Get the SLP instance as arg.
4928 During analysis remember the SLP node with the PHIs in the
4929 instance. Simplify getting at the vectorized reduction PHIs.
4930 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
4931 through SLP instance.
4932 (vect_slp_analyze_operations): Likewise.
4933 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
4934 (vect_transform_stmt): Likewise.
4936 2017-07-20 Tom de Vries <tom@codesourcery.com>
4938 PR tree-optimization/81489
4939 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
4940 read of phi arg location to before loop that modifies phi.
4942 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4944 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
4947 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4950 * except.c (execute): Fix ordering issue.
4952 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4954 PR rtl-optimization/81423
4955 * combine.c (make_compound_operation_int): Don't try to optimize
4956 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
4958 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4960 PR rtl-optimization/81423
4961 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
4962 with a constant that is -1 in the truncated to mode.
4964 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4966 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
4967 (determine_unlikely_bbs): ... here.
4968 * predict.h (propagate_unlikely_bbs_forward): Declare.
4969 * cfgexpand.c (pass_expand::execute): Use it.
4970 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
4972 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
4973 propagate_unlikely_bbs_forward.
4975 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4978 * except.c (maybe_add_nop_after_section_switch): New function.
4981 2017-07-19 Tom de Vries <tom@codesourcery.com>
4983 * gimple.h (gimple_phi_set_arg): Make assert more strict.
4985 2017-07-19 Tom de Vries <tom@codesourcery.com>
4987 * gimple.h (gimple_phi_arg): Make assert more strict.
4989 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
4991 * config.gcc (powerpc*-*-*): Add mmintrin.h.
4992 * config/rs6000/mmintrin.h: New file.
4993 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
4995 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4997 PR tree-optimization/81346
4998 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5000 2017-07-19 Tom de Vries <tom@codesourcery.com>
5002 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5004 2017-07-19 Tom de Vries <tom@codesourcery.com>
5006 * config/nvptx/nvptx-modes.def: Add V2DImode.
5007 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5008 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5009 (nvptx_output_mov_insn): Handle lack of mov.b128.
5010 (nvptx_print_operand): Handle 'H' and 'L' codes.
5011 (nvptx_vector_mode_supported): Allow V2DImode.
5012 (nvptx_preferred_simd_mode): New function.
5013 (nvptx_data_alignment): New function.
5014 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5015 nvptx_preferred_simd_mode.
5016 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5018 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5020 2017-07-19 Tom de Vries <tom@codesourcery.com>
5022 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5023 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5024 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5025 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5026 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5027 (mov<VECIM>_insn): New define_insn.
5028 (define_expand "mov<VECIM>): New define_expand.
5030 2017-07-19 Tom de Vries <tom@codesourcery.com>
5032 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5034 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5036 PR tree-optimization/81346
5037 * fold-const.h (fold_div_compare, range_check_type): Declare.
5038 * fold-const.c (range_check_type): New function.
5039 (build_range_check): Use range_check_type.
5040 (fold_div_compare): No longer static, rewritten into
5041 a match.pd helper function.
5042 (fold_comparison): Don't call fold_div_compare here.
5043 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5046 2017-07-19 Nathan Sidwell <nathan@acm.org>
5048 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5049 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5050 * tree.c (find_decls_types_r, verify_type): Use
5051 TYPE_{MIN,MAX}_VALUE_RAW.
5052 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5053 (hash_tree): Likewise.
5054 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5056 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5059 2017-07-18 Tom de Vries <tom@codesourcery.com>
5062 * omp-expand.c (expand_omp_for_static_chunk): Handle
5063 equal-argument loop exit phi.
5065 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5068 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5069 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5070 operand 2 predicate.
5071 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5072 operand 2 predicate.
5073 (ror,rol -> rorx splitters): Use const_int_operand as
5074 operand 2 predicate.
5076 2017-06-18 Richard Biener <rguenther@suse.de>
5078 PR tree-optimization/81410
5079 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5080 the gap in the ! slp_perm SLP case after each group.
5082 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5085 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5088 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5091 * predict.c (set_even_probabilities): Cleanup; do not affect
5092 probabilities that are already known.
5093 (combine_predictions_for_bb): Call even when count is set.
5095 2017-07-18 Nathan Sidwell <nathan@acm.org>
5097 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5100 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5103 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5104 optimization for loop niter analysis.
5106 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5109 * config/avr/avr.c (avr_optimize_casesi): Don't use
5110 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5112 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5114 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5115 body_cost_vec from _vect_peel_extended_info.
5116 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5117 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5120 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5122 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5124 2017-07-18 Richard Biener <rguenther@suse.de>
5126 PR tree-optimization/80620
5127 PR tree-optimization/81403
5128 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5129 info when re-using a VN table entry.
5131 2017-07-18 Richard Biener <rguenther@suse.de>
5133 PR tree-optimization/81418
5134 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5135 vectype_in. Verify that with lane-reducing reduction operations
5136 we have a single def-use cycle.
5138 2017-07-17 Carl Love <cel@us.ibm.com>
5140 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5142 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5143 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5144 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5145 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5146 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5147 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5148 VMULOSW): New enum "unspec" values.
5149 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5150 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5151 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5152 altivec_vmulosw): New patterns.
5153 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5154 VMULOSW): Add definitions.
5155 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5157 * config/alpha/alpha.c: Include predict.h.
5159 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5161 * tree-vrp.c (compare_assert_loc): Fix comparison function
5162 to return predictable results.
5164 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5166 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5169 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5170 * doc/invoke.texi (mexpand-adddi): Update text.
5172 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5174 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5175 that also clobbers the CC register. The old expand code is moved
5177 (*arc_clzsi2): ... here.
5178 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5179 the CC register. The old expand code is moved to ...
5180 (arc_ctzsi2): ... here.
5182 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5184 * config/arc/arc.opt (mindexed-loads): Use initial value
5185 TARGET_INDEXED_LOADS_DEFAULT.
5186 (mauto-modify-reg): Use initial value
5187 TARGET_AUTO_MODIFY_REG_DEFAULT.
5188 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5189 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5190 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5191 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5193 2017-07-17 Martin Liska <mliska@suse.cz>
5196 * opts.c (finish_options): Do not allow -fgnu-tm
5197 w/ -fsanitize={kernel-,}address. Say sorry.
5199 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5202 * tree-loop-distribution.c (classify_partition): Only assert on
5203 numer of iterations.
5204 (merge_dep_scc_partitions): Delete prameter. Update function call.
5205 (distribute_loop): Remove code handling loop with unknown niters.
5206 (pass_loop_distribution::execute): Skip loop with unknown niters.
5208 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5211 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5212 function sort_partitions_by_post_order.
5214 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5216 PR tree-optimization/81374
5217 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5218 the max index of basic blocks, rather than number of basic blocks.
5220 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5222 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5224 (arc_legitimate_pic_operand_p): Likewise.
5225 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5227 (arc_needs_pcl_p): Likewise.
5228 (arc_legitimate_pc_offset_p): Likewise.
5229 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5230 function is also used in constrains.md.
5231 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5232 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5233 PLUS. Only return true/false in known cases, otherwise assert.
5234 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5235 is already called in arc_legitimate_constant_p.
5236 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5238 (LEGITIMATE_PIC_OPERAND_P): Use
5239 arc_raw_symbolic_reference_mentioned_p function.
5240 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5245 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5246 Andrew Burgess <andrew.burgess@embecosm.com>
5248 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5249 (arc_return_address_register): New function.
5250 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5251 (arc_handle_fndecl_attribute): Add naked attribute.
5252 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5253 (TARGET_WARN_FUNC_RETURN): Likewise.
5254 (arc_allocate_stack_slots_for_args): New function.
5255 (arc_warn_func_return): Likewise.
5256 (machine_function): Change type fn_type.
5257 (arc_compute_function_type): Consider new naked function type,
5258 change function return type.
5259 (arc_must_save_register): Adapt to handle new
5260 arc_compute_function_type's return type.
5261 (arc_expand_prologue): Likewise.
5262 (arc_expand_epilogue): Likewise.
5263 (arc_return_address_regs): Delete.
5264 (arc_return_address_register): New function.
5265 (arc_epilogue_uses): Use above function.
5266 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5267 (arc_function_type): Change encoding, add naked type.
5268 (ARC_INTERRUPT_P): Change to handle the new encoding.
5269 (ARC_FAST_INTERRUPT_P): Likewise.
5270 (ARC_NORMAL_P): Define.
5271 (ARC_NAKED_P): Likewise.
5272 (arc_compute_function_type): Delete prototype.
5273 * config/arc/arc.md (in_ret_delay_slot): Use
5274 arc_return_address_register function.
5275 (simple_return): Likewise.
5276 (p_return_i): Likewise.
5278 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5280 PR tree-optimization/81428
5281 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5282 can't be built for those types.
5284 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5286 Remove stuff dead since r239246.
5288 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5289 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5290 (avr_inform_devices): Remove dead stuff.
5292 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5294 * config/arm/arm_neon.h: Fix softp typo.
5296 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5298 PR tree-optimization/81365
5299 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5300 aggregate moves onto bb predecessor edges, make sure there are no
5301 loads that could alias the lhs in between the start of bb and the
5304 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5307 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5308 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5309 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5311 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5313 PR tree-optimization/81396
5314 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5315 (init_symbolic_number): Initialize it to 1.
5316 (perform_symbolic_merge): Add n_ops from both operands into the new
5318 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5319 without base_addr as useless if they need more than one operation.
5320 (bswap_replace): Handle !bswap case for NULL base_addr.
5322 2017-07-17 Tom de Vries <tom@codesourcery.com>
5325 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5328 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5330 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5331 conditional builtin define __FIX_LEON3FT_B2BST.
5333 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5335 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5336 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5339 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5341 PR rtl-optimization/81424
5342 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5343 to remove potential trapping from operands if -fnon-call-exceptions.
5345 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5347 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5348 profile_proability for scalling.
5349 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5351 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5353 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5355 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5357 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5358 fixpoint arithmetics.
5360 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5362 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5363 fixpoint arithmetics.
5365 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5367 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5368 fixpoint arithmetics.
5370 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5372 * profile-count.h (profile_probability::from_reg_br_prob_note,
5373 profile_probability::to_reg_br_prob_note): New functions.
5374 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5375 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5376 * predict.c (probability_reliable_p): Update.
5377 (edge_probability_reliable_p): Update.
5378 (br_prob_note_reliable_p): Update.
5379 (invert_br_probabilities): Update.
5380 (add_reg_br_prob_note): New function.
5381 (combine_predictions_for_insn): Update.
5382 * asan.c (asan_clear_shadow): Update.
5383 * cfgbuild.c (compute_outgoing_frequencies): Update.
5384 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5385 (update_br_prob_note): Update.
5386 (rtl_verify_edges): Update.
5387 (purge_dead_edges): Update.
5388 (fixup_reorder_chain): Update.
5389 * emit-rtl.c (try_split): Update.
5390 * ifcvt.c (cond_exec_process_insns): Update.
5391 (cond_exec_process_if_block): Update.
5392 (dead_or_predicable): Update.
5393 * internal-fn.c (expand_addsub_overflow): Update.
5394 (expand_neg_overflow): Update.
5395 (expand_mul_overflow): Update.
5396 * loop-doloop.c (doloop_modify): Update.
5397 * loop-unroll.c (compare_and_jump_seq): Update.
5398 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5399 * predict.h: Update.
5400 * reorg.c (mostly_true_jump): Update.
5402 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5403 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5404 * config/arc/arc.c: (emit_unlikely_jump): Update.
5405 * config/arm/arm.c: (emit_unlikely_jump): Update.
5406 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5407 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5408 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5409 (ix86_print_operand): Update.
5410 (ix86_split_fp_branch): Update.
5411 (predict_jump): Update.
5412 * config/ia64/ia64.c (ia64_print_operand): Update.
5413 * config/mmix/mmix.c (mmix_print_operand): Update.
5414 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5415 (rs6000_expand_split_stack_prologue): Update.
5416 * config/rs6000/rs6000.c: Update.
5417 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5418 (s390_expand_vec_movstr): Update.
5419 (s390_expand_cs_tdsi): Update.
5420 (s390_expand_split_stack_prologue): Update.
5421 * config/sh/sh.c (sh_print_operand): Update.
5422 (expand_cbranchsi4): Update.
5423 (expand_cbranchdi4): Update.
5424 * config/sparc/sparc.c (output_v9branch): Update.
5425 * config/spu/spu.c (get_branch_target): Update.
5426 (ea_load_store_inline): Update.
5427 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5428 * config/tilepro/tilepro.c: Update.
5430 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5432 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5433 (gimplify_save_expr): Likewise.
5435 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5437 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5439 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5441 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5443 * timevar.def (TV_IPA_FNSUMMARY): Define.
5445 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5447 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5448 to back store errata sensitive sequence from being generated.
5449 (sqrtdf2_fix): Likewise.
5451 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5453 * tree-ssa-threadupdate.c (compute_path_counts,
5454 update_joiner_offpath_counts): Use profile_probability.
5456 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5459 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5461 * config/arm/arm-c.c (arm_cpu_builtins): Define
5462 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5464 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5466 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5467 array entries to represent __ieee128 versions of the
5468 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5469 scalar_extract_sig, and scalar_insert_exp built-in functions.
5470 (altivec_resolve_overloaded_builtin): Add special case handling
5471 for the __builtin_scalar_insert_exp function, as represented by
5472 the P9V_BUILTIN_VEC_VSIEDP constant.
5473 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5474 exponent support for __ieee128 argument.
5475 (VSESQP): Add scalar extract signature support for __ieee128
5477 (VSTDCNQP): Add scalar test negative support for __ieee128
5479 (VSIEQP): Add scalar insert exponent support for __int128 argument
5480 with __ieee128 result.
5481 (VSIEQPF): Add scalar insert exponent support for __ieee128
5482 argument with __ieee128 result.
5483 (VSTDCQP): Add scalar test data class support for __ieee128
5485 (VSTDCNQP): Add overload support for scalar test negative with
5487 (VSTDCQP): Add overload support for scalar test data class
5489 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5491 (UNSPEC_VSX_SIEXPQP): New constant.
5492 (xsxexpqp): New insn for VSX scalar extract exponent quad
5494 (xsxsigqp): New insn for VSX scalar extract significand quad
5496 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5497 precision with floating point argument.
5498 (xststdcqp): New expand for VSX scalar test data class quad
5500 (xststdcnegqp): New expand for VSX scalar test negative quad
5502 (xststdcqp): New insn to match expansions for VSX scalar test data
5503 class quad precision and VSX scalar test negative quad precision.
5504 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5505 special case operand checking to enforce that second operand of
5506 VSX scalar test data class with quad precision argument is a 7-bit
5508 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5509 prototypes and descriptions of __ieee128 versions of
5510 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5511 scalar_test_data_class, and scalar_test_neg built-in functions.
5513 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5515 PR tree-optimization/81162
5516 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5517 replace a negate with an add.
5519 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5521 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5523 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5525 * config/arm/arm-c.c (arm_cpu_builtins): Define
5526 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5528 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5530 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5531 (armv8-r): Set ARM Cortex-R52 as default CPU.
5532 * config/arm/arm-tables.opt: Regenerate.
5533 * config/arm/arm-tune.md: Regenerate.
5534 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5536 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5537 extension for -mcpu=cortex-r52.
5539 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5541 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5542 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5543 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5544 (fp-armv8): Define it as FP_ARMv8 only.
5545 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5546 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5548 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5549 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5550 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5551 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5552 than TARGET_FPU_ARMV8.
5553 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5554 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5555 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5557 * config/arm/neon.md (neon_vrint): Likewise.
5558 (neon_vcvt): Likewise.
5559 (neon_<fmaxmin_op><mode>): Likewise.
5560 (<fmaxmin><mode>3): Likewise.
5561 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5562 * config/arm/predicates.md (arm_cond_move_operator): Check against
5563 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5565 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5567 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5570 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5572 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5573 loop in comment with memset.
5575 2017-07-14 Martin Liska <mliska@suse.cz>
5577 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5578 * dwarf2out.c (is_java): Remove the function.
5579 (output_pubname): Remove usage of the function.
5580 (lower_bound_default): Remove usage of DW_LANG_Java.
5581 (gen_compile_unit_die): Likewise.
5582 * gcc.c: Remove compiler defaults for .java and .zip files.
5583 * gimple-expr.c (remove_suffix): Change as there's no longer
5584 extension than 4-letter one.
5585 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5586 (gimplify_save_expr): Likewise.
5587 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5588 as it's possible even for other languages than Java.
5589 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5590 * lto-opts.c (lto_write_options): Remove reference to Java.
5591 * opts.c (strip_off_ending): Update file extension handling.
5592 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5593 * tree-eh.c (lower_resx): Likewise.
5594 * tree.c (free_lang_data_in_type): Remove dead code.
5595 (find_decls_types_r): Likewise.
5596 (build_common_builtin_nodes): Remove Java from a comment.
5597 (verify_type): Remove dead code.
5598 * varasm.c (assemble_external): Remove Java from a comment.
5600 2017-07-14 Martin Liska <mliska@suse.cz>
5602 * opts.c (finish_options): Add quotes.
5603 (common_handle_option): Likewise.
5605 2017-07-14 Martin Liska <mliska@suse.cz>
5607 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5608 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5610 * dwarf2out.c (lower_bound_default): Do not handle
5612 (gen_compile_unit_die): Likewise.
5613 * gcc.c: Remove default extension binding for GNU Pascal.
5614 * stmt.c: Remove Pascal language from a comment.
5615 * xcoffout.c: Likewise.
5617 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5620 * diagnostic-show-locus.c (fixit_cmp): New function.
5621 (layout::layout): Sort m_fixit_hints.
5622 (column_range::column_range): Assert that the values are valid.
5623 (struct char_span): New struct.
5624 (correction::overwrite): New method.
5625 (struct source_line): New struct.
5626 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5627 calls in terms of classes source_line and char_span, and
5628 correction::overwrite.
5629 (selftest::test_overlapped_fixit_printing_2): New function.
5630 (selftest::diagnostic_show_locus_c_tests): Call it.
5632 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5634 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5635 early if there is no lhs.
5637 2017-07-13 Martin Liska <mliska@suse.cz>
5639 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5640 (gen_reference_type_die): Likewise.
5641 * stor-layout.c: Remove Pascal-related comment.
5643 2017-07-13 Martin Liska <mliska@suse.cz>
5645 * opts.c (finish_options): Add quotes to error messages.
5646 (parse_sanitizer_options): Likewise.
5648 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5650 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5652 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5654 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5656 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5658 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5660 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5662 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5665 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5666 provides the hardware capability bits, define the macro
5667 __BUILTIN_CPU_SUPPORTS__.
5668 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5669 if GLIBC does not provide the hardware capability bits. Add a
5670 gcc_unreachable call if the built-in cpu function is neither
5671 __builtin_cpu_is nor __builtin_cpu_supports.
5672 (rs6000_get_function_versions_dispatcher): Change the warning
5673 that an old GLIBC is used which does not export the capability
5674 bits to be an error.
5675 * doc/extend.texi (target_clones attribute): Document the
5676 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5677 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5678 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5679 the macros defined by GCC if the newer GLIBC is available.
5681 2017-07-12 Jeff Law <law@redhat.com>
5683 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5684 remaining includes slightly.
5685 * config/riscv/riscv-builtins.c: Include profile-count.h.
5687 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5690 * config/avr/avr.c (avr_set_current_function): In diagnostic
5691 messages: Quote keywords and (parts of) identifiers.
5692 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5695 2017-07-12 Carl Love <cel@us.ibm.com>
5697 * config/rs6000/rs6000-c.c: Add support for built-in functions
5698 vector bool char vec_revb (vector bool char);
5699 vector bool short vec_revb (vector short char);
5700 vector bool int vec_revb (vector bool int);
5701 vector bool long long vec_revb (vector bool long long);
5702 * doc/extend.texi: Update the built-in documentation file for the
5703 new built-in functions.
5705 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5707 * config/s390/s390.md: Remove movcc splitter.
5709 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5711 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5712 load/store on condition.
5714 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5717 * config/avr/avr.c (avr_encode_section_info)
5718 [progmem && !TREE_READONLY]: Error if progmem object needs
5721 2017-07-11 Michael Collison <michael.collison@arm.com>
5723 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5726 2017-07-11 Carl Love <cel@us.ibm.com>
5728 * config/rs6000/rs6000-c.c: Add support for builtins
5729 vector unsigned int vec_parity_lsbb (vector signed int);
5730 vector unsigned int vec_parity_lsbb (vector unsigned int);
5731 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5732 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5733 vector unsigned long long vec_parity_lsbb (vector signed long long);
5734 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5735 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5736 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5737 * doc/extend.texi: Update the built-in documentation file for the
5738 new built-in functions.
5740 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5742 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5743 (layout::m_primary_loc): New field.
5744 (layout::layout): Initialize new field. Move location filtering
5745 logic from here to...
5746 (layout::maybe_add_location_range): ...this new method. Add
5747 support for filtering to just the lines already specified by other
5749 (layout::will_show_line_p): New method.
5750 (gcc_rich_location::add_location_if_nearby): New method.
5751 (selftest::test_add_location_if_nearby): New test function.
5752 (selftest::diagnostic_show_locus_c_tests): Call it.
5753 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5756 2017-07-11 Tom de Vries <tom@codesourcery.com>
5758 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5759 (bb_first_real_insn): New function.
5760 (nvptx_single): Add extra initialization of broadcasted condition
5763 2017-07-11 Nathan Sidwell <nathan@acm.org>
5765 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5767 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5769 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5770 Binutils doc as TEXI will mess them up.
5771 * doc/invoke.texi (AVR Options): Same here.
5773 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5775 * config/sparc/sparc.opt (mfix-ut700): New option.
5776 (mfix-gr712rc): Likewise.
5777 (sparc_fix_b2bst): New variable.
5778 * doc/invoke.texi (SPARC options): Document them.
5779 (ARM options): Fix warnings.
5780 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5781 instructions to prevent sequences that can trigger the store-store
5782 errata for certain LEON3FT processors.
5783 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
5784 (sparc_option_override): Set sparc_fix_b2bst appropriately.
5785 * config/sparc/sparc.md (fix_b2bst): New attribute.
5786 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
5788 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
5791 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
5793 (*rsqrtsf2_sse): Ditto.
5795 (div<mode>3): Macroize insn from divdf3 and divsf3
5796 using MODEF mode iterator.
5798 2017-07-10 Martin Sebor <msebor@redhat.com>
5800 PR tree-optimization/80397
5801 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
5802 instead of testing for equality to INTEGER_TYPE.
5804 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
5806 * config.gcc: Remove uclibc from arc target spec.
5808 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
5810 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
5812 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5815 * lto-wrapper.c (remove_option): New function.
5816 (merge_and_complain): Merge PIC/PIE options more realistically.
5818 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5820 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
5824 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
5825 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
5826 * config.in: Regenerate.
5827 * configure: Regenerate.
5828 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
5829 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
5830 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
5831 (TARGET_GASISR_PROLOGUES): ...target mask.
5832 * common/config/avr/avr-common.c
5833 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
5834 Set -mgas-isr-prologues.
5835 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
5836 INSERT_PASS_BEFORE for it.
5837 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
5838 * config/avr/avr.c (avr_option_override)
5839 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
5840 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
5841 (avr_attribute_table) <no_gccisr>: Add new function attribute.
5842 (avr_set_current_function) <is_no_gccisr>: Init machine field.
5843 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
5845 (make_avr_pass_pre_proep): New function.
5846 (emit_push_sfr) <treg>: Add argument to function and use it
5848 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
5849 and set machine->gasisr.yes.
5850 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
5851 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
5852 __gcc_isr.n_pushed to .L__stack_usage.
5853 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
5854 (avr_asm_final_postscan_insn): ...this new static function.
5855 * config/avr/avr.h (machine_function)
5856 <is_no_gccisr, use_L__stack_usage>: New fields.
5857 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
5858 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
5859 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
5860 (gasisr, *gasisr): New expander and insn.
5861 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5862 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
5863 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
5865 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
5867 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
5870 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5872 Move jump-tables out of .text again.
5875 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
5876 (ASM_OUTPUT_ADDR_VEC): New function.
5877 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
5878 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
5879 INSN_ADDRESSes as asm comment.
5880 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
5881 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
5882 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
5883 * config/avr/avr.md (*tablejump): Adjust comment.
5884 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
5885 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
5887 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
5888 (avr_output_addr_vec): New proto.
5889 (avr_log_t) <insn_addresses>: New field.
5891 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
5894 * config/i386/i386.c (ix86_function_arg_advance): Set
5895 outgoing_args_on_stack to true if there are outgoing arguments
5897 (ix86_function_arg): Likewise.
5898 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
5899 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
5900 * config/i386/i386.h (machine_function): Add
5901 outgoing_args_on_stack.
5903 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
5905 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5907 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
5909 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
5911 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
5912 (REAL_H): Remove $(MACHMODE_H).
5913 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
5915 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
5916 $(MACHMODE_H) and double-int.h.
5917 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
5919 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
5922 2017-07-07 Andrew Pinski <apinski@cavium.com>
5924 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
5925 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
5927 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5929 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
5930 Add warning if GCC was not configured to link against a GLIBC that
5931 exports the hardware capability bits.
5932 (make_resolver_func): Make resolver function private and not a
5933 COMDAT function. Create the name with clone_function_name instead
5934 of make_unique_name.
5937 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
5938 correct operand in doing the split.
5940 2017-07-07 Carl Love <cel@us.ibm.com>
5942 * config/rs6000/rs6000-c: Add support for built-in function
5943 vector unsigned short vec_pack_to_short_fp32 (vector float,
5945 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
5946 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
5947 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
5948 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
5949 (convert_4f32_8i16): Add define_expand.
5950 * doc/extend.texi: Update the built-in documentation file for the
5951 new built-in function.
5953 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5955 * config/sparc/m8.md: New file.
5956 * config/sparc/sparc.md: Include m8.md.
5958 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5960 * config/sparc/sparc.opt: New option -mvis4b.
5961 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
5962 (sparc_option_override): Handle VIS4B.
5963 (enum sparc_builtins): Define
5964 SPARC_BUILTIN_DICTUNPACK{8,16,32},
5965 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
5966 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
5967 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
5968 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
5969 (check_constant_argument): New function.
5970 (sparc_vis_init_builtins): Define builtins
5971 __builtin_vis_dictunpack{8,16,32},
5972 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
5973 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
5974 __builtin_vis_fpcmpde{8,16,32}shl and
5975 __builtin_vis_fpcmpur{8,16,32}shl.
5976 (sparc_expand_builtin): Check that the constant operands to
5977 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
5978 constant and in range.
5979 * config/sparc/sparc-c.c (sparc_target_macros): Handle
5981 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
5982 (SPARC_IMM5_P): Likewise.
5983 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
5984 (enabled): Handle vis4b.
5985 (UNSPEC_DICTUNPACK): New unspec.
5986 (UNSPEC_FPCMPSHL): Likewise.
5987 (UNSPEC_FPUCMPSHL): Likewise.
5988 (UNSPEC_FPCMPDESHL): Likewise.
5989 (UNSPEC_FPCMPURSHL): Likewise.
5990 (cpu_feature): New CPU feature `vis4b'.
5991 (dictunpack{8,16,32}): New insns.
5992 (FPCSMODE): New mode iterator.
5993 (fpcscond): New code iterator.
5994 (fpcsucond): Likewise.
5995 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
5996 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
5997 (fpcmpde{8,16,32}{si,di}shl): Likewise.
5998 (fpcmpur{8,16,32}{si,di}shl): Likewise.
5999 * config/sparc/constraints.md: Define constraints `q' for unsigned
6000 2-bit integer constants and `t' for unsigned 5-bit integer
6002 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6004 (imm5_operand_dictunpack16): Likewise.
6005 (imm5_operand_dictunpack32): Likewise.
6006 (imm2_operand): Likewise.
6007 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6008 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6009 ditunpack* and fpcmp*shl builtins.
6011 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6013 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6014 * config.in: Add HAVE_AS_SPARC6 define.
6015 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6017 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6019 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6020 (CPP_CPU_SPEC): Handle m8.
6021 (ASM_CPU_SPEC): Likewise.
6022 * config/sparc/sparc-opts.h (enum processor_type): Add
6024 * config/sparc/sparc.c (m8_costs): New struct.
6025 (sparc_option_override): Handle TARGET_CPU_m8.
6026 (sparc32_initialize_trampoline): Likewise.
6027 (sparc64_initialize_trampoline): Likewise.
6028 (sparc_issue_rate): Likewise.
6029 (sparc_register_move_cost): Likewise.
6030 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6031 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6032 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6033 (CPP_CPU_SPEC): Handle M8.
6034 (ASM_CPU_SPEC): Likewise.
6035 (AS_M8_FLAG): Define.
6036 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6037 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6038 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6040 * configure: Regenerate.
6041 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6044 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6046 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6048 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6049 ("*movdi_insn_sp32"): Do not set v3pipe.
6050 ("*movsi_insn"): Likewise.
6051 ("*movdi_insn_sp64"): Likewise.
6052 ("*movsf_insn"): Likewise.
6053 ("*movdf_insn_sp32"): Likewise.
6054 ("*movdf_insn_sp64"): Likewise.
6055 ("*zero_extendsidi2_insn_sp64"): Likewise.
6056 ("*sign_extendsidi2_insn"): Likewise.
6057 ("*mov<VM32:mode>_insn"): Likewise.
6058 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6059 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6060 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6061 ("<vlop:code><VL:mode>3"): Likewise.
6062 ("*not_<vlop:code><VL:mode>3"): Likewise.
6063 ("*nand<VL:mode>_vis"): Likewise.
6064 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6065 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6066 ("one_cmpl<VL:mode>2"): Likewise.
6067 ("faligndata<VM64:mode>_vis"): Likewise.
6068 ("alignaddrsi_vis"): Likewise.
6069 ("alignaddrdi_vis"): Likweise.
6070 ("alignaddrlsi_vis"): Likewise.
6071 ("alignaddrldi_vis"): Likewise.
6072 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6073 ("bmaskdi_vis"): Likewise.
6074 ("bmasksi_vis"): Likewise.
6075 ("bshuffle<VM64:mode>_vis"): Likewise.
6076 ("cmask8<P:mode>_vis"): Likewise.
6077 ("cmask16<P:mode>_vis"): Likewise.
6078 ("cmask32<P:mode>_vis"): Likewise.
6079 ("pdistn<P:mode>_vis"): Likewise.
6080 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6082 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6084 * config/sparc/sparc.md ("subtype"): New insn attribute.
6085 ("*wrgsr_sp64"): Set insn subtype.
6086 ("*rdgsr_sp64"): Likewise.
6087 ("alignaddrsi_vis"): Likewise.
6088 ("alignaddrdi_vis"): Likewise.
6089 ("alignaddrlsi_vis"): Likewise.
6090 ("alignaddrldi_vis"): Likewise.
6091 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6092 ("fexpand_vis"): Likewise.
6093 ("fpmerge_vis"): Likewise.
6094 ("faligndata<VM64:mode>_vis"): Likewise.
6095 ("bshuffle<VM64:mode>_vis"): Likewise.
6096 ("cmask8<P:mode>_vis"): Likewise.
6097 ("cmask16<P:mode>_vis"): Likewise.
6098 ("cmask32<P:mode>_vis"): Likewise.
6099 ("fchksm16_vis"): Likewise.
6100 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6101 ("fmean16_vis"): Likewise.
6102 ("fp<plusminus_insn>64_vis"): Likewise.
6103 ("<plusminus_insn>v8qi3"): Likewise.
6104 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6105 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6106 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6107 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6108 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6109 ("*movqi_insn"): Likewise.
6110 ("*movhi_insn"): Likewise.
6111 ("*movsi_insn"): Likewise.
6112 ("movsi_pic_gotdata_op"): Likewise.
6113 ("*movdi_insn_sp32"): Likewise.
6114 ("*movdi_insn_sp64"): Likewise.
6115 ("movdi_pic_gotdata_op"): Likewise.
6116 ("*movsf_insn"): Likewise.
6117 ("*movdf_insn_sp32"): Likewise.
6118 ("*movdf_insn_sp64"): Likewise.
6119 ("*zero_extendhisi2_insn"): Likewise.
6120 ("*zero_extendqihi2_insn"): Likewise.
6121 ("*zero_extendqisi2_insn"): Likewise.
6122 ("*zero_extendqidi2_insn"): Likewise.
6123 ("*zero_extendhidi2_insn"): Likewise.
6124 ("*zero_extendsidi2_insn_sp64"): Likewise.
6125 ("ldfsr"): Likewise.
6126 ("prefetch_64"): Likewise.
6127 ("prefetch_32"): Likewise.
6128 ("tie_ld32"): Likewise.
6129 ("tie_ld64"): Likewise.
6130 ("*tldo_ldub_sp32"): Likewise.
6131 ("*tldo_ldub1_sp32"): Likewise.
6132 ("*tldo_ldub2_sp32"): Likewise.
6133 ("*tldo_ldub_sp64"): Likewise.
6134 ("*tldo_ldub1_sp64"): Likewise.
6135 ("*tldo_ldub2_sp64"): Likewise.
6136 ("*tldo_ldub3_sp64"): Likewise.
6137 ("*tldo_lduh_sp32"): Likewise.
6138 ("*tldo_lduh1_sp32"): Likewise.
6139 ("*tldo_lduh_sp64"): Likewise.
6140 ("*tldo_lduh1_sp64"): Likewise.
6141 ("*tldo_lduh2_sp64"): Likewise.
6142 ("*tldo_lduw_sp32"): Likewise.
6143 ("*tldo_lduw_sp64"): Likewise.
6144 ("*tldo_lduw1_sp64"): Likewise.
6145 ("*tldo_ldx_sp64"): Likewise.
6146 ("*mov<VM32:mode>_insn"): Likewise.
6147 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6148 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6150 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6152 * config/sparc/sparc.md ("type"): New insn type viscmp.
6153 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6155 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6156 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6157 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6158 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6160 ("n7_vis_logical_11cycle"): Likewise.
6161 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6162 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6163 * config/sparc/niagara.md ("niag_vis"): Likewise.
6164 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6165 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6167 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6169 * config/sparc/sparc.md: New instruction type `bmask'.
6170 (bmaskdi_vis): Use the `bmask' type.
6171 (bmasksi_vis): Likewise.
6172 * config/sparc/ultra3.md (us3_array): Likewise.
6173 * config/sparc/niagara7.md (n7_array): Likewise.
6174 * config/sparc/niagara4.md (n4_array): Likewise.
6175 * config/sparc/niagara2.md (niag2_vis): Likewise.
6176 (niag3_vis): Likewise.
6177 * config/sparc/niagara.md (niag_vis): Likewise.
6179 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6181 * ipa-comdats.c: Remove optimize check from gate.
6182 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6183 for functions not optimized.
6184 (ipa_fn_summary_read): Skip optimize check.
6185 (ipa_fn_summary_write): Likewise.
6186 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6188 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6190 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6191 (check_callers): Give up on caller which is not optimized.
6192 (inline_small_functions): Likewise.
6193 (ipa_inline): Do not give up when not optimizing.
6194 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6195 away unoptimizes cdtors.
6196 (whole_program_function_and_variable_visibility): Do
6197 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6198 * ipa.c (process_references): Do not check optimize.
6199 (symbol_table::remove_unreachable_nodes): Update optimize check.
6200 (set_writeonly_bit): Update optimize check.
6201 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6202 (pass_ipa_single_use::gate): Remove.
6204 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6206 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6207 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6208 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6209 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6210 permute_load, permute_store, adjust_extract, adjust_splat,
6211 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6212 replace_swap_with_copy, dump_swap_insn_table,
6213 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6214 recombine_lvx_pattern, recombine_stvx_pattern,
6215 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6216 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6217 to file rs6000-p8swap.c.
6218 * config/rs6000/rs6000-p8swap.c: New file.
6219 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6220 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6221 and rs6000*-*-* targets.
6223 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6225 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6227 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6229 * lto-wrapper.c (merge_and_complain): Do not merge
6230 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6231 fsigned_zeros, ftrapping_math, fwrapv.
6232 (append_compiler_options): Do not track these options.
6233 (append_linker_options): Likewie
6235 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6237 * cgraphunit.c (cgraph_node::finalize_function): When
6238 !flag_toplevel_reorde set no_reorder flag.
6239 (varpool_node::finalize_decl): Likewise.
6240 (symbol_table::compile): Drop no toplevel reorder path.
6242 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6244 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6245 edges; zero probability is not better than uninitialized.
6247 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6249 * asan.h (asan_sanitize_allocas_p): Declare.
6250 * asan.c (asan_sanitize_allocas_p): New function.
6251 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6252 (handle_builtin_alloca): Likewise.
6253 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6254 if !asan_sanitize_allocas_p.
6255 * params.def (asan-instrument-allocas): Add new option.
6256 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6257 * opts.c (common_handle_option): Disable allocas sanitization for
6260 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6262 * asan.c: Include gimple-fold.h.
6263 (get_last_alloca_addr): New function.
6264 (handle_builtin_stackrestore): Likewise.
6265 (handle_builtin_alloca): Likewise.
6266 (asan_emit_allocas_unpoison): Likewise.
6267 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6268 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6269 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6270 (instrument_builtin_call): Pass gimple iterator to
6271 get_mem_refs_of_builtin_call.
6272 (last_alloca_addr): New global.
6273 * asan.h (asan_emit_allocas_unpoison): Declare.
6274 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6275 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6276 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6277 if function calls alloca.
6278 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6279 * gimple-fold.h (replace_call_with_value): Declare.
6280 * internal-fn.c: Include asan.h.
6281 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6282 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6284 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6286 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6287 (C_SELFTEST_FLAGS): New.
6288 (CPP_SELFTEST_FLAGS): New.
6289 (SELFTEST_DEPS): New, from deps of s-selftest.
6290 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6291 (CPP_SELFTEST_DEPS): New.
6292 (selftest): Add dependency on s-selftest-c++.
6293 (s-selftest): Rename to...
6294 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6295 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6296 than SELFTEST_FLAGS.
6297 (selftest-gdb): Rename to...
6298 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6300 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6301 (selftest-valgrind): Rename to...
6302 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6304 (selftest-valgrind): Reintroduce as an alias for
6305 selftest-c-valgrind.
6306 (s-selftest-c++): New.
6307 (selftest-c++-gdb): New.
6308 (selftest-c++-valgrind): New.
6310 2017-07-06 Olivier Hainque <hainque@adacore.com>
6312 * gcc.c (process_command): When deciding if undefined variables
6313 should be ignored when processing specs, accept "gcc -v" as well.
6315 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6317 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6318 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6320 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6322 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6323 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6324 * config/arm/arm-tables.opt: Regenerate.
6325 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6327 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6329 2017-07-06 Carl Love <cel@us.ibm.com>
6331 * ChangeLog: Clean up from mid air collision
6333 2017-07-06 Carl Love <cel@us.ibm.com>
6335 * config/rs6000/rs6000-c.c: Add support for built-in functions
6336 vector signed int vec_subc (vector signed int, vector signed int);
6337 vector signed __int128 vec_subc (vector signed __int128,
6338 vector signed __int128);
6339 vector unsigned __int128 vec_subc (vector unsigned __int128,
6340 vector unsigned __int128);
6341 vector signed int vec_sube (vector signed int, vector signed int,
6343 vector unsigned int vec_sube (vector unsigned int,
6344 vector unsigned int,
6345 vector unsigned int);
6346 vector signed __int128 vec_sube (vector signed __int128,
6347 vector signed __int128,
6348 vector signed__int128);
6349 vector unsigned __int128 vec_sube (vector unsigned __int128,
6350 vector unsigned __int128,
6351 vector unsigned __int128);
6352 vector signed int vec_subec (vector signed int, vector signed int,
6354 vector unsigned int vec_subec (vector unsigned int,
6355 vector unsigned int,
6356 vector unsigned int);
6357 vector signed __int128 vec_subec (vector signed __int128,
6358 vector signed __int128,
6359 vector signed__int128);
6360 vector unsigned __int128 vec_subec (vector unsigned __int128,
6361 vector unsigned __int128,
6362 vector unsigned __int128);
6363 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6364 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6365 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6366 BU_ALTIVEC_OVERLOAD_X definitions.
6367 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6368 * doc/extend.texi: Update the built-in documentation file for the new
6371 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6374 * diagnostic-show-locus.c (layout::layout): Use start and finish
6375 spelling location for the start and finish of each range.
6376 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6377 Add unused aspect param.
6378 * input.c (expand_location_1): Add "aspect" param, and use it
6379 to access the correct part of the location.
6380 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6382 (expand_location_to_spelling_point): Likewise.
6383 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6384 param, and pass it to expand_location_1.
6386 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6388 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6389 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6390 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6391 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6392 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6393 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6394 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6395 _mm_maskz_getmant_ss): New intrinsics.
6396 (__builtin_ia32_getexpss128_mask): Changed to ...
6397 __builtin_ia32_getexpss128_round ... this.
6398 (__builtin_ia32_getexpsd128_mask): Changed to ...
6399 __builtin_ia32_getexpsd128_round ... this.
6400 * config/i386/i386-builtin-types.def
6401 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6402 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6403 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6404 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6405 __builtin_ia32_getmantss_mask_round): New builtins.
6406 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6407 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6408 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6409 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6410 * config/i386/sse.md
6411 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6412 avx512f_sgetexp<mode><mask_scalar_name>
6413 <round_saeonly_scalar_name> ... this.
6414 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6415 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6416 vgetexp<ssescalarmodesuffix>
6417 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6418 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6419 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6420 avx512f_vgetmant<mode><mask_scalar_name>
6421 <round_saeonly_scalar_name> ... this.
6422 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6423 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6424 vgetmant<ssescalarmodesuffix>
6425 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6426 %0<mask_scalar_operand4>, %1, %2
6427 <round_saeonly_scalar_mask_op4>, %3} ... this.
6428 * config/i386/subst.md (mask_scalar_operand4,
6429 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6430 round_saeonly_scalar_nimm_predicate): New subst attributes.
6432 2017-07-06 Julia Koval <julia.koval@intel.com>
6434 * config/i386/i386.c (ix86_erase_embedded_rounding):
6435 Remove code for old rounding pattern.
6437 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6439 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6441 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6443 * doc/sourcebuild.texi (Test Directives, Variants of
6444 dg-require-support): Add documentation for dg-require-stack-check.
6446 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6448 * config/i386/subst.md (mask_scalar, round_scalar,
6449 round_saeonly_scalar): New meta-templates.
6450 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6451 round_scalar_mask_operand3, round_scalar_mask_op3,
6452 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6453 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6454 round_saeonly_scalar_constraint,
6455 round_saeonly_scalar_prefix): New subst attribute.
6456 * config/i386/sse.md
6457 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6458 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6459 <round_scalar_name> ... this.
6460 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6461 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6462 <round_scalar_name> ... this.
6463 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6464 <sse>_vm<code><mode>3<mask_scalar_name>
6465 <round_saeonly_scalar_name> ... this.
6466 (v<plusminus_mnemonic><ssescalarmodesuffix>
6467 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6468 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6469 v<plusminus_mnemonic><ssescalarmodesuffix>
6470 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6471 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6472 (v<multdiv_mnemonic><ssescalarmodesuffix>
6473 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6474 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6475 v<multdiv_mnemonic><ssescalarmodesuffix>
6476 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6477 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6478 (v<maxmin_float><ssescalarmodesuffix>
6479 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6480 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6481 v<maxmin_float><ssescalarmodesuffix>
6482 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6483 %0<mask_scalar_operand3>, %1, %<iptr>2
6484 <round_saeonly_scalar_mask_op3>} ... this.
6486 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6488 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6489 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6491 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6492 Alan Hayward <alan.hayward@arm.com>
6493 David Sherwood <david.sherwood@arm.com>
6495 * combine.c (simplify_if_then_else): Remove "enum" before
6497 * compare-elim.c (can_eliminate_compare): Likewise.
6498 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6500 (aarch64_lookup_simd_builtin_type): Likewise.
6501 (aarch64_simd_builtin_type): Likewise.
6502 (aarch64_init_simd_builtin_types): Likewise.
6503 (aarch64_simd_expand_args): Likewise.
6504 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6506 (aarch64_reverse_mask): Likewise.
6507 (aarch64_simd_emit_reg_reg_move): Likewise.
6508 (aarch64_gen_adjusted_ldpstp): Likewise.
6509 (aarch64_ccmp_mode_to_code): Likewise.
6510 (aarch64_operands_ok_for_ldpstp): Likewise.
6511 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6512 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6514 (aarch64_min_divisions_for_recip_mul): Likewise.
6515 (aarch64_reassociation_width): Likewise.
6516 (aarch64_get_condition_code_1): Likewise.
6517 (aarch64_simd_emit_reg_reg_move): Likewise.
6518 (aarch64_simd_attr_length_rglist): Likewise.
6519 (aarch64_reverse_mask): Likewise.
6520 (aarch64_operands_ok_for_ldpstp): Likewise.
6521 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6522 (aarch64_gen_adjusted_ldpstp): Likewise.
6523 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6525 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6526 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6527 (arm_lookup_simd_builtin_type): Likewise.
6528 (arm_simd_builtin_type): Likewise.
6529 (arm_init_simd_builtin_types): Likewise.
6530 (arm_expand_builtin_args): Likewise.
6531 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6532 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6533 (ft32_setup_incoming_varargs): Likewise.
6534 (ft32_function_arg): Likewise.
6535 (ft32_function_arg_advance): Likewise.
6536 (ft32_pass_by_reference): Likewise.
6537 (ft32_arg_partial_bytes): Likewise.
6538 (ft32_valid_pointer_mode): Likewise.
6539 (ft32_addr_space_pointer_mode): Likewise.
6540 (ft32_addr_space_legitimate_address_p): Likewise.
6541 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6543 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6544 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6545 (iamcu_alignment): Likewise.
6546 (canonicalize_vector_int_perm): Likewise.
6547 (ix86_noce_conversion_profitable_p): Likewise.
6548 (ix86_mpx_bound_mode): Likewise.
6549 (ix86_operands_ok_for_move_multiple): Likewise.
6550 * config/microblaze/microblaze-protos.h
6551 (microblaze_expand_conditional_branch_reg): Likewise.
6552 * config/microblaze/microblaze.c
6553 (microblaze_expand_conditional_branch_reg): Likewise.
6554 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6556 (rs6000_reassociation_width): Likewise.
6557 (rs6000_invalid_binary_op): Likewise.
6558 (fusion_p9_p): Likewise.
6559 (emit_fusion_p9_load): Likewise.
6560 (emit_fusion_p9_store): Likewise.
6561 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6563 (riscv_hard_regno_mode_ok_p): Likewise.
6564 (riscv_address_insns): Likewise.
6565 (riscv_split_symbol): Likewise.
6566 (riscv_legitimize_move): Likewise.
6567 (riscv_function_value): Likewise.
6568 (riscv_hard_regno_nregs): Likewise.
6569 (riscv_expand_builtin): Likewise.
6570 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6571 (riscv_build_integer): Likewise.
6572 (riscv_split_integer): Likewise.
6573 (riscv_legitimate_constant_p): Likewise.
6574 (riscv_cannot_force_const_mem): Likewise.
6575 (riscv_regno_mode_ok_for_base_p): Likewise.
6576 (riscv_valid_base_register_p): Likewise.
6577 (riscv_valid_offset_p): Likewise.
6578 (riscv_valid_lo_sum_p): Likewise.
6579 (riscv_classify_address): Likewise.
6580 (riscv_legitimate_address_p): Likewise.
6581 (riscv_address_insns): Likewise.
6582 (riscv_load_store_insns): Likewise.
6583 (riscv_force_binary): Likewise.
6584 (riscv_split_symbol): Likewise.
6585 (riscv_force_address): Likewise.
6586 (riscv_legitimize_address): Likewise.
6587 (riscv_move_integer): Likewise.
6588 (riscv_legitimize_const_move): Likewise.
6589 (riscv_legitimize_move): Likewise.
6590 (riscv_address_cost): Likewise.
6591 (riscv_subword): Likewise.
6592 (riscv_output_move): Likewise.
6593 (riscv_canonicalize_int_order_test): Likewise.
6594 (riscv_emit_int_order_test): Likewise.
6595 (riscv_function_arg_boundary): Likewise.
6596 (riscv_pass_mode_in_fpr_p): Likewise.
6597 (riscv_pass_fpr_single): Likewise.
6598 (riscv_pass_fpr_pair): Likewise.
6599 (riscv_get_arg_info): Likewise.
6600 (riscv_function_arg): Likewise.
6601 (riscv_function_arg_advance): Likewise.
6602 (riscv_arg_partial_bytes): Likewise.
6603 (riscv_function_value): Likewise.
6604 (riscv_pass_by_reference): Likewise.
6605 (riscv_setup_incoming_varargs): Likewise.
6606 (riscv_print_operand): Likewise.
6607 (riscv_elf_select_rtx_section): Likewise.
6608 (riscv_save_restore_reg): Likewise.
6609 (riscv_for_each_saved_reg): Likewise.
6610 (riscv_register_move_cost): Likewise.
6611 (riscv_hard_regno_mode_ok_p): Likewise.
6612 (riscv_hard_regno_nregs): Likewise.
6613 (riscv_class_max_nregs): Likewise.
6614 (riscv_memory_move_cost): Likewise.
6615 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6616 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6617 (rl78_addr_space_address_mode): Likewise.
6618 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6620 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6621 (rs6000_reassociation_width): Likewise.
6622 (rs6000_invalid_binary_op): Likewise.
6623 (fusion_p9_p): Likewise.
6624 (emit_fusion_p9_load): Likewise.
6625 (emit_fusion_p9_store): Likewise.
6626 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6627 (ok_for_simple_move_operands): Likewise.
6628 (ok_for_simple_move_strict_operands): Likewise.
6629 (ok_for_simple_arith_logic_operands): Likewise.
6630 (visium_legitimize_reload_address): Likewise.
6631 (visium_select_cc_mode): Likewise.
6632 (output_cbranch): Likewise.
6633 (visium_split_double_move): Likewise.
6634 (visium_expand_copysign): Likewise.
6635 (visium_expand_int_cstore): Likewise.
6636 (visium_expand_fp_cstore): Likewise.
6637 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6638 (visium_function_arg): Likewise.
6639 (visium_function_arg_advance): Likewise.
6640 (visium_libcall_value): Likewise.
6641 (visium_setup_incoming_varargs): Likewise.
6642 (visium_legitimate_constant_p): Likewise.
6643 (visium_legitimate_address_p): Likewise.
6644 (visium_legitimize_address): Likewise.
6645 (visium_secondary_reload): Likewise.
6646 (visium_register_move_cost): Likewise.
6647 (visium_memory_move_cost): Likewise.
6648 (prepare_move_operands): Likewise.
6649 (ok_for_simple_move_operands): Likewise.
6650 (ok_for_simple_move_strict_operands): Likewise.
6651 (ok_for_simple_arith_logic_operands): Likewise.
6652 (visium_function_value_1): Likewise.
6653 (rtx_ok_for_offset_p): Likewise.
6654 (visium_legitimize_reload_address): Likewise.
6655 (visium_split_double_move): Likewise.
6656 (visium_expand_copysign): Likewise.
6657 (visium_expand_int_cstore): Likewise.
6658 (visium_expand_fp_cstore): Likewise.
6659 (visium_split_cstore): Likewise.
6660 (visium_select_cc_mode): Likewise.
6661 (visium_split_cbranch): Likewise.
6662 (output_cbranch): Likewise.
6663 (visium_print_operand_address): Likewise.
6664 * expmed.c (flip_storage_order): Likewise.
6665 * expmed.h (emit_cstore): Likewise.
6666 (flip_storage_order): Likewise.
6667 * genrecog.c (validate_pattern): Likewise.
6668 * hsa-gen.c (gen_hsa_addr): Likewise.
6669 * internal-fn.c (expand_arith_overflow): Likewise.
6670 * ira-color.c (allocno_copy_cost_saving): Likewise.
6671 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6672 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6673 (process_invariant_for_inheritance): Likewise.
6674 * lra-eliminations.c (move_plus_up): Likewise.
6675 * omp-low.c (lower_oacc_reductions): Likewise.
6676 * simplify-rtx.c (simplify_subreg): Likewise.
6677 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6678 (TARGET_CHKP_BOUND_MODE): Likewise..
6679 * targhooks.c (default_chkp_bound_mode): Likewise.
6680 (default_setup_incoming_vararg_bounds): Likewise.
6681 * targhooks.h (default_chkp_bound_mode): Likewise.
6682 (default_setup_incoming_vararg_bounds): Likewise.
6683 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6684 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6685 (have_whole_vector_shift): Likewise.
6686 * tree-vect-stmts.c (vectorizable_load): Likewise.
6687 * doc/tm.texi: Regenerate.
6689 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6691 Graceful degrade if Binutils PR21472 is not available.
6694 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6695 .rodata in flash test fails.
6696 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6697 * confgure: Regenerate.
6698 * config.in: Regenerate.
6699 * config/avr/avr.c (avr_asm_named_section)
6700 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6701 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6702 (avr_asm_init_sections): Same.
6704 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6706 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6707 (fma<VH:mode>4_intrinsic): Likewise.
6708 (*fmsub<VCVTF:mode>4): Likewise.
6709 (*fmsub<VH:mode>4_intrinsic): Likewise.
6711 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6714 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6715 Don't depend on "optimize > 0".
6716 (out_movhi_r_mr, out_movqi_mr_r): Same.
6717 (out_movhi_mr_r, out_movqi_r_mr): Same.
6718 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6719 io_address_operand on "optimize > 0".
6721 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6723 * tree-loop-distribution.c: Add general explanantion on the pass.
6724 (generate_loops_for_partition): Mark distributed loop.
6725 (pg_add_dependence_edges): New parameter. Handle alias data
6726 dependence specially and record it in the parameter if asked.
6727 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6728 (init_partition_graph_vertices, add_partition_graph_edge): New.
6729 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6730 (free_partition_graph_vdata, build_partition_graph): New.
6731 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6732 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6733 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6734 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6735 (version_for_distribution_p, finalize_partitions): New.
6736 (distribute_loop): Handle alias data dependence specially. Factor
6737 out loop fusion code as functions and call these functions.
6739 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6741 * tree-loop-distribution.c (classify_partition): New parameter and
6742 better handle reduction statement.
6743 (rdg_build_partitions): Revise comment.
6744 (distribute_loop): Compute statements in all partitions and pass it
6745 to classify_partition.
6747 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6749 * tree-loop-distribution.c (enum partition_type): New.
6750 (struct partition): New field type.
6751 (partition_merge_into): Add parameter. Update partition type.
6752 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6753 (build_rdg_partition_for_vertex): Compute partition type.
6754 (rdg_build_partitions): Dump partition type.
6755 (distribute_loop): Update calls to partition_merge_into.
6757 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6759 * tree-loop-distribution.c (struct ddr_hasher): New.
6760 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6762 (classify_partition): Call get_data_dependence.
6763 (pg_add_dependence_edges): Ditto.
6764 (distribute_loop): Release data dependence hash table.
6766 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6768 * tree-loop-distribution.c (ref_base_address): Delete.
6769 (similar_memory_accesses): Rename ...
6770 (share_memory_accesses): ... to this. Check if partitions access
6771 the same memory reference.
6772 (distribute_loop): Call share_memory_accesses.
6774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6776 * tree-loop-distribution.c (struct partition): New field recording
6778 (partition_alloc, partition_free): Init and release data refs.
6779 (partition_merge_into): Merge data refs.
6780 (build_rdg_partition_for_vertex): Collect data refs for partition.
6781 (pg_add_dependence_edges): Change parameters from vector to bitmap.
6783 (distribute_loop): Remve data refs from vertice data of partition
6786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6788 * tree-loop-distribution.c (params.h): Include header file.
6789 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
6790 (datarefs_vec): New global var.
6791 (create_rdg_vertices): Use datarefs_vec directly.
6792 (free_rdg): Don't free data references.
6793 (build_rdg): Update use. Don't free data references.
6794 (distribute_loop): Compute global variable for data references.
6795 Bail out if there are too many data references.
6797 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6799 * tree-loop-distribution.c (loop_nest): New global var.
6800 (build_rdg): Use loop directly, rather than loop nest.
6801 (pg_add_dependence_edges): Remove loop nest parameter. Use global
6803 (distribute_loop): Compute global variable loop nest. Update use.
6805 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6807 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
6808 (partition_merge_into): New parameter. Dump reason for fusion.
6809 (distribute_loop): Update use of partition_merge_into.
6811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6813 * tree-loop-distribution.c (bb_top_order_index): New.
6814 (bb_top_order_index_size, bb_top_order_cmp): New.
6815 (stmts_from_loop): Use topological order.
6816 (pass_loop_distribution::execute): Compute and release topological
6817 order for basic blocks.
6819 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6821 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
6824 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6826 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
6827 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
6828 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
6829 * internal-fn.def (LOOP_DIST_ALIAS): New.
6830 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
6831 (fold_loop_internal_call): ... this.
6832 (vect_loop_dist_alias_call): New function.
6833 (set_uid_loop_bbs): Call fold_loop_internal_call.
6834 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
6837 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6840 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
6841 Require dead FLAGS_REG at the beginning of a peephole.
6843 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6846 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
6847 arguments in the call to __builtin_ia32_sbb_u32.
6848 (_subborrow_u64): Swap _X and _Y arguments in the call to
6849 __builtin_ia32_sbb_u64.
6851 2017-07-04 Jakub Jelinek <jakub@redhat.com>
6854 * tree-vrp.c (compare_assert_loc): Turn into a function template
6855 with stable template parameter. Only test if a->e is NULL,
6856 !a->e == !b->e has been verified already. Use e == NULL or
6857 e != NULL instead of e or ! e tests. If stable is true, don't use
6858 iterative_hash_expr, on the other side allow a or b or both NULL
6859 and sort the NULLs last.
6860 (process_assert_insertions): Sort using compare_assert_loc<false>
6861 instead of compare_assert_loc, later sort using
6862 compare_assert_loc<true> before calling process_assert_insertions_for
6863 in a loop. Use break instead of continue once seen NULL pointer.
6865 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6867 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6868 Cortex-R7 and Cortex-R8 processors.
6870 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6872 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
6873 uninitialized while src is not.
6875 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
6877 * common/config/arm/arm-common.c: Adjust include path for
6879 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
6880 (arm-cpu.h): Create in build directory. Adjust dependency rules.
6881 (arm-cpu-data.h): Likewise.
6882 (arm-cpu-cdata.h): Likewise.
6883 * config/arm/arm-cpu.h: Delete.
6884 * config/arm/arm-cpu-cdata.h: Delete.
6885 * config/arm/arm-cpu-data.h: Delete.
6887 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
6889 * config/arm/arm-cpus.in (cortex-a55): New.
6890 (cortex-a75): Likewise.
6891 (cortex-a75.cortex-a55): Likewise.
6892 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
6894 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
6895 * config/arm/arm-cpu-cdata.h: Regenerate.
6896 * config/arm/arm-cpu-data.h: Regenerate.
6897 * config/arm/arm-cpu.h: Regenerate.
6898 * config/arm/arm-tables.opt: Regenerate.
6899 * config/arm/arm-tune.md: Regenerate.
6901 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6903 * haifa-sched.c (sched_create_recovery_edges): Update profile.
6905 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6907 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
6910 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
6912 PR tree-optimization/81292
6913 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
6914 full_string_p, also call adjust_related_strinfos if the adjustment
6915 is simple, otherwise invalidate related strinfos.
6917 2017-07-04 Martin Liska <mliska@suse.cz>
6920 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
6921 newly created variable as DECL_IGNORED_P.
6923 2017-07-04 Martin Liska <mliska@suse.cz>
6926 * ipa-inline.c (inline_small_functions):
6927 Use xstrdup_for_dump.
6929 2017-07-04 Tom de Vries <tom@codesourcery.com>
6931 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
6933 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6936 * config/darwin.c (darwin_function_switched_text_sections):
6939 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
6941 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
6943 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6945 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
6947 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6949 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
6950 min_profitable_iters, and th as inclusive lower bounds.
6951 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
6952 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
6953 for min_profitable_iters and min_profitable_estimate.
6954 (vect_transform_loop): Treat th as an inclusive lower bound.
6955 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6957 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6960 * config/darwin.c (darwin_function_switched_text_sections):
6961 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
6962 in two pieces, and suppress the use of buf.
6964 2017-07-03 Nathan Sidwell <nathan@acm.org>
6966 * hash-table.h (hash_table_mod1): Fix indentation.
6968 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6971 * predict.c (force_edge_cold): Be more careful about propagation
6973 * profile-count.h (profile_probability::guessed,
6974 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
6976 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
6978 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
6980 * doc/invoke.texi (rcpc architecture extension): Document it.
6982 2017-07-03 Richard Biener <rguenther@suse.de>
6984 PR tree-optimization/60510
6985 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
6986 the scalar reduction PHI and use it.
6987 (vectorizable_reduction): Properly guard the single_defuse_cycle
6988 path for non-SLP reduction chains where we cannot use it.
6989 Rework reduc_def/index and vector type deduction. Rework
6990 vector operand gathering during reduction op code-gen.
6991 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
6992 chains dissolve the chain and leave it to non-SLP reduction
6995 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6997 * tree-data-ref.h (dr_alignment): Declare.
6998 * tree-data-ref.c (dr_alignment): New function.
6999 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7000 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7002 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7004 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7006 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7007 and base_misalignment fields.
7008 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7009 * tree-data-ref.c: Include builtins.h.
7010 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7011 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7012 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7013 * tree-vect-data-refs.c: Include tree-cfg.h.
7014 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7015 fields instead of calculating an alignment here.
7016 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7017 innermost_loop_behavior fields.
7019 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7021 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7023 (DR_STEP_ALIGNMENT): New macro.
7024 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7025 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7026 (create_data_ref): Print it.
7027 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7028 to tell whether the step preserves vector (mis)alignment.
7029 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7030 Move the check for an integer step and generalise to all INTEGER_CST.
7031 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7032 Print the outer step alignment.
7034 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7036 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7037 with offset_alignment.
7038 (DR_ALIGNED_TO): Delete.
7039 (DR_OFFSET_ALIGNMENT): New macro.
7040 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7041 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7042 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7043 (create_data_ref): Likewise.
7044 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7045 (vect_analyze_data_refs): Likewise.
7046 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7047 creating dummy innermost behavior.
7049 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7051 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7052 with a "innermost_loop_behavior *" and refeence tree.
7053 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7054 (create_data_ref): Update call accordingly.
7055 * tree-predcom.c (find_looparound_phi): Likewise.
7057 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7059 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7060 fields with dr_wrt_vec_loop.
7061 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7062 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7063 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7064 (vect_dr_behavior): New function.
7065 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7066 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7067 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7068 track whether the step preserves the misalignment.
7069 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7070 Use vect_dr_behavior.
7071 (vect_setup_realignment): Update call accordingly.
7072 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7073 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7074 call to vect_create_addr_base_for_vector_ref.
7075 (vect_create_cond_for_align_checks): Likewise.
7076 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7077 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7078 (vect_recog_mask_conversion_pattern): Likewise.
7079 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7080 (new_stmt_vec_info): Remove redundant zeroing.
7082 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7084 * common/config/arm/arm-common.c (arm_be8_option): New function.
7085 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7086 (ISA_ARMv6): Add isa_bit_be8.
7087 * config/arm/arm.h (arm_be8_option): Add prototype.
7088 (BE8_SPEC_FUNCTION): New define.
7089 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7090 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7091 (mlittle-endian): Similarly.
7092 (mbe8, mbe32): New options.
7093 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7094 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7096 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7098 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7100 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7102 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7103 (cleanup_tree_cfg_bb): Use it.
7104 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7106 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7108 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7111 * loop-doloop.c (add_test): Update profile.
7113 2017-07-03 Martin Liska <mliska@suse.cz>
7116 * sanopt.c (rewrite_usage_of_param): New function.
7117 (sanitize_rewrite_addressable_params): Likewise.
7118 (pass_sanopt::execute): Call rewrite_usage_of_param.
7120 2017-07-03 Richard Biener <rguenther@suse.de>
7122 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7123 back to using VIEW_CONVERT_EXPR.
7125 2017-07-03 Martin Liska <mliska@suse.cz>
7128 * doc/extend.texi: Document when a resolver function is
7129 generated for target_clones.
7131 2017-07-03 Martin Liska <mliska@suse.cz>
7133 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7134 and shadow memory of auto variables which are subject of
7135 use-after-scope sanitization.
7136 (asan_expand_mark_ifn): Add do set only when is_poison.
7138 2016-07-03 Richard Biener <rguenther@suse.de>
7140 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7142 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7143 (vectorizable_reduction): Perform reduction PHI creation when
7144 visiting a reduction PHI and adjust and simplify code generation
7145 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7146 (vect_transform_loop): Visit reduction PHIs.
7147 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7148 defs into the SLP tree.
7149 (vect_build_slp_tree): Reduction defs terminate the recursion.
7150 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7152 (vect_get_vec_defs_for_stmt_copy): Export.
7153 (vect_get_vec_defs): Likewise.
7154 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7156 (vect_get_vec_defs_for_stmt_copy): Declare.
7157 (vect_get_vec_defs): Likewise.
7159 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7161 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7162 parameter with a "loop" parameter and use it instead of the
7163 loop containing DR_STMT. Don't check simple_iv when doing
7164 BB analysis. Describe the two analysis modes in the comment.
7166 2017-07-03 Tom de Vries <tom@codesourcery.com>
7168 PR tree-optimization/69468
7169 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7170 (find_same_succ_bb): Handle ignore_edge_flags.
7172 2017-07-03 Tom de Vries <tom@codesourcery.com>
7174 PR tree-optimization/81192
7175 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7177 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7179 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7181 2017-07-03 Tom de Vries <tom@codesourcery.com>
7183 PR tree-optimization/81192
7184 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7185 BB_SAME_SUCC (bb) == NULL.
7187 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7189 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7194 * dumpfile.c: Include profile-count.h
7195 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7197 (insert_cond_bb): Update profile.
7198 * tree-cfg.h (insert_cond_bb): Update prototype.
7199 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7200 * tree-dump.c: Do not include tree-cfg.
7202 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7204 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7206 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7208 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7211 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7213 * tree-complex.c (expand_complex_div_wide): update profile.
7215 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7216 Alan Hayward <alan.hayward@arm.com>
7217 David Sherwood <david.sherwood@arm.com>
7219 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7220 (CORETYPES_H): New define.
7221 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7222 (insn-modes-inline.h, s-modes-inline-h): New rules.
7223 (generated_files): Add insn-modes-inline.h.
7224 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7225 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7226 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7227 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7228 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7229 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7230 (build/gencodes.o, build/genconditions.o): Likewise.
7231 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7232 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7233 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7234 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7235 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7236 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7237 * coretypes.h: Include everything up to real.h for generators.
7238 Include insn-modes.h first. Include wide-int-print.h after
7239 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7240 * machmode.h: Don't include insn-modes.h here.
7241 * function-tests.c: Remove includes of signop.h, machmode.h,
7242 double-int.h and wide-int.h.
7244 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7246 * optc-save-gen.awk: Likewise.
7247 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7248 * godump.c: Remove include of wide-int-print.h.
7249 * pretty-print.h: Likewise.
7250 * wide-int-print.cc: Likewise.
7251 * wide-int.cc: Likewise.
7252 * hash-map-tests.c: Remove include of signop.h.
7253 * hash-set-tests.c: Likewise.
7254 * rtl-tests.c: Likewise.
7255 * mkconfig.sh: Remove include of machmode.h.
7256 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7258 (emit_insn_modes_inline_h): ...this new function. Emit the code
7259 into an insn-modes-inline.h header file, adding appropriate
7260 include guards and end comments.
7261 (emit_insn_modes_c_header): Remove include of machmode.h.
7262 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7264 (main): Handle -i flag and call emit_insn_modes_inline_h when
7267 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7269 * tree-ssa-strlen.c (strinfo): Rename the length field to
7270 nonzero_chars. Add a full_string_p field.
7271 (compare_nonzero_chars, zero_length_string_p): New functions.
7272 (get_addr_stridx): Add an offset_out parameter.
7273 Use compare_nonzero_chars.
7274 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7275 (new_strinfo): Update after above changes to strinfo.
7276 (set_endptr_and_length): Set full_string_p.
7277 (get_string_length): Update after above changes to strinfo.
7278 (unshare_strinfo): Update call to new_strinfo.
7279 (maybe_invalidate): Likewise.
7280 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7281 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7282 as a uhwi instead of an shwi. Update after above changes to
7283 strinfo and new_strinfo.
7284 (zero_length_string): Assert that chainsi contains full strings.
7285 Use zero_length_string_p. Update call to new_strinfo.
7286 (adjust_related_strinfos): Update after above changes to strinfo.
7287 Copy full_string_p from origsi.
7288 (adjust_last_stmt): Use zero_length_string_p.
7289 (handle_builtin_strlen): Update after above changes to strinfo and
7290 new_strinfo. Install the lhs as the string length if the previous
7291 entry didn't describe a full string.
7292 (handle_builtin_strchr): Update after above changes to strinfo
7294 (handle_builtin_strcpy): Likewise.
7295 (handle_builtin_strcat): Likewise.
7296 (handle_builtin_malloc): Likewise.
7297 (handle_pointer_plus): Likewise.
7298 (handle_builtin_memcpy): Likewise. Track nonzero characters
7299 that aren't necessarily followed by a nul terminator.
7300 (handle_char_store): Likewise.
7302 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7304 PR tree-optimization/80769
7305 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7306 for malloc and calloc. Document the new invariant that all related
7307 strinfos have delayed lengths or none do.
7308 (verify_related_strinfos): Move earlier in file.
7309 (set_endptr_and_length): New function, split out from...
7310 (get_string_length): ...here. Also set the lengths of related
7312 (zero_length_string): Assert that chainsi has known (rather than
7314 (adjust_related_strinfos): Likewise.
7316 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7318 PR tree-optimization/81136
7319 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7320 assert that two references with the same misalignment have the same
7321 compile-time misalignment if those compile-time misalignments
7324 2017-07-01 Andi Kleen <ak@linux.intel.com>
7326 * print-tree.c (print_node): Print all attributes.
7328 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7330 * cfg.c (scale_bbs_frequencies): New function.
7331 * cfg.h (scale_bbs_frequencies): Declare it.
7332 * cfgloopanal.c (single_likely_exit): Cleanup.
7333 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7335 (scale_loop_profile): Likewise.
7336 (loop_version): Likewise.
7337 (create_empty_loop_on_edge): Update.
7338 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7339 scale_loop_frequencies, scale_loop_profile, loopify,
7340 loop_version): Update prototypes.
7341 * modulo-sched.c (sms_schedule): Update.
7342 * predict.c (unlikely_executed_edge_p): Also check probability.
7343 (probably_never_executed_edge_p): Fix typo.
7344 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7345 * tree-parloops.c (gen_parallel_loop): Update.
7346 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7347 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7348 * tree-ssa-loop-split.c (split_loop): Update.
7349 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7350 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7351 (vect_loop_versioning): Update.
7352 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7354 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7356 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7358 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7360 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7361 to keep profile consistent.
7363 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7365 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7366 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7367 * profile-count.h (max_safe_multiplier): Make unsigned.
7368 (profile_count::guessed_zero): New.
7370 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7372 * bb-reorder.c (fix_up_crossing_landing_pad,
7373 fix_crossing_conditional_branches): Use make_single_succ_edge
7374 to keep profile consistent.
7376 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7378 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7381 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7384 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7385 the right scopes, make sure cond_jump isn't preserved between multiple
7386 iterations. Search for fallthru edge whenever there are 3+ edges and
7387 use find_fallthru_edge for it.
7389 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7391 Patch by Alexander Monakov <amonakov@ispras.ru>
7392 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7393 probabilities consistently.
7395 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7397 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7398 emit_cmp_and_jump_insns.
7400 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7403 * tree-inline.c (expand_call_inline): Combine profile statuses.
7405 2017-06-30 Andrew Pinski <apinski@cavium.com>
7407 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7408 fold_stmt returned true.
7410 2017-06-30 Nathan Sidwell <nathan@acm.org>
7412 * ggc.h (empty_string): Delete.
7413 * cfgexpand.c (expand_asm_stmt): Use plain "".
7414 * optabs.c (expand_asm_memory_barrier): Likewise.
7415 * stringpool.c (empty_string): Delete.
7416 (digit_vector, digit_string): Delete.
7417 (ggc_alloc_string): Use plain "", don't optimize single digit
7418 strings. Use ggc_alloc_atomic.
7420 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7422 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7423 comparison set and one other set, use the cost of the non-comparison
7426 2017-06-30 Nathan Sidwell <nathan@acm.org>
7428 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7431 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7433 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7434 loops. Remove now unneeded calls to gimple_switch_set_label() that
7435 just set removed labels to NULL_TREE.
7437 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7439 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7440 (set_range_info): ...here. Only call set_range_info_raw if domain
7442 (set_nonzero_bits): Call set_range_info_raw.
7443 * tree-ssanames.h (set_range_info_raw): New.
7445 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7448 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7449 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7450 of nonimmediate_operand and <store_mask_constraint> instead of m
7451 for the input operand. For V8FI iterator, always split if input
7452 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7453 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7454 <store_mask_predicate> instead of register_operand and
7455 <store_mask_constraint> instead of v for the input operand. Make
7456 sure both operands aren't MEMs for if not <mask_applied>.
7458 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7460 * lto-wrapper.c (copy_file) Close both file descriptors before
7463 2017-06-30 Martin Liska <mliska@suse.cz>
7466 * multiple_target.c (create_dispatcher_calls): Make ifunc
7467 also for function that don't have calls or are not referenced.
7469 2017-06-30 Richard Biener <rguenther@suse.de>
7471 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7472 analyze the first scalar stmt. Move vector type computation
7473 for the BB case here from ...
7474 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7475 live operation processing in the SLP case properly.
7477 2017-06-30 Richard Biener <rguenther@suse.de>
7479 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7481 2017-06-30 Martin Liska <mliska@suse.cz>
7484 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7485 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7487 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7489 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7492 2017-06-30 Richard Biener <rguenther@suse.de>
7494 PR tree-optimization/81249
7495 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7496 condition reduction result to original scalar type.
7498 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7500 * profile-count.h (enum profile_quality): Fix typos and whitespace
7503 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7505 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7506 type for branch probabilities.
7508 2017-06-29 Julian Brown <julian@codesourcery.com>
7509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7511 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7512 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7513 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7514 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7516 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7518 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7519 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7520 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7521 CC usage from generic code to here.
7522 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7523 CC usage into the target macros.
7525 2017-06-29 Maya Rashish <coypu@sdf.org>
7527 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7530 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7532 * arm/arm-builtins.c: Include profile-count.h
7533 * except.c (sjlj_emit_function_enter): Use
7534 profile_probability::unlikely.
7536 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7538 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7539 and tocrel_offset be pointer args rather than implicitly using
7541 (legitimate_constant_pool_address_p, rs6000_emit_move,
7542 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7543 tocrel_offset and use in toc_relative_expr_p call.
7544 (print_operand, print_operand_address): Use static tocrel_base_oac
7545 and tocrel_offset_oac.
7546 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7549 2017-06-29 Maya Rashish <coypu@sdf.org>
7551 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7553 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7555 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7556 objects, take into account only the alignment of 'op0' and 'mode1' if
7559 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7561 * ccmp.c (ccmp_tree_comparison_p): New function.
7562 (ccmp_candidate_p): Update to use above function.
7563 (get_compare_parts): New function.
7564 (expand_ccmp_next): Update to use new functions.
7565 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7567 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7568 take mode as argument.
7569 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7570 * expr.c (expand_expr_real_1): Pass mode as argument.
7572 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7574 * combine.c (combine_instructions): Print insns to dump_file, together
7577 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7579 * asan.c (asan_emit_stack_protection): Update.
7580 (create_cond_insert_point): Update.
7581 * auto-profile.c (afdo_propagate_circuit): Update.
7582 * basic-block.h (struct edge_def): Turn probability to
7583 profile_probability.
7584 (EDGE_FREQUENCY): Update.
7585 * bb-reorder.c (find_traces_1_round): Update.
7586 (better_edge_p): Update.
7587 (sanitize_hot_paths): Update.
7588 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7589 (make_single_succ_edge): Update.
7590 (check_bb_profile): Update.
7591 (dump_edge_info): Update.
7592 (update_bb_profile_for_threading): Update.
7593 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7594 probabilitycount to 0.
7595 * cfgbuild.c (compute_outgoing_frequencies): Update.
7596 * cfgcleanup.c (try_forward_edges): Update.
7597 (outgoing_edges_match): Update.
7598 (try_crossjump_to_edge): Update.
7599 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7600 (expand_gimple_tailcall): Update.
7601 (construct_init_block): Use make_single_succ_edge.
7602 (construct_exit_block): Use make_single_succ_edge.
7603 * cfghooks.c (verify_flow_info): Update.
7604 (redirect_edge_succ_nodup): Update.
7605 (split_edge): Update.
7606 (account_profile_record): Update.
7607 * cfgloopanal.c (single_likely_exit): Update.
7608 * cfgloopmanip.c (scale_loop_profile): Update.
7609 (set_zero_probability): Remove.
7610 (duplicate_loop_to_header_edge): Update.
7611 * cfgloopmanip.h (loop_version): Update prototype.
7612 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7613 (force_nonfallthru_and_redirect): Update.
7614 (update_br_prob_note): Update.
7615 (rtl_verify_edges): Update.
7616 (purge_dead_edges): Update.
7617 (rtl_lv_add_condition_to_bb): Update.
7618 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7619 * cgraphunit.c (init_lowered_empty_function): Update.
7620 (cgraph_node::expand_thunk): Update.
7621 * cilk-common.c: Include profile-count.h
7622 * dojump.c (inv): Remove.
7623 (jumpifnot): Update.
7624 (jumpifnot_1): Update.
7625 (do_jump_1): Update.
7627 (do_jump_by_parts_greater_rtx): Update.
7628 (do_compare_rtx_and_jump): Update.
7629 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7630 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7631 * dwarf2cfi.c: Include profile-count.h
7632 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7633 (sjlj_emit_dispatch_table): Likewise.
7634 * explow.c: Include profile-count.h
7635 * expmed.c (emit_store_flag_force): Update.
7636 (do_cmp_and_jump): Update.
7637 * expr.c (compare_by_pieces_d::generate): Update.
7638 (compare_by_pieces_d::finish_mode): Update.
7639 (emit_block_move_via_loop): Update.
7640 (store_expr_with_bounds): Update.
7641 (store_constructor): Update.
7642 (expand_expr_real_2): Update.
7643 (expand_expr_real_1): Update.
7644 * expr.h (try_casesi, try_tablejump): Update prototypes.
7645 * gimple-pretty-print.c (dump_probability): Update.
7646 (dump_profile): New.
7647 (dump_gimple_label): Update.
7648 (dump_gimple_bb_header): Update.
7649 * graph.c (draw_cfg_node_succ_edges): Update.
7650 * hsa-gen.c (convert_switch_statements): Update.
7651 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7652 (find_if_case_1): Update.
7653 (find_if_case_2): Update.
7654 * internal-fn.c (expand_arith_overflow_result_store): Update.
7655 (expand_addsub_overflow): Update.
7656 (expand_neg_overflow): Update.
7657 (expand_mul_overflow): Update.
7658 (expand_vector_ubsan_overflow): Update.
7659 * ipa-cp.c (good_cloning_opportunity_p): Update.
7660 * ipa-split.c (split_function): Use make_single_succ_edge.
7661 * ipa-utils.c (ipa_merge_profiles): Update.
7662 * loop-doloop.c (add_test): Update.
7663 (doloop_modify): Update.
7664 * loop-unroll.c (compare_and_jump_seq): Update.
7665 (unroll_loop_runtime_iterations): Update.
7666 * lra-constraints.c (lra_inheritance): Update.
7667 * lto-streamer-in.c (input_cfg): Update.
7668 * lto-streamer-out.c (output_cfg): Update.
7669 * mcf.c (adjust_cfg_counts): Update.
7670 * modulo-sched.c (sms_schedule): Update.
7671 * omp-expand.c (expand_omp_for_init_counts): Update.
7672 (extract_omp_for_update_vars): Update.
7673 (expand_omp_ordered_sink): Update.
7674 (expand_omp_for_ordered_loops): Update.
7675 (expand_omp_for_generic): Update.
7676 (expand_omp_for_static_nochunk): Update.
7677 (expand_omp_for_static_chunk): Update.
7678 (expand_cilk_for): Update.
7679 (expand_omp_simd): Update.
7680 (expand_omp_taskloop_for_outer): Update.
7681 (expand_omp_taskloop_for_inner): Update.
7682 * omp-simd-clone.c (simd_clone_adjust): Update.
7683 * optabs.c (expand_doubleword_shift): Update.
7684 (expand_abs): Update.
7685 (emit_cmp_and_jump_insn_1): Update.
7686 (expand_compare_and_swap_loop): Update.
7687 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7688 * predict.c (predictable_edge_p): Update.
7689 (edge_probability_reliable_p): Update.
7690 (set_even_probabilities): Update.
7691 (combine_predictions_for_insn): Update.
7692 (combine_predictions_for_bb): Update.
7693 (propagate_freq): Update.
7694 (estimate_bb_frequencies): Update.
7695 (force_edge_cold): Update.
7696 * profile-count.c (profile_count::dump): Add missing space into dump.
7697 (profile_count::debug): Add newline.
7698 (profile_count::differs_from_p): Explicitly convert to unsigned.
7699 (profile_count::stream_in): Update.
7700 (profile_probability::dump): New member function.
7701 (profile_probability::debug): New member function.
7702 (profile_probability::differs_from_p): New member function.
7703 (profile_probability::differs_lot_from_p): New member function.
7704 (profile_probability::stream_in): New member function.
7705 (profile_probability::stream_out): New member function.
7706 * profile-count.h (profile_count_quality): Rename to ...
7707 (profile_quality): ... this one.
7708 (profile_probability): New.
7709 (profile_count): Update.
7710 * profile.c (compute_branch_probabilities): Update.
7711 * recog.c (peep2_attempt): Update.
7712 * sched-ebb.c (schedule_ebbs): Update.
7713 * sched-rgn.c (find_single_block_region): Update.
7714 (compute_dom_prob_ps): Update.
7715 (schedule_region): Update.
7716 * sel-sched-ir.c (compute_succs_info): Update.
7717 * stmt.c (struct case_node): Update.
7718 (do_jump_if_equal): Update.
7719 (get_outgoing_edge_probs): Update.
7720 (conditional_probability): Update.
7721 (emit_case_dispatch_table): Update.
7722 (expand_case): Update.
7723 (expand_sjlj_dispatch_table): Update.
7724 (emit_case_nodes): Update.
7725 * targhooks.c: Update.
7726 * tracer.c (better_p): Update.
7727 (find_best_successor): Update.
7728 * trans-mem.c (expand_transaction): Update.
7729 * tree-call-cdce.c: Update.
7730 * tree-cfg.c (gimple_split_edge): Upate.
7731 (move_sese_region_to_fn): Upate.
7732 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7733 * tree-eh.c (lower_resx): Upate.
7734 (cleanup_empty_eh_move_lp): Upate.
7735 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7736 * tree-inline.c (copy_edges_for_bb): Update.
7737 (copy_cfg_body): Update.
7738 * tree-parloops.c (gen_parallel_loop): Update.
7739 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7740 (gimple_gen_time_profiler): Update.
7741 * tree-ssa-dce.c (remove_dead_stmt): Update.
7742 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7743 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7744 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7745 (unloop_loops): Update.
7746 (try_peel_loop): Update.
7747 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7748 * tree-ssa-loop-split.c (connect_loops): Update.
7749 (split_loop): Update.
7750 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7751 (hoist_guard): Update.
7752 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7753 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7754 (value_replacement): Update.
7755 * tree-ssa-reassoc.c (branch_fixup): Update.
7756 * tree-ssa-tail-merge.c (replace_block_by): Update.
7757 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7758 (create_edge_and_update_destination_phis): Update.
7759 (compute_path_counts): Update.
7760 (recompute_probabilities): Update.
7761 (update_joiner_offpath_counts): Update.
7762 (freqs_to_counts_path): Update.
7763 (duplicate_thread_path): Update.
7764 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7765 (struct switch_conv_info): Update.
7766 (gen_inbound_check): Update.
7767 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7768 (vect_do_peeling): Update.
7769 (vect_loop_versioning): Update.
7770 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7771 (optimize_mask_stores): Update.
7772 * ubsan.c (ubsan_expand_null_ifn): Update.
7773 * value-prof.c (gimple_divmod_fixed_value): Update.
7774 (gimple_divmod_fixed_value_transform): Update.
7775 (gimple_mod_pow2): Update.
7776 (gimple_mod_pow2_value_transform): Update.
7777 (gimple_mod_subtract): Update.
7778 (gimple_mod_subtract_transform): Update.
7779 (gimple_ic): Update.
7780 (gimple_stringop_fixed_value): Update.
7781 (gimple_stringops_transform): Update.
7782 * value-prof.h: Update.
7784 2017-06-29 Carl Love <cel@us.ibm.com>
7786 * config/rs6000/rs6000-c.c: Add support for built-in functions
7787 vector signed int vec_signed (vector float);
7788 vector signed long long vec_signed (vector double);
7789 vector signed int vec_signed2 (vector double, vector double);
7790 vector signed int vec_signede (vector double);
7791 vector signed int vec_signedo (vector double);
7792 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
7793 instruction generator.
7794 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7795 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
7796 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
7798 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
7799 vunsignede_v2df): Add define_expands.
7800 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
7801 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
7802 VEC_UNSIGNEDO): Add definitions.
7803 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7804 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
7805 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
7806 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
7807 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
7808 * config/rs6000/altivec.h (vec_signed, vec_signed2,
7809 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
7810 vec_unsignede, vec_unsignedo): Add builtin defines.
7811 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
7813 * doc/extend.texi: Update the built-in documentation file for the
7814 new built-in functions.
7816 2017-06-29 Richard Biener <rguenther@suse.de>
7818 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
7819 reduction chains to LOOP_VINFO_REDUCTIONS.
7820 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
7821 SLP reductions after processing reduction chains.
7823 2017-06-29 Nathan Sidwell <nathan@acm.org>
7825 * builtins.c (fold_builtin_FUNCTION): Use
7826 lang_hooks.decl_printable_name.
7828 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
7831 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
7832 with only one label.
7833 * stmt.c (expand_case): Assert NCASES is greater than one.
7835 2017-06-29 Richard Biener <rguenther@suse.de>
7837 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
7839 (group_case_labels): Likewise.
7840 (find_taken_edge): Push sanity checking on val to workers...
7841 (find_taken_edge_cond_expr): ... here
7842 (find_taken_edge_switch_expr): ... and here, handle cases
7843 with just a default label.
7844 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
7845 (group_case_labels): Likewise.
7846 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
7847 group_case_labels does anything cleanup the CFG again.
7849 2017-06-29 Bin Cheng <bin.cheng@arm.com>
7851 PR tree-optimization/81196
7852 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
7853 exit condition comparing two IVs.
7855 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
7857 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
7858 profile to the dummy entry at the end of the list of architectures.
7859 * config/arm/arm-cpu-cdata.h: Regenerated.
7861 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7862 Michael Collison <michael.collison@arm.com>
7865 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
7867 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
7868 (*aarch64_reg_<mode>3_minus_mask): New pattern.
7869 (*aarch64_<optab>_reg_di3_mask2): New pattern.
7870 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
7871 of shift when the shift amount is masked with constant equal to
7872 the size of the mode.
7873 * config/aarch64/predicates.md (subreg_lowpart_operator): New
7876 2017-06-29 Martin Liska <mliska@suse.cz>
7878 * config/i386/i386.opt: Change range from [1,5] to [0,5].
7880 2017-06-29 Yury Gribov <tetra2005@gmail.com>
7883 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
7886 (edge_growth_cache_entry::edge_growth_cache_entry): New
7888 (reset_edge_growth_cache): Update to use constructor.
7890 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7892 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
7893 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7894 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7896 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
7898 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
7899 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
7901 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
7903 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
7904 (*-linux-uclibc*): Add t-uclibc tmake_file.
7905 * config/t-musl: New.
7906 * config/t-uclibc: New.
7908 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
7910 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
7912 (gen_comm_data): Emit architectural setting of arch_prof.
7913 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
7915 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
7916 (armv8-m.base, armv8-m.main): Likewise.
7917 * arm-protos.h (arm_build_target): Add profile field.
7918 (arch_option): Likewise.
7919 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
7921 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
7922 arm_active_target.profile.
7924 2017-06-28 Richard Biener <rguenther@suse.de>
7927 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
7928 TYPE_OVERFLOW_WRAPS.
7929 * match.pd (negate_expr_p): Likewise.
7930 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
7931 fold_build2, not fold_binary.
7933 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7935 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
7936 Convert memory address to Pmode.
7937 (aarch64_print_operand): Assert MEM operands are always Pmode.
7939 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7942 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
7943 Remove redundant if.
7944 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
7945 * config/arm/aarch-common-protos.h
7946 (aarch_forward_to_shift_is_not_shifted_re): Remove.
7947 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
7949 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
7952 * multiple_target.c (create_dispatcher_calls): Set the default
7953 clone to be static, not public.
7955 2017-06-28 Richard Biener <rguenther@suse.de>
7957 * tree-vect-loop.c (vectorizable_reduction): Move special
7958 cond reduction IV var creation ...
7959 (vect_create_epilog_for_reduction): ... here. Remove induction_index
7960 parameter. Use STMT_VINFO_VECTYPE.
7961 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
7964 2017-06-28 Martin Liska <mliska@suse.cz>
7967 * ipa-visibility.c (non_local_p): Handle visibility.
7969 2017-06-28 Martin Liska <mliska@suse.cz>
7972 * common.opt: Add IntegerRange to various options.
7973 * opt-functions.awk (integer_range_info): New function.
7974 * optc-gen.awk: Add integer_range_info to cl_options struct.
7975 * opts-common.c (decode_cmdline_option): Handle
7976 CL_ERR_INT_RANGE_ARG.
7977 (cmdline_handle_error): Likewise.
7978 * opts.c (print_filtered_help): Show valid interval in
7979 when --help is provided.
7980 * opts.h (struct cl_option): Add range_min and range_max fields.
7981 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
7983 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
7985 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
7986 (x * C EQ/NE y * C): New transformation.
7988 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
7990 * genmultilib (combination_space): Accept '+' in option names.
7992 2017-06-28 Martin Liska <mliska@suse.cz>
7995 * asan.c (instrument_derefs): Bail out inner references
7996 that are hard register variables.
7998 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8001 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8002 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8004 2017-06-28 Richard Biener <rguenther@suse.de>
8006 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8007 (vect_get_slp_defs): Adjust.
8008 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8010 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8012 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8013 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8014 (vectorizable_reduction): Adjust.
8015 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8017 (vect_get_slp_defs): Likewise.
8018 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8019 (vectorizable_bswap): Adjust.
8020 (vectorizable_call): Likewise.
8021 (vectorizable_conversion): Likewise.
8022 (vectorizable_assignment): Likewise.
8023 (vectorizable_shift): Likewise.
8024 (vectorizable_operation): Likewise.
8025 (vectorizable_store): Likewise.
8026 (vectorizable_condition): Likewise.
8027 (vectorizable_comparison): Likewise.
8029 2017-06-28 Michael Collison <michael.collison@arm.com>
8032 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8034 (arm_gen_movmemqi): Removed unused variable 'i'.
8035 Convert 'for' loop into 'while' loop.
8036 (arm_expand_prologue): Remove last unnecessary set of insn.
8037 (thumb_pop): Remove unused variable 'pushed_words'.
8038 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8040 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8042 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8043 * config/s390/s390-protos.h: Add prototype of
8044 s390_rel_address_ok_p.
8045 * config/s390/s390.c (s390_got_symbol): New function.
8046 (s390_rel_address_ok_p): New function.
8047 (legitimize_pic_address): Use s390_rel_address_ok_p.
8048 (s390_load_got): Use s390_got_symbol.
8049 (s390_option_override): Issue error if
8050 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8051 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8053 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8055 2017-06-27 Andrew Pinski <apinski@cavium.com>
8057 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8058 (X * copysign (1.0, X)): New pattern.
8059 (X * copysign (1.0, -X)): New pattern.
8060 (copysign (-1.0, CST)): New pattern.
8062 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8064 * genmultilib (combination_space): Remove variable.
8065 Validate reuse rules against regular expression for any sequence
8066 of multilib options in any order.
8068 2017-06-27 Michael Collison <michael.collison@arm.com>
8070 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8071 call aarch64_split_simd_combine.
8072 * (aarch64_combine_internal<mode>): Delete pattern.
8073 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8074 Allow register and subreg operands.
8076 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8078 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8079 specific need, just fallback on defaults.
8080 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8082 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8083 Olivier Hainque <hainque@adacore.com>
8085 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8087 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8088 targets. Pick a default if no particular attempt applied.
8089 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8092 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8094 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8095 (x86_64-wrs-vxworks7): Likewise.
8097 2017-06-27 Marek Polacek <polacek@redhat.com>
8100 * ubsan.c (instrument_null): Check get_base_address's result for null.
8102 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8104 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8106 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8108 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8109 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8110 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8111 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8113 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8114 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8115 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8116 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8117 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8118 BUILT_IN_FEUPDATEENV): New builtins.
8119 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8120 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8121 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8122 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8124 (builtin_structptr_types): Adjust size.
8125 * tree.c (builtin_structptr_types): Add four entries.
8127 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8128 Olivier Hainque <hainque@adacore.com>
8130 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8131 (TLS_SYM): New local macro, forcing reference to __tls__ on
8132 link command lines for VxWorks 7 RTPs, triggering initialization
8134 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8135 OS features TLS support, true for RTPs on VxWorks 7.
8136 * config/vxworks.c (vxworks_override_options): Setup emutls
8139 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8141 * predict.c (test_prediction_value_range): Use -1U instead of -1
8142 to avoid narrowing conversion warning.
8143 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8144 to avoid narrowing conversion warning.
8145 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8147 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8149 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8151 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8152 64bit configurations.
8153 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8154 (SIZE_TYPE): Likewise.
8155 * config/vxworks.c (vxworks_emutls_var_fields): Use
8156 long_unsigned_type_node instead of unsigned_type_node as the offset
8157 field type, which is "pointer" mode in emutls.c.
8159 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8162 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8165 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8166 gimple_vuse copying separately from gimple_vdef copying.
8168 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8170 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8173 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8174 Olivier Hainque <hainque@adacore.com>
8176 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8177 for all vxworks7 targets.
8178 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8179 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8180 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8181 variations for VX6/VX7 and 32/64bits later on in ...
8182 (VXWORKS_LIB_SPEC): Leverage new macros.
8183 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8184 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8186 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8188 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8189 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8191 2017-06-26 Carl Love <cel@us.ibm.com>
8193 * config/rs6000/rs6000-c.c: Add support for built-in functions
8194 vector bool char vec_reve (vector bool char);
8195 vector signed char vec_reve (vector signed char);
8196 vector unsigned char vec_reve (vector unsigned char);
8197 vector bool int vec_reve (vector bool int);
8198 vector signed int vec_reve (vector signed int);
8199 vector unsigned int vec_reve (vector unsigned int);
8200 vector bool long long vec_reve (vector bool long long);
8201 vector signed long long vec_reve (vector signed long long);
8202 vector unsigned long long vec_reve (vector unsigned long long);
8203 vector bool short vec_reve (vector bool short);
8204 vector signed short vec_reve (vector signed short);
8205 vector double vec_reve (vector double);
8206 vector float vec_reve (vector float);
8207 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8208 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8209 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8210 (altivec_vreve): New pattern.
8211 * config/rs6000/altivec.h (vec_reve): New define.
8212 * doc/extend.texi (vec_rev): Update the built-in documentation file
8213 for the new built-in functions.
8215 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8217 PR tree-optimization/71815
8218 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8220 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8222 (slsr_process_phi): Likewise.
8223 (replace_uncond_cands_and_profitable_phis): Don't replace a
8224 multiply candidate with a stride of 1 (copy or cast).
8225 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8227 (lowest_cost_path): Likewise.
8228 (total_savings): Likewise.
8230 2017-06-26 Richard Biener <rguenther@suse.de>
8233 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8234 Use def_builtin_pure for all gather builtins.
8236 2017-06-26 Richard Biener <rguenther@suse.de>
8238 PR tree-optimization/81203
8239 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8242 2017-06-26 Marek Polacek <polacek@redhat.com>
8245 * doc/invoke.texi: Document -Wmultistatement-macros.
8247 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8249 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8250 arm_neon_ok_no_float_abi effective target.
8252 2017-06-26 Richard Biener <rguenther@suse.de>
8254 PR tree-optimization/80928
8255 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8256 (copy_bbs): Set BB_DUPLICATED flag early.
8257 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8259 (execute_on_shrinking_pred): Likewise.
8260 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8261 BB_DUPLICATED blocks.
8262 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8263 iterate over all PHIs considering removal of *gsi.
8265 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8267 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8270 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8272 * config/rs6000/rs6000-string.c: (expand_block_clear,
8273 do_load_for_compare, select_block_compare_mode,
8274 compute_current_alignment, expand_block_compare,
8275 expand_strncmp_align_check, expand_strn_compare,
8276 expand_block_move, rs6000_output_load_multiple)
8277 Move functions related to string/block move/compare
8279 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8280 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8281 for this function which is now used in two files.
8282 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8283 * config.gcc: Add rs6000-string.o to extra_objs for
8284 targets powerpc*-*-* and rs6000*-*-*.
8286 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8289 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8290 32-bit, since indexed is not valid for DImode.
8291 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8292 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8293 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8294 (define_peephole2 for Altivec d-form store): Likewise.
8297 * multiple_target.c (create_dispatcher_calls): Only create the
8298 dispatcher call if the function is the default clone of a
8301 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8304 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8305 a call, force the call to not be a tail call.
8307 2017-06-23 Jeff Law <law@redhat.com>
8309 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8312 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8314 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8315 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8316 handling for early expansion of vector shifts (sl,sr,sra,rl).
8317 (builtin_function_type): Add vector shift right instructions
8318 to the unsigned argument list.
8320 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8322 rtl-optimizatoin/79286
8323 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8324 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8325 trap. PIC register plus a const unspec without offset can never trap.
8327 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8329 * tree.h (builtin_structptr_type): New type.
8330 (builtin_structptr_types): Declare new array.
8331 * tree.c (builtin_structptr_types): New array.
8332 (free_lang_data, build_common_tree_nodes): Use it.
8334 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8337 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8340 2017-06-22 Matt Turner <mattst88@gmail.com>
8342 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8343 Lake models to skylake case. Assume skylake for unknown
8344 models with clflushopt.
8346 2017-06-22 Jeff Law <law@redhat.com>
8348 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8349 frame sizes that do not satisfy aarch64_uimm12_shift.
8351 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8353 * profile-count.h (apply_probability,
8354 apply_scale, probability_in): Fix checks for zero.
8356 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8358 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8359 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8361 2016-06-22 Richard Biener <rguenther@suse.de>
8363 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8364 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8365 REDUC_MAX_EXPR support.
8366 (vectorizable_reduction): Likewise.
8367 (vect_create_epilog_for_reduction): Likewise.
8369 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8371 * match.pd (A / (1 << B) -> A >> B): New.
8372 * generic-match-head.c: Include optabs-tree.h.
8373 * gimple-match-head.c: Likewise.
8374 * optabs-tree.h (target_supports_op_p): New.
8375 * optabs-tree.c (target_supports_op_p): New.
8377 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8379 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8380 $gcc_cv_ld --help output.
8381 (gcc_cv_ld_demangle): Likewise.
8382 (gcc_cv_ld_eh_frame_hdr): Likewise.
8383 (gcc_cv_ld_pie): Likewise.
8384 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8385 (gcc_cv_ld_buildid): Likewise.
8386 (gcc_cv_ld_sysroot): Likewise.
8387 (ld_bndplt_support): Likewise.
8388 (ld_pushpopstate_support): Likewise.
8389 * configure: Regenerate.
8390 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8392 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8395 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8396 operands indexes to avoid gap between operands and match_dups.
8398 2017-06-21 Andrew Pinski <apinski@cavium.com>
8400 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8401 Increment Arith_shift and Arith_shift_reg by 1.
8402 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8404 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8405 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8406 (aarch64_strip_extend): Add new argument and test for it.
8407 (aarch64_cheap_mult_shift_p): New function.
8408 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8409 add a cost if it is true.
8410 Update calls to aarch64_strip_extend.
8411 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8413 2017-06-21 Andrew Pinski <apinski@cavium.com>
8415 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8417 (thunderxt88): Likewise.
8418 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8419 (thunderx_prefetch_tune): New variable.
8420 (thunderx2t99_prefetch_tune): Update for the correct values.
8421 (thunderxt88_tunings): New variable.
8422 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8423 generic_prefetch_tune.
8424 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8426 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8428 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8429 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8430 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8431 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8432 (aarch64_atomic_cas<mode>, GPI): Likewise.
8434 2017-06-21 Martin Liska <mliska@suse.cz>
8436 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8437 statements on cold and hot labels.
8438 * predict.c (tree_estimate_probability_bb): Remove the
8439 prediction from this place.
8441 2017-06-21 Martin Liska <mliska@suse.cz>
8443 PR tree-optimization/79489
8444 * gimplify.c (maybe_add_early_return_predict_stmt): New
8446 (gimplify_return_expr): Call the function.
8447 * predict.c (tree_estimate_probability_bb): Remove handling
8449 * predict.def: Update comment about early return predictor.
8450 * gimple-predict.h (is_gimple_predict): New function.
8451 * predict.def: Change default value of early return to 66.
8452 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8454 * passes.def: Put pass_strip_predict_hints to the beginning of
8457 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8459 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8460 FUNCTION_DECL declarations.
8461 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8463 (dwaf2out_decl): Likewise.
8464 * godump.c (go_early_global_decl): Skip call to the real debug hook
8465 for FUNCTION_DECL declarations.
8466 * passes.c (rest_of_decl_compilation): Skip call to the
8467 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8468 -fdump-go-spec is passed.
8470 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8472 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8473 Reorder for compactness.
8474 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8475 (def_builtin_pure, def_builtin_pure2): New functions.
8476 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8478 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8480 * match.pd (nop_convert): New predicate.
8481 ((A +- CST1) +- CST2): Allow some NOP conversions.
8483 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8486 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8487 with ctors/dtors if GOVD_SHARED is set.
8489 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8491 * config/aarch64/aarch64.md (movti_aarch64):
8492 Emit mov rather than orr.
8493 (movtf_aarch64): Likewise.
8494 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8495 Emit mov rather than orr.
8497 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8499 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8500 Swap alternatives, make integer dup more expensive.
8502 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8504 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8505 Return true for non-tls symbols.
8507 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8509 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8510 (cortex-a75): Likewise.
8511 (cortex-a75.cortex-a55): Likewise.
8512 * config/aarch64/aarch64-tune.md: Regenerate.
8513 * doc/invoke.texi (-mtune): Document new values for -mtune.
8515 2017-06-21 Tom de Vries <tom@codesourcery.com>
8517 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8519 (Effective-Target Keywords, Other attributes): Suggest using
8520 dg-add-options stack_size feature to get stack limit in stack_size
8521 effective target documentation.
8523 2017-06-21 Julian Brown <julian@codesourcery.com>
8524 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8526 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8527 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8528 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8530 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8531 attribute type list for neon_multiply.
8532 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8533 attribute type list for neon_multiply.
8534 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8535 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8536 attribute type list for neon_multiply.
8537 * config/arm/types.md (crypto_pmull): Add.
8538 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8539 attribute type list.
8541 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8543 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8546 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8548 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8549 to make sure not to dereference a NULL cost_classes_ptr pointer.
8551 2017-06-20 Carl Love <cel@us.ibm.com>
8553 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8554 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8555 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8556 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8557 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8558 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8559 VMULOSW): New enum "unspec" values.
8560 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8561 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8562 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8563 altivec_vmulosw): New patterns.
8564 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8565 VMULOSW): Add definitions.
8567 2017-06-20 Julia Koval <julia.koval@intel.com>
8569 * config/i386/i386.c: Fix rounding expand for new pattern.
8570 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8572 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8574 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8575 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8577 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8579 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8582 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8584 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8585 architecture, then by implementer ID.
8586 * config/aarch64/aarch64-tune.md: Regenerate.
8588 2017-06-20 Richard Biener <rguenther@suse.de>
8591 * fold-const.c (split_tree): Fold to type before negating.
8593 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8595 * diagnostic-show-locus.c
8596 (selftest::test_fixit_deletion_affecting_newline): New function.
8597 (selftest::diagnostic_show_locus_c_tests): Call it.
8599 2017-06-20 Andreas Schwab <schwab@suse.de>
8602 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8605 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8607 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8608 __ARM_FEATURE_COPROC according to support.
8610 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8612 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8613 Rewritten to avoid overflow for > 32-bit pointers.
8616 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8617 by removing enum keyword.
8618 (ubsan_type_descriptor): Likewise. Formatting fix.
8621 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8622 splitter): Require TARGET_SSE2 in the condition.
8624 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8627 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8628 for doing vector set of SFmode on ISA 3.0.
8629 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8630 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8632 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8633 SFmode value into a V4SF variable that was extracted from another
8634 V4SF variable without converting the element to double precision
8635 and back to single precision vector format.
8636 (vsx_insert_extract_v4sf_p9_2): Likewise.
8638 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8640 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8641 in UWHI to avoid undefined overflow.
8644 * ubsan.h (enum ubsan_encode_value_phase): New.
8645 (ubsan_encode_value): Change second argument to
8646 enum ubsan_encode_value_phase with default value of
8647 UBSAN_ENCODE_VALUE_GENERIC.
8648 * ubsan.c (ubsan_encode_value): Change second argument to
8649 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8650 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8651 create_tmp_var_raw instead of create_tmp_var and use a
8653 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8654 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8655 ubsan_encode_value callers.
8658 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8659 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8660 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8662 2017-06-19 Richard Biener <rguenther@suse.de>
8665 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8666 estimates if we changed anything.
8668 2017-06-19 Richard Biener <rguenther@suse.de>
8670 PR tree-optimization/80887
8671 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8672 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8673 simplified lookups, then reset mprts_hook.
8674 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8676 (try_to_simplify): Likewise.
8678 2017-06-19 Martin Liska <mliska@suse.cz>
8681 * gimplify.c (gimplify_switch_expr):
8682 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8684 2017-06-19 Martin Liska <mliska@suse.cz>
8686 * doc/install.texi: Document that PGO runs in 4 stages.
8688 2017-06-19 Martin Liska <mliska@suse.cz>
8691 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8692 to dispatcher function name.
8693 * multiple_target.c (replace_function_decl): New function.
8694 (create_dispatcher_calls): Redirect both edges and references.
8696 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8698 * profile-count.c (profile_count::dump): Dump quality.
8699 (profile_count::differs_from_p): Update for unsigned val.
8700 * profile-count.h (profile_count_quality): New enum.
8701 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8703 2017-06-19 Richard Biener <rguenther@suse.de>
8705 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8706 struct function as arg.
8707 (estimate_numbers_of_iterations): Export overload with loop arg.
8708 (free_numbers_of_iterations_estimates_loop): Use an overload of
8709 free_numbers_of_iterations_estimates instead.
8710 * tree-cfg.c (remove_bb): Adjust.
8711 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8712 * tree-parloops.c (gen_parallel_loop): Likewise.
8713 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8715 (tree_unroll_loops_completely): Likewise.
8716 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8717 Use an overload instead and export.
8718 (estimated_loop_iterations): Adjust.
8719 (max_loop_iterations): Likewise.
8720 (likely_max_loop_iterations): Likewise.
8721 (estimate_numbers_of_iterations): Take struct function as arg
8723 (loop_exits_before_overflow): Adjust.
8724 (free_numbers_of_iterations_estimates_loop): Use an overload.
8725 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8726 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8728 2017-06-19 Richard Biener <rguenther@suse.de>
8731 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8732 RANGE_EXPR conservatively.
8734 2017-06-16 Carl Love <cel@us.ibm.com>
8736 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8737 definitions for vec_float, vec_float2, vec_floato,
8738 vec_floate built-ins.
8739 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8740 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8742 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8743 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8744 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8745 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8746 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8747 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8748 vec_floato): Add builtin defines.
8749 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8750 Update the built-in documentation file for the new built-in
8753 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8755 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8756 (mthumb): Mark as the negative of -marm.
8758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8760 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8762 (ARM Options, -mtune): Document that this accepts the same
8763 extension options as -mcpu.
8764 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8768 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8769 permitted extensions.
8771 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8773 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8774 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8775 (armv8-m.main): Add option +nodsp.
8776 * config/arm/arm-cpu-cdata.h: Regenerated.
8778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8780 * config/arm/t-fuchsia: New file.
8781 * config.gcc (arm*-*-fuchsia*): Use it.
8783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8785 * config/arm/t-symbian: Rewrite for new option infrastructure.
8787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8789 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
8790 (MULTILIB_REQUIRED): Likewise.
8792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8794 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
8795 (MULTILIB_RESUE): Likewise.
8796 (MULTILIB_MATCHES): Likewise.
8797 (MULTLIB_REQUIRED): Likewise.
8799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8801 * config/arm/t-rtems: Rewrite for new option framework.
8803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8805 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
8806 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
8807 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
8808 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
8809 * config/arm/t-multilib: ... here.
8810 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
8811 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
8812 armv7-a and armv8*-a when A-profile libraries have not been built.
8813 * config/arm/t-rmprofile: Rewrite.
8815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8817 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
8818 with a backslash. Remove the backslash after substituting unescaped
8820 * doc/fragments.texi (MULTILIB_REUSE): Document it.
8822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8824 * config.gcc: (arm*-*-*): When building a-profile libraries, force
8825 the driver to pass through the default setting of -mfloat-abi.
8826 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
8828 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
8829 (all_feat_combs): New rule.
8830 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
8832 * config/arm/t-aprofile: Rewrite.
8834 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8836 * config/arm/arm.h (FPUTYPE_AUTO): Define.
8837 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
8838 fpu is not specified by the user/command-line.
8839 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
8840 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
8841 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
8842 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
8843 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
8844 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
8846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8848 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
8849 * config/arm/t-arm-elf: Rewritten.
8851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8853 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
8854 have some floating-point instructions.
8855 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
8856 (TARGET_MAYBE_HARD_FLOAT): New macro.
8857 * config/arm/arm-builtins.c (arm_init_builtins): Use
8858 TARGET_MAYBE_HARD_FLOAT.
8859 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
8861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8863 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
8864 (configargs.h): Include it.
8865 (arm_print_hint_for_fpu_option): New function.
8866 (arm_parse_fpu_option): New function.
8867 (candidate_extension): New class.
8868 (arm_canon_for_multilib): New function.
8869 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
8870 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
8871 (ARCH_CANONICAL_SPECS): New macro.
8872 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
8874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8876 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
8877 are set after handling multilib fragments. Set target_cpu_default2
8880 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8882 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
8884 (arm*-*-*): Set target_cpu_default2 to a quoted string.
8885 * config/arm/parsecpu.awk (check_cpu): Validate any extension
8887 (check_arch): Likewise.
8888 * config/arm/arm.c (arm_configure_build_target): Handle
8889 TARGET_CPU_DEFAULT being a string constant. Scan any feature
8890 options in the default.
8892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8894 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
8895 when an option is an alias of another.
8896 * config/arm/parsecpu.awk (optalias): New parser token.
8897 (gen_comm_data): Mark non-alias options as such. Emit entries
8898 for extension aliases.
8899 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
8900 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
8901 (armv6kz, armv6zk, armv6t2): Likewise.
8902 (armv7): Make vfpv3-d16 an alias.
8903 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
8905 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
8906 Sort in canonical order.
8907 (armv8-a): Sort in canonical order.
8908 (armv8.1-a, armv8.2-a): Likewise.
8909 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
8911 (cortex-a9): Sort in canonical order.
8912 * config/arm/arm.c (selftests.h): Include it.
8913 (arm_test_cpu_arch_data): New function.
8914 (arm_run_self_tests): New function.
8915 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
8916 (targetm): Move declaration to the end of the file.
8917 * arm-cpu-cdata.h: Regenerated.
8919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8921 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
8922 call to target_mode_check describing the type of option passed.
8923 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
8924 (arm_target_thumb_only): Use arm_parse_arch_option_name or
8925 arm_parse_cpu_option_name to match parameters against list of
8927 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
8928 arm_arch_core_flags data structure.
8929 * config/arm/arm-cpu_cdata.h: Regenerated.
8931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8933 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
8935 (arm_print_hint_for_cpu_option): Likewise.
8936 (arm_print_hint_for_arch_option): Likewise.
8937 (arm_parse_cpu_option_name): Likewise.
8938 (arm_parse_arch_option_name): Likewise.
8939 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
8940 of entries in the all_fpus list.
8941 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
8942 (arm_parse_cpu_option_name): Declare.
8943 (arm_parse_arch_option_name): Declare.
8944 (arm_parse_option_features): Declare.
8945 (arm_intialize_isa): Declare.
8946 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
8948 (gen_comm_data): ... here. Make definitions non-static.
8949 * config/arm/arm-cpu-data.h: Regenerated.
8950 * config/arm/arm-cpu-cdata.h: Regenerated.
8952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8954 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
8955 (cpu_arch_extension): New structure.
8956 (cpu_arch_option, arch_option, cpu_option): New structures.
8957 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
8959 (gen_data): Generate new format data tables.
8960 * config/arm/arm.c (cpu_tune): New structure.
8961 (cpu_option, processors): Delete.
8962 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
8963 (arm_print_hint_for_cpu_option): ... this and ...
8964 (arm_print_hint_for_arch_option): ... this.
8965 (arm_parse_arch_cpu_name): Delete. Replace with ...
8966 (arm_parse_cpu_option_name): ... this and ...
8967 (arm_parse_arch_option_name): ... this.
8968 (arm_unrecognized_feature): Change type of target parameter to
8970 (arm_parse_arch_cpu_features): Delete. Replace with ...
8971 (arm_parse_option_features): ... this.
8972 (arm_configure_build_target): Rework to use new configuration data
8974 (arm_print_tune_info): Rework for new configuration data tables.
8975 * config/arm/arm-cpu-data.h: Regenerated.
8976 * config/arm/arm-cpu.h: Regenerated.
8978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8980 * Makefile.in (OBJS): Move sbitmap.o from here ...
8981 (OBJS-libcommon): ... to here.
8983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8985 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
8986 (ISA_ALL_CRYPTO): New macro.
8987 (ISA_ALL_SIMD): New macro
8988 (ISA_ALL_FP): New macro.
8989 * config/arm/arm.c (fpu_bitlist): Update initializer.
8990 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
8992 (arm9e): Add fpu. Add option for nofp
8993 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
8994 (arm926ej-s, arm1026ej-s): Likewise.
8995 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
8996 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
8997 neon-fp16, neon-vfpv4, nofp and nosimd.
8998 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
8999 (cortex-a8): Add fpu. Add option for nofp.
9000 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9001 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9002 (cortex-r4f): Add fpu.
9003 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9004 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9006 (cortex-r8): Likewise.
9007 (cortex-m4): Add fpu. Add option for nofp.
9008 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9009 (cortex-a17.cortex-a7): Likewise.
9010 (cortex-a32): Add fpu. Add options for crypto and nofp.
9011 (cortex-a35, cortex-a53): Likewise.
9012 (cortex-a57): Add fpu. Add option for crypto.
9013 (cortex-a72, cortex-a73): Likewise.
9014 (exynos-m1): Likewise.
9015 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9016 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9017 (cortex-m33): Add fpu. Add option for nofp.
9018 * config/arm/arm-cpu-cdata.h: Regenerated
9019 * config/arm/arm-cpu-data.h: Regenerated.
9021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9023 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9024 (armv5te, armv5tej): Likewise.
9025 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9026 (armv7): Add options fp and vfpv3-d16.
9027 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9028 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9030 (armv7ve): Likewise.
9031 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9032 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9033 (armv8-a): Add nocrypto option.
9034 (armv8.1-a, armv8.2-a): Likewise.
9035 (armv8-m.main): add options fp, fp.dp and nofp.
9037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9039 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9041 (armv8-a+crc): Delete.
9042 (armv8.1-a): Add options simd, crypto and nofp.
9043 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9044 (armv8.2-a+fp16): Delete.
9045 (armv8-m.main): Add option dsp.
9046 (armv8-m.main+dsp): Delete.
9047 (cortex-a8): Add fpu. Add nofp option.
9048 (cortex-a9): Add fpu. Add nofp and nosimd options.
9049 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9050 link to main cpu and architecture data structures.
9051 (gen_comm_data): Only put isa attributes from the main architecture
9053 (option): New statement for architecture and CPU entries.
9054 * arm.c (struct cpu_option): New structure.
9055 (struct processors): Add entry for options.
9056 (arm_unrecognized_feature): New function.
9057 (arm_parse_arch_cpu_name): Ignore any characters after the first
9059 (arm_parse_arch_cpu_feature): New function.
9060 (arm_configure_build_target): Separate out any CPU and architecture
9061 features and parse separately. Don't error out if -mfpu=auto is
9062 used with only an architecture string.
9063 (arm_print_asm_arch_directives): New function.
9064 (arm_file_start): Call it.
9065 * config/arm/arm-cpu-cdata.h: Regenerated.
9066 * config/arm/arm-cpu-data.h: Likewise.
9067 * config/arm/arm-tables.opt: Likewise.
9069 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9071 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9072 assembler when it is not -mfpu=auto.
9074 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9076 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9077 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9078 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9079 (ASM_CPU_SPEC): Rewrite.
9080 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9081 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9082 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9083 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9084 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9085 copied string is NUL-terminated. Also strip any characters prefixed
9087 (arm_rewrite_selected_arch): New function.
9088 (arm_rewrite_march): New function.
9090 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9092 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9093 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9094 (march, mcpu, mtune): Convert to string-based options.
9095 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9096 (arm_parse_arch_cpu_name): New function.
9097 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9098 identify selected architecture or CPU.
9099 (arm_option_save): New function.
9100 (TARGET_OPTION_SAVE): Redefine.
9101 (arm_option_restore): Restore string options.
9102 (arm_option_print): Print string options.
9104 2017-06-16 Martin Sebor <msebor@redhat.com>
9106 PR tree-optimization/80933
9107 PR tree-optimization/80934
9108 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9109 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9110 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9111 (gimple_fold_builtin): Call them.
9113 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9115 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9116 as unlikely; update profile.
9118 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9120 * predict.c (force_edge_cold): Handle declaring edges impossible
9123 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9125 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9127 (try_unroll_loop_completely): Fix reporting.
9129 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9131 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9133 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9136 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9137 if given a non-constant argument for an intrinsic which requires a
9140 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9142 * profile.c (compare_freqs): New function.
9143 (branch_prob): Sort edge list.
9144 (find_spanning_tree): Assume that the list is priority sorted.
9146 2017-06-16 Richard Biener <rguenther@suse.de>
9148 PR tree-optimization/81090
9149 * passes.def (pass_record_bounds): Remove.
9150 * tree-pass.h (make_pass_record_bounds): Likewise.
9151 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9152 make_pass_record_bounds): Likewise.
9153 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9154 not free niter estimates at the beginning but at the end.
9155 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9157 2017-06-16 Richard Biener <rguenther@suse.de>
9159 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9160 initializer to workaround ICE in host GCC 4.8.
9162 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9164 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9166 (clone_inlined_nodes): Update.
9168 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9170 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9171 prefetch settings, and enable prefetching by default at -O3.
9173 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9175 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9176 Set flag_prefetch_loop_arrays according to tuning data.
9178 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9180 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9182 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9183 [Unrelated to main purpose of the patch] Place the pointer field last
9184 to enable type checking errors when tune structure are wrongly merged.
9185 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9186 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9187 (thunderx2t99_prefetch_tune): New tune constants.
9188 (tune_params *_tunings): Update all tunings (no functional change).
9189 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9190 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9191 from tunings structures.
9193 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9196 * ubsan.c (instrument_null): Add T argument, use it instead
9197 of computing it based on IS_LHS.
9198 (instrument_object_size): Likewise.
9199 (pass_ubsan::execute): Adjust instrument_null and
9200 instrument_object_size callers to pass gimple_get_lhs or
9201 gimple_assign_rhs1 result to it. Use instrument_null instead of
9202 calling get_base_address and instrument_mem_ref. Handle
9203 aggregate call arguments for object-size sanitization.
9205 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9207 PR tree-optimization/81089
9208 * tree-vrp.c (is_masked_range_test): Validate operands of
9211 2017-06-15 Martin Sebor <msebor@redhat.com>
9214 * dumpfile.c (dump_register): Avoid calling memset to initialize
9215 a class with a default ctor.
9216 * gcc.c (struct compiler): Remove const qualification.
9217 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9218 * hash-table.h: Ditto.
9219 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9221 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9222 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9224 * params.h (struct param_info): Make struct members non-const.
9225 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9226 with default initialization.
9227 * vec.h (vec_copy_construct, vec_default_construct): New helper
9229 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9230 with vec_copy_construct.
9231 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9232 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9233 * doc/invoke.texi (-Wclass-memaccess): Document.
9235 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9237 * emit-rtl.h (is_leaf): Update comment about local
9240 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9243 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9244 for a variable to have a section before checking if the section has a
9246 Set section to.persistent if persistent attribute is set.
9247 Warn if .persistent attribute is used on an automatic variable.
9249 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9251 PR rtl-optimization/80474
9252 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9254 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9256 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9259 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9261 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9262 (casesi): Emit bounds checking as RTL.
9263 (casesi_internal_mips16_<mode>): Remove bounds checking.
9265 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9267 * config/xtensa/xtensa.c (xtensa_option_override): Append
9268 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9269 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9270 xtensa_doloop_hooks): Define unconditionally.
9271 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9273 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9274 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9275 for it in xtensa_option_override.
9276 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9277 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9279 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9281 * doc/cppopts.texi: Document '-' special value to -MF.
9283 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9285 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9286 (cortex_a53_fconst): Likewise.
9287 (cortex_a53_fpmul): Likewise.
9288 (cortex_a53_f_load_64): Likewise.
9289 (cortex_a53_f_load_many): Likewise.
9290 (cortex_a53_advsimd_alu): Likewise.
9291 (cortex_a53_advsimd_alu_q): Likewise.
9292 (cortex_a53_advsimd_mul): Likewise.
9293 (cortex_a53_advsimd_mul_q): Likewise.
9294 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9295 Add missing fmul, r2f_cvt and fconst cases.
9297 2017-06-14 Richard Biener <rguenther@suse.de>
9300 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9302 (fold_binary_loc): When associating do not treat pre-existing
9303 TREE_OVERFLOW on literal constants as a reason to allow
9304 TREE_OVERFLOW on associated literal constants.
9306 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9308 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9309 (MASK_FEATURES): New macro.
9310 * config/sparc/sparc.c (sparc_option_override): Remove the special
9311 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9313 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9315 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9316 a division of 0 if non-call exceptions are enabled.
9318 2017-06-14 Andrew Pinski <apinski@cavium.com>
9319 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9322 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9323 Improve vector initialization code gen for only variable case.
9325 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9327 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9329 2017-06-14 Richard Biener <rguenther@suse.de>
9331 PR tree-optimization/81083
9332 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9335 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9337 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9338 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9339 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9340 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9341 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9342 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9344 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9346 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9347 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9349 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9351 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9353 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9355 * config/rs6000/t-rtems: Don't handle SPE.
9357 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9359 * config/rs6000/t-linux: Don't handle SPE.
9361 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9363 * config/rs6000/eabispe.h: Delete file.
9365 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9367 * config/rs6000/t-spe: Delete file.
9369 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9371 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9372 (rs6000_legitimate_offset_address_p): Return false for anything in
9373 V2SImode or V2SFmode.
9375 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9377 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9378 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9379 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9381 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9382 (rs6000_legitimate_offset_address_p): Ditto.
9383 (rs6000_emit_move): Ditto.
9384 (rs6000_init_builtins): Remove V4HI_type_node.
9386 2017-06-13 Martin Liska <mliska@suse.cz>
9389 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9390 (gate_asan): Likewise.
9391 * asan.h (asan_no_sanitize_address_p): Remove the function.
9392 (sanitize_flags_p): New function.
9393 * builtins.def: Fix coding style.
9394 * common.opt: Use renamed enum value.
9395 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9396 * doc/extend.texi: Document no_sanitize attribute.
9397 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9398 to SANITIZE_UNDEFINED_NONDEFAULT.
9399 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9400 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9401 Use sanitize_flags_p.
9402 * gimplify.c (gimplify_function_tree): Likewise.
9403 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9404 * opts.c (parse_no_sanitize_attribute): New function.
9405 (common_handle_option): Use renamed enum value.
9406 * opts.h (parse_no_sanitize_attribute): Declare.
9407 * tree.c (sanitize_flags_p): New function.
9408 * tree.h: Declared here.
9409 * tsan.c: Use sanitize_flags_p.
9410 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9411 (instrument_mem_ref): Likewise.
9412 (instrument_bool_enum_load): Likewise.
9413 (do_ubsan_in_current_function): Remove the function.
9414 (pass_ubsan::execute): Use sanitize_flags_p.
9415 * ubsan.h: Remove do_ubsan_in_current_function
9416 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9417 (dump_function_to_file): Use it here.
9419 2017-06-13 Martin Jambor <mjambor@suse.cz>
9421 PR tree-optimization/80803
9422 PR tree-optimization/81063
9423 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9424 (propagate_subaccesses_across_link): Enqueue subtree whenever
9425 necessary instead of relying on the caller.
9427 2017-06-13 Martin Jambor <mjambor@suse.cz>
9429 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9430 that have a first_link.
9431 (sort_and_splice_var_accesses): Do not check first_link before
9433 (subtree_mark_written_and_enqueue): Likewise.
9434 (propagate_all_subaccesses): Likewise and do not stop at first
9435 parent with a first_link.
9437 2017-06-13 Martin Jambor <mjambor@suse.cz>
9439 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9442 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9444 * match.pd: New pattern.
9446 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9448 * tree-vrp.c (is_masked_range_test): New function.
9449 (register_edge_assert_for): Determine ranges for
9452 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9454 PR tree-optimization/67328
9455 * fold-const.c (maskable_range_p): New function.
9456 (build_range_check): Generate bittests if possible.
9458 2017-06-13 Martin Liska <mliska@suse.cz>
9460 * gimple-pretty-print.c (dump_probability): Add new argument.
9461 (dump_edge_probability): Dump both probability and count.
9462 (dump_gimple_label): Likewise.
9463 (dump_gimple_bb_header): Likewise.
9465 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9468 * config/avr/avr-devices.c: Fix indentation.
9469 * config/avr/gen-avr-mmcu-specs.c: Dito.
9471 2017-06-13 Richard Biener <rguenther@suse.de>
9473 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9474 instead get vector type from stmt_info.
9475 (vectorizable_reduction): Adjust. Remove dead code.
9477 2017-06-13 Richard Biener <rguenther@suse.de>
9480 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9481 case of C * (x * C2 + C3).
9482 (fold_addr_of_array_ref_difference): Properly fold index difference.
9484 2017-06-12 David S. Miller <davem@davemloft.net>
9487 * config/sparc/sparc.md (return expander): Emit frame blockage if
9488 function uses alloca.
9490 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9492 * combine.c (make_field_assignment): Check len rather than the mode
9493 precision when calling force_to_mode.
9495 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9497 Support multilibs and devices that see flash in RAM address range.
9500 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9501 (avr_mcu_t) <flash_pm_offset>: New field.
9502 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9503 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9504 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9505 (AVR_TINY_PM_OFFSET): Remove macro.
9506 * config/avr/avr.opt (-mshort-calls): New option.
9507 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9508 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9509 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9510 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9511 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9512 instead of avr_arch->have_jmp_call.
9513 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9514 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9515 avr_arch->flash_pm_offset to define.
9516 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9517 new field flash_pm_offset. Add entry for avrxmega3.
9518 (avr_texinfo): Add entry for avrxmega3.
9519 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9520 attiny212, attiny214,
9521 attiny412, attiny414, attiny416, attiny417,
9522 attiny814, attiny816, attiny817,
9523 attiny1614, attiny1616, attiny1617,
9524 attiny3214, attiny3216, attiny3217.
9525 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9526 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9527 (avr_print_operand_address) [AVR_TINY]: Same.
9528 (avr_asm_init_sections) <readonly_data_section>: Only patch
9529 callback if avr_arch->flash_pm_offset = 0.
9530 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9531 for rodata if avr_arch->flash_pm_offset != 0.
9532 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9533 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9534 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9535 (m_options): Append opt_rcall.
9536 (m_dirnames): Append dir_rcall.
9537 * config/avr/t-multilib: Regenerate.
9539 * configure.ac [target=avr]: Check whether avrxmega3 default
9540 linker description file works as needed.
9541 * configure: Regenerate.
9542 * doc/avr-mmcu.texi: Regenerate.
9543 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9544 <__AVR_ARCH__>: Document avrxmega3 and 103.
9545 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9546 <__AVR_SHORT_CALLS__>: Document it.
9547 <__AVR_PM_BASE_ADDRESS__>: Document it.
9548 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9549 (AVR Variable Attributes) <progmem>: Document this is
9550 not needed for avrxmega3.
9551 (AVR Named Address Spaces) <__flash>: Dito.
9553 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9555 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9557 2017-06-12 Doug Rupp <rupp@adacore.com>
9559 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9560 Append vxworks-stdint.h to the tm_file list.
9561 * config/vxworks-stdint.h: New file.
9563 2017-06-12 Martin Liska <mliska@suse.cz>
9565 PR tree-optimization/81041
9566 * tree-profile.c (gimple_gen_ic_func_profiler):
9567 Create an extra BB in profile-generate
9568 (gimple_gen_time_profiler): Likewise.
9570 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9572 PR tree-optimization/81003
9573 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9574 (update_range_test): Use it instead of force_gimple_operand_gsi.
9576 2017-06-12 Richard Biener <rguenther@suse.de>
9578 PR tree-optimization/81053
9579 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9580 with backedge value not defined in loop. Simplify def stmt
9583 2017-06-11 Tom de Vries <tom@codesourcery.com>
9586 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9588 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9589 nvptx_cannot_force_const_mem.
9591 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9593 * opts.c (finish_options): Move test for flag_split_stack after
9594 it has been initialized.
9596 2017-06-11 Jason Merrill <jason@redhat.com>
9598 * tree.h (id_equal): New.
9599 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9600 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9601 instead of strcmp of IDENTIFIER_POINTER.
9603 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9605 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9606 (mark_all_inlined_calls_cdtor): Fix formating.
9607 (inline_transform): Rescale profile before inlining.
9609 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9611 * cgraph.h (cgraph_edge::clone): Update prototype.
9612 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9613 (cgraph_node::create_clone): Update.
9614 (cgraph_node::create_version_clone): Update.
9615 * tree-inline.c (copy_bb): Update.
9616 (expand_call_inline): Update.
9618 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9620 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9621 factored out from ...
9622 (rs6000_emit_prologue): ... here.
9624 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9626 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9627 factored out from ...
9628 (rs6000_emit_prologue): ... here.
9630 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9632 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9634 (handle_missing_profiles): Fix computation of tp_first_run.
9635 (counts_to_freqs): Do not touch freqs when count is 0.
9637 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9639 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9642 2017-06-10 Tom de Vries <tom@codesourcery.com>
9644 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9645 attributes): Document signal effective target.
9647 2017-06-10 Tom de Vries <tom@codesourcery.com>
9649 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9650 Document effective target stack_size.
9652 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9654 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9655 to the edit_context if they can be auto-applied.
9657 2017-06-9 Ian Lance Taylor <iant@golang.org>
9659 * opts.c (finish_options): If -fsplit-stack, disable implicit
9660 -forder-blocks-and-partition.
9661 * doc/invoke.texi (Optimize Options): Document that when using
9662 -fsplit-stack -forder-blocks-and-partition is not implicitly
9665 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9667 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9668 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9669 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9670 * builtins.def (abort, trap, unreachable): Declare cold.
9671 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9672 * tree-core.h (ECF_COLD): New.
9673 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9674 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9676 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9678 * predict.c (unlikely_executed_stmt_p): Cleanup.
9680 2017-06-09 Richard Biener <rguenther@suse.de>
9682 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9683 model if the ref is always written to.
9685 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9687 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9689 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9691 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9694 2017-06-09 Tom de Vries <tom@codesourcery.com>
9697 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9698 "target cannot support label values" when encountering LABEL_REF.
9700 2017-06-09 Martin Liska <mliska@suse.cz>
9702 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9703 (gimple_gen_ic_func_profiler): Emit direct comparison
9704 of __gcov_indirect_call_callee with NULL.
9705 (gimple_gen_time_profiler): Change probability from
9706 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9708 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9710 * profile.c (edge_gcov_counts): Turn to pointer.
9711 (compute_branch_probabilities, compute_branch_probabilities): Update.
9712 (branch_prob): Do not clear edge_gcov_count.
9713 * profile.h (edge_gcov_counts): Turn to pointer.
9714 (edge_gcov_count): Update.
9716 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9718 * gimple.h (gimple_check_failed): Mark cold.
9720 2017-06-09 Richard Biener <rguenther@suse.de>
9722 PR tree-optimization/66623
9723 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9724 refactor check_reduction into two parts, properly computing
9725 whether we have to check reduction validity for outer loop
9728 2017-06-09 Richard Biener <rguenther@suse.de>
9730 PR tree-optimization/79483
9731 * graphite-scop-detection.c (order): New global.
9732 (get_order): Compute bb to order mapping that satisfies code
9733 generation constraints.
9734 (cmp_pbbs): New helper.
9735 (build_scops): Start domwalk at entry block, sort generated
9738 2017-06-09 Richard Biener <rguenther@suse.de>
9741 * ipa-polymorphic-call.c
9742 (ipa_polymorphic_call_context::restrict_to_inner_class):
9743 Skip FIELD_DECLs with error_mark_node type.
9744 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9747 2017-06-09 Martin Liska <mliska@suse.cz>
9749 * predict.c (struct branch_predictor): New struct.
9750 (test_prediction_value_range): New test.
9751 (predict_c_tests): New function.
9752 * selftest-run-tests.c (selftest::run_tests): Run the function.
9753 * selftest.h: Declare new tests.
9755 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9758 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9759 gen_add3_insn did not fail.
9760 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9761 r0, construct that number in a temporary reg and add that reg to r0.
9762 If asked to put the result in r0 as well, fail.
9764 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9767 for early expansion of vec_eqv.
9769 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9772 * ubsan.c (instrument_null): Avoid pointless code temporary.
9773 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9775 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9779 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
9782 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
9785 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9787 * predict.c (unlikely_executed_edge_p): Move ahead.
9788 (probably_never_executed_edge_p): Use it.
9790 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9793 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
9794 gimple_call_builtin_p call.
9796 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9798 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
9799 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
9800 rtl_check_failed_type2, rtl_check_failed_code1,
9801 rtl_check_failed_code2, rtl_check_failed_code_mode,
9802 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
9803 rtvec_check_failed_bounds, rtl_check_failed_flag,
9804 _fatal_insn_not_found, _fatal_insn): Likewise.
9805 * tree.h (tree_contains_struct_check_failed,
9806 tree_check_failed, tree_not_check_failed,
9807 tree_class_check_failed, tree_range_check_failed,
9808 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
9809 tree_vec_elt_check_failed, phi_node_elt_check_failed,
9810 tree_operand_check_failed, omp_clause_check_failed,
9811 omp_clause_operand_check_failed, omp_clause_range_check_failed):
9814 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9816 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
9817 flag_branch_probabilities.
9818 * ipa-inline.c (edge_badness): Likewise.
9819 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9820 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
9821 * predict.c (maybe_hot_frequency_p): Likewise.
9822 (probably_never_executed): Likewise.
9823 * sched-ebb.c (schedule_ebbs): Likewise.
9824 * sched-rgn.c (find_single_block_region): Likewise.
9825 * tracer.c (tail_duplicate): Likewise.
9827 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9829 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
9830 longer requires x_flag_profile_use.
9832 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9834 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
9835 instead of flag_reorder_blocks_and_partition.
9836 * dbxout.c (dbxout_function_end): Likewise.
9837 * dwarf2out.c (gen_subprogram_die): Likewise.
9838 * haifa-sched.c (sched_create_recovery_edges): Likewise.
9839 * hw-doloop.c (reorg_loops): Likewise.
9840 * varasm.c (assemble_start_function,
9841 assemble_end_function): Likewise.
9842 (decide_function_section): Do not check for
9843 flag_reorder_blocks_and_partition.
9845 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9847 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9849 (chkp_get_hard_register_fake_addr_expr): Ditto.
9850 (chkp_build_addr_expr): Add check for hard reg case.
9851 (chkp_parse_array_and_component_ref): Ditto.
9852 (chkp_find_bounds_1): Ditto.
9853 (chkp_process_stmt): Don't generate bounds store for
9856 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9858 * predict.c (maybe_hot_bb_p): Do not check profile status.
9859 (maybe_hot_edge_p): Likewise.
9860 (probably_never_executed): Check for zero counts even if profile
9862 (unlikely_executed_edge_p): New function.
9863 (unlikely_executed_stmt_p): New function.
9864 (unlikely_executed_bb_p): New function.
9865 (set_even_probabilities): Use unlikely predicates.
9866 (combine_predictions_for_bb): Likewise.
9867 (predict_paths_for_bb): Likewise.
9868 (predict_paths_leading_to_edge): Likewise.
9869 (determine_unlikely_bbs): New function.
9870 (estimate_bb_frequencies): Use it.
9871 (compute_function_frequency): Use zero counts even if profile is
9873 * profile-count.h: Fix typo.
9875 2017-08-08 Julia Koval <julia.koval@intel.com>
9877 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
9878 _mm512_mask_cvtsepi16_storeu_epi8,
9879 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
9880 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
9881 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
9882 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
9883 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
9884 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
9885 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
9886 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
9887 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
9888 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
9889 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
9890 __builtin_ia32_pmovuswb256mem_mask,
9891 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
9892 __builtin_ia32_pmovwb512mem_mask): New builtins.
9894 2017-08-08 Julia Koval <julia.koval@intel.com>
9896 PR target/73350,80862
9897 * config/i386/subst.md (round): Fix round pattern.
9898 * config/i386/i386.c (ix86_erase_embedded_rounding):
9899 Fix erasing rounding for the fixed pattern.
9901 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9903 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
9905 2017-06-08 Martin Liska <mliska@suse.cz>
9907 PR gcov-profile/80911
9908 * gcov.c (block_info::block_info): New constructor.
9910 2017-06-07 Carl Love <cel@us.ibm.com>
9912 * config/rs6000/rs6000-c: The return type of the following
9913 built-in functions was implemented as int not long long. Fix sign
9914 of return value for the unsigned version of vec_mulo and vec_mule.
9915 vector unsigned long long vec_bperm (vector unsigned long long,
9916 vector unsigned char)
9917 vector signed long long vec_mule (vector signed int,
9919 vector unsigned long long vec_mule (vector unsigned int,
9920 vector unsigned int)
9921 vector signed long long vec_mulo (vector signed int,
9923 vector unsigned long long vec_mulo (vector unsigned int,
9924 vector unsigned int)
9925 * doc/extend.texi: Fix the documentation for the built-in
9928 2017-06-07 Carl Love <cel@us.ibm.com>
9931 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
9934 2017-06-07 Carl Love <cel@us.ibm.com>
9936 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
9937 support, Generate doublehv for signed int/float for BE case only.
9939 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
9941 * doc/invoke.texi (mcx16): Rewrite.
9943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9945 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
9946 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
9947 *mov<mode>_softfloat, and an anonymous splitter): Use
9948 nonimmediate_operand instead of rs6000_nonimmediate_operand.
9950 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9952 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
9954 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9955 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
9956 (rs6000_debug_reg_global): Adjust.
9957 (rs6000_init_hard_regno_mode_ok): Adjust.
9958 (rs6000_dbx_register_number): Adjust.
9959 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
9960 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9961 Remove SPE_ACC and SPEFSCR.
9962 (REG_ALLOC_ORDER): Ditto.
9963 (FRAME_POINTER_REGNUM): Change to 111.
9964 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
9965 (REG_CLASS_NAMES): Ditto.
9966 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
9967 (REGISTER_NAMES): Ditto.
9968 (ADDITIONAL_REG_NAMES): Ditto.
9969 (rs6000_reg_names): Ditto.
9970 * config/rs6000/rs6000.md: Renumber some register number
9973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9975 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
9977 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9978 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
9980 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
9981 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9982 Delete the SPE high registers.
9983 (REG_ALLOC_ORDER): Ditto.
9984 (enum reg_class): Remove SPE_HIGH_REGS.
9985 (REG_CLASS_NAMES): Ditto.
9986 (REG_CLASS_CONTENTS): Delete the SPE high registers.
9987 (REGISTER_NAMES): Ditto.
9988 (rs6000_reg_names): Ditto.
9989 * doc/tm.texi.in: Remove SPE as example.
9990 * doc/tm.texi: Regenerate.
9992 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9994 * config/rs6000/8540.md (ppc8540_brinc): Delete.
9995 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
9996 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
9997 * config/rs6000/rs6000.md (type): Remove "brinc".
9999 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10001 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10002 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10003 * config/rs6000/linuxspe.h: Delete file.
10004 * config/rs6000/rs6000.md: Don't include spe.md.
10005 * config/rs6000/spe.h: Delete file.
10006 * config/rs6000/spe.md: Delete file.
10007 * config/rs6000/t-rs6000: Remove spe.md.
10009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10011 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10012 (reg_or_none500mem_operand): Delete.
10013 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10014 instead of reg_or_none500mem_operand.
10016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10018 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10019 handling of SPE flags.
10020 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10024 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10026 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10027 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10028 paired_divv2sf3): Similar.
10029 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10030 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10031 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10033 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10034 Rename the paired_* instruction patterns.
10035 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10037 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10038 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10039 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10040 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10041 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10042 PAIRED_VECTOR_MODE.
10043 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10044 (spe_func_has_64bit_regs_p): Delete.
10045 (spe_expand_predicate_builtin): Delete.
10046 (spe_expand_evsel_builtin): Delete.
10047 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10048 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10049 (invalid_e500_subreg): Delete.
10050 (rs6000_legitimize_address): Always force_reg op2 as well, for
10051 paired single memory accesses.
10052 (rs6000_member_type_forces_blk): Delete.
10053 (rs6000_spe_function_arg): Delete.
10054 (rs6000_expand_unop_builtin): Delete SPE handling.
10055 (rs6000_expand_binop_builtin): Ditto.
10056 (spe_expand_stv_builtin): Delete.
10057 (bdesc_2arg_spe): Delete.
10058 (spe_expand_builtin): Delete.
10059 (spe_expand_predicate_builtin): Delete.
10060 (spe_expand_evsel_builtin): Delete.
10061 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10062 (spe_init_builtins): Delete.
10063 (spe_func_has_64bit_regs_p): Delete.
10064 (savres_routine_name): Delete "info" parameter. Adjust callers.
10065 (rs6000_emit_stack_reset): Ditto.
10066 (rs6000_dwarf_register_span): Delete.
10067 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10068 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10069 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10071 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10073 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10074 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10075 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10076 mulv2sf3, divv2sf3): Delete expanders.
10078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10080 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10084 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10085 * config/rs6000/rs6000.c: Ditto.
10087 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10089 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10090 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10091 comparison_operator.
10093 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10095 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10096 * config/rs6000/rs6000.opt: Ditto.
10097 * config/rs6000/t-rtems: Ditto.
10099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10101 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10102 TARGET_E500_SINGLE by 0, simplify.
10103 * config/rs6000/rs6000.c: Ditto.
10104 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10105 (spe_build_register_parallel): Delete.
10106 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10107 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10108 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10109 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10110 (E500_CONVERT): Delete.
10111 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10113 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10115 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10116 * config/rs6000/dfp.md: Ditto.
10117 (negdd2, *negdd2_fpr): Merge.
10118 (absdd2, *absdd2_fpr): Merge.
10119 (negtd2, *negtd2_fpr): Merge.
10120 (abstd2, *abstd2_fpr): Merge.
10121 * config/rs6000/e500.h: Delete file.
10122 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10123 TARGET_FPRS by 1 and simplify.
10124 * config/rs6000/rs6000-c.c: Ditto.
10125 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10126 TARGET_DF_SPE by 0.
10127 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10129 * config/rs6000/rs6000.md: Ditto.
10130 (floatdidf2, *floatdidf2_fpr): Merge.
10131 (move_from_CR_gt_bit): Delete.
10132 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10133 (E500_CR_IOR_COMPARE): Delete.
10134 (All patterns that require !TARGET_FPRS): Delete.
10135 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10137 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10139 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10141 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10143 * graphds.c (add_edge): Intitialize edge's attached data.
10144 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10145 pointer parameter. Call pointed function on each edge during
10146 graph traversing. Skip traversing the edge when the function
10148 (graphds_dfs, graphds_scc): Ditto.
10149 (for_each_edge): New parameter. Pass the new parameter to callback
10151 * graphds.h (skip_edge_callback): New function pointer type.
10152 (graphds_dfs, graphds_scc): New function pointer parameter.
10153 (graphds_edge_callback, for_each_edge): New parameter.
10155 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10157 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10158 out code checking if runtime alias check is possible to below ...
10159 Call the new function.
10160 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10161 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10163 2017-06-07 Marek Polacek <polacek@redhat.com>
10166 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10167 TYPE_OVERFLOW_WRAPS check.
10169 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10171 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10172 if versioning is required.
10173 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10174 peeling with the check for versioning.
10176 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10178 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10179 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10180 Set true to new parameter if new ssa variable is defined.
10181 (vect_gen_vector_loop_niters): Refactor. Set range information
10182 for the new vector loop bound variable.
10183 (vect_do_peeling): Ditto.
10185 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10187 * tree-affine.c (ssa.h): Include header file.
10188 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10189 has wrapping overflow behavior.
10191 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10193 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10195 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10197 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10198 (tree_to_aff_combination): ... here.
10200 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10202 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10203 reg_pressure model function.
10204 (ivopts_global_cost_for_size): Delete.
10205 (determine_set_costs, iv_ca_recount_cost): Call new model function
10206 ivopts_estimate_reg_pressure.
10208 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10210 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10211 expensive than udiv. Remove floating point cases from mod.
10213 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10215 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10216 Increase idiv cost.
10218 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10220 * config/aarch64/aarch64.md
10221 (copysignsf3): Fix mask generation.
10223 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10225 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10227 (class dump_manager): Add register_dumps method.
10228 * dumpfile.c: Include langhooks.h.
10229 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10230 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10231 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10232 (dump_manager::dump_register): Start with 512 entries instead of 32.
10233 (dump_manager::register_dumps): New method.
10234 * toplev.c (general_init): Instead of invoking register_dumps
10235 langhook, invoke register_dumps method on the dump manager.
10236 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10239 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10241 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10242 Say that # requires an associated define_split to exist, and that
10243 the define_split must be suitable for use after register allocation.
10245 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10247 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10248 (compute_outgoing_frequencies): Also initialize zero counts.
10249 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10250 around loops; preserve more of profile when nothing changes.
10252 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10254 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10256 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10257 * config/arm/arm-cpu-cdata.h: Regenerate.
10258 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10259 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10260 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10261 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10262 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10264 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10266 * config/arm/t-rmprofile: Likewise.
10267 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10269 2017-06-06 David S. Miller <davem@davemloft.net>
10272 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10273 blockage if function uses alloca.
10275 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10277 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10278 New "uid" fields to hold pretty-print IDs of group and ref.
10279 Memory references are now identified as <group_id>:<ref_id>
10280 instead of using [random] addresses.
10281 (dump_mem_details): Simplify, no functional change.
10282 (dump_mem_ref): Simplify and make output more concise.
10283 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10284 (find_or_create_group): Initialize group uid.
10285 (record_ref): Initialize ref uid. Improve debug output.
10286 (prune_group_by_reuse, should_issue_prefetch_p,)
10287 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10288 (mark_nontemporal_store, determine_loop_nest_reuse):
10289 Improve debug output.
10291 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10293 * dbgcnt.def (prefetch): New debug counter.
10294 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10295 (schedule_prefetches): Stop issueing prefetches if debug counter
10298 2017-06-06 Tom de Vries <tom@codesourcery.com>
10300 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10301 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10303 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10305 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10306 Use aarch64_reg_or_zero predicate for operand 4.
10307 (aarch64_compare_and_swap<mode> define_insn_and_split):
10308 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10309 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10311 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10313 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10314 (arm_compute_save_core_reg_mask): This.
10315 (thumb1_compute_save_reg_mask): Rename into ...
10316 (thumb1_compute_save_core_reg_mask): This.
10317 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10318 (arm_compute_frame_layout): Likewise.
10320 2017-06-06 Richard Biener <rguenther@suse.de>
10322 PR tree-optimization/80974
10323 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10324 keep or clear leaders SSA info.
10326 2017-06-06 Tom de Vries <tom@codesourcery.com>
10328 * config/nvptx/nvptx.c (split_mode_p): New function.
10329 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10331 2017-06-06 Tom de Vries <tom@codesourcery.com>
10333 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10335 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10338 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10341 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10343 * shrink-wrap.c (handle_simple_exit): Update profile.
10344 (try_shrink_wrapping): Upate profile.
10346 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10348 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10349 (tree_guess_outgoing_edge_probabilities): New.
10350 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10351 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10353 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10355 * ipa-split.c (split_function): Initialize return bb profile.
10357 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10359 * profile.c (compute_branch_probabilities): Also initialize
10360 EXIT_BLOCK profile.
10362 2017-06-06 Richard Biener <rguenther@suse.de>
10364 PR tree-optimization/80928
10365 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10366 (vect_analyze_loop_operations): Properly guard analysis for
10368 (vect_transform_loop): Likewise.
10369 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10370 (vect_model_induction_cost): Do not cost for pure SLP.
10371 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10372 of induction in inner loop vectorization.
10373 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10374 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10375 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10377 (vect_analyze_slp_cost_1): Cost induction.
10378 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10379 (vect_get_slp_vect_defs): Likewise.
10380 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10381 (vect_transform_stmt): Handle SLP reductions.
10382 * tree-vectorizer.h (vectorizable_induction): Adjust.
10384 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10386 * config/rs6000/rs6000.c (make_resolver_func): Update
10387 init_lowered_empty_function call.
10389 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10391 * doc/invoke.texi: Document the -fprofile-abs-path option.
10392 * common.opt (fprofile-abs-path): New option.
10393 * gcov-io.h (gcov_write_filename): Declare.
10394 * gcov-io.c (gcov_write_filename): New function.
10395 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10396 * profile.c (output_location): Likewise.
10398 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10400 * shring-wrap.c: Revert accidental commit.
10402 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10404 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10406 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10408 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10410 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10411 profile in callgraph edge.
10412 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10413 (apply_scale): Likewise.
10414 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10415 Also scale profile when inlining function with zero profile.
10416 (initialize_cfun): Update exit block profile even when it is zero.
10417 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10418 when profile is read.
10420 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10422 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10423 (CLONE_*): New constants to define the processors we can generate
10424 code for with the target_clone attribute.
10425 (rs6000_clone_map): New array to identify which clone processors
10426 the current program is running on.
10427 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10428 target_clone attribute.
10429 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10430 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10431 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10432 (cpu_expand_builtin): Add support for target_clone attribute.
10433 (rs6000_valid_attribute_p): Allow "default" attribute.
10434 (get_decl_name): New debug function to simplify printing the
10435 current function name in debugging statements.
10436 (rs6000_clone_priority): New functions to support the target_clone
10437 attribute, and be able to generate code to switch between ISA 2.05
10438 through ISA 3.0 (power6 through power9).
10439 (rs6000_compare_version_priority): Likewise.
10440 (rs6000_get_function_versions_dispatcher): Likewise.
10441 (make_resolver_func): Likewise.
10442 (add_condition_to_bb): Likewise.
10443 (dispatch_function_versions): Likewise.
10444 (rs6000_generate_version_dispatcher_body): Likewise.
10445 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10446 (fusion_gpr_load_p): Fix a spacing issue.
10447 * doc/extend.texi (Common Function Attributes): Document that the
10448 PowerPC supports the target_clone attribute.
10450 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10452 * config/arm/arm.h: explain F symbol found in description of ARM
10453 register allocation in its legend.
10455 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10457 * config/mips/frame-header-opt.c: Include profile-count.h.
10458 * config/riscv/riscv.c: Include profile-count.h
10460 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10462 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10464 (sm_set_flag_if_changed): Add bbs field.
10465 (execute_sm_if_changed_flag_set): Pass BBS.
10466 (execute_sm): Update.
10468 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10470 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10473 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10475 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10476 (peephole2): New peephole2 to emit the above.
10477 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10479 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10481 * config/aarch64/aarch64.c (define_peephole2 above
10482 *sub_<shift>_<mode>): New peephole.
10484 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10486 * config/i386/i386.c (make_resolver_func): Update.
10487 * Makefile.in: Add profile-count.h and profile-count.o
10488 * auto-profile.c (afdo_indirect_call): Update to new API.
10489 (afdo_set_bb_count): Update.
10490 (afdo_propagate_edge): Update.
10491 (afdo_propagate_circuit): Update.
10492 (afdo_calculate_branch_prob): Update.
10493 (afdo_annotate_cfg): Update.
10494 * basic-block.h: Include profile-count.h
10495 (struct edge_def): Turn count to profile_count.
10496 (struct basic_block_def): Likewie.
10497 (REG_BR_PROB_BASE): Move to profile-count.h
10498 (RDIV): Move to profile-count.h
10499 * bb-reorder.c (max_entry_count): Turn to profile_count.
10500 (find_traces): Update.
10501 (rotate_loop):Update.
10502 (connect_traces):Update.
10503 (sanitize_hot_paths):Update.
10504 * bt-load.c (migrate_btr_defs): Update.
10505 * cfg.c (RDIV): Remove.
10506 (init_flow): Use alloc_block.
10507 (alloc_block): Uninitialize count.
10508 (unchecked_make_edge): Uninitialize count.
10509 (check_bb_profile): Update.
10510 (dump_edge_info): Update.
10511 (dump_bb_info): Update.
10512 (update_bb_profile_for_threading): Update.
10513 (scale_bbs_frequencies_int): Update.
10514 (scale_bbs_frequencies_gcov_type): Update.
10515 (scale_bbs_frequencies_profile_count): New.
10516 * cfg.h (update_bb_profile_for_threading): Update.
10517 (scale_bbs_frequencies_profile_count): Declare.
10518 * cfgbuild.c (compute_outgoing_frequencies): Update.
10519 (find_many_sub_basic_blocks): Update.
10520 * cfgcleanup.c (try_forward_edges): Update.
10521 (try_crossjump_to_edge): Update.
10522 * cfgexpand.c (expand_gimple_tailcall): Update.
10523 (construct_exit_block): Update.
10524 * cfghooks.c (verify_flow_info): Update.
10525 (dump_bb_for_graph): Update.
10526 (split_edge): Update.
10527 (make_forwarder_block): Update.
10528 (duplicate_block): Update.
10529 (account_profile_record): Update.
10530 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10531 (get_estimated_loop_iterations): Update.
10532 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10533 (single_likely_exit): Update.
10534 * cfgloopmanip.c (scale_loop_profile): Update.
10536 (set_zero_probability): Update.
10537 (lv_adjust_loop_entry_edge): Update.
10538 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10539 (purge_dead_edges): Update.
10540 (rtl_account_profile_record): Update.
10541 * cgraph.c (cgraph_node::create): Uninitialize count.
10542 (symbol_table::create_edge): Uninitialize count.
10543 (cgraph_update_edges_for_call_stmt_node): Update.
10544 (cgraph_edge::dump_edge_flags): Update.
10545 (cgraph_node::dump): Update.
10546 (cgraph_edge::maybe_hot_p): Update.
10547 * cgraph.h: Include profile-count.h
10548 (create_clone), create_edge, create_indirect_edge): Update.
10549 (cgraph_node): Turn count to profile_count.
10550 (cgraph_edge0: Likewise.
10551 (make_speculative, clone): Update.
10552 (create_edge): Update.
10553 (init_lowered_empty_function): Update.
10554 * cgraphclones.c (cgraph_edge::clone): Update.
10555 (duplicate_thunk_for_node): Update.
10556 (cgraph_node::create_clone): Update.
10557 * cgraphunit.c (cgraph_node::analyze): Update.
10558 (cgraph_node::expand_thunk): Update.
10559 * final.c (dump_basic_block_info): Update.
10560 * gimple-streamer-in.c (input_bb): Update.
10561 * gimple-streamer-out.c (output_bb): Update.
10562 * graphite.c (print_global_statistics): Update.
10563 (print_graphite_scop_statistics): Update.
10564 * hsa-brig.c: Include basic-block.h.
10565 * hsa-dump.c: Include basic-block.h.
10566 * hsa-gen.c (T sum_slice): Update.
10567 (convert_switch_statements):Update.
10568 * hsa-regalloc.c: Include basic-block.h.
10569 * ipa-chkp.c (chkp_produce_thunks): Update.
10570 * ipa-cp.c (struct caller_statistics): Update.
10571 (init_caller_stats): Update.
10572 (gather_caller_stats): Update.
10573 (ipcp_cloning_candidate_p): Update.
10574 (good_cloning_opportunity_p): Update.
10575 (get_info_about_necessary_edges): Update.
10576 (dump_profile_updates): Update.
10577 (update_profiling_info): Update.
10578 (update_specialized_profile): Update.
10579 (perhaps_add_new_callers): Update.
10580 (decide_about_value): Update.
10581 (ipa_cp_c_finalize): Update.
10582 * ipa-devirt.c (struct odr_type_warn_count): Update.
10583 (struct decl_warn_count): Update.
10584 (struct final_warning_record): Update.
10585 (possible_polymorphic_call_targets): Update.
10586 (ipa_devirt): Update.
10587 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10588 * ipa-icf.c (sem_function::merge): Update.
10589 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10590 * ipa-inline.c (compute_uninlined_call_time): Update.
10591 (compute_inlined_call_time): Update.
10592 (want_inline_small_function_p): Update.
10593 (want_inline_self_recursive_call_p): Update.
10594 (edge_badness): Update.
10595 (lookup_recursive_calls): Update.
10596 (recursive_inlining): Update.
10597 (inline_small_functions): Update.
10598 (dump_overall_stats): Update.
10599 (dump_inline_stats): Update.
10600 * ipa-profile.c (ipa_profile_generate_summary): Update.
10601 (ipa_propagate_frequency): Update.
10602 (ipa_profile): Update.
10603 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10604 * ipa-utils.c (ipa_merge_profiles): Update.
10605 * loop-doloop.c (doloop_modify): Update.
10606 * loop-unroll.c (report_unroll): Update.
10607 (unroll_loop_runtime_iterations): Update.
10608 * lto-cgraph.c (lto_output_edge): Update.
10609 (lto_output_node): Update.
10610 (input_node): Update.
10611 (input_edge): Update.
10612 (merge_profile_summaries): Update.
10613 * lto-streamer-in.c (input_cfg): Update.
10614 * lto-streamer-out.c (output_cfg): Update.
10615 * mcf.c (create_fixup_graph): Update.
10616 (adjust_cfg_counts): Update.
10617 (sum_edge_counts): Update.
10618 * modulo-sched.c (sms_schedule): Update.
10619 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10620 * predict.c (maybe_hot_count_p): Update.
10621 (probably_never_executed): Update.
10622 (dump_prediction): Update.
10623 (combine_predictions_for_bb): Update.
10624 (propagate_freq): Update.
10625 (handle_missing_profiles): Update.
10626 (counts_to_freqs): Update.
10627 (rebuild_frequencies): Update.
10628 (force_edge_cold): Update.
10629 * predict.h: Include profile-count.h
10630 (maybe_hot_count_p, counts_to_freqs): UPdate.
10631 * print-rtl-function.c: Do not include cfg.h
10632 * print-rtl.c: Include basic-block.h
10633 * profile-count.c: New file.
10634 * profile-count.h: New file.
10635 * profile.c (is_edge_inconsistent): Update.
10636 (correct_negative_edge_counts): Update.
10637 (is_inconsistent): Update.
10638 (set_bb_counts): Update.
10639 (read_profile_edge_counts): Update.
10640 (compute_frequency_overlap): Update.
10641 (compute_branch_probabilities): Update; Initialize and deinitialize
10643 (branch_prob): Update.
10644 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10645 (edge_gcov_count): New.
10646 (bb_gcov_count): New.
10647 * shrink-wrap.c (try_shrink_wrapping): Update.
10648 * tracer.c (better_p): Update.
10649 * trans-mem.c (expand_transaction): Update.
10650 (ipa_tm_insert_irr_call): Update.
10651 (ipa_tm_insert_gettmclone_call): Update.
10652 * tree-call-cdce.c: Update.
10653 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10654 (gimple_duplicate_sese_tail): Update.
10655 (gimple_account_profile_record): Update.
10656 (execute_fixup_cfg): Update.
10657 * tree-inline.c (copy_bb): Update.
10658 (copy_edges_for_bb): Update.
10659 (initialize_cfun): Update.
10660 (freqs_to_counts): Update.
10661 (copy_cfg_body): Update.
10662 (expand_call_inline): Update.
10663 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10664 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10665 (try_unroll_loop_completely): Update.
10666 (try_peel_loop): Update.
10667 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10668 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10669 * tree-ssa-loop-split.c (connect_loops): Update.
10670 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10671 * tree-ssa-reassoc.c (branch_fixup): Update.
10672 * tree-ssa-tail-merge.c (replace_block_by): Update.
10673 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10674 (compute_path_counts): Update.
10675 (update_profile): Update.
10676 (recompute_probabilities): Update.
10677 (update_joiner_offpath_counts): Update.
10678 (estimated_freqs_path): Update.
10679 (freqs_to_counts_path): Update.
10680 (clear_counts_path): Update.
10681 (ssa_fix_duplicate_block_edges): Update.
10682 (duplicate_thread_path): Update.
10683 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10684 (struct switch_conv_info): Update.
10685 * tree-tailcall.c (decrease_profile): Update.
10686 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10687 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10688 * value-prof.c (check_counter): Update.
10689 (gimple_divmod_fixed_value): Update.
10690 (gimple_mod_pow2): Update.
10691 (gimple_mod_subtract): Update.
10692 (gimple_ic_transform): Update.
10693 (gimple_stringop_fixed_value): Update.
10694 * value-prof.h (gimple_ic): Update.
10696 2017-06-02 Carl Love <cel@us.ibm.com>
10698 * config/rs6000/rs6000-c: Add support for built-in functions
10699 vector double vec_doublee (vector signed int);
10700 vector double vec_doublee (vector unsigned int);
10701 vector double vec_doublee (vector float);
10702 vector double vec_doubleh (vector signed int);
10703 vector double vec_doubleh (vector unsigned int);
10704 vector double vec_doubleh (vector float);
10705 vector double vec_doublel (vector signed int);
10706 vector double vec_doublel (vector unsigned int);
10707 vector double vec_doublel (vector float);
10708 vector double vec_doubleo (vector signed int);
10709 vector double vec_doubleo (vector unsigned int);
10710 vector double vec_doubleo (vector float);.
10711 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10712 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10714 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10715 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10716 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10718 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10719 vec_doublel, vec_doubleh.
10720 * doc/extend.texi: Update the built-in documentation file for the
10721 new built-in functions.
10723 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10726 * ipa-inline-analysis.c (free_growth_caches): Set
10727 edge_removal_hook_holder to NULL after removing it.
10729 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10731 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10732 comparision with zero.
10734 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10735 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10736 for early expansion of vec_min and vec_max builtins.
10737 (builtin_function_type): Add min/max unsigned variants to those
10738 identified as having unsigned arguments.
10740 2017-06-02 Olivier Hainque <hainque@adacore.com>
10742 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10744 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10746 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10747 Use VALL_F16 iterator rather than VALL.
10749 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10751 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10752 Emit CBNZ inside loop when doing a strong exchange and comparing
10753 against zero. Generate the CC flags after the loop.
10755 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10757 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10758 (dl_section_ref): New.
10759 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10760 On AIX, append an expression to subtract the size of the
10761 section length to dl_section_ref.
10763 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10765 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10766 for early expansion of vector absolute builtins.
10768 2017-06-02 Richard Biener <rguenther@suse.de>
10770 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10771 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10773 2017-06-02 Richard Biener <rguenther@suse.de>
10775 PR tree-optimization/80948
10776 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10777 stmt order as well.
10779 2017-06-02 Richard Biener <rguenther@suse.de>
10781 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
10783 * tree-vect-stmts.c (process_use): Do not mark backedge defs
10784 for inductions as relevant.
10786 2017-06-02 Richard Biener <rguenther@suse.de>
10788 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
10789 (vectorizable_induction): ... this. Remove dead code.
10791 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
10793 * builtins. (expand_builtin_alloca): Remove second parameter and
10794 infer its value from the first parameter instead.
10795 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
10797 2017-06-02 Jakub Jelinek <jakub@redhat.com>
10799 PR rtl-optimization/80903
10800 * loop-doloop.c (add_test): Unshare sequence.
10802 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
10804 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
10806 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
10808 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
10810 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
10811 xlogue_layout::get_instance, logue_layout::xlogue_layout,
10812 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
10813 (xlogue_layout::get_stub_rtx): Make static.
10814 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
10815 (xlogue_layout::compute_stub_managed_regs): Rename to...
10816 (xlogue_layout::count_stub_managed_regs): ...this.
10817 (xlogue_layout::is_stub_managed_reg): New function.
10818 (xlogue_layout::m_stub_names): Rename to...
10819 (xlogue_layout::s_stub_names): ...this, make static.
10820 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
10821 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
10822 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
10823 xlogue_layout::s_stub_names): Instantiate statics.
10824 (stub_managed_regs): Remove.
10825 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
10826 (disable_call_ms2sysv_xlogues): Rename to...
10827 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
10828 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
10830 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
10831 change after reload_completed.
10832 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
10834 (ix86_expand_prologue): Likewise.
10835 (ix86_expand_epilogue): Likewise.
10836 (ix86_expand_split_stack_prologue): Likewise.
10837 (ix86_compute_frame_layout): Remove frame parameter ...
10838 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
10839 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
10841 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
10842 (ix86_frame): Move from here ...
10843 * config/i386/i386.h (ix86_frame): ... to here.
10844 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
10845 complete ix86_frame data structure instead. Remove some_ld_name.
10847 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10849 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
10850 symbols that hold a DECL_VALUE_EXPR.
10852 2017-06-01 Martin Jambor <mjambor@suse.cz>
10854 PR tree-optimization/80898
10855 * tree-sra.c (process_subtree_disqualification): Removed.
10856 (disqualify_candidate): Do not acll
10857 process_subtree_disqualification.
10858 (subtree_mark_written_and_enqueue): New function.
10859 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
10860 RHS has been disqualified and re-queue LHS if necessary. Apart
10861 from that, ignore disqualified RHS.
10863 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10865 * config/s390/s390.c (s390_emit_epilogue): Disable early return
10866 address fetch for z10 or later.
10868 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10870 * config/arc/arc.md (tst_movb): Add guard when splitting.
10872 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10874 * config/arc/arc.c (arc_can_eliminate): Test against
10875 arc_frame_pointer_needed.
10877 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10879 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
10880 to prevent store reordering.
10881 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
10882 (type): Add block type.
10883 (stack_tie): Define special instruction to be used in
10886 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10888 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
10889 constraint. It is not valid for the pattern.
10890 (noncommutative_binary_comparison): Likewise.
10892 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10894 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
10897 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10899 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
10900 be used by the reg-alloc.
10902 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10904 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
10905 reg-alloc when having mul64 or mul32x16 instructions.
10906 (mulsidi3): Likewise.
10907 (umulsidi3): Likewise.
10908 (mulsi32x16): New pattern.
10909 (mulsi64): Likewise.
10910 (mulsidi64): Likewise.
10911 (umulsidi64): Likewise.
10912 (MUL32x16_REG): Define.
10913 (mul64_600): Use MUL32x16_REG.
10914 (mac64_600): Likewise.
10915 (umul64_600): Likewise.
10916 (umac64_600): Likewise.
10918 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10920 * config/arc/arc.md (mulsi3_700): Make it commutative.
10922 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
10924 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
10926 (*sign_extendsidi2_insn): Likewise for movstosw.
10928 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10930 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
10931 the type of the input discriminant value. Convert the
10932 discriminant value of signedness vary.
10934 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
10936 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
10937 Add to -Wall section.
10939 2017-06-01 Richard Biener <rguenther@suse.de>
10941 PR middle-end/66313
10942 * fold-const.c (fold_plusminus_mult_expr): If the factored
10943 factor may be zero use a wrapping type for the inner operation.
10944 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
10945 and handle moved defs.
10946 (process_assignment): Properly guard the unary op case. Return a
10947 tri-state indicating that moving the stmt before the call may allow
10948 to continue. Pass through to_move.
10949 (find_tail_calls): Handle moving unrelated defs before
10952 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10955 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
10956 splitter result in the canonical way.
10958 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
10960 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
10961 also for 32bit target. Update insn attributes.
10962 (zero-extendsidi2 splitter): Allow all registers for operand 1.
10964 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
10966 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
10967 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
10968 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
10969 (_mm_maskz_min_ss): New intrinsics.
10971 2017-05-31 Martin Liska <mliska@suse.cz>
10973 * tree-vect-loop.c (vect_create_epilog_for_reduction):
10974 Change comment style to one we normally use.
10975 (vectorizable_reduction): Likewise.
10976 (vectorizable_induction): Likewise.
10977 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10978 (vectorizable_call): Likewise.
10979 (vectorizable_simd_clone_call): Likewise.
10980 (vectorizable_conversion): Likewise.
10981 (vectorizable_assignment): Likewise.
10982 (vectorizable_shift): Likewise.
10983 (vectorizable_operation): Likewise.
10984 (vectorizable_store): Likewise.
10985 (vectorizable_load): Likewise.
10986 * tree-vectorizer.h: Likewise.
10988 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
10990 * passes.c (emergency_dump_function): New.
10991 * tree-pass.h (emergency_dump_function): Declare.
10992 * plugin.c (plugins_internal_error_function): Remove.
10993 * plugin.h (plugins_internal_error_function): Remove declaration.
10994 * toplev.c (internal_error_function): New static function. Use it...
10995 (general_init): ...here.
10997 2017-05-31 Graham Markall <graham.markall@embecosm.com>
10999 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11000 (arc_rtx_costs): Add costs for new patterns.
11001 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11002 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11004 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11006 * tree-ssa-strlen.c (get_next_strinfo): New function.
11007 (get_stridx_plus_constant): Use it.
11008 (zero_length_string): Likewise.
11009 (adjust_related_strinfos): Likewise.
11010 (adjust_last_stmt): Likewise.
11012 2017-05-31 Richard Biener <rguenther@suse.de>
11015 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11016 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11018 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11020 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11021 loop_vinfo argument and use of dependence distance vectors.
11022 Check instead whether the two references differ only in their
11023 initial value and assume that they have the same alignment if the
11024 difference is a multiple of the vector alignment.
11025 (vect_analyze_data_refs_alignment): Update call accordingly.
11027 2017-05-31 Martin Liska <mliska@suse.cz>
11030 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11032 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11034 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11035 (create_intersect_range_checks): Move from ...
11036 * tree-data-ref.c (create_intersect_range_checks_index)
11037 (create_intersect_range_checks): ... to here.
11038 (create_runtime_alias_checks): New function factored from ...
11039 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11040 here. Call above function.
11041 * tree-data-ref.h (create_runtime_alias_checks): New function.
11043 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11045 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11046 segment length for dr_b and compute it in wide_int.
11048 2017-05-31 Richard Biener <rguenther@suse.de>
11050 PR tree-optimization/80906
11051 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11052 and pass through iv_map.
11053 (copy_bb_and_scalar_dependences): Adjust.
11054 (translate_pending_phi_nodes): Likewise.
11055 (copy_loop_close_phi_args): Handle code-generating IVs instead
11058 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11060 * diagnostic-color.c (color_dict): Add "type-diff".
11061 (parse_gcc_colors): Update comment.
11062 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11063 -fdiagnostics-show-template-tree and -fno-elide-type.
11064 (GCC_COLORS): Add type-diff to example.
11066 (-fdiagnostics-show-template-tree): New.
11067 (-fno-elide-type): New.
11068 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11069 the pp_format_decoder callback. Call any m_format_postprocessor's
11071 (pretty_printer::pretty_printer): Initialize
11072 m_format_postprocessor.
11073 (pretty_printer::~pretty_printer): Delete any
11074 m_format_postprocessor.
11075 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11076 (class format_postprocessor): New class.
11077 (struct pretty_printer::format_decoder): Document the new parameters.
11078 (struct pretty_printer::m_format_postprocessor): New field.
11079 * tree-diagnostic.c (default_tree_printer): Update for new
11080 bool and const char ** params.
11081 * tree-diagnostic.h (default_tree_printer): Likewise.
11083 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11085 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11086 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11087 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11088 rs6000_gen_cell_microcode code.
11089 (rs6000_final_prescan_insn): Delete.
11090 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11091 "warn-cell-microcode" entries.
11092 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11093 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11094 throughout. Change cc_reg_not_micro_cr0_operand to
11095 cc_reg_not_cr0_operand throughout.
11096 (*extendhi<mode>2_noload): Delete.
11097 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11098 (mwarn-cell-microcode): Delete.
11099 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11100 -mgen-cell-microcode and -mwarn-cell-microcode.
11102 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11105 * config/i386/constraints.md (Yd): New constraint.
11107 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11108 and (?Yd, r) alternatives. Update insn attributes.
11109 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11110 and (?*Yd, r) alternatives. Update insn attributes.
11111 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11113 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11115 * gimplify.c (gimplify_modify_expr): Don't create a
11116 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11119 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11121 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11123 2017-05-30 Richard Biener <rguenther@suse.de>
11125 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11126 and reduc_def fields.
11127 (STMT_VINFO_REDUC_TYPE): New define.
11128 (STMT_VINFO_REDUC_DEF): Likewise.
11129 (vect_force_simple_reduction): Adjust prototype.
11130 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11131 (vect_is_simple_reduction): Remove check_reduction argument.
11132 (vect_force_simple_reduction): Adjust and set
11133 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11134 (vectorizable_reduction): Do not re-do reduction analysis
11135 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11136 * tree-parloops.c (gather_scalar_reductions): Adjust.
11138 2017-05-30 Richard Biener <rguenther@suse.de>
11140 PR middle-end/80901
11141 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11144 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11146 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11147 Introduce unknown_misalignment parameter and remove vf.
11148 (vect_peeling_hash_get_lowest_cost):
11149 Pass unknown_misalignment parameter.
11150 (vect_enhance_data_refs_alignment):
11151 Fix unsupportable data ref treatment.
11153 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11155 * tree-vect-data-refs.c (vect_get_data_access_cost):
11156 Workaround for SLP handling.
11157 (vect_enhance_data_refs_alignment):
11158 Compute costs for doing no peeling at all, compare to the best
11159 peeling costs so far and avoid peeling if cheaper.
11161 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11163 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11164 Return peeling info and set costs to zero for unlimited cost
11166 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11167 with unknown misalignment. Compute and costs for unknown
11168 misalignment, compare them to the costs for known misalignment
11169 and choose the cheapest for peeling.
11171 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11173 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11174 (vect_get_peeling_costs_all_drs): Create function.
11175 (vect_peeling_hash_get_lowest_cost):
11176 Use vect_get_peeling_costs_all_drs.
11177 (vect_peeling_supportable): Create function.
11178 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11180 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11182 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11183 DR_HAS_NEGATIVE_STEP.
11184 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11185 (vect_enhance_data_refs_alignment): Use.
11186 (vect_duplicate_ssa_name_ptr_info): Use.
11187 * tree-vectorizer.h (dr_misalignment): Use.
11188 (known_alignment_for_access_p): Use.
11190 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11193 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11195 (has_section_name): New function.
11197 2017-05-30 Martin Liska <mliska@suse.cz>
11200 * auto-profile.c (get_function_decl_from_block): Fix
11203 2017-05-30 Richard Biener <rguenther@suse.de>
11205 PR middle-end/80876
11206 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11208 2017-05-30 Martin Liska <mliska@suse.cz>
11210 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11211 * dumpfile.h (struct dump_file_info): Remove ctors.
11213 2017-05-30 Martin Liska <mliska@suse.cz>
11215 * predict.def: Fix GNU coding style.
11217 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11219 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11220 Mark 'to' argument with ATTRIBUTE_UNUSED.
11222 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11224 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11225 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11226 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11229 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11231 * doc/install.texi (Options specification): Restore entry of
11232 --enable-sjlj-exceptions.
11234 2017-05-27 Michael Eager <eager@eagercon.com>
11237 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11239 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11241 * config/microblaze/microblaze.h
11242 (FIXED_REGISTERS): Update in macro.
11243 (CALL_USED_REGISTERS): Update in macro.
11245 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11247 * doc/install.texi: Add links to macOS binary distributions.
11249 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11253 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11255 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11257 2017-05-26 Martin Liska <mliska@suse.cz>
11259 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11261 2017-05-26 Martin Liska <mliska@suse.cz>
11263 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11264 always leading ';; '.
11265 (dump_bb_info): Likewise.
11266 (brief_dump_cfg): Likewise.
11267 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11268 * dumpfile.c: Remove usage of TDF_VERBOSE.
11269 * dumpfile.h (enum dump_kind): Likewise.
11270 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11271 * print-tree.c (debug_verbose): Remove.
11272 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11273 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11274 * tree-diagnostic.c (default_tree_printer): Replace
11275 TDF_DIAGNOSTIC with TDF_SLIM.
11277 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11279 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11280 in parameter loop, rather than loop_vinfo.
11281 (create_intersect_range_checks): Ditto.
11282 (vect_create_cond_for_alias_checks): Update call to above functions.
11284 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11286 PR tree-optimization/80815
11287 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11288 for merging runtime alias checks. Handle negative DR_STEPs.
11290 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11292 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11294 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11295 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11296 out code pruning runtime alias checks.
11297 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11298 factored out from above.
11299 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11301 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11303 (prune_runtime_alias_test_list): New decalaration.
11305 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11307 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11308 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11309 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11310 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11311 (operator==, comp_dr_with_seg_len_pair): Ditto.
11312 (vect_prune_runtime_alias_test_list): Ditto.
11314 2017-05-26 Martin Liska <mliska@suse.cz>
11317 * params.def: Bound partial-inlining-entry-probability param.
11319 2017-05-26 Marek Polacek <polacek@redhat.com>
11322 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11325 2017-05-26 Richard Biener <rguenther@suse.de>
11327 PR tree-optimization/80842
11328 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11331 2017-05-26 Richard Biener <rguenther@suse.de>
11333 PR tree-optimization/80844
11334 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11336 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11338 * doc/md.texi (Machine Constraints): Update x86 family
11339 machine constraints section to match 'config/i386/constraints.md'.
11341 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11343 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11345 2017-05-25 Nathan Sidwell <nathan@acm.org>
11347 * doc/invoke.texi (--enable-languages): Update documentation.
11349 2017-05-25 Martin Liska <mliska@suse.cz>
11351 * dumpfile.c: Add TDF_FOLDING.
11352 * dumpfile.h (enum dump_kind): Likewise.
11353 * genmatch.c (dt_simplify::gen_1): Use it.
11355 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11357 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11359 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11361 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11362 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11364 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11366 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11367 * match.pd (X == C): Rewrite it here.
11368 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11369 with_certain_nonzero_bits2): New predicates.
11370 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11372 2017-05-24 Nathan Sidwell <nathan@acm.org>
11374 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11377 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11380 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11382 * config/powerpcspe: New port. Files are copied from the rs6000
11383 port, with "rs6000" in filenames replaced by "powerpcspe".
11385 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11387 PR rtl-optimization/80754
11388 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11390 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11392 * config/sparc/sparc.md (length): Return the correct value for -mflat
11393 sibcalls to match output_sibcall.
11395 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11399 * config/rs6000/rs6000.c (struct machine_function): Add new field
11401 (rs6000_get_separate_components): Init that field, use it.
11402 (rs6000_components_for_bb): Use the field.
11404 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11406 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11408 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11410 PR middle-end/80823
11411 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11413 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11416 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11417 address against address_operand predicate.
11418 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11420 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11422 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11423 parameters passed indirectly.
11425 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11427 * config/i386/i386.md (*movdi_internal): Remove SSE4
11428 alternative 18 (?r, *v). Update insn attributes.
11429 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11430 Update insn attributes.
11431 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11432 Update insn attributes.
11433 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11434 alternative 1 (r, v). Remove isa attribute.
11435 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11436 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11437 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11439 2017-05-23 Tom de Vries <tom@codesourcery.com>
11441 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11444 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11446 * cgraphunit.c (symbol_table::process_new_functions): Update.
11447 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11448 (inline_generate_summary): Rename to ...
11449 (ipa_fn_summary_generate): ... this one.
11450 (inline_read_summary): Rename to ...
11451 (ipa_fn_summary_read): ... this one.
11452 (inline_write_summary): Rename to ...
11453 (ipa_fn_summary_write): ... this one.
11454 (inline_free_summary): Rename to ...
11455 (ipa_free_fn_summary): ... this one.
11456 (pass_data_local_fn_summary, pass_local_fn_summary,
11457 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11458 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11459 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11460 make_pass_ipa_fn_summary): New.
11461 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11462 inline_write_summary, inline_free_summary): Remove.
11463 (ipa_free_fn_summary) : New.
11464 * ipa-inline.c (ipa_inline): Update.
11465 (pass_ipa_inline): Do not generate summaries.
11466 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11468 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11469 and add pass_ipa_fn_summary.
11470 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11472 (make_pass_inline_parameters): Remove.
11474 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11476 * omp-low.c (struct omp_context): Remove "default_kind" member.
11479 * omp-offload.c (execute_oacc_device_lower): Remove the
11480 parallelism dimensions function attributes for unparallelized
11481 OpenACC kernels constructs.
11483 2017-05-23 Martin Liska <mliska@suse.cz>
11485 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11487 (cgraph_edge::make_speculative): Likewise.
11488 (cgraph_edge::resolve_speculation): Likewise.
11489 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11490 (cgraph_node::dump): Likewise.
11491 * cgraph.h: Likewise.
11492 * cgraphunit.c (analyze_functions): Likewise.
11493 (symbol_table::compile): Likewise.
11494 * ipa-cp.c (print_all_lattices): Likewise.
11495 (determine_versionability): Likewise.
11496 (initialize_node_lattices): Likewise.
11497 (ipcp_verify_propagated_values): Likewise.
11498 (estimate_local_effects): Likewise.
11499 (update_profiling_info): Likewise.
11500 (create_specialized_node): Likewise.
11501 (perhaps_add_new_callers): Likewise.
11502 (decide_about_value): Likewise.
11503 (decide_whether_version_node): Likewise.
11504 (identify_dead_nodes): Likewise.
11505 (ipcp_store_bits_results): Likewise.
11506 * ipa-devirt.c (dump_targets): Likewise.
11507 (ipa_devirt): Likewise.
11508 * ipa-icf.c (sem_item::dump): Likewise.
11509 (sem_function::equals): Likewise.
11510 (sem_variable::equals): Likewise.
11511 (sem_item_optimizer::read_section): Likewise.
11512 (sem_item_optimizer::execute): Likewise.
11513 (congruence_class::dump): Likewise.
11514 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11515 (dump_inline_summary): Likewise.
11516 (estimate_node_size_and_time): Likewise.
11517 (inline_analyze_function): Likewise.
11518 * ipa-inline-transform.c (inline_call): Likewise.
11519 * ipa-inline.c (report_inline_failed_reason): Likewise.
11520 (want_early_inline_function_p): Likewise.
11521 (edge_badness): Likewise.
11522 (update_edge_key): Likewise.
11523 (inline_small_functions): Likewise.
11524 * ipa-profile.c (ipa_profile): Likewise.
11525 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11526 (ipa_make_edge_direct_to_target): Likewise.
11527 (remove_described_reference): Likewise.
11528 (ipa_impossible_devirt_target): Likewise.
11529 (propagate_controlled_uses): Likewise.
11530 (ipa_print_node_params): Likewise.
11531 (ipcp_transform_function): Likewise.
11532 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11533 (propagate_pure_const): Likewise.
11534 * ipa-reference.c (generate_summary): Likewise.
11535 (read_write_all_from_decl): Likewise.
11536 (propagate): Likewise.
11537 (ipa_reference_read_optimization_summary): Likewise.
11538 * ipa-utils.c (ipa_merge_profiles): Likewise.
11539 * ipa.c (walk_polymorphic_call_targets): Likewise.
11540 (symbol_table::remove_unreachable_nodes): Likewise.
11541 (ipa_single_use): Likewise.
11542 * passes.c (execute_todo): Likewise.
11543 * predict.c (drop_profile): Likewise.
11544 * symtab.c (symtab_node::get_dump_name): New function.
11545 (symtab_node::dump_name): Likewise.
11546 (symtab_node::dump_asm_name): Likewise.
11547 (symtab_node::dump_references): Likewise.
11548 (symtab_node::dump_referring): Likewise.
11549 (symtab_node::dump_base): Likewise.
11550 (symtab_node::debug_symtab): Likewise.
11551 * tree-sra.c (convert_callers_for_node): Likewise.
11552 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11553 * value-prof.c (init_node_map): Likewise.
11555 2017-05-23 Martin Liska <mliska@suse.cz>
11557 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11558 and symtab_node::debug_symtab to symbol_table::debug.
11559 * cgraphunit.c (analyze_functions): Use the renamed function.
11560 (symbol_table::compile): Likewise.
11561 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11562 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11563 * passes.c (execute_todo): Likewise.
11564 * symtab.c (symbol_table::dump): New function.
11565 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11567 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11569 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11570 that nonconst implies exec.
11572 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11574 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11575 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11576 (inline_edge_summary_vec): Turn into ...
11577 (ipa_call_summaries): ... this one.
11578 (redirect_to_unreachable, edge_set_predicate,
11579 evaluate_properties_for_edge, inline_summary_alloc,
11580 reset_ipa_call_summary, reset_inline_summary,
11581 inline_summary_t::duplicate): Update.
11582 (inline_edge_duplication_hook): Turn to ...
11583 (ipa_call_summary_t::duplicate): ... this one.
11584 (inline_edge_removal_hook): Turn to ...
11585 (ipa_call_summary_t::remove): ... this one.
11586 (dump_inline_edge_summary): Turn to ...
11587 (dump_ipa_call_summary): ... this one.
11588 (estimate_function_body_sizes): Update.
11589 (inline_update_callee_summaries): Update.
11590 (remap_edge_change_prob): Update.
11591 (remap_edge_summaries): Update.
11592 (inline_merge_summary): Update.
11593 (do_estimate_edge_time): Update.
11594 (inline_generate_summary): Update.
11595 (inline_read_section): Update.
11596 (inline_read_summary): Update.
11597 (inline_free_summary): Update.
11598 * ipa-inline.c (can_inline_edge_p): Update.
11599 (compute_inlined_call_time): Update.
11600 (want_inline_small_function_p): Update.
11601 (edge_badness): Update.
11602 (early_inliner): Update.
11603 * ipa-inline.h (inline_edge_summary): Turn to ...
11604 (ipa_call_summary): ... this one.
11605 (ipa_call_summary_t): New class.
11606 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11607 (ipa_call_summaries): New.
11608 (inline_edge_summary): Remove.
11609 (estimate_edge_growth): Update.
11610 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11611 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11612 * ipa-split.c (execute_split_functions): Update.
11613 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11615 2017-05-23 Tom de Vries <tom@codesourcery.com>
11617 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11618 attributes): Document rdrand effective target.
11620 2017-05-23 Tom de Vries <tom@codesourcery.com>
11622 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11623 attributes): Sort alphabetically.
11625 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11627 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11629 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11632 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11633 V2DF/V2DI splat into two separate patterns, one that handles
11634 registers, and the other that only handles memory. Drop support
11635 for splatting from a GPR on ISA 2.07 and then splitting the
11636 splat into direct move and splat.
11637 (vsx_splat_<mode>_reg): Likewise.
11638 (vsx_splat_<mode>_mem): Likewise.
11640 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11642 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11644 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11646 PR middle-end/80809
11647 * omp-low.c (finish_taskreg_remap): New function.
11648 (finish_taskreg_scan): If unit size of ctx->record_type
11649 is non-constant, unshare the size expression and replace
11650 decls in it with possible outer var refs.
11652 PR middle-end/80809
11653 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11654 GOVD_SHARED rather than GOVD_PRIVATE with it.
11655 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11656 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11658 PR middle-end/80853
11659 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11660 as last argument to build_outer_var_ref for pointer bases of array
11661 section reductions.
11663 2017-05-19 Martin Sebor <msebor@redhat.com>
11665 * print-tree.c (print_node): Print DECL_READ_P flag.
11667 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11669 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11670 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11671 * cgraph.c: Likewise.
11672 * cgraphunit.c: Likewise.
11673 * gengtype.c: Likewise.
11674 * ipa-cp.c: Likewise.
11675 * ipa-devirt.c: Likewise.
11676 * ipa-icf.c: Likewise.
11677 * ipa-predicate.c: Likewise.
11678 * ipa-profile.c: Likewise.
11679 * ipa-prop.c: Likewise.
11680 * ipa-split.c: Likewise.
11682 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11683 edge_predicate_pool, dump_inline_hints,
11684 inline_summary::account_size_time, redirect_to_unreachable,
11685 edge_set_predicate, set_hint_predicate,
11686 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11687 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11688 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11689 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11690 ipa_call_summary_t::remove, initialize_growth_caches,
11691 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11692 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11693 mark_modified, unmodified_parm_1, unmodified_parm,
11694 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11695 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11696 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11697 will_be_nonconstant_predicate, record_modified_bb_info,
11698 get_minimal_bb, record_modified, param_change_prob,
11699 phi_result_unknown_predicate, predicate_for_phi_result,
11700 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11701 estimate_function_body_sizes, compute_inline_parameters,
11702 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11703 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11704 inline_update_callee_summaries, remap_edge_change_prob,
11705 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11706 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11707 inline_analyze_function, inline_summary_t::insert,
11708 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11709 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11710 inline_free_summary): Move to ipa-fnsummary.h
11711 (predicate_t): Remove.
11712 * ipa-fnsummary.c: New file.
11713 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11714 (enum inline_hints_vals, inline_hints, agg_position_info,
11715 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11716 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11717 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11718 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11719 inline_read_summary, inline_write_summary, inline_free_summary,
11720 inline_analyze_function, initialize_inline_failed,
11721 inline_merge_summary, inline_update_overall_summary,
11722 compute_inline_parameters): Move to ipa-fnsummary.h
11723 * ipa-fnsummary.h: New file.
11724 * ipa-inline-transform.h: Include ipa-inline.h.
11725 * ipa-inline.c: LIkewise.
11727 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11729 * ipa-inline.c (edge_badness): Use inlined_time instead of
11730 inline_summaries->get.
11732 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11734 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11736 2017-05-22 Nathan Sidwell <nathan@acm.org>
11738 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11739 (fdump-lang): Document 'raw' option.
11740 * dumpfile.h (TDI_tu): Delete.
11741 * dumpfile.c (dump_files): Remove translation-unit.
11742 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11744 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11746 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11747 command option from $(AWK) call.
11748 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11749 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11750 [FORMAT]: Remove handling of variable.
11751 * config/avr/t-multilib: Regenerate.
11753 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11755 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11757 (dump_inline_summary): Do not print self_time.
11758 (estimate_function_body_sizes): Do not set self_time.
11759 (compute_inline_parameters): Likewise.
11760 (inline_read_section, inline_write_summary): Do not stream self_time.
11761 * ipa-inline.h (inline_summary): Drop self_time.
11763 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11765 * ipa-inline-analysis.c (account_size_time): Rename to ...
11766 (inline_summary::account_size_time): ... this one.
11767 (reset_ipa_call_summary): Turn to ...
11768 (ipa_call_summary::reset): ... this one.
11769 (reset_inline_summary): Turn to ...
11770 (inline_summary::reset): ... this one.
11771 (inline_summary_t::remove): Update.
11772 (inline_summary_t::duplicate): Update.
11773 (ipa_call_summary_t::remove): Update.
11774 (dump_inline_summary): Update.
11775 (estimate_function_body_sizes): Update.
11776 (compute_inline_parameters): Update.
11777 (estimate_node_size_and_time): Update.
11778 (inline_merge_summary): Update.
11779 (inline_update_overall_summary): Update.
11780 (inline_read_section): Update.
11781 (inline_write_summary): Update.
11782 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
11783 add account_size_time and reset member functions.
11784 (ipa_call_summary): Add reset function.
11785 * ipa-predicate.h (predicate::operator &): Constify.
11787 2017-05-22 Richard Biener <rguenther@suse.de>
11789 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
11791 2017-05-19 Jason Merrill <jason@redhat.com>
11793 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
11795 2017-05-19 Marek Polacek <polacek@redhat.com>
11798 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
11799 TYPE_OVERFLOW_WRAPS checks.
11801 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
11803 * tree-core.h (enum omp_clause_default_kind): Add
11804 "OMP_CLAUSE_DEFAULT_PRESENT".
11805 * tree-pretty-print.c (dump_omp_clause): Handle it.
11806 * gimplify.c (enum gimplify_omp_var_data): Add
11807 "GOVD_MAP_FORCE_PRESENT".
11808 (gimplify_adjust_omp_clauses_1): Map it to
11809 "GOMP_MAP_FORCE_PRESENT".
11810 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
11812 * gimplify.c (oacc_default_clause): Clarify.
11814 2017-05-19 Nathan Sidwell <nathan@acm.org>
11816 LANG_HOOK_REGISTER_DUMPS
11817 * toplev.c (general_init): Call register dump lang hook.
11818 * doc/invoke.texi: Document -fdump-lang option family.
11819 * dumpfile.c (dump_files): Remove class dump here.
11820 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
11821 * dumpfile.h (tree_dump_index): Remove TDI_class.
11822 * langhooks-def.h (lhd_register_dumps): Declare.
11823 (LANG_HOOKS_REGISTER_DUMPS): Define.
11824 (LANG_HOOKS_INITIALIZER): Add it.
11825 * langhooks.c (lhd_register_dumps): Define.
11826 * langhooks.h (struct lang_hooks): Add register_dumps.
11828 2017-05-19 Nathan Sidwell <nathan@acm.org>
11830 * context.h (context::set_passes): New.
11831 * context.c (context::context): Do not create pass manager.
11832 * toplev.c (general_init): Create pass manager here.
11834 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
11836 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
11837 use this splitter if two add or or instructions would also work for
11838 the constant we want to generate.
11840 2017-05-19 Richard Biener <rguenther@suse.de>
11843 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
11844 predicate evaluation.
11846 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11848 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
11850 * ipa-inline.c (want_inline_small_function_p): Do not cast to
11853 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11855 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11856 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11857 (inline_edge_summary_vec): Turn into ...
11858 (ipa_call_summaries): ... this one.
11859 (redirect_to_unreachable, edge_set_predicate,
11860 evaluate_properties_for_edge, inline_summary_alloc,
11861 reset_ipa_call_summary, reset_inline_summary,
11862 inline_summary_t::duplicate): Update.
11863 (inline_edge_duplication_hook): Turn to ...
11864 (ipa_call_summary_t::duplicate): ... this one.
11865 (inline_edge_removal_hook): Turn to ...
11866 (ipa_call_summary_t::remove): ... this one.
11867 (dump_inline_edge_summary): Turn to ...
11868 (dump_ipa_call_summary): ... this one.
11869 (estimate_function_body_sizes): Update.
11870 (inline_update_callee_summaries): Update.
11871 (remap_edge_change_prob): Update.
11872 (remap_edge_summaries): Update.
11873 (inline_merge_summary): Update.
11874 (do_estimate_edge_time): Update.
11875 (inline_generate_summary): Update.
11876 (inline_read_section): Update.
11877 (inline_read_summary): Update.
11878 (inline_free_summary): Update.
11879 * ipa-inline.c (can_inline_edge_p): Update.
11880 (compute_inlined_call_time): Update.
11881 (want_inline_small_function_p): Update.
11882 (edge_badness): Update.
11883 (early_inliner): Update.
11884 * ipa-inline.h (inline_edge_summary): Turn to ...
11885 (ipa_call_summary): ... this one.
11886 (ipa_call_summary_t): New class.
11887 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11888 (ipa_call_summaries): New.
11889 (inline_edge_summary): Remove.
11890 (estimate_edge_growth): Update.
11891 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11892 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11893 * ipa-split.c (execute_split_functions): Update.
11894 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11896 2017-05-19 Richard Biener <rguenther@suse.de>
11898 PR middle-end/80764
11899 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
11901 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
11903 * config/rs6000/rs6000.c (struct machine_function): Add field
11904 fpr_is_wrapped_separately.
11905 (rs6000_get_separate_components): Use 64 components. Handle the
11906 new FPR components.
11907 (rs6000_components_for_bb): Handle the FPR components.
11908 (rs6000_emit_prologue_components): Handle the FPR components.
11909 (rs6000_emit_epilogue_components): Handle the FPR components.
11910 (rs6000_set_handled_components): Handle the FPR components.
11911 (rs6000_emit_prologue): Don't output prologue code for those FPRs
11912 that are already separately shrink-wrapped.
11913 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
11914 that are already separately shrink-wrapped.
11916 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11919 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
11922 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
11923 (define_peephole2 for Altivec d-form load): Add peepholes to catch
11924 cases where the register allocator uses a move and an offsettable
11925 memory operation to/from a FPR register on ISA 2.06/2.07.
11926 (define_peephole2 for Altivec d-form store): Likewise.
11928 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
11931 * config/i386/mmx.md (*mov<mode>_internal): Enable
11932 alternatives 11, 12, 13 and 14 also for 32bit targets.
11933 Remove alternatives 15, 16, 17 and 18.
11934 * config/i386/sse.md (vec_concatv2di): Change
11935 alternative (!x, *y) to (x, ?!*Yn).
11937 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
11939 * dumpfile.h (enum dump_kind): Remove stray comma.
11941 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11943 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
11944 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
11945 predicate::num_conditions
11946 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
11947 (CHANGED): turn into predicate::changed.
11948 (agg_position_info): Move to ipa-predicate.h
11949 (add_condition, predicate::add_clause, predicate::operator &=,
11950 predicate::or_with, predicate::evaluate, predicate::probability,
11951 dump_condition, dump_clause, predicate::dump,
11952 predicate::remap_after_duplication, predicate::remap_after_inlining,
11953 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
11954 (evaluate_conditions_for_known_args): Update.
11955 (set_cond_stmt_execution_predicate): Update.
11956 * ipa-inline.h: Include ipa-predicate.h
11957 (condition, inline_param_summary, conditions, agg_position_info,
11958 predicate): Move to ipa-predicate.h
11959 * ipa-predicate.c: New file.
11960 * ipa-predicate.h: New file.
11962 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
11964 * final.c (leaf_function_p): Check we are not in a sequence.
11966 2017-05-18 Martin Liska <mliska@suse.cz>
11968 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
11969 * dumpfile.c (dump_register): Use new enum dump_kind.
11970 (get_dump_file_name): Likewise.
11971 (dump_enable_all): Likewise.
11972 (dump_switch_p_1): Likewise.
11973 (enable_rtl_dump_file): Remove usage of TDF_RTL.
11974 * dumpfile.h (enum dump_kind): New enum type.
11975 (struct dump_file_info): Create constructor and
11976 format fields and comments.
11977 * passes.c (pass_manager::register_one_dump_file):
11979 * statistics.c (statistics_early_init): Likewise.
11980 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
11981 TDF_TREE with TDF_SLIM.
11982 (gather_memory_references_ref): Likewise.
11984 2017-05-18 Martin Liska <mliska@suse.cz>
11986 * vec.h (struct vnull): Use it.
11988 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11990 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
11991 (true_predicate, false_predicate, true_predicate_p,
11992 false_predicate_p): Remove.
11993 (single_cond_predicate, not_inlined_predicate): Turn to member function
11995 (add_condition): Update.
11996 (add_clause): Turn to...
11997 (predicate::add_clause): ... this one; update; allow passing NULL
11999 (and_predicates): Turn to ...
12000 (predicate::operator &=): ... this one.
12001 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12002 (or_predicates): Turn to ...
12003 (predicate::or_with): ... this one.
12004 (evaluate_predicate): Turn to ...
12005 (predicate::evaluate): ... this one.
12006 (predicate_probability): Turn to ...
12007 (predicate::probability): ... this one.
12008 (dump_condition): Update.
12009 (dump_predicate): Turn to ...
12010 (predicate::dump): ... this one.
12011 (account_size_time): Update.
12012 (edge_set_predicate): Update.
12013 (set_hint_predicate): UPdate.
12014 (evaluate_conditions_for_known_args): Update.
12015 (evaluate_properties_for_edge): Update.
12016 (remap_predicate_after_duplication): Turn to...
12017 (predicate::remap_after_duplication): ... this one.
12018 (remap_hint_predicate_after_duplication): Update.
12019 (inline_summary_t::duplicate): UPdate.
12020 (dump_inline_edge_summary): Update.
12021 (dump_inline_summary): Update.
12022 (set_cond_stmt_execution_predicate): Update.
12023 (set_switch_stmt_execution_predicate): Update.
12024 (compute_bb_predicates): Update.
12025 (will_be_nonconstant_expr_predicate): Update.
12026 (will_be_nonconstant_predicate): Update.
12027 (phi_result_unknown_predicate): Update.
12028 (predicate_for_phi_result): Update.
12029 (array_index_predicate): Update.
12030 (estimate_function_body_sizes): Update.
12031 (estimate_node_size_and_time): Update.
12032 (estimate_ipcp_clone_size_and_time): Update.
12033 (remap_predicate): Rename to ...
12034 (predicate::remap_after_inlining): ... this one.
12035 (remap_hint_predicate): Update.
12036 (inline_merge_summary): Update.
12037 (inline_update_overall_summary): Update.
12038 (estimate_size_after_inlining): Update.
12039 (read_predicate): Rename to ...
12040 (predicate::stream_in): ... this one.
12041 (read_inline_edge_summary): Update.
12042 (write_predicate): Rename to ...
12043 (predicate::stream_out): ... this one.
12044 (write_inline_edge_summary): Update.
12045 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12046 (clause_t): Turn to uint32_t
12047 (predicate): Turn to class; implement constructor and operators
12049 (size_time_entry): Update.
12050 (inline_summary): Update.
12051 (inline_edge_summary): Update.
12053 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12055 * fold-const.c (fold_binary_loc): Move transformation...
12056 * match.pd (C - X CMP X): ... here.
12058 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12060 * config/sparc/sparc.c (sparc_option_override): Set function
12061 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12062 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12064 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12066 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12068 2017-05-18 Marek Polacek <polacek@redhat.com>
12071 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12072 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12073 gimple_assign_load_p.
12075 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12077 PR middle-end/80692
12078 * real.c (do_compare): Give decimal_do_compare preference over
12079 comparing just the signs.
12081 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12083 * doc/md.texi (Canonicalization of Instructions): Describe the
12084 canonical form of instructions that inherently set a condition
12087 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12089 PR middle-end/80775
12090 * tree-cfg.c: Move deletion of unreachable case statements to after
12091 the merging of consecutive case labels.
12093 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12095 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12096 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12097 restoring of callee-saved registers.
12099 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12101 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12102 * config/visium/visium.c (single_set_and_flags): Likewise.
12103 * config/visium/visium.md (Substitutions): Likewise.
12105 2017-05-17 Martin Liska <mliska@suse.cz>
12107 * cfg.c: Introduce dump_flags_t type and
12108 use it instead of int type.
12110 * cfghooks.c: Likewise.
12111 * cfghooks.h (struct cfg_hooks): Likewise.
12112 * cfgrtl.c: Likewise.
12113 * cfgrtl.h: Likewise.
12114 * cgraph.c (cgraph_node::get_body): Likewise.
12115 * coretypes.h: Likewise.
12116 * domwalk.c: Likewise.
12117 * domwalk.h: Likewise.
12118 * dumpfile.c (struct dump_option_value_info): Likewise.
12119 (dump_enable_all): Likewise.
12120 (dump_switch_p_1): Likewise.
12121 (opt_info_switch_p): Likewise.
12122 * dumpfile.h (enum tree_dump_index): Likewise.
12123 (struct dump_file_info): Likewise.
12124 * genemit.c: Likewise.
12125 * generic-match-head.c: Likewise.
12126 * gengtype.c (open_base_files): Likewise.
12127 * gimple-pretty-print.c: Likewise.
12128 * gimple-pretty-print.h: Likewise.
12129 * graph.c (print_graph_cfg): Likewise.
12130 * graphite-scop-detection.c (dot_all_sese): Likewise.
12131 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12132 * loop-unroll.c (report_unroll): Likewise.
12133 * passes.c (pass_manager::register_one_dump_file): Likewise.
12134 * print-tree.c: Likewise.
12135 * statistics.c: Likewise.
12136 * tree-cfg.c: Likewise.
12137 * tree-cfg.h: Likewise.
12138 * tree-dfa.c: Likewise.
12139 * tree-dfa.h: Likewise.
12140 * tree-dump.c (dump_function): Likewise.
12141 * tree-dump.h (struct dump_info): Likewise.
12142 * tree-pretty-print.c: Likewise.
12143 * tree-pretty-print.h: Likewise.
12144 * tree-ssa-live.c: Likewise.
12145 * tree-ssa-live.h: Likewise.
12146 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12147 * tree-vect-loop.c: Likewise.
12148 * tree-vect-slp.c: Likewise.
12150 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12151 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12153 PR tree-optimization/80457
12154 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12155 of all arguments to a statement as scalar_to_vec operations.
12156 (vectorizable_call): Adjust call to vect_model_simple_cost for
12158 (vectorizable_conversion): Likewise.
12159 (vectorizable_assignment): Likewise.
12160 (vectorizable_shift): Likewise.
12161 (vectorizable_operation): Likewise.
12162 (vectorizable_comparison): Likewise.
12163 (vect_is_simple_cond): Record the def types for operands.
12164 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12165 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12166 for statement argument count.
12168 2017-05-16 Carl Love <cel@us.ibm.com>
12170 * config/rs6000/rs6000-c: Add support for built-in functions
12171 vector unsigned long long vec_bperm (vector unsigned long long,
12172 vector unsigned char)
12173 vector signed long long vec_mule (vector signed int,
12175 vector unsigned long long vec_mule (vector unsigned int,
12176 vector unsigned int)
12177 vector signed long long vec_mulo (vector signed int,
12179 vector unsigned long long vec_mulo (vector unsigned int,
12180 vector unsigned int)
12181 vector signed char vec_sldw (vector signed char,
12182 vector signed char,
12184 vector unsigned char vec_sldw (vector unsigned char,
12185 vector unsigned char,
12187 vector signed short vec_sldw (vector signed short,
12188 vector signed short,
12190 vector unsigned short vec_sldw (vector unsigned short,
12191 vector unsigned short,
12193 vector signed int vec_sldw (vector signed int,
12196 vector unsigned int vec_sldw (vector unsigned int,
12197 vector unsigned int,
12199 vector signed long long vec_sldw (vector signed long long,
12200 vector signed long long,
12202 vector unsigned long long vec_sldw (vector unsigned long long,
12203 vector unsigned long long,
12205 * config/rs6000/rs6000-c: Add support for built-in functions
12206 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12207 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12208 * doc/extend.texi: Update the built-in documentation for the
12209 new built-in functions.
12211 2017-05-16 Marek Polacek <polacek@redhat.com>
12215 * tree.c (save_expr): Don't fold the expression.
12217 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12219 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12220 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12221 and (?*y,m). Update insn attributes.
12223 2017-05-16 Martin Liska <mliska@suse.cz>
12225 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12226 flags argument of print_gimple_stmt, print_gimple_expr,
12227 print_generic_stmt and print_generic_expr.
12228 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12229 * coretypes.h: Likewise.
12230 * except.c (dump_eh_tree): Likewise.
12231 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12232 * gimple-pretty-print.h: Likewise.
12233 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12234 (backprop::push_to_worklist): Likewise.
12235 (backprop::pop_from_worklist): Likewise.
12236 (backprop::process_use): Likewise.
12237 (backprop::intersect_uses): Likewise.
12238 (note_replacement): Likewise.
12239 * gimple-ssa-store-merging.c
12240 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12241 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12242 (pass_store_merging::execute): Likewise.
12243 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12244 (ssa_base_cand_dump_callback): Likewise.
12245 (dump_incr_vec): Likewise.
12246 (replace_refs): Likewise.
12247 (replace_mult_candidate): Likewise.
12248 (create_add_on_incoming_edge): Likewise.
12249 (create_phi_basis): Likewise.
12250 (insert_initializers): Likewise.
12251 (all_phi_incrs_profitable): Likewise.
12252 (introduce_cast_before_cand): Likewise.
12253 (replace_one_candidate): Likewise.
12254 * gimplify.c (gimplify_expr): Likewise.
12255 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12256 (set_rename): Likewise.
12257 (rename_uses): Likewise.
12258 (copy_loop_phi_nodes): Likewise.
12259 (add_close_phis_to_merge_points): Likewise.
12260 (copy_loop_close_phi_args): Likewise.
12261 (copy_cond_phi_args): Likewise.
12262 (graphite_copy_stmts_from_block): Likewise.
12263 (translate_pending_phi_nodes): Likewise.
12264 * graphite-poly.c (print_pdr): Likewise.
12265 (dump_gbb_cases): Likewise.
12266 (dump_gbb_conditions): Likewise.
12267 (print_scop_params): Likewise.
12268 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12269 (build_cross_bb_scalars_use): Likewise.
12270 (gather_bbs::before_dom_children): Likewise.
12271 * hsa-dump.c (dump_hsa_immed): Likewise.
12272 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12273 (get_replacement_map): Likewise.
12274 * ipa-inline-analysis.c (dump_condition): Likewise.
12275 (estimate_function_body_sizes): Likewise.
12276 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12277 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12278 * ipa-prop.c (ipa_dump_param): Likewise.
12279 (ipa_print_node_jump_functions_for_edge): Likewise.
12280 (ipa_modify_call_arguments): Likewise.
12281 (ipa_modify_expr): Likewise.
12282 (ipa_dump_param_adjustments): Likewise.
12283 (ipa_dump_agg_replacement_values): Likewise.
12284 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12285 * ipa-pure-const.c (check_stmt): Likewise.
12286 (pass_nothrow::execute): Likewise.
12287 * ipa-split.c (execute_split_functions): Likewise.
12288 * omp-offload.c (dump_oacc_loop_part): Likewise.
12289 (dump_oacc_loop): Likewise.
12290 * trans-mem.c (tm_log_emit): Likewise.
12291 (tm_memopt_accumulate_memops): Likewise.
12292 (dump_tm_memopt_set): Likewise.
12293 (dump_tm_memopt_transform): Likewise.
12294 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12295 (print_loop): Likewise.
12296 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12297 (chkp_gather_checks_info): Likewise.
12298 (chkp_get_check_result): Likewise.
12299 (chkp_remove_check_if_pass): Likewise.
12300 (chkp_use_outer_bounds_if_possible): Likewise.
12301 (chkp_reduce_bounds_lifetime): Likewise.
12302 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12303 (chkp_mark_completed_bounds): Likewise.
12304 (chkp_register_incomplete_bounds): Likewise.
12305 (chkp_mark_invalid_bounds): Likewise.
12306 (chkp_maybe_copy_and_register_bounds): Likewise.
12307 (chkp_build_returned_bound): Likewise.
12308 (chkp_get_bound_for_parm): Likewise.
12309 (chkp_build_bndldx): Likewise.
12310 (chkp_get_bounds_by_definition): Likewise.
12311 (chkp_generate_extern_var_bounds): Likewise.
12312 (chkp_get_bounds_for_decl_addr): Likewise.
12313 * tree-chrec.c (chrec_apply): Likewise.
12314 * tree-data-ref.c (dump_data_reference): Likewise.
12315 (dump_subscript): Likewise.
12316 (dump_data_dependence_relation): Likewise.
12317 (analyze_overlapping_iterations): Likewise.
12318 * tree-inline.c (expand_call_inline): Likewise.
12319 (tree_function_versioning): Likewise.
12320 * tree-into-ssa.c (dump_defs_stack): Likewise.
12321 (dump_currdefs): Likewise.
12322 (dump_names_replaced_by): Likewise.
12323 (dump_update_ssa): Likewise.
12324 (update_ssa): Likewise.
12325 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12326 * tree-parloops.c (build_new_reduction): Likewise.
12327 (try_create_reduction_list): Likewise.
12328 (ref_conflicts_with_region): Likewise.
12329 (oacc_entry_exit_ok_1): Likewise.
12330 (oacc_entry_exit_single_gang): Likewise.
12331 * tree-pretty-print.h: Likewise.
12332 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12333 (get_scalar_evolution): Likewise.
12334 (add_to_evolution): Likewise.
12335 (get_loop_exit_condition): Likewise.
12336 (analyze_evolution_in_loop): Likewise.
12337 (analyze_initial_condition): Likewise.
12338 (analyze_scalar_evolution): Likewise.
12339 (instantiate_scev): Likewise.
12340 (number_of_latch_executions): Likewise.
12341 (gather_chrec_stats): Likewise.
12342 (final_value_replacement_loop): Likewise.
12343 (scev_const_prop): Likewise.
12344 * tree-sra.c (dump_access): Likewise.
12345 (disqualify_candidate): Likewise.
12346 (create_access): Likewise.
12347 (reject): Likewise.
12348 (maybe_add_sra_candidate): Likewise.
12349 (create_access_replacement): Likewise.
12350 (analyze_access_subtree): Likewise.
12351 (analyze_all_variable_accesses): Likewise.
12352 (sra_modify_assign): Likewise.
12353 (initialize_constant_pool_replacements): Likewise.
12354 (find_param_candidates): Likewise.
12355 (decide_one_param_reduction): Likewise.
12356 (replace_removed_params_ssa_names): Likewise.
12357 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12358 * tree-ssa-copy.c (dump_copy_of): Likewise.
12359 (copy_prop_visit_cond_stmt): Likewise.
12360 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12361 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12362 (record_equivalences_from_stmt): Likewise.
12363 * tree-ssa-dse.c (compute_trims): Likewise.
12364 (delete_dead_call): Likewise.
12365 (delete_dead_assignment): Likewise.
12366 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12367 (forward_propagate_into_cond): Likewise.
12368 (pass_forwprop::execute): Likewise.
12369 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12370 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12372 (move_computations_worker): Likewise.
12373 (execute_sm): Likewise.
12374 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12375 (remove_exits_and_undefined_stmts): Likewise.
12376 (remove_redundant_iv_tests): Likewise.
12377 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12378 (adjust_iv_update_pos): Likewise.
12379 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12380 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12381 (value_replacement): Likewise.
12382 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12383 * tree-ssa-pre.c (print_pre_expr): Likewise.
12384 (get_representative_for): Likewise.
12385 (create_expression_by_pieces): Likewise.
12386 (insert_into_preds_of_block): Likewise.
12387 (eliminate_insert): Likewise.
12388 (eliminate_dom_walker::before_dom_children): Likewise.
12389 (eliminate): Likewise.
12390 (remove_dead_inserted_code): Likewise.
12391 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12392 * tree-ssa-reassoc.c (get_rank): Likewise.
12393 (eliminate_duplicate_pair): Likewise.
12394 (eliminate_plus_minus_pair): Likewise.
12395 (eliminate_not_pairs): Likewise.
12396 (undistribute_ops_list): Likewise.
12397 (eliminate_redundant_comparison): Likewise.
12398 (update_range_test): Likewise.
12399 (optimize_range_tests_var_bound): Likewise.
12400 (optimize_vec_cond_expr): Likewise.
12401 (rewrite_expr_tree): Likewise.
12402 (rewrite_expr_tree_parallel): Likewise.
12403 (linearize_expr): Likewise.
12404 (break_up_subtract): Likewise.
12405 (linearize_expr_tree): Likewise.
12406 (attempt_builtin_powi): Likewise.
12407 (attempt_builtin_copysign): Likewise.
12408 (transform_stmt_to_copy): Likewise.
12409 (transform_stmt_to_multiply): Likewise.
12410 (dump_ops_vector): Likewise.
12411 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12412 (print_scc): Likewise.
12413 (set_ssa_val_to): Likewise.
12414 (visit_reference_op_store): Likewise.
12415 (visit_use): Likewise.
12416 (sccvn_dom_walker::before_dom_children): Likewise.
12417 (run_scc_vn): Likewise.
12418 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12420 (expr_hash_elt::print): Likewise.
12421 (const_and_copies::pop_to_marker): Likewise.
12422 (const_and_copies::record_const_or_copy_raw): Likewise.
12423 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12424 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12425 (dump_predicates): Likewise.
12426 (find_uninit_use): Likewise.
12427 (warn_uninitialized_phi): Likewise.
12428 (pass_late_warn_uninitialized::execute): Likewise.
12429 * tree-ssa.c (verify_vssa): Likewise.
12430 (verify_ssa): Likewise.
12431 (maybe_optimize_var): Likewise.
12432 * tree-vrp.c (dump_value_range): Likewise.
12433 (dump_all_value_ranges): Likewise.
12434 (dump_asserts_for): Likewise.
12435 (register_edge_assert_for_2): Likewise.
12436 (vrp_visit_cond_stmt): Likewise.
12437 (vrp_visit_switch_stmt): Likewise.
12438 (vrp_visit_stmt): Likewise.
12439 (vrp_visit_phi_node): Likewise.
12440 (simplify_cond_using_ranges_1): Likewise.
12441 (fold_predicate_in): Likewise.
12442 (evrp_dom_walker::before_dom_children): Likewise.
12443 (evrp_dom_walker::push_value_range): Likewise.
12444 (evrp_dom_walker::pop_value_range): Likewise.
12445 (execute_early_vrp): Likewise.
12447 2017-05-16 Richard Biener <rguenther@suse.de>
12449 * dwarf2out.c (loc_list_from_tree_1): Do not create
12450 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12452 2017-05-16 Richard Biener <rguenther@suse.de>
12454 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12456 (note_variable_value_in_expr): If we resolved the decl ref
12457 do not push to the stack.
12459 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12461 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12462 operations in fast-math mode.
12463 (vaddq_f16): Likewise.
12464 (vmul_f16): Likewise.
12465 (vmulq_f16): Likewise.
12466 (vsub_f16): Likewise.
12467 (vsubq_f16): Likewise.
12468 * config/arm/neon.md (add<mode>3): New.
12470 (fma:<VH:mode>3): New. Also remove outdated comment.
12473 2017-05-16 Martin Liska <mliska@suse.cz>
12477 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12478 (odr_types_equivalent_p): Likewise.
12480 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12482 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12484 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12487 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12488 non-interunit SSE move alternatives with '?'.
12489 (zero-extendsidi peephole2): New peephole to skip intermediate
12490 general register in SSE zero-extend sequence.
12492 2017-05-15 Jeff Law <law@redhat.com>
12494 * reorg.c (relax_delay_slots): Create a new variable to hold
12495 the temporary target rather than clobbering TARGET_LABEL.
12497 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12498 missing argument to extract_bit_field call.
12499 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12501 2017-05-15 Martin Liska <mliska@suse.cz>
12504 * gcc.c (process_command): Do not allow empty argument of -o option.
12506 2017-05-15 Renlin Li <renlin.li@arm.com>
12508 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12509 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12510 * config/aarch64/constraints.md (Usf): Add long call check.
12511 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12512 (call_value): Likewise.
12513 (sibcall): Likewise.
12514 (sibcall_value): Likewise.
12516 (call_value_insn): New.
12517 (sibcall_insn): Update rtx pattern.
12518 (sibcall_value_insn): Likewise.
12519 (call_internal): Remove.
12520 (call_value_internal): Likewise.
12521 (sibcall_internal): Likewise.
12522 (sibcall_value_internal): Likewise.
12523 (call_reg): Likewise.
12524 (call_symbol): Likewise.
12525 (call_value_reg): Likewise.
12526 (call_value_symbol): Likewise.
12528 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12531 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12533 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12535 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12536 compatible with CCGOCmode and with CCZmode.
12538 2017-05-14 Martin Sebor <msebor@redhat.com>
12540 PR middle-end/77671
12541 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12542 (gimple_fold_builtin_snprintf): Same.
12543 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12544 (gimple_fold_builtin_snprintf): Same.
12545 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12546 of character types.
12547 (is_call_safe): New function.
12548 (try_substitute_return_value): Call it.
12549 (try_simplify_call): New function.
12550 (pass_sprintf_length::handle_gimple_call): Call it.
12552 2017-05-14 Martin Sebor <msebor@redhat.com>
12554 PR middle-end/80669
12555 * builtins.c (expand_builtin_stpncpy): Simplify.
12557 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12559 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12560 * config/i386/i386.h
12561 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12562 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12563 (struct machine_function): Add new members call_ms2sysv,
12564 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12565 (struct machine_frame_state): New fields sp_realigned and
12566 sp_realigned_offset.
12567 * config/i386/i386.c
12568 (enum xlogue_stub): New enum.
12569 (enum xlogue_stub_sets): New enum.
12570 (class xlogue_layout): New class.
12571 (struct ix86_frame): New fields stack_realign_allocate_offset,
12572 stack_realign_offset and outlined_save_offset. Modify comments to
12573 detail stack layout when using out-of-line stubs.
12574 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12575 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12576 -mcall-ms2sysv-xlogues.
12577 (stub_managed_regs): New static variable.
12578 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12579 registers managed by out-of-line stub.
12580 (disable_call_ms2sysv_xlogues): New function.
12581 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12582 m->call_ms2sysv when appropriate and compute frame layout for
12584 (sp_valid_at, fp_valid_at): New inline functions.
12585 (choose_basereg): New function.
12586 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12588 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12589 Use align parameter of choose_baseaddr to generated aligned SSE movs
12591 (pro_epilogue_adjust_stack): Modify to track
12592 machine_frame_state::sp_realigned.
12593 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12594 (ix86_nsaved_sseregs): Likewise.
12595 (ix86_emit_save_regs): Likewise.
12596 (ix86_emit_save_regs_using_mov): Likewise.
12597 (ix86_emit_save_sse_regs_using_mov): Likewise.
12598 (get_scratch_register_on_entry): Likewise.
12599 (gen_frame_set): New function.
12600 (gen_frame_load): Likewise.
12601 (gen_frame_store): Likewise.
12602 (emit_outlined_ms2sysv_save): Likewise.
12603 (emit_outlined_ms2sysv_restore): Likewise.
12604 (ix86_expand_prologue): Modify stack re-alignment code and call
12605 emit_outlined_ms2sysv_save when appropriate.
12606 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12607 parameter rtx_insn *insn, which allows the function to be used to only
12608 generate the notes.
12609 (ix86_expand_epilogue): Modify validity checks of frame and stack
12610 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12611 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12612 * config/i386/predicates.md
12613 (save_multiple): New predicate.
12614 (restore_multiple): Likewise.
12615 * config/i386/sse.md
12616 (save_multiple<mode>): New pattern.
12617 (save_multiple_realign<mode>): Likewise.
12618 (restore_multiple<mode>): Likewise.
12619 (restore_multiple_and_return<mode>): Likewise.
12620 (restore_multiple_leave_return<mode>): Likewise.
12621 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12623 2017-05-14 Julia Koval <julia.koval@intel.com>
12625 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12626 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12627 (__builtin_ia32_xsetbv): New builtins.
12628 * config/i386/i386.c (ix86_expand_special_args_builtin):
12630 (ix86_expand_builtin): Special expand for new intrinsics.
12631 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12632 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12633 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12635 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12637 * cfganal.c (inverted_post_order_compute): Change argument type
12639 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12640 * df-core.c (rest_of_handle_df_initialize): Adjust.
12641 (rest_of_handle_df_finish): Likewise.
12642 (df_analyze_1): Likewise.
12643 (df_analyze): Likewise.
12644 (loop_inverted_post_order_compute): Change argument to be a vec *.
12645 (df_analyze_loop): Adjust.
12646 (df_get_n_blocks): Likewise.
12647 (df_get_postorder): Likewise.
12648 * df.h (struct df_d): Change field to be a vec.
12649 * lcm.c (compute_laterin): Adjust.
12650 (compute_available): Likewise.
12651 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12652 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12653 * tree-ssa-pre.c (compute_antic): Likewise.
12655 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12657 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12658 (depth_first_search::depth_first_search): Change structure init
12659 function to this constructor.
12660 (depth_first_search::add_bb): Rename function to this member.
12661 (depth_first_search::execute): Likewise.
12662 (flow_dfs_compute_reverse_finish): Adjust.
12664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12666 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12667 (longest_simple_path): Likewise.
12668 * shrink-wrap.c (spread_components): Likewise.
12669 (disqualify_problematic_components): Likewise.
12670 (emit_common_heads_for_components): Likewise.
12671 (emit_common_tails_for_components): Likewise.
12672 (insert_prologue_epilogue_for_components): Likewise.
12674 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12676 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12679 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12681 * df-core.c (df_set_blocks): Start using auto_bitmap.
12682 (df_compact_blocks): Likewise.
12683 * df-problems.c (df_rd_confluence_n): Likewise.
12684 * df-scan.c (df_insn_rescan_all): Likewise.
12685 (df_process_deferred_rescans): Likewise.
12686 (df_update_entry_block_defs): Likewise.
12687 (df_update_exit_block_uses): Likewise.
12688 (df_entry_block_bitmap_verify): Likewise.
12689 (df_exit_block_bitmap_verify): Likewise.
12690 (df_scan_verify): Likewise.
12691 * lra-constraints.c (lra_constraints): Likewise.
12692 (undo_optional_reloads): Likewise.
12693 (lra_undo_inheritance): Likewise.
12694 * lra-remat.c (calculate_gen_cands): Likewise.
12695 (do_remat): Likewise.
12696 * lra-spills.c (assign_spill_hard_regs): Likewise.
12697 (spill_pseudos): Likewise.
12698 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12699 (bitmap_set_subtract_values): Likewise.
12701 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12703 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12704 management with auto_bitmap.
12705 (fix_inter_tick): Likewise.
12706 (fix_recovery_deps): Likewise.
12707 * ira.c (add_store_equivs): Likewise.
12708 (find_moveable_pseudos): Likewise.
12709 (split_live_ranges_for_shrink_wrap): Likewise.
12710 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12711 (rtx_reuse_manager::seen_def_p): Likewise.
12712 (rtx_reuse_manager::set_seen_def): Likewise.
12713 * print-rtl.h (class rtx_reuse_manager): Likewise.
12715 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12717 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12719 (migrate_btr_def): Likewise.
12720 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12721 * df-core.c (loop_post_order_compute): Likewise.
12722 (loop_inverted_post_order_compute): Likewise.
12723 * hsa-common.h: Likewise.
12724 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12725 * init-regs.c (initialize_uninitialized_regs): Likewise.
12726 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12727 (inline_small_functions): Likewise.
12728 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12729 * ira.c (combine_and_move_insns): Likewise.
12730 (build_insn_chain): Likewise.
12731 * loop-invariant.c (find_invariants): Likewise.
12732 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12733 * predict.c (tree_predict_by_opcode): Likewise.
12734 (predict_paths_leading_to): Likewise.
12735 (predict_paths_leading_to_edge): Likewise.
12736 (estimate_loops_at_level): Likewise.
12737 (estimate_loops): Likewise.
12738 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12739 (spread_components): Likewise.
12740 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12741 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12742 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12743 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12744 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12745 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12746 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12747 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12748 (mark_threaded_blocks): Likewise.
12749 (thread_through_all_blocks): Likewise.
12750 * tree-ssa.c (verify_ssa): Likewise.
12751 (execute_update_addresses_taken): Likewise.
12752 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12756 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12758 (post_order_compute): Likewise.
12759 (inverted_post_order_compute): Likewise.
12760 (pre_and_rev_post_order_compute_fn): Likewise.
12762 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12764 * genrecog.c (int_set::int_set): Explicitly construct our
12765 auto_vec base class.
12766 * vec.h (auto_vec::auto_vec): New constructor.
12768 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12770 * bitmap.h (class auto_bitmap): New constructor taking
12771 bitmap_obstack * argument.
12773 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12775 * bitmap.h (class auto_bitmap): Change type of m_bits to
12776 bitmap_head, and adjust ctor / dtor and member operators.
12778 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12780 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12781 when returned register mode doesn't match original mode.
12783 2017-05-12 Jeff Law <law@redhat.com>
12784 Jakub Jelinek <jakub@redhat.com>
12786 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
12787 we look for cc setter after the compare-elim changes.
12788 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
12789 within the vector to match what compare-elim now expects.
12790 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
12791 (xorsi3_flags, one_cmplsi2_flags): Likewise.
12793 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
12794 after the compare-elim changes.
12795 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
12796 the vector to match what compare-elim now expects.
12797 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
12798 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
12799 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
12800 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
12801 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
12803 * config/visium/visium.c (single_set_and_flags): Fix where
12804 we look for cc setter after the compare-elim changes.
12805 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
12806 with the vector to match what compare-elim now expects.
12807 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
12808 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
12809 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
12810 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
12811 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
12812 (neg<mode>2_insn_set_overflow): Likewise.
12814 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
12816 PR middle-end/79794
12817 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
12818 maybe_expand_insn call, set ops[0].target. If still set after call,
12819 set alt_rtl. Add extra arg to recursive calls.
12820 (extract_bit_field): Add alt_rtl argument. Pass to
12822 * expmed.h (extract_bit_field): Fix prototype.
12823 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
12824 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
12825 to extract_bit_field_calls.
12826 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
12827 Pass alt_rtl to extract_bit_field calls.
12828 * calls.c (store_unaligned_arguments_into_psuedos)
12829 load_register_parameters): Pass extra NULL to extract_bit_field calls.
12830 * optabs.c (maybe_legitimize_operand): Clear op->target when call
12832 * optabs.h (struct expand_operand): Add target bitfield.
12834 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12836 * compare-elim.c (try_eliminate_compare): Canonicalize
12837 operation with embedded compare to
12838 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
12839 (set (reg) (operation)].
12841 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
12843 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12846 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
12847 cost of adding a carry flag for ADC instruction.
12848 [case MINUS]: Ignore the cost of subtracting a carry flag
12849 for SBB instruction.
12851 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
12853 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
12855 * config/rs6000/bmiintrin.h: New file.
12856 * config/rs6000/bmi2intrin.h: New file.
12857 * config/rs6000/x86intrin.h: New file.
12859 2017-05-12 Jeff Law <law@redhat.com>
12861 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
12864 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
12866 PR middle-end/80707
12867 * tree-cfg.c: Remove cfg edges of unreachable case statements.
12869 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12872 early expansion of vector divide builtins.
12873 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
12874 builtins identified as having unsigned arguments.
12876 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12878 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
12879 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
12880 expansion of vector logical operations (and, andc, or, xor,
12883 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12885 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
12886 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
12888 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12890 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12891 early GIMPLE expansion of vector multiplies.
12893 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12895 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
12896 TARGET_HAVE_MOVT conditional.
12897 (movt splitter): Likewise.
12899 2017-05-12 Richard Biener <rguenther@suse.de>
12901 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
12902 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12903 Fold all stmts not inplace.
12905 2017-05-12 Richard Biener <rguenther@suse.de>
12907 PR tree-optimization/80713
12908 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
12909 inserted_exprs bit for not removed stmts.
12911 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
12913 PR middle-end/69921
12914 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
12915 parallelized" attribute for parallelized OpenACC kernels.
12916 * omp-offload.c (execute_oacc_device_lower): Use it.
12918 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
12919 Set "oacc kernels" attribute.
12920 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
12921 parameter. Adjust all users.
12922 (oacc_fn_attrib_kernels_p): Remove function.
12923 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
12924 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
12925 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
12926 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
12927 assert "oacc kernels" attribute is set.
12929 2017-05-11 Carl Love <cel@us.ibm.com>
12931 * config/rs6000/rs6000-c: Add support for built-in functions
12932 vector unsigned char vec_popcnt (vector signed char)
12933 vector unsigned char vec_popcnt (vector unsigned char)
12934 vector unsigned short vec_popcnt (vector signed short)
12935 vector unsigned short vec_popcnt (vector unsigned short)
12936 vector unsigned int vec_popcnt (vector signed int)
12937 vector unsigned int vec_popcnt (vector unsigned int)
12938 vector unsigned long long vec_popcnt (vector signed long long)
12939 vector unsigned long long vec_popcnt (vector unsigned long long)
12940 vector signed long long vec_slo (vector signed long long,
12941 vector signed char)
12942 vector signed long long vec_slo (vector signed long long,
12943 vector unsigned char)
12944 vector unsigned long long vec_slo (vector unsigned long long,
12945 vector signed char)
12946 vector unsigned long long vec_slo (vector unsigned long long,
12947 vector unsigned char)
12948 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
12949 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
12950 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
12951 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
12952 * doc/extend.texi: Update the built-in documentation file for the
12953 new built-in functions.
12955 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12957 * attribs.h (sorted_attr_string): Move machine independent
12958 functions for target clone support from the i386 port to common
12959 code. Rename ix86_function_versions to common_function_versions.
12960 Rename make_name to make_unique_name.
12961 (common_function_versions): Likewise.
12962 (make_unique_name): Likewise.
12963 (make_dispatcher_decl): Likewise.
12964 (is_function_default_version): Likewise.
12965 * attribs.c (attr_strcmp): Likewise.
12966 (sorted_attr_string): Likewise.
12967 (common_function_versions): Likewise.
12968 (make_unique_name): Likewise.
12969 (make_dispatcher_decl): Likewise.
12970 (is_function_default_version): Likewise.
12971 * config/i386/i386.c (attr_strcmp): Likewise.
12972 (sorted_attr_string): Likewise.
12973 (ix86_function_versions): Likewise.
12974 (make_name): Likewise.
12975 (make_dispatcher_decl): Likewise.
12976 (is_function_default_version): Likewise.
12977 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
12979 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12982 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12983 Account for direct move costs for vec_construct of integer
12986 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
12989 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
12990 (UNSPEC_STX_ATOMIC): Ditto.
12991 (loaddi_via_sse): New insn.
12992 (storedi_via_sse): Ditto.
12993 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
12994 Update corresponding peephole2 patterns.
12995 (atomic_storedi_fpu): Ditto.
12997 2017-05-11 Julia Koval <julia.koval@intel.com>
12999 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13000 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13002 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13003 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13004 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13006 2017-05-11 Nathan Sidwell <nathan@acm.org>
13008 * graphite-poly.c: Include dumpfile.h.
13010 * dumpfle.h (dump_function): Declare here ...
13011 * tree-dump.h (dump_function): ... not here.
13012 * dumpfile.c: #include tree-cfg.h.
13013 (dump_function): Move here from ...
13014 * tree-dump.c (dump_function): ... here.
13015 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13016 * graphite-poly.c: Don't include tree-dump.h.
13017 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13018 * print-tree.c: Likewise.
13019 * stor-layout.c: Likewise.
13020 * tree-nested.c: Likewise.
13022 * dumpfile.c (dump_start): Use TDF_FLAGS.
13023 (dump_enable_all): Fix TDF_KIND check thinko.
13025 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13027 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13028 array entries to represent two legal parameterizations of the
13029 overloaded __builtin_cmpb function, as represented by the
13030 P6_OV_BUILTIN_CMPB constant.
13031 (altivec_resolve_overloaded_builtin): Add special case handling
13032 for the __builtin_cmpb function, as represented by the
13033 P6_OV_BUILTIN_CMPB constant.
13034 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13035 (BU_P6_64BIT_2): New macro.
13036 (BU_P6_OVERLOAD_2): New macro
13037 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13038 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13039 (CMPB): Add overload support to represent both 32-bit and 64-bit
13040 compare-bytes function.
13041 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13042 support for TARGET_CMPB.
13043 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13044 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13045 documentation of the __builtin_cmpb overloaded built-in function.
13047 2017-05-11 Richard Biener <rguenther@suse.de>
13049 PR tree-optimization/80705
13050 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13051 bases are not vectorizable.
13053 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13055 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13056 when counting register pressure.
13058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13060 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13061 (iv_ca_more_deps): Renamed to ...
13062 (iv_ca_compare_deps): ... this.
13063 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13065 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13067 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13069 (determine_group_iv_costs): ... here.
13070 (find_inv_vars_cb): Record inv var if it's not recorded before.
13072 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13074 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13075 (get_shiftadd_cost): Ditto.
13077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13079 * tree-ssa-address.c: Include header file.
13080 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13082 (add_to_parts): Refactor.
13083 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13084 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13087 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13089 PR tree-optimization/53090
13090 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13092 (extract_cond_operands): Detect condition with IV on both sides
13093 and return COMP_IV_EXPR_2.
13094 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13095 (rewrite_use_compare): Simplify by removing call to function
13096 extract_cond_operands.
13098 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13100 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13101 (extract_cond_operands): Detect condition comparing against non-
13102 invariant bound and return appropriate enum value.
13103 (find_interesting_uses_cond): Update use of extract_cond_operands.
13104 Handle its return value accordingly.
13105 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13109 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13110 nonlinear iv_use computation in loop invariant sensitive way.
13112 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13114 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13115 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13117 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13119 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13120 (dump_cand): Support iv_cand.inv_exprs.
13121 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13123 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13126 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13128 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13130 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13131 as local function. Include necessary header files.
13132 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13134 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13136 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13138 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13140 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13141 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13142 RSHIFT_EXPR and BIT_NOT_EXPR.
13144 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13146 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13147 (adjust_setup_cost): New parameter supporting round up adjustment.
13148 (struct address_cost_data): Delete.
13149 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13150 (split_address_cost, ptr_difference_cost): Delete.
13151 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13152 (struct ainc_cost_data): New struct.
13153 (get_address_cost_ainc): New function.
13154 (get_address_cost, get_computation_cost): Reimplement.
13155 (determine_group_iv_cost_address): Record inv_expr for all uses of
13157 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13158 (iv_ca_has_deps): Reimplemented to ...
13159 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13161 (iv_ca_extend): Call iv_ca_more_deps.
13163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13165 * tree-ssa-address.c (struct mem_address): Move to header file.
13166 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13167 * tree-ssa-address.h (struct mem_address): Move from C file.
13168 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13170 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13172 * tree-affine.h (aff_combination_type): New interface.
13173 (aff_combination_zero_p): Remove static.
13174 (aff_combination_const_p): New interface.
13175 (aff_combination_singleton_var_p): New interfaces.
13177 2017-05-11 Richard Biener <rguenther@suse.de>
13179 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13180 Skip unreachable blocks and destinations.
13181 (eliminate): Move stmt removal and fixup ...
13182 (fini_eliminate): ... here. Skip inserted exprs.
13183 (pass_pre::execute): Move fini_pre after fini_eliminate.
13184 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13185 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13186 PRE to get rid of dead code that has invalid SSA form and
13187 split critical edges again.
13189 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13191 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13193 2017-05-11 Richard Biener <rguenther@suse.de>
13195 * passes.c (execute_function_todo): Verify loops if they are
13196 said to be up-to-date.
13197 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13198 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13200 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13203 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13204 handle calling assemble_external ourself.
13207 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13208 modes with zero size. Enhance comment.
13210 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13212 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13213 built-ins for vec_xl and vec_xst with short and char pointer
13216 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13218 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13219 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13220 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13221 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13222 (_mm_maskz_min_round_ss): New intrinsics.
13223 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13224 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13225 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13226 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13227 (__builtin_ia32_minss_mask_round): New builtins.
13228 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13229 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13230 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13232 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13233 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13235 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13238 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13240 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13241 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13242 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13243 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13244 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13245 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13246 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13247 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13248 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13249 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13250 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13251 (__builtin_ia32_mulss_mask_round): New builtins.
13252 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13253 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13254 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13256 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13257 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13259 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13262 2017-05-10 Julia Koval <julia.koval@intel.com>
13264 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13265 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13266 (_mm256_setr_m128i): New intrinsics.
13268 2017-05-10 Julia Koval <julia.koval@intel.com>
13270 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13271 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13272 (_mm_maskz_rcp14_ss): New intrinsics.
13273 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13274 (__builtin_ia32_rcp14ss_mask): New builtins.
13275 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13277 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13279 PR tree-optimization/51513
13280 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13281 (assert_unreachable_fallthru_edge_p): Use it.
13282 (group_case_labels_stmt): Likewise.
13283 * tree-cfg.h: Prototype it.
13284 * stmt.c: Include cfghooks.h and tree-cfg.h.
13285 (emit_case_dispatch_table) <gap_label>: New local variable.
13286 Use it to fill dispatch table gaps.
13287 Test for default_label before updating probabilities.
13288 (expand_case) <default_label>: Remove unneeded initialization.
13289 Test for unreachable default case statement and remove its edge.
13290 Set default_label accordingly.
13291 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13293 2017-05-10 Carl Love <cel@us.ibm.com>
13295 * config/rs6000/rs6000-c: Add support for built-in functions
13296 vector signed char vec_neg (vector signed char)
13297 vector signed short int vec_neg (vector short int)
13298 vector signed int vec_neg (vector signed int)
13299 vector signed long long vec_neg (vector signed long long)
13300 vector float vec_neg (vector float)
13301 vector double vec_neg (vector double)
13302 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13304 * config/rs6000/altivec.h: Add define for vec_neg
13305 * doc/extend.texi: Update the built-in documentation for the
13306 new built-in functions.
13308 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13310 PR tree-optimization/77644
13311 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13313 2017-05-10 Nathan Sidwell <nathan@acm.org>
13315 * dumpfile.h (TDI_lang_all): New.
13316 (TDF_KIND): New. Renumber others
13317 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13319 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13321 (get_dump_file_name): Adjust suffix generation.
13322 (dump_enable_all): Use TDF_KIND.
13323 * doc/invoke.texi (-fdump-lang-all): Document.
13325 * dumpfile.h: Tabify.
13327 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13330 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13331 Move member access before delete.
13333 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13335 * tree-inline.c (expand_call_inline): Split block at stmt
13338 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13341 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13342 are now unused after splitting mov{sf,sd}_hardfloat.
13343 (f32_lr2): Likewise.
13344 (f32_lm): Likewise.
13345 (f32_lm2): Likewise.
13346 (f32_li): Likewise.
13347 (f32_li2): Likewise.
13348 (f32_lv): Likewise.
13349 (f32_sr): Likewise.
13350 (f32_sr2): Likewise.
13351 (f32_sm): Likewise.
13352 (f32_sm2): Likewise.
13353 (f32_si): Likewise.
13354 (f32_si2): Likewise.
13355 (f32_sv): Likewise.
13356 (f32_dm): Likewise.
13357 (f32_vsx): Likewise.
13358 (f32_av): Likewise.
13359 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13360 For movsf, order stores so the VSX stores occur before the GPR
13361 store which encourages the register allocator to use a traditional
13362 FPR instead of a GPR. For movsd, order the stores so that the GPR
13363 store comes before the VSX stores to allow the power6 to work.
13364 This is due to the power6 not having a 32-bit integer store
13365 instruction from a FPR.
13366 (movsf_hardfloat): Likewise.
13367 (movsd_hardfloat): Likewise.
13369 2017-05-09 Martin Sebor <msebor@redhat.com>
13371 PR translation/80280
13372 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13375 PR translation/80280
13376 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13377 data member added in r247778.
13378 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13380 2017-05-09 Nathan Sidwell <nathan@acm.org>
13382 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13384 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13387 2017-05-09 Marek Polacek <polacek@redhat.com>
13389 * doc/invoke.texi: Fix typo.
13391 2017-05-09 Richard Biener <rguenther@suse.de>
13393 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13394 (vrp_val_is_min): Likewise.
13395 (set_value_range_to_value): Likewise.
13396 (set_value_range_to_nonnegative): Likewise.
13397 (gimple_assign_nonzero_p): Likewise.
13398 (gimple_stmt_nonzero_p): Likewise.
13399 (vrp_int_const_binop): Likewise. Remove unreachable case.
13400 (adjust_range_with_scev): Adjust comments.
13401 (compare_range_with_value): Likewise.
13402 (extract_range_from_phi_node): Likewise.
13403 (test_for_singularity): Likewise.
13405 2017-05-09 Richard Biener <rguenther@suse.de>
13407 * tree-vrp.c (get_single_symbol): Add assert that we don't
13408 get overflowed constants as invariant part.
13409 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13410 checks. Use wi::cmp instead of recursing for integer constants.
13411 (compare_values): Just ignore whether we assumed undefined
13412 overflow instead of failing the compare.
13413 (extract_range_for_var_from_comparison_expr): Add comment before the
13414 TREE_NO_WARNING sets.
13415 (test_for_singularity): Likewise.
13416 (extract_range_from_comparison): Do not disable optimization
13417 when we assumed undefined overflow.
13418 (extract_range_basic): Remove init of unused var.
13420 2017-05-09 Richard Biener <rguenther@suse.de>
13422 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13423 (extract_range_from_multiplicative_op_1): Adjust.
13424 (extract_range_from_binary_expr_1): Use int_const_binop.
13426 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13429 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13430 rs6000_store_data_bypass_p in seven define_bypass directives and
13431 in several comments.
13432 * config/rs6000/rs6000-protos.h: Add prototype for
13433 rs6000_store_data_bypass_p function.
13434 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13435 function implements slightly different (rs6000-specific) semantics
13436 than store_data_bypass_p, returning false rather than aborting
13437 with assertion error when arguments do not satisfy the
13438 requirements of store data bypass.
13439 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13440 rs6000_store_data_bypass_p.
13442 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13444 * config/xtensa/xtensa-protos.h
13445 (xtensa_initial_elimination_offset): New declaration.
13446 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13447 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13448 macro definition, add case for FRAME_POINTER_REGNUM when
13449 FRAME_GROWS_DOWNWARD.
13450 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13451 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13452 xtensa_initial_elimination_offset.
13454 2017-05-08 Nathan Sidwell <nathan@acm.org>
13456 * doc/invoke.texi: Alphabetize -fdump options.
13458 2017-05-08 Martin Sebor <msebor@redhat.com>
13460 PR translation/80280
13461 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13463 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13465 * target.def (compute_frame_layout): New optional target hook.
13466 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13467 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13468 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13470 * reload1.c (verify_initial_elim_offsets): Likewise.
13471 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13472 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13473 (arm_get_frame_offsets): Split up into this ...
13474 (arm_compute_frame_layout): ... and this function.
13476 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13478 * config/aarch64/constraints.md (Usa): New constraint.
13479 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13481 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13483 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13484 with_multilib_list after it has been checked.
13486 2017-05-08 Richard Biener <rguenther@suse.de>
13488 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13489 (bitmap_set_subtract_values): Likewise.
13491 2017-05-08 Richard Biener <rguenther@suse.de>
13493 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13494 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13496 (gimple_stmt_nonzero_warnv_p): Rename to ...
13497 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13499 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13500 (extract_range_basic): Adjust, do not disable propagation on
13501 strict overflow sensitive simplification.
13502 (vrp_visit_cond_stmt): Likewise.
13504 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13506 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13507 body size unconditionally.
13509 2017-05-07 Jeff Law <law@redhat.com>
13512 2017-05-06 Jeff Law <law@redhat.com>
13513 PR tree-optimization/78496
13514 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13517 PR tree-optimization/78496
13518 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13519 (simplify_stmt_using_ranges): Call it.
13520 (vrp_dom_walker::before_dom_children): Extract equivalences
13521 from an ASSERT_EXPR with an equality comparison against a
13524 2017-05-06 Jeff Law <law@redhat.com>
13526 PR tree-optimization/78496
13527 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13530 PR tree-optimization/78496
13531 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13532 (simplify_stmt_using_ranges): Call it.
13533 (vrp_dom_walker::before_dom_children): Extract equivalences
13534 from an ASSERT_EXPR with an equality comparison against a
13537 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13539 * lra-constraints.c (lra_copy_reg_equiv): New function.
13540 (split_reg): Use it to copy equivalence information from the
13541 original register to the spill register.
13543 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13545 PR rtl-optimization/75964
13546 * simplify-rtx.c (simplify_const_relational_operation): Remove
13547 invalid handling of comparisons of integer ABS.
13549 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13551 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13552 initialize to zero.
13553 (init_regs): Remove declaration.
13554 (function_arg_advance_32): Initialize error_p as boolean variable.
13556 2017-05-05 Nathan Sidwell <nathan@acm.org>
13558 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13559 lines. Use for (;;).
13561 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13563 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13564 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13565 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13566 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13567 VF=2 that require versioning.
13569 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13571 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13574 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13576 * diagnostic.h (diagnostic_override_option_index): Convert from
13577 macro to inline function.
13579 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13581 * diagnostic.c (last_module_changed_p): New function.
13582 (set_last_module): New function.
13583 (diagnostic_report_current_module): Convert macro usage to
13584 the above functions.
13585 * diagnostic.h (diagnostic_context::last_module): Strengthen
13586 from const line_map * to const line_map_ordinary *.
13587 (diagnostic_last_module_changed): Delete macro.
13588 (diagnostic_set_last_module): Delete macro.
13590 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13592 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13593 with diagnostic_report_diagnostic.
13594 (diagnostic_n_impl_richloc): Likewise.
13595 * diagnostic.h (report_diagnostic): Delete macro.
13596 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13597 with diagnostic_report_diagnostic.
13598 * substring-locations.c (format_warning_va): Likewise.
13600 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13602 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13603 save/restor of format_spec. Move option-printing code to...
13604 (print_option_information): ...this new function, and
13605 reimplement by simply printing to the pretty_printer,
13606 rather than appending to the format string.
13608 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13610 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13611 handling logic into...
13612 (update_effective_level_from_pragmas): ...this new function.
13614 2017-05-04 Andrew Waterman <andrew@sifive.com>
13616 * config/riscv/riscv.opt (mstrict-align): New option.
13617 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13618 (SLOW_UNALIGNED_ACCESS): Define.
13619 (riscv_slow_unaligned_access): Declare.
13620 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13622 (riscv_slow_unaligned_access): New variable.
13623 (rocket_tune_info): Set slow_unaligned_access to true.
13624 (optimize_size_tune_info): Set slow_unaligned_access to false.
13625 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13626 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13627 (riscv_option_override): Set riscv_slow_unaligned_access.
13628 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13630 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13632 * config/riscv/riscv.md: Unify indentation.
13634 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13639 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13641 (extendsi<mode>2): Add support for doing sign extension via
13642 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13643 don't have ISA 3.0 instructions.
13644 (extendsi<mode>2 splitter): Likewise.
13645 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13646 generate the normal insns since SImode can now go in vector
13647 registers. Disallow the special UNSPECs needed for previous
13648 machines to hide SImode being used. Add new insns
13649 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13650 (fix_trunc<mode>si2_stfiwx): Likewise.
13651 (fix_trunc<mode>si2_internal): Likewise.
13652 (fixuns_trunc<mode>si2): Likewise.
13653 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13654 (fctiw<u>z_<mode>_smallint): Likewise.
13655 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13656 of floating point to 32-bit integer from doing a direct move to
13657 the GPR registers to do a store.
13658 (fctiwz_<mode>): Break long line.
13660 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13662 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13663 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13664 (addr_list, addr_offset_valid_p): New.
13665 (split_address_groups): Check offset validity with above function.
13666 (gt-tree-ssa-loop-ivopts.h): Include header file.
13668 2017-05-05 Nathan Sidwell <nathan@acm.org>
13670 * config.gcc (arm*-*-*): Add missing 'fi'.
13672 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13674 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13675 included in -fopt-info does not matter.
13676 * doc/optinfo.texi (-fopt-info): Fix description of default
13677 behavour. Explicitly say order of options included in -fopt-info
13680 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13682 * config.gcc: Allow combinations of aprofile and rmprofile values for
13683 --with-multilib-list.
13684 * config/arm/t-multilib: New file.
13685 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13686 variables. Remove setting of ISA and floating-point ABI in
13687 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13688 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13689 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13690 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13692 * config/arm/t-rmprofile: Likewise except for the matches changes.
13693 * doc/install.texi (--with-multilib-list): Document the combination of
13694 aprofile and rmprofile values and warn about pitfalls in doing that.
13696 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13698 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13699 (movdi_aarch64): Likewise.
13701 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13703 PR tree-optimization/80632
13704 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13706 (build_arrays): Initialize it for virtual phis.
13707 (fix_phi_nodes): Use it for virtual phis.
13709 PR tree-optimization/80558
13710 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13711 [x, y] op z into [x op, y op z] for op & or | if conditions
13714 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13715 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13718 * config/arm/arm.md (use_literal_pool): Remove.
13719 (64-bit immediate split): No longer takes cost into consideration
13720 if arm_disable_literal_pool is enabled.
13721 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13722 used when arm_disable_literal_pool is enabled.
13723 (arm_max_const_double_inline_cost): Remove use of
13724 arm_disable_literal_pool.
13725 (push_minipool_fix): Add assert.
13726 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13727 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13728 (no_literal_pool_sf_immediate): New.
13730 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13732 PR tree-optimization/80613
13733 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13734 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13736 2017-05-05 Richard Biener <rguenther@suse.de>
13738 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13740 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13742 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13743 of this flag from insn conditions due to removal from r247495.
13745 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13747 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13749 (arm_early_store_addr_dep_ptr): Likewise.
13750 * config/arm/aarch-common-protos.h
13751 (arm_early_load_addr_dep_ptr): Add prototype.
13752 (arm_early_store_addr_dep_ptr): Likewise.
13753 * config/arm/cortex-a53.md: Add new bypasses.
13755 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13757 * tree.c (next_type_uid): Change type to unsigned.
13758 (type_hash_canon): Decrement back next_type_uid if
13759 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13760 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13763 2017-05-04 Martin Sebor <msebor@redhat.com>
13765 * builtins.c: Fix a trivial typo in a comment.
13767 PR middle-end/79234
13768 * builtins.c (check_sizes): Adjust to handle reading past the end.
13769 Avoid printing excessive upper bound of ranges. Use %E to print
13770 tree nodes instead of converting them to %wu.
13771 (expand_builtin_memchr): New function.
13772 (compute_dest_size): Rename...
13773 (compute_objsize): ...to this.
13774 (expand_builtin_memcpy): Adjust.
13775 (expand_builtin_mempcpy): Adjust.
13776 (expand_builtin_strcat): Adjust.
13777 (expand_builtin_strcpy): Adjust.
13778 (check_strncat_sizes): Adjust.
13779 (expand_builtin_strncat): Adjust.
13780 (expand_builtin_strncpy): Adjust and simplify.
13781 (expand_builtin_memset): Adjust.
13782 (expand_builtin_bzero): Adjust.
13783 (expand_builtin_memcmp): Adjust.
13784 (expand_builtin): Handle memcmp.
13785 (maybe_emit_chk_warning): Check strncat just once.
13787 2017-05-04 Martin Sebor <msebor@redhat.com>
13789 PR preprocessor/79214
13790 PR middle-end/79222
13791 PR middle-end/79223
13792 * builtins.c (check_sizes): Add inlining context and issue
13793 warnings even when -Wno-system-headers is set.
13794 (check_strncat_sizes): Same.
13795 (expand_builtin_strncat): Same.
13796 (expand_builtin_memmove): New function.
13797 (expand_builtin_stpncpy): Same.
13798 (expand_builtin): Handle memmove and stpncpy.
13800 2017-05-04 Bin Cheng <bin.cheng@arm.com>
13802 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
13803 which is not used any more.
13805 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13807 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
13809 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13811 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
13812 (cortexa53_tunings): Likewise.
13813 (cortexa57_tunings): Likewise.
13814 (cortexa72_tunings): Likewise.
13815 (cortexa73_tunings): Likewise.
13817 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13819 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
13820 Set loop alignment to 8.
13822 2017-05-04 Martin Sebor <msebor@redhat.com>
13824 PR translation/80280
13825 * builtins.c (expand_builtin_object_size): Add missing quoting to
13826 %D and like directives.
13827 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
13828 (hsa_type_for_tree_type): Same.
13829 (verify_function_arguments): Same.
13830 * symtab.c (symbol_table::change_decl_assembler_name): Same.
13831 * varasm.c (get_section): Same.
13834 2017-05-04 Martin Sebor <msebor@redhat.com>
13836 PR translation/80280
13837 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
13839 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13841 * config/aarch64/aarch64.c (generic_addrcost_table):
13842 Change HI/TI mode setting.
13844 2017-05-04 Martin Jambor <mjambor@suse.cz>
13846 PR tree-optimization/80622
13847 * tree-sra.c (comes_initialized_p): New function.
13848 (build_accesses_from_assign): Only set write lazily when
13849 comes_initialized_p is false.
13850 (analyze_access_subtree): Use comes_initialized_p.
13851 (propagate_subaccesses_across_link): Assert !comes_initialized_p
13852 instead of testing for PARM_DECL.
13854 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13856 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
13857 constraint on operand 0 to allow more general addressing modes.
13858 Adjust output template.
13859 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
13861 * config/aarch64/aarch64-protos.h
13862 (aarch64_address_valid_for_prefetch_p): Declare prototype.
13863 * config/aarch64/constraints.md (Dp): New address constraint.
13864 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
13867 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
13869 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
13870 update use of estimate_ipcp_clone_size_and_time.
13871 (estimate_local_effects): Update use of
13872 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
13873 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
13874 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
13875 Return nonspecialized time.
13877 2017-05-04 Richard Biener <rguenther@suse.de>
13879 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
13880 for the last VUSE which def dominates the PHI. Directly call
13882 (get_continuation_for_phi_1): Remove.
13884 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
13886 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
13887 to explain the use of truncating division. Cap the number of
13888 iterations to the maximum given by nb_iterations_upper_bound,
13891 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13893 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
13894 * configure: Regenerate.
13895 * config.in: Regenerate.
13896 * config/i386/driver-mingw32.c: new file.
13897 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
13898 * config.host: Link driver-mingw32.o on MinGW host.
13899 * doc/install.texi: Document new --enable-mingw-wildcard configure
13902 2017-05-04 Marek Polacek <polacek@redhat.com>
13904 PR tree-optimization/80612
13905 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
13907 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13908 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
13910 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
13911 (movt splitter): Likewise.
13912 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
13913 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
13914 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
13915 block for Thumb-1 with MOVT.
13916 (thumb2_legitimate_address_p): Move code block ...
13917 (can_avoid_literal_pool_for_label_p): ... into this new function.
13918 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
13920 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
13921 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
13922 "M-profile targets with the MOVT instruction".
13924 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13926 * config/arm/arm-builtins.c (arm_init_builtins): Rename
13927 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
13928 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
13930 2017-05-04 Martin Liska <mliska@suse.cz>
13932 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
13933 variable cond_code.
13935 2017-05-04 Richard Biener <rguenther@suse.de>
13937 * tree.c (array_at_struct_end_p): Handle arrays at struct
13938 end with flexarrays more conservatively. Refactor and treat
13939 arrays of arrays or aggregates more strict. Fix
13940 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
13941 * tree.c (array_at_struct_end_p): Adjust prototype.
13942 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13943 * gimple-fold.c (get_range_strlen): Likewise.
13944 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
13946 2017-05-04 Richard Biener <rguenther@suse.de>
13948 PR tree-optimization/31130
13949 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
13951 (supports_overflow_infinity): Likewise.
13952 (is_negative_overflow_infinity): Likewise.
13953 (is_positive_overflow_infinity): Likewise.
13954 (is_overflow_infinity): Likewise.
13955 (stmt_overflow_infinity): Likewise.
13956 (overflow_infinity_range_p): Likewise.
13957 (usable_range_p): Remove as always returning true.
13958 (make_overflow_infinity): Remove.
13959 (negative_overflow_infinity): Likewise.
13960 (positive_overflow_infinity): Likewise.
13961 (avoid_overflow_infinity): Likewise.
13962 (set_value_range): Adjust accordingly.
13963 (set_value_range_to_nonnegative): Likewise, remove now unused
13964 overflow_infinity arg.
13965 (vrp_operand_equal_p): Adjust.
13966 (update_value_range): Likewise.
13967 (range_int_cst_singleton_p): Likewise.
13968 (operand_less_p): Likewise.
13969 (compare_values_warnv): Likewise.
13970 (extract_range_for_var_from_comparison_expr): Likewise.
13971 (vrp_int_const_binop): Likewise.
13972 (zero_nonzero_bits_from_vr): Likewise.
13973 (extract_range_from_multiplicative_op_1): Likewise.
13974 (extract_range_from_binary_expr_1): Likewise.
13975 (extract_range_from_unary_expr): Likewise.
13976 (extract_range_from_comparison): Likewise.
13977 (extract_range_basic): Likewise.
13978 (adjust_range_with_scev): Likewise.
13979 (compare_ranges): Likewise.
13980 (compare_range_with_value): Likewise.
13981 (dump_value_range): Likewise.
13982 (test_for_singularity): Likewise, remove strict_overflow_p parameter
13984 (simplify_cond_using_ranges): Adjust.
13986 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13988 * brig-builtins.def: Added a builtin for class_f64.
13989 * builtin-types.def: Added a builtin type needed by class_f64.
13991 2017-05-03 Jason Merrill <jason@redhat.com>
13993 * timevar.def: Add TV_CONSTEXPR.
13995 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13997 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
13999 2017-05-03 Martin Jambor <mjambor@suse.cz>
14001 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14002 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14003 * ipa-cp.c (ipcp_propagate_stage): Do not call
14004 ipa_update_after_lto_read.
14005 * ipa-inline.c (ipa_inline): Likewise.
14007 2017-05-03 Martin Jambor <mjambor@suse.cz>
14009 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14010 tag. Added a default constructor and a destructor.
14011 (ipa_edge_args_sum_t): New class;
14012 (ipa_edge_args_sum): Declare.
14013 (ipa_edge_args_vector): Remove declaration.
14014 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14015 (ipa_free_edge_args_substructures): Remove declaration.
14016 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14017 (ipa_edge_args_info_available_for_edge_p): Likewise.
14018 * ipa-prop.c (ipa_edge_args_vector): Removed.
14019 (edge_removal_hook_holder): Likewise.
14020 (edge_duplication_hook_holder): Likewise.
14021 (ipa_edge_args_sum): New variable.
14022 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14023 ipa_edge_args_vector.
14024 (ipa_free_edge_args_substructures): Likewise.
14025 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14026 ipa_edge_args_vector.
14027 (ipa_edge_removal_hook): Turned into method
14028 ipa_edge_args_sum_t::remove.
14029 (ipa_edge_duplication_hook): Turned into method
14030 ipa_edge_args_sum_t::duplicate.
14031 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14032 registering edge hooks.
14033 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14034 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14035 ipa_edge_args_sum instead of ipa_edge_args_vector.
14036 * ipa-profile.c (ipa_profile): Likewise.
14038 2017-05-03 Martin Jambor <mjambor@suse.cz>
14040 * symbol-summary.h (function_summary): New method exists.
14041 (function_summary::symtab_removal): Deallocate through release.
14042 (call_summary): New class.
14043 (gt_ggc_mx): New overload.
14044 (gt_pch_nx): Likewise.
14045 (gt_pch_nx): Likewise.
14047 2017-05-03 Jeff Law <law@redhat.com>
14049 PR tree-optimization/78496
14050 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14051 from simplify_cond_using_ranges. Split off code to walk
14052 backwards through casts into ...
14053 (simplify_cond_using_ranges_2): New function.
14054 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14055 (execute_vrp): After identifying jump threads, call
14056 simplify_cond_using_ranges_2.
14058 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14061 * ipa-inline.h (inline_summary): Add ctor.
14062 (create_ggc): Do not use ggc_cleared_alloc.
14064 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14065 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14067 * gcc.c (handle_braces): Support escaping in switch matching
14069 * doc/invoke.texi (Spec Files): Document it.
14070 Remove superfluous @code markup in items.
14072 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14074 * diagnostic-show-locus.c (struct column_range): New struct.
14075 (get_affected_columns): New function.
14076 (get_printed_columns): New function.
14077 (struct correction): New struct.
14078 (correction::ensure_capacity): New function.
14079 (correction::ensure_terminated): New function.
14080 (struct line_corrections): New struct.
14081 (line_corrections::~line_corrections): New dtor.
14082 (line_corrections::add_hint): New function.
14083 (layout::print_trailing_fixits): Reimplement in terms of the new
14085 (selftest::test_overlapped_fixit_printing): New function.
14086 (selftest::diagnostic_show_locus_c_tests): Call it.
14088 2017-05-03 Nathan Sidwell <nathan@acm.org>
14090 Canonicalize canonical type hashing
14091 * tree.h (type_hash_canon_hash): Declare.
14092 * tree.c (type_hash_list, attribute_hash_list): Move into
14093 type_hash_canon_hash.
14094 (build_type_attribute_qual_variant): Break out hash code calc into
14095 type_hash_canon_hash.
14096 (type_hash_canon_hash): New. Generic type hash computation.
14097 (build_range_type_1, build_array_type_1, build_function_type,
14098 build_method_type_directly, build_offset_type, build_complex_type,
14099 make_vector_type): Call it.
14101 2017-05-03 Richard Biener <rguenther@suse.de>
14103 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14104 When all DRs have unknown misaligned do not always peel
14105 when there is a store but apply the same costing model as if
14106 there were only loads.
14108 2017-05-03 Richard Biener <rguenther@suse.de>
14111 PR tree-optimization/80492
14112 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14113 compare_base_decls returning dont-know properly.
14115 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14117 * config/arm/iterators.md (CCSI): New mode iterator.
14118 (arch): New mode attribute.
14119 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14120 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14121 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14122 code iterator for success result mode.
14123 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14124 the corresponding new insn generators.
14126 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14129 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14130 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14132 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14134 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14135 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14136 (DDR_A): Wrap DDR argument in brackets.
14137 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14138 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14139 (DDR_REVERSED_P): Likewise.
14141 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14143 PR tree-optimization/79472
14144 * tree-switch-conversion.c (struct switch_conv_info): Add
14145 contiguous_range and default_case_nonstandard fields.
14146 (collect_switch_conv_info): Compute contiguous_range and
14147 default_case_nonstandard fields, don't clear final_bb if
14148 contiguous_range and only the default case doesn't have the required
14150 (check_all_empty_except_final): Set default_case_nonstandard instead
14151 of failing if contiguous_range and the default case doesn't have empty
14153 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14154 and only the default case doesn't have the required constants. Skip
14156 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14157 if default_case_nonstandard.
14158 (build_constructors): Build constant 1 just once. Assert that default
14159 values aren't inserted in between cases if contiguous_range. Skip
14161 (build_arrays): Skip virtual phis.
14162 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14163 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14164 Handle virtual phis.
14165 (gen_inbound_check): Handle default_case_nonstandard case.
14166 (process_switch): Adjust check_final_bb caller. Call
14167 gather_default_values with the first non-default case instead of
14168 default case if default_case_nonstandard.
14170 2017-05-02 Nathan Sidwell <nathan@acm.org>
14172 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14173 check. Fix formatting.
14175 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14177 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14178 errors when comparing specialized and unspecialized times.
14180 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14182 * diagnostic-show-locus.c
14183 (layout::should_print_annotation_line_p): Make private.
14184 (layout::print_annotation_line): Make private.
14185 (layout::annotation_line_showed_range_p): Make private.
14186 (layout::show_ruler): Make private.
14187 (layout::print_source_line): Make private. Pass in line and
14188 line_width, rather than calling location_get_source_line. Drop
14190 (layout::print_leading_fixits): New method.
14191 (layout::print_any_fixits): Rename to...
14192 (layout::print_trailing_fixits): ...this, and make private.
14193 Don't print newline fixits.
14194 (diagnostic_show_locus): Move logic for printing one row into...
14195 (layout::print_line): ...this new function. Move the
14196 location_get_source_line call and error-handling from
14197 print_source_line to here. Call print_leading_fixits, and rename
14198 print_any_fixits to print_trailing_fixits.
14199 (selftest::test_fixit_insert_containing_newline): Update now that
14200 newlines are partially supported.
14201 (selftest::test_fixit_insert_containing_newline_2): New test.
14202 (selftest::test_fixit_replace_containing_newline): Update comments.
14203 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14204 * edit-context.c (class added_line): New class.
14205 (class edited_line): Describe newline handling in comment.
14206 (edited_line::actually_edited_p): New method.
14207 (edited_line::print_content): Delete redundant decl.
14208 (edited_line::m_predecessors): New field.
14209 (edited_file::print_content): Call edited_line::print_content.
14210 (edited_file::print_diff): Update to support newlines.
14211 (edited_file::print_diff_hunk): Likewise.
14212 (edited_file::print_run_of_changed_lines): New function.
14213 (edited_file::print_diff_line): Convert to...
14214 (print_diff_line): ...this.
14215 (edited_file::get_effective_line_count): New function.
14216 (edited_line::edited_line): Initialize new field m_predecessors.
14217 (edited_line::~edited_line): Clean up m_predecessors.
14218 (edited_line::apply_fixit): Handle newlines.
14219 (edited_line::get_effective_line_count): New function.
14220 (edited_line::print_content): New function.
14221 (edited_line::print_diff_lines): New function.
14222 (selftest::test_applying_fixits_insert_containing_newline): New
14224 (selftest::test_applying_fixits_replace_containing_newline): New
14226 (selftest::insert_line): New function.
14227 (selftest::test_applying_fixits_multiple_lines): Add example of
14229 (selftest::edit_context_c_tests): Call the new tests.
14231 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14233 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14234 parameter cand. Update dump information.
14235 (get_computation_cost): Update uses.
14237 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14239 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14240 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14241 (get_computation_at, rewrite_use_address): Update use of
14242 get_computation_aff.
14244 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14246 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14247 (get_computation): Delete.
14248 (get_computation_cost): Implement like get_computation_cost_at.
14249 Use get_computation_at.
14250 (get_computation_cost_at): Delete.
14251 (rewrite_use_nonlinear_expr): Use get_computation_at.
14252 (rewrite_use_compare, remove_unused_ivs): Ditto.
14254 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14256 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14258 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14260 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14261 (ivopts_global_cost_for_size): Rename parameter and update uses.
14262 (iv_ca_recount_cost): Update uses.
14263 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14264 candidates seperately in n_invs and n_cands.
14265 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14269 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14270 (find_inv_vars_cb): New.
14271 (find_depends): Renamed to ...
14272 (find_inv_vars): ... this.
14273 (add_candidate_1, force_var_cost): Call find_inv_vars.
14274 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14276 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14278 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14279 inv_vars. Add inv_exprs.
14280 (struct iv_cand): Rename depends_on to inv_vars.
14281 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14282 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14283 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14284 (dump_cand): Dump inv_vars.
14285 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14286 (record_invariant, find_depends, add_candidate_1): Ditto.
14287 (set_group_iv_cost, force_var_cost): Ditto.
14288 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14289 (get_computation_cost_at, get_computation_cost): Ditto.
14290 (determine_group_iv_cost_generic): Ditto.
14291 (determine_group_iv_cost_address): Ditto.
14292 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14293 (determine_group_iv_costs): Ditto.
14294 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14295 (iv_ca_set_remove_invariants): Renamed to ...
14296 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14297 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14298 (iv_ca_set_add_invariants): Renamed to ...
14299 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14300 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14301 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14302 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14303 (create_new_ivs): Remove useless dump.
14305 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14307 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14309 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14310 (iv_ca_set_no_cp, create_new_iv): Ditto.
14312 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14314 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14316 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14318 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14319 function tree_check2.
14321 2017-05-02 Martin Liska <mliska@suse.cz>
14323 * doc/gcov.texi: Add missing preposition.
14324 * gcov.c (function_info::function_info): Properly fill up
14325 all member variables.
14327 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14329 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14331 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14333 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14335 2017-05-02 Martin Liska <mliska@suse.cz>
14338 * lto-streamer-in.c (lto_read_tree_1): Remove
14339 LTO_STREAMER_DEBUG.
14340 * lto-streamer.c (struct tree_hash_entry): Likewise.
14341 (struct tree_entry_hasher): Likewise.
14342 (tree_entry_hasher::hash): Likewise.
14343 (tree_entry_hasher::equal): Likewise.
14344 (lto_streamer_init): Likewise.
14345 (lto_orig_address_map): Likewise.
14346 (lto_orig_address_get): Likewise.
14347 (lto_orig_address_remove): Likewise.
14348 * lto-streamer.h: Likewise.
14349 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14350 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14352 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14354 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14355 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14356 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14357 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14358 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14359 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14360 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14361 (mm_maskz_sub_ss): New intrinsics.
14362 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14363 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14364 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14365 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14366 (__builtin_ia32_subss_mask_round): New builtins.
14367 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14368 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14369 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14371 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14372 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14374 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14377 2017-05-02 Martin Jambor <mjambor@suse.cz>
14379 PR tree-optimization/78687
14380 * tree-sra.c (access): New field parent.
14381 (process_subtree_disqualification): New function.
14382 (disqualify_candidate): Call it.
14383 (build_accesses_from_assign): Reset write flag if creating an
14385 (build_access_subtree): Fill in parent field and also prpagate
14386 down grp_write flag.
14387 (create_artificial_child_access): New parameter set_grp_write, set
14388 grp_write to its value.
14389 (propagate_subaccesses_across_link): Also propagate grp_write flag
14391 (propagate_all_subaccesses): Push the closest parent back to work
14392 queue if add_access_to_work_queue returned true.
14394 2017-05-02 Richard Biener <rguenther@suse.de>
14396 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14397 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14398 -fstrict-overflow documentation.
14399 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14400 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14401 flag_strict_overflow.
14402 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14403 * lto-opts.c (lto_write_options): Do not stream it.
14404 * lto-wrapper.c (merge_and_complain): Do not handle it.
14405 * opts.c (default_options_table): Do not set -fstrict-overflow.
14406 (finish_options): Likewise do not clear it when sanitizing.
14407 * simplify-rtx.c (simplify_const_relational_operation): Do not
14408 test flag_strict_overflow.
14410 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14412 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14413 using enabled attribute.
14414 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14415 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14416 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14417 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14418 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14419 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14420 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14421 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14422 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14423 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14425 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14427 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14429 2017-05-02 Richard Biener <rguenther@suse.de>
14431 PR tree-optimization/80591
14433 2017-04-10 Richard Biener <rguenther@suse.de>
14435 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14438 2017-05-02 Richard Biener <rguenther@suse.de>
14440 PR tree-optimization/80549
14441 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14442 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14443 headers if they do not have a preheader.
14445 2017-05-02 Martin Liska <mliska@suse.cz>
14448 * common.opt: Fix typo.
14449 * doc/invoke.texi: Likewise.
14451 2017-05-01 Jan Beulich <jbeulich@suse.com>
14453 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14454 swapping, add (x,x,m,x,n) alternative.
14456 2017-05-01 Nathan Sidwell <nathan@acm.org>
14458 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14459 unnecessary unadjusted_alignment check.
14461 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14464 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14466 * gimplify.c (gimplify_cilk_detach): New function.
14467 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14468 * tree-core.h: Document EXPR_CILK_SPAWN.
14469 * tree.h (EXPR_CILK_SPAWN): Define.
14471 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14473 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14474 to use new fixit_hint representation, using the "replace" logic.
14475 (get_line_span_for_fixit_hint): Likewise.
14476 (layout::print_any_fixits): Likewise.
14477 (selftest::test_one_liner_many_fixits): Rename to...
14478 (selftest::test_one_liner_many_fixits_1): ...this, and update
14479 comment and expected output to reflect that the multiple fix-it
14480 hints are now consolidated into one insertion.
14481 (selftest::test_one_liner_many_fixits_2): New test.
14482 (selftest::test_diagnostic_show_locus_one_liner): Update for
14484 (selftest::test_fixit_consolidation): Update for fix-it API
14486 * diagnostic.c (print_parseable_fixits): Likewise.
14487 * edit-context.c (edited_line::m_line_events): Convert from
14488 auto_vec <line_event *> to auto_vec <line_event>.
14489 (class line_event): Convert from abstract base class to a concrete
14490 class, taking over the role of replace_event.
14491 (class insert_event): Delete.
14492 (class replace_event): Rename to class line_event. Convert to
14494 (edit_context::add_fixits): Reimplement.
14495 (edit_context::apply_insert): Delete.
14496 (edit_context::apply_replace): Rename to...
14497 (edit_context::apply_fixit): ...this. Convert to half-open range.
14498 (edited_file::apply_insert): Delete.
14499 (edited_file::apply_replace): Rename to...
14500 (edited_file::apply_fixit): ...this.
14501 (edited_line::~edited_line): Drop deletion of events.
14502 (edited_line::apply_insert): Delete.
14503 (edited_line::apply_replace): Rename to...
14504 (edited_line::apply_fixit): ...this. Convert to half-open range.
14505 Update for change to type of m_line_events.
14506 * edit-context.h (edit_context::apply_insert): Delete.
14507 (edit_context::apply_replace): Rename to...
14508 (edit_context::apply_fixit): ...this.
14510 2017-05-01 Martin Sebor <msebor@redhat.com>
14512 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14515 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14518 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14519 __get_cpuid_max returns 0.
14520 (__get_cpuid_count): Ditto.
14522 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14524 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14525 replacement expression is another instance of one of its arguments.
14527 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14530 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14531 check for stack push/pop autoinc.
14532 * config/i386/i386.c (ix86_agi_dependent): Return false
14533 if the only reason why modified_in_p returned true is that
14534 addr is SP based and set_insn is a push or pop.
14536 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14538 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14541 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14544 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14545 (account_size_time): Use two predicates - exec_pred and
14547 (evaluate_conditions_for_known_args): Compute both clause and
14549 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14550 (inline_summary_t::duplicate): Update.
14551 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14553 (compute_inline_parameters): Likewise.
14554 (estimate_edge_size_and_time): Update caluclation of time.
14555 (estimate_node_size_and_time): Compute both time and nonspecialized
14557 (estimate_ipcp_clone_size_and_time): Update.
14558 (inline_merge_summary): Update.
14559 (do_estimate_edge_time): Update.
14560 (do_estimate_edge_size): Update.
14561 (do_estimate_edge_hints): Update.
14562 (inline_read_section, inline_write_summary): Stream both new predicates.
14563 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14565 (compute_inlined_call_time): Cleanup.
14566 (big_speedup_p): Update.
14567 (edge_badness): Update.
14568 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14569 (size_time_entry): Replace predicate by exec_predicate and
14570 nonconst_predicate.
14571 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14572 (estimate_edge_time): Return also nonspec_time.
14573 (reset_edge_growth_cache): Update.
14575 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14577 PR rtl-optimization/80491
14578 * ifcvt.c (noce_process_if_block): When looking for x setter
14579 with missing else_bb, don't check only the insn right before
14580 cond_earliest, but look for the last insn that x is modified in
14581 within the same bb.
14583 PR rtl-optimization/80491
14584 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14586 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14588 PR tree-optimization/80487
14589 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14591 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14593 PR tree-optimization/79697
14594 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14595 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14596 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14598 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14599 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14601 2017-04-28 Martin Sebor <msebor@redhat.com>
14603 PR tree-optimization/80523
14604 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14605 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14607 (maybe_warn, format_directive, parse_directive): Use new functions.
14608 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14610 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14612 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14614 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14616 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14617 target_header_dir): Set correctly.
14618 * configure: Regenerated.
14619 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14620 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14621 instead of SYSTEM_HEADER_DIR.
14623 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14625 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14626 (estimate_local_effects): Likewise.
14627 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14628 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14629 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14630 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14631 do_estimate_edge_time, estimate_edge_time): Likewise.
14632 * ipa-inline-analysis.c (estimate_node_size_and_time,
14633 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14634 (estimate_time_after_inlining): Remove.
14636 2017-04-28 Martin Liska <mliska@suse.cz>
14638 * doc/gcov.texi: Enhance documentation of gcov.
14640 2017-04-28 Martin Liska <mliska@suse.cz>
14642 * doc/gcov.texi: Sort options in alphabetic order.
14643 * doc/gcov-dump.texi: Likewise.
14644 * doc/gcov-tool.texi: Likewise.
14645 * gcov.c (print_usage): Likewise.
14646 * gcov-dump.c (print_usage): Likewise.
14647 * gcov-tool.c (print_merge_usage_message): Likewise.
14648 (print_rewrite_usage_message): Likewise.
14649 (print_overlap_usage_message): Likewise.
14651 2017-04-28 Martin Liska <mliska@suse.cz>
14653 PR gcov-profile/53915
14654 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14656 2017-04-28 Martin Liska <mliska@suse.cz>
14658 PR gcov-profile/79891
14659 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14660 is marked by compiler as living on a line.
14661 (get_cycles_count): Remove usage of the union.
14662 (output_intermediate_file): Likewise.
14663 (find_source): Fix GNU coding style.
14664 (accumulate_line_counts): Remove old non-all block mode.
14665 (output_lines): Remove usage of the union.
14666 * profile.c (output_location): Include all BBs, even if
14667 belonging to a same line (and file) as a previous BB.
14669 2017-04-28 Martin Liska <mliska@suse.cz>
14671 * gcov.c (process_args): Handle new argument 'w'.
14672 (read_graph_file): Assign ID to BBs.
14673 (output_branch_count): Display BB # if verbose flag is set.
14674 (output_lines): Likewise for arcs.
14675 (print_usage): Add '--verbose' option help.
14676 * doc/gcov.texi: Document --verbose (-w) option.
14678 2017-04-28 Martin Liska <mliska@suse.cz>
14680 * gcov.c (struct block_location_info): New struct.
14681 (process_file): Fill up the new structure.
14682 (read_graph_file): Replace usage of encoding by the newly added
14684 (add_line_counts): Likewise.
14685 (accumulate_line_counts): Remove usage of the union.
14686 (function_info::function_info): New function.
14687 (function_info::~function_info): Likewise.
14688 (process_file): Call delete instead of release_function.
14689 (release_function): Release the function.
14690 (release_structures): Call delete instead of release_function.
14691 (solve_flow_graph): Replace usage of num_blocks.
14692 (find_exception_blocks): Likewise.
14693 (output_lines): Fix GNU coding style.
14695 2017-04-28 Martin Liska <mliska@suse.cz>
14698 * coverage.c (coverage_remove_note_file): New function.
14699 * coverage.h: Declare the function.
14700 * toplev.c (finalize): Clean if an error has been seen.
14702 2017-04-28 Martin Liska <mliska@suse.cz>
14704 PR gcov-profile/80031
14705 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14706 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14707 * gcov.c (read_graph_file): Read just number of blocks.
14708 * profile.c (branch_prob): Do not stream 0 flags per a basic
14711 2017-04-28 Martin Liska <mliska@suse.cz>
14713 * gcov-dump.c (tag_*): Add new argument to declarations.
14714 (dump_gcov_file): Likewise.
14715 (tag_blocks): Add and use new argument depth.
14716 (tag_arcs): Likewise.
14717 (tag_lines): Likewise.
14718 (tag_counters): Likewise.
14719 (tag_summary): Likewise.
14720 (dump_working_sets): Use depth to do a proper indentation.
14722 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14725 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14726 * symtab.c (symtab_node::debug_symtab): Move definition here.
14728 2017-04-28 Richard Biener <rguenther@suse.de>
14730 * lto-streamer.h (LTO_major_version): Bump to 7.
14732 2017-04-28 Richard Biener <rguenther@suse.de>
14734 * tree-vrp.c (assert_info): New struct.
14735 (add_assert_info): New helper.
14736 (register_edge_assert_for_2): Refactor to add asserts to a vector
14738 (register_edge_assert_for_1): Likewise.
14739 (register_edge_assert_for): Likewise.
14740 (finish_register_edge_assert_for): New helper actually registering
14741 asserts where live on edge.
14742 (find_conditional_asserts): Adjust.
14743 (find_switch_asserts): Likewise.
14744 (evrp_dom_walker::try_find_new_range): Generalize.
14745 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14747 2017-04-27 Marek Polacek <polacek@redhat.com>
14750 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14751 arg10 and arg11 to itype.
14753 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14755 * doc/extend.texi (Object Size Checking): Improve grammar.
14757 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14760 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14761 that the logic for permitting reciprocal estimates matches that
14764 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14767 * tree.c (type_cache_hasher::equal): Only compare
14768 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14769 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14770 non-aggregate element types.
14771 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14772 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14774 2017-04-27 Richard Biener <rguenther@suse.de>
14776 PR middle-end/80533
14777 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14778 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14779 keeping a reference to a trailing array.
14781 2017-04-27 Richard Biener <rguenther@suse.de>
14783 PR middle-end/80539
14784 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
14785 being in loop-closed SSA form conservatively.
14786 (chrec_fold_multiply_poly_poly): Likewise.
14788 2017-04-27 Tamar Christina <tamar.christina@arm.com>
14790 PR middle-end/79665
14791 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
14792 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
14794 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14797 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
14798 (aarch64_function_arg_alignment): Return unsigned int again, but still
14799 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
14800 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14801 Don't emit -Wpsabi note.
14802 (aarch64_function_arg_boundary): Likewise.
14803 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14806 2017-04-26 Nathan Sidwell <nathan@acm.org>
14808 * tree.h (crc32_unsigned_n): Declare.
14809 (crc32_unsigned, crc32_unsigned): Make inline.
14810 * tree.c (crc32_unsigned_bits): Replace with ...
14811 (crc32_unsigned_n): ... this.
14812 (crc32_unsigned, crc32_byte): Remove.
14813 (crc32_string): Remove unnecessary braces.
14815 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14817 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
14818 * ipa-inline-analysis.c (MAX_TIME): Remove.
14819 (account_size_time): Use sreal for time.
14820 (dump_inline_summary): Update.
14821 (estimate_function_body_sizes): Update.
14822 (estimate_edge_size_and_time): Update.
14823 (estimate_calls_size_and_time): Update.
14824 (estimate_node_size_and_time): Update.
14825 (inline_merge_summary): Update.
14826 (inline_update_overall_summary): Update.
14827 (estimate_time_after_inlining): Update.
14828 (inline_read_section): Update.
14829 (inline_write_summary): Update.
14830 * ipa-inline.c (compute_uninlined_call_time): Update.
14831 (compute_inlined_call_time): Update.
14832 (recursive_inlining): Update.
14833 (inline_small_functions): Update.
14834 (dump_overall_stats): Update.
14835 * ipa-inline.h: Include sreal.h.
14836 (size_time_entry): Turn time to sreal.
14837 (inline_summary): Turn self_time nad time to sreal.
14839 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14841 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
14843 (sreal::stream_out, sreal::stream_in): New.
14844 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
14846 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14848 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
14851 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
14854 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
14856 (dimode_scalar_chain::compute_convert_gain): Ditto.
14857 (dimode_scalar_chain::make_vector_copies): Ditto.
14858 (dimode_scalar_chain::convert_reg): Ditto.
14859 (dimode_scalar_chain::convert_insn): Ditto.
14860 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
14861 (VI248_AVX512BW_1): New mode iterator.
14862 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
14863 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
14866 2017-04-25 Martin Sebor <msebor@redhat.com>
14868 PR tree-optimization/80497
14869 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
14870 constants are representable in HOST_WIDE_INT.
14871 (parse_directive): Ditto.
14873 2017-04-25 Martin Sebor <msebor@redhat.com>
14876 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
14877 (new_zero_array): Adjust signature.
14878 (dom_info::dom_init): Used unsigned rather that size_t.
14879 (dom_info::dom_info): Same.
14881 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14882 Jakub Jelinek <jakub@redhat.com>
14885 * config/arm/arm.c: Include gimple.h.
14886 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14887 returns negative, increment ncrn only if it returned positive.
14888 (arm_needs_doubleword_align): Return int instead of bool,
14889 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
14890 members, but if there is any such non-FIELD_DECL
14891 > PARM_BOUNDARY aligned decl, return -1 instead of false.
14892 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14893 returns negative, increment nregs only if it returned positive.
14894 (arm_setup_incoming_varargs): Likewise.
14895 (arm_function_arg_boundary): Emit -Wpsabi note if
14896 arm_needs_doubleword_align returns negative, return
14897 DOUBLEWORD_ALIGNMENT only if it returned positive.
14899 2017-04-25 Marek Polacek <polacek@redhat.com>
14902 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
14903 first argument to type.
14905 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
14908 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
14909 type checks to test for compatibility instead of equality.
14911 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14912 Jakub Jelinek <jakub@redhat.com>
14915 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
14917 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
14918 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
14919 the alignment computation, but return their maximum in warn_alignment.
14920 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14921 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
14923 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
14924 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14927 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14929 * config/arc/simdext.md (dmpyh): Fix typo.
14931 2017-04-25 Richard Biener <rguenther@suse.de>
14933 PR tree-optimization/80492
14934 * alias.c (compare_base_decls): Handle registers with asm
14935 specification conservatively.
14936 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14937 compare_base_decls returning dont-know properly.
14939 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14941 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
14942 (legitimate_offset_address_p): New function.
14943 (arc_legitimate_address_p): Use above function.
14945 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14947 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
14949 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14951 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
14952 ACCH registers whenever they are available.
14954 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14956 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
14957 double regs fix when not used.
14959 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14961 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
14963 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
14964 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
14966 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14968 * config/arc/arc.c (arc_output_addsi): Check for h-register class
14969 when emitting short ADD instructions.
14971 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14973 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
14975 (cmpsi_cc_c_insn): Likewise.
14976 (cbranchsi4_scratch): Compute proper instruction length using
14977 compact_hreg_operand.
14978 * config/arc/predicates.md (compact_hreg_operand): New predicate.
14980 2017-04-25 Richard Biener <rguenther@suse.de>
14982 PR middle-end/80509
14983 * passes.c (pass_manager::pass_manager): Initialize
14984 m_name_to_pass_map.
14986 2017-04-25 Richard Biener <rguenther@suse.de>
14988 PR tree-optimization/79201
14989 * tree-ssa-sink.c (statement_sink_location): Handle calls.
14991 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14994 * config/s390/vector.md: Split MEM->GPR vector moves for
14995 non-s_operand addresses.
14997 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15000 * config/s390/predicates.md (reload_const_wide_int_operand): New
15002 * config/s390/s390.md ("movti"): Remove d/P alternative.
15003 ("movti_bigconst"): New pattern definition.
15005 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15008 * s390-protos.h (s390_expand_cs_hqi): Removed.
15009 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15010 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15011 modes as well as CCZ1mode and CCZmode.
15012 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15013 signature of s390_emit_compare_and_swap.
15014 (s390_expand_cs_hqi): Likewise, make static.
15015 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15016 compare-and-swap, in some cases.
15017 (s390_expand_cs): Wrapper function.
15018 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15020 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15021 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15022 patterns for small and large integers. Forbid symref memory operands.
15023 Move expander to s390.c. Require cc register.
15024 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15025 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15026 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15027 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15028 symref memory operands. Remove CC mode and call s390_match_ccmode
15030 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15032 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15034 * config/s390/s390.md (define_peephole2): New peephole to help
15035 combining the load-and-test pattern with volatile memory.
15037 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15039 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15040 with CCZmode for TARGET_Z196.
15042 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15044 PR rtl-optimization/80501
15045 * combine.c (make_compound_operation_int): Set subreg_code to SET
15046 even for AND with mask of the sign bit of mode.
15048 PR rtl-optimization/80500
15049 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15050 sum's initial value.
15052 2017-04-25 Julian Brown <julian@codesourcery.com>
15053 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15055 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15057 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15059 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15061 2017-04-25 Julian Brown <julian@codesourcery.com>
15062 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15064 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15065 (thunderx2t99_sha): New Reservation.
15067 2017-04-25 Julian Brown <julian@codesourcery.com>
15068 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15070 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15071 type for 1-element load.
15073 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15075 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15077 2017-04-24 Martin Jambor <mjambor@suse.cz>
15079 PR tree-optimization/80293
15080 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15081 char arrays not totally scalarizable if it is false.
15082 (analyze_all_variable_accesses): Pass correct value in the new
15083 parameter. Add a statistics counter.
15085 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15087 PR middle-end/79931
15088 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15090 2017-04-24 Richard Biener <rguenther@suse.de>
15092 PR tree-optimization/80494
15093 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15094 out for complex types.
15096 2017-04-24 Richard Biener <rguenther@suse.de>
15098 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15099 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15100 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15101 (DFS): Adjust and never fail.
15102 (sccvn_dom_walker::fail): Remove.
15103 (sccvn_dom_walker::before_dom_children): Adjust.
15104 (run_scc_vn): Likewise and never fail.
15105 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15106 (pass_fre::execute): Likewise.
15108 2017-04-24 Richard Biener <rguenther@suse.de>
15110 PR tree-optimization/79725
15111 * tree-ssa-sink.c (statement_sink_location): Return whether
15112 failure reason was zero uses. Move that check later.
15113 (sink_code_in_bb): Deal with zero uses by removing the stmt
15116 2017-04-24 Richard Biener <rguenther@suse.de>
15119 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15120 pointer-based references.
15122 2017-04-24 Richard Biener <rguenther@suse.de>
15125 * pass_manager.h (pass_manager::operator new): Remove.
15126 (pass_manager::operator delete): Likewise.
15127 * passes.c (pass_manager::operator new): Remove.
15128 (pass_manager::operator delete): Likewise.
15129 (pass_manager::pass_manager): Zero individual pass members.
15131 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15134 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15135 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15136 Check "XEXP (src, 1)" operand here.
15137 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15138 Check "XEXP (src, 1)" operand here.
15139 (dimode_scalar_chain::make_vector_copies): Detect count register
15140 of a shift instruction. Zero extend count register from QImode
15141 to DImode to satisfy vector shift pattern count operand predicate.
15142 Substitute vector shift count operand with a DImode copy.
15143 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15146 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15148 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15149 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15150 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15151 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15152 (UNSPEC_NOREX_MEM): Remove definition.
15154 2017-04-21 Richard Biener <rguenther@suse.de>
15156 PR tree-optimization/79547
15157 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15158 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15159 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15160 without any constraints.
15162 2017-04-21 Richard Biener <rguenther@suse.de>
15164 PR tree-optimization/78847
15165 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15167 2017-04-21 Richard Biener <rguenther@suse.de>
15169 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15170 (build_distinct_type_copy): Likewise.
15171 (build_variant_type_copy): Likewise.
15172 * tree.c (build_qualified_type): Pass down mem-stat info.
15173 (build_distinct_type_copy): Likewise.
15174 (build_variant_type_copy): Likewise.
15176 2017-04-21 Richard Biener <rguenther@suse.de>
15178 PR tree-optimization/80237
15179 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15181 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15182 for a simplified result.
15184 2016-04-21 Richard Biener <rguenther@suse.de>
15186 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15187 sth as strict as a simple_iv but a chrec without symbols and an
15188 operand defined in the loop we are peeling (and not some subloop).
15189 (propagate_constants_for_unrolling): Propagate all constants.
15191 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15194 * config/i386/i386.c (print_reg): Remove assert for disalowed
15195 regno values, call output_operand_lossage instead.
15197 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15200 * config/i386/constraints.md (Yc): New register constraint.
15201 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15202 Use Yc constraint for alternative 2 of operand 0. Remove
15203 preferred_for_speed attribute.
15205 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15207 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15208 lastprivate clauses in SIMT case.
15210 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15212 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15214 2017-04-20 Richard Biener <rguenther@suse.de>
15216 PR tree-optimization/57796
15217 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15218 as N scalar stores.
15219 (vect_model_load_cost): Cost gathers as N scalar loads.
15221 2017-04-20 Richard Biener <rguenther@suse.de>
15223 * ggc-page.c (ggc_allocated_p): Rename to ...
15224 (safe_lookup_page_table_entry): ... this and return the lookup
15226 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15228 2017-04-20 Richard Biener <rguenther@suse.de>
15230 PR tree-optimization/80453
15231 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15232 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15233 from the conditions.
15234 (vn_phi_eq): Pass them down.
15235 (vn_phi_lookup): Record them.
15236 (vn_phi_insert): Likewise.
15238 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15240 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15241 uninitialized variable warning to avoid buffer overrun.
15243 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15246 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15247 is suppressed for '{ 0 }' in C.
15249 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15251 * BASE-VER: Set to 8.0.0.
15253 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15255 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15256 priority .init_array and .fini_array section with SECTION_NOTYPE
15259 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15261 PR middle-end/80423
15262 * tree.h (build_array_type): Add typeless_storage default argument.
15263 * tree.c (type_cache_hasher::equal): Also compare
15264 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15265 (build_array_type): Add typeless_storage argument, set
15266 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15268 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15269 (build_array_type): Likewise. Add typeless_storage argument.
15271 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15272 Jakub Jelinek <jakub@redhat.com>
15274 PR tree-optimization/80426
15275 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15276 operation on symbolic operands, also compute the overflow for the
15277 invariant part when the operation degenerates into a negation.
15279 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15282 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15283 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15286 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15288 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15291 * config/avr/avr.c (tree.h): Include it.
15292 (cgraph.h): Include it.
15293 (avr_encode_section_info): Don't warn for uninitialized progmem
15294 variable if it's just an alias.
15296 2017-04-19 Richard Biener <rguenther@suse.de>
15299 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15300 when needed by AutoPGO.
15302 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15305 * doc/lto.texi: Remove an extra 'that'.
15307 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15309 PR rtl-optimization/80429
15310 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15311 are only used in debug insns.
15313 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15314 Vladimir Makarov <vmakarov@redhat.com>
15316 * config/sparc/predicates.md (input_operand): Add comment. Return
15317 true for any memory operand when LRA is in progress.
15318 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15320 2017-04-18 Jeff Law <law@redhat.com>
15323 * mips.md ({return,simple_return}_internal): Do not overwrite
15326 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15328 PR tree-optimization/80443
15329 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15330 instead of adding 1, subtract -1 and similarly instead of subtracting
15333 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15335 PR rtl-optimization/80357
15336 * haifa-sched.c (tmp_bitmap): New variable.
15337 (model_recompute): Handle duplicate use records.
15338 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15339 (free_global_sched_pressure_data): Free it.
15341 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15344 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15345 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15346 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15347 instead of SYSTEM_HEADER_DIR.
15349 2017-04-18 Jeff Law <law@redhat.com>
15351 PR middle-end/80422
15352 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15353 predecessors after walking up the insn chain.
15355 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15358 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15359 sizetype type into debug info.
15361 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15364 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15365 unneeded test for TARGET_UPPER_REGS_SF.
15366 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15368 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15371 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15372 instead of gsi_after_labels.
15374 2017-04-18 Jeff Law <law@redhat.com>
15376 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15380 2017-04-13 Jeff Law <law@redhat.com>
15381 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15382 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15384 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15387 * config/avr/avr.c (intl.h): Include it.
15388 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15390 2017-04-18 Martin Liska <mliska@suse.cz>
15392 PR gcov-profile/78783
15393 * gcov-tool.c (gcov_output_files): Validate that destination
15394 file is either removed by the tool or by a user.
15396 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15397 Guy Benyei <guybe@mellanox.com>
15399 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15400 block, and do not negate it, the stored id is already negative.
15402 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15404 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15406 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15409 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15410 masks of options that should be turned off if the VSX vector
15411 options are turned off.
15412 (OTHER_P8_VECTOR_MASKS): Likewise.
15413 (OTHER_VSX_VECTOR_MASKS): Likewise.
15414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15415 rs6000_disable_incompatible_switches to validate no type switches
15417 (rs6000_incompatible_switch): New function to disallow turning on
15418 other vector options if -mno-vsx, -mno-power8-vector, or
15419 -mno-power9-vector are specified.
15421 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15423 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15425 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15427 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15428 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15429 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15430 (ARG_POINTER_CFA_OFFSET): Likewise.
15432 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15434 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15435 conditions to take advantage of various optimizations.
15437 2017-04-13 Jeff Law <law@redhat.com>
15439 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15440 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15441 (zero_extendsidi2_dext): Likewise.
15443 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15446 * fold-const.c (fold_ternary_loc): Revert
15447 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15450 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15452 PR rtl-optimization/80343
15453 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15454 new scratch pseudo.
15456 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15459 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15460 to ubsan_encode_value.
15462 2017-04-13 Jeff Law <law@redhat.com>
15464 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15465 appearing in DEBUG_INSNs.
15467 2017-04-13 Martin Liska <mliska@suse.cz>
15469 PR gcov-profile/80413
15470 * gcov-io.c (gcov_write_string): Copy to buffer just when
15471 allocated size is greater than zero.
15473 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15476 * dwarf2out.c (decls_for_scope): Ignore declarations of
15477 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15479 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15482 * ipa-visibility.c (non_local_p): Fix typos.
15483 (localize_node): When localizing symbol in same comdat group,
15484 dissolve the group only when we know external symbols are going
15486 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15488 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15490 PR tree-optimization/79390
15491 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15492 order does not result in usable sequence, retry with reversed operand
15498 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15499 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15500 op0 instead of fold_convert_loc (loc, type, arg0).
15502 2017-04-12 Jeff Law <law@redhat.com>
15504 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15505 has a delay slot in the generated code.
15507 * config/cris/cris.md (cris_preferred_reload_class): Return
15508 GENNONACR_REGS rather than GENERAL_REGS.
15510 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15513 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15514 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15515 signedness of the result type.
15517 2017-04-12 Richard Biener <rguenther@suse.de>
15518 Jeff Law <law@redhat.com>
15520 PR tree-optimization/80359
15521 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15522 trim stores to TARGET_MEM_REFs.
15524 2017-04-12 Richard Biener <rguenther@suse.de>
15526 PR tree-optimization/79390
15527 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15528 threading case even more.
15530 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15533 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15534 for quad_address_p for TImode, instead of just not indexed_address.
15536 2017-04-12 Richard Biener <rguenther@suse.de>
15537 Bernd Edlinger <bernd.edlinger@hotmail.de>
15539 PR middle-end/79671
15540 * alias.c (component_uses_parent_alias_set_from): Handle
15541 TYPE_TYPELESS_STORAGE.
15542 (get_alias_set): Likewise.
15543 * tree-core.h (tree_type_common): Add typeless_storage flag.
15544 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15545 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15546 for types containing members with TYPE_TYPELESS_STORAGE.
15547 (place_field): Likewise.
15548 (layout_type): Likewise for ARRAY_TYPE.
15549 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15550 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15551 TYPE_TYPELESS_STORAGE.
15552 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15554 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15557 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15558 first argument to type.
15560 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15564 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15565 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15566 (rs6000_expand_binop_builtin): Likewise.
15567 (rs6000_expand_ternop_builtin): Likewise; also add missing
15568 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15569 vshasigma built-ins.
15570 * doc/extend.texi: Document that vec_xxpermdi's third argument
15571 must be a constant.
15573 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15575 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15576 Use shift_const cost parameter when calculating gain of STV shifts.
15578 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15580 PR rtl-optimization/70478
15581 * lra-constraints.c (process_alt_operands): Check memory for
15582 disfavoring memory insn operand.
15584 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15586 PR middle-end/80100
15587 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15588 left shift in unsigned HOST_WIDE_INT type.
15590 PR rtl-optimization/80385
15591 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15592 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15595 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15596 if they have any depend clauses.
15598 2017-04-11 Martin Liska <mliska@suse.cz>
15601 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15602 * ipa-split.c (split_function): Create a local comdat symbol
15603 if caller is in a comdat group.
15605 2017-04-11 Martin Liska <mliska@suse.cz>
15608 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15611 2017-04-11 Martin Sebor <msebor@redhat.com>
15613 PR middle-end/80364
15614 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15615 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15617 (directive::set_width, directive::set_precision, format_character):
15619 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15622 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15625 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15626 conflict, set target->arch_name instead of target->cpu_name.
15628 2017-04-11 Richard Biener <rguenther@suse.de>
15630 PR tree-optimization/80374
15631 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15632 build_zero_cst, remove fold_convertible_p check again.
15634 2017-04-11 Martin Liska <mliska@suse.cz>
15637 * ubsan.c (instrument_object_size): Do not instrument register
15640 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15643 * config/i386/i386-builtin-types.def
15644 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15645 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15646 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15647 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15648 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15649 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15650 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15651 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15652 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15653 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15654 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15655 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15656 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15657 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15658 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15659 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15660 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15661 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15662 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15663 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15664 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15665 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15666 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15667 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15668 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15669 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15670 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15671 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15672 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15673 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15674 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15675 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15676 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15677 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15678 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15679 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15680 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15681 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15682 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15683 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15684 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15685 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15686 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15687 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15688 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15689 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15691 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15692 flag to second_arg_count, handle 4 argument function type _COUNT
15693 aliases, handle second_arg_count on second argument rather than last.
15695 2017-04-10 Jeff Law <law@redhat.com>
15697 PR tree-optimization/80374
15698 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15699 record anything if we can not convert integer_zero_node to the
15702 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15705 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15706 Enhance special handling given to the TARGET_P9_MINMAX option in
15707 relation to certain other options.
15709 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15711 PR tree-optimization/80153
15712 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15713 remove POINTER_PLUS_EXPR's base part directly, rather than through
15716 2017-04-10 Richard Biener <rguenther@suse.de>
15717 Bin Cheng <bin.cheng@arm.com>
15719 PR tree-optimization/80153
15720 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15721 the first element of pointer type aff_tree. Build result expr in
15723 (add_elt_to_tree): Convert to type unconditionally. Remove other
15724 fold_convert calls.
15725 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15726 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15728 2017-04-10 Richard Biener <rguenther@suse.de>
15730 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15733 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15735 PR rtl-optimization/70478
15736 * lra-constraints.c (curr_small_class_check): New.
15737 (update_and_check_small_class_inputs): New.
15738 (process_alt_operands): Update curr_small_class_check. Disfavor
15739 alternative insn memory operands. Check available regs for small
15742 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15745 * config/mips/mips.opt (-mvirt): Update description.
15746 * doc/invoke.texi (-mvirt): Likewise.
15748 2017-04-10 Richard Biener <rguenther@suse.de>
15750 PR middle-end/80362
15751 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15752 looking for NEGATE_EXPR in -A / -B to A / B folding.
15754 2017-04-10 Martin Liska <mliska@suse.cz>
15756 PR gcov-profile/80224
15757 * gcov.c (print_usage): Fix usage string.
15758 (get_gcov_intermediate_filename): Remove.
15759 (output_gcov_file): Use both for normal and intermediate format.
15760 (generate_results): Do not initialize special file for
15761 intermediate format.
15763 2017-04-10 Richard Biener <rguenther@suse.de>
15765 PR tree-optimization/80304
15766 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15769 2017-04-10 Nathan Sidwell <nathan@acm.org>
15772 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15773 (rs6000_init_builtins): Use it.
15775 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15777 * config/arm/arm.md (<mrc>): Add mode to SET source.
15778 (<mrrc>): Likewise.
15780 2017-04-10 Richard Biener <rguenther@suse.de>
15782 PR middle-end/80344
15783 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
15785 2017-04-10 Jakub Jelinek <jakub@redhat.com>
15788 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
15789 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
15790 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
15791 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
15792 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
15793 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
15794 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
15795 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
15796 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
15797 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
15798 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
15799 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
15800 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
15801 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
15802 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
15803 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
15804 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
15805 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
15806 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
15807 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
15808 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
15809 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
15810 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
15812 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
15814 PR rtl-optimization/70478
15815 * lra-constraints.c: Reverse the last patch.
15817 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
15819 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
15820 Add comment for WCHAR_T.
15822 2017-04-08 Martin Liska <mliska@suse.cz>
15825 2017-04-07 Martin Liska <mliska@suse.cz>
15828 * ipa-split.c (split_function): Add function part to a same comdat
15831 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15834 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
15836 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
15838 * rs6000/rs6000.c (vec_load_pendulum): Rename...
15839 (vec_pairing): ...to this.
15840 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
15841 (rs6000_sched_init): Adjust for name change.
15842 (struct rs6000_sched_context): Likewise.
15843 (rs6000_init_sched_context): Likewise.
15844 (rs6000_set_sched_context): Likewise.
15846 2017-04-07 Jakub Jelinek <jakub@redhat.com>
15852 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
15854 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
15855 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
15856 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
15858 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
15860 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
15862 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15864 PR rtl-optimization/70703
15865 * ira-color.c (update_conflict_hard_regno_costs): Use
15866 int64_t instead of HOST_WIDE_INT.
15868 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15870 PR rtl-optimization/70478
15871 * lra-constraints.c (process_alt_operands): Disfavor alternative
15872 insn memory operands.
15874 2017-04-07 Jeff Law <law@redhat.com>
15876 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
15877 CALL and NOTE_INSN_CALL_ARG_LOCATION.
15879 2017-04-07 Martin Liska <mliska@suse.cz>
15882 * config/aarch64/aarch64.c (aarch64_process_target_attr):
15883 Show error message instead of an ICE.
15885 2017-04-07 Martin Liska <mliska@suse.cz>
15888 * ipa-split.c (split_function): Add function part to a same comdat
15891 2017-04-07 Richard Biener <rguenther@suse.de>
15893 PR middle-end/80341
15894 * tree.c (get_unwidened): Also handle ! for_type case for
15896 * convert.c (do_narrow): Split out from ...
15897 (convert_to_integer_1): ... here. Do not pass final truncation
15898 type to get_unwidened for TRUNC_DIV_EXPR.
15900 2017-04-07 Richard Biener <rguenther@suse.de>
15902 * tree-affine.c (wide_int_ext_for_comb): Take type rather
15904 (aff_combination_const): Adjust.
15905 (aff_combination_scale): Likewise.
15906 (aff_combination_add_elt): Likewise.
15907 (aff_combination_add_cst): Likewise.
15908 (aff_combination_convert): Likewise.
15909 (add_elt_to_tree): Likewise. Remove unused argument.
15910 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
15912 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
15914 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
15916 * config/arm/arm.c (arm_default_short_enums): Use
15917 ARM_DEFAULT_SHORT_ENUMS.
15918 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
15920 2017-04-06 Jakub Jelinek <jakub@redhat.com>
15923 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
15924 members with redundant out-of-class redeclaration.
15926 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15929 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
15930 * config/i386/i386.md (*zero_extendsidi2):
15931 Add (?*x,*x) and (?*v,*v) alternatives.
15933 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15936 * config/i386/i386.c (ix86_expand_builtin)
15937 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
15938 mode from insn data. Convert operands to insn operand mode.
15939 Copy operands that don't satisfy insn predicate to a register.
15941 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
15943 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
15946 2017-04-06 Richard Biener <rguenther@suse.de>
15948 PR tree-optimization/80334
15949 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
15950 preserve alignment of accesses.
15952 2017-04-06 Richard Biener <rguenther@suse.de>
15954 PR tree-optimization/80262
15955 * tree-sra.c (build_ref_for_offset): Preserve address-space
15957 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
15958 Drop useless address-space information on MEM_REF offsets.
15960 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
15962 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
15964 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15966 PR rtl-optimization/70703
15967 * ira-color.c (update_conflict_hard_regno_costs): Use
15968 HOST_WIDE_INT instead of long.
15970 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
15973 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
15974 not defined for x86_64 target. Add -mmmx target option when __SSE2__
15976 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
15977 for x86_64 target. Handle -m3dnowa option.
15979 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15981 PR rtl-optimization/70703
15982 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
15983 (update_conflict_hard_regno_costs): Use long instead of unsigned
15984 arithmetic for cost calculation.
15986 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15987 Bernd Edlinger <bernd.edlinger@hotmail.de>
15990 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
15993 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
15996 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
15997 ptr_mode with Pmode throughout.
15998 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
15999 into probe_stack_range and use DImode.
16001 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16004 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16005 call_eh_return is true.
16007 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16009 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16010 Initialize last_match_fntype_index.
16012 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16015 * tree-nvr.c: Include internal-fn.h.
16016 (pass_return_slot::execute): Ignore internal calls without
16019 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16020 Richard Biener <rguenther@suse.de>
16023 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16024 captures used multiple times, except for the last use.
16025 * generic-match-head.c: Include gimplify.h.
16027 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16029 PR tree-optimization/79390
16030 * target.h (struct noce_if_info): Declare.
16031 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16032 * target.def (noce_conversion_profitable_p): New target hook.
16033 * ifcvt.h (struct noce_if_info): New type, moved from ...
16034 * ifcvt.c (struct noce_if_info): ... here.
16035 (noce_conversion_profitable_p): Renamed to ...
16036 (default_noce_conversion_profitable_p): ... this. No longer
16038 (noce_try_store_flag_constants, noce_try_addcc,
16039 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16040 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16041 instead of noce_conversion_profitable_p.
16042 * config/i386/i386.c: Include ifcvt.h.
16043 (ix86_option_override_internal): Don't override
16044 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16045 (ix86_noce_conversion_profitable_p): New function.
16046 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16047 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16048 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16049 * doc/tm.texi: Regenerated.
16051 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16053 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16056 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16059 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16060 instructions for small multiply cores.
16062 2017-04-04 Jeff Law <law@redhat.com>
16064 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16066 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16067 that are not set by the loop over the elements.
16069 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16072 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16073 int mode, convert_modes it to mode as unsigned, otherwise use
16074 lowpart_subreg to mode rather than SImode.
16075 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16076 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16077 Use DImode instead of SImode for the shift count operand.
16078 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16081 2017-04-04 Richard Biener <rguenther@suse.de>
16083 PR middle-end/80281
16084 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16085 arithmetic done for the negate or the plus. Simplify.
16086 (A - (-B) -> A + B): Likewise.
16087 * fold-const.c (split_tree): Make sure to not negate pointers.
16089 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16091 PR rtl-optimization/60818
16092 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16093 a compare of comparisons with the thing compared if this results
16094 in a different machine mode.
16096 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16098 * alias.c (base_alias_check): Fix typo in comment.
16099 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16100 * cgraphunit.c (symbol_table::compile): Likewise.
16101 * collect2.c (maybe_run_lto_and_relink): Likewise.
16102 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16103 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16104 * config/avr/avr.c (avr_map_op_t): Likewise.
16105 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16106 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16107 * config/epiphany/epiphany.md (movcc): Likewise.
16108 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16109 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16111 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16112 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16113 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16114 * config/sh/sh.c (sh_rtx_costs): Likewise.
16115 * fold-const.c (fold_truth_andor): Likewise.
16116 * genautomata.c (collapse_flag): Likewise.
16117 * gengtype.h (struct type::u::s): Likewise.
16118 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16119 * input.c (FORMAT_AMOUNT): Likewise.
16120 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16121 (known_aggs_to_agg_replacement_list): Likewise.
16122 * ipa-inline-analysis.c: Likewise.
16123 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16124 * ipa-polymorphic-call.c
16125 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16126 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16127 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16129 * modulo-sched.c (apply_reg_moves): Likewise.
16130 * omp-expand.c (build_omp_regions_1): Likewise.
16131 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16132 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16133 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16134 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16135 * value-prof.c: Likewise.
16136 * var-tracking.c (val_reset): Likewise.
16138 2017-04-03 Richard Biener <rguenther@suse.de>
16140 PR tree-optimization/80275
16141 * fold-const.c (split_address_to_core_and_offset): Handle
16144 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16146 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16147 descriptors is at least equal to that of functions.
16149 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16151 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16153 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16156 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16157 (mov<IMOD4:mode>): New expander.
16158 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16160 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16162 PR rtl-optimization/79405
16163 * fwprop.c (propagations_left): New variable.
16164 (forward_propagate_into): Decrement it.
16165 (fwprop_init): Initialize it.
16166 (fw_prop): If the variable has reached zero, stop propagating.
16167 (fwprop_addr): Ditto.
16169 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16172 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16173 a FUNCTION_DECL, pass it as decl instead of origin to
16176 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16178 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16181 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16184 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16185 TARGET_VSX_SMALL_INTEGER.
16187 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16189 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16190 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16192 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16194 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16195 extraction from odd-numbered MSA register.
16197 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16199 PR middle-end/80173
16200 * expmed.c (store_bit_field_1): Don't attempt to create
16201 a word subreg out of hard registers wider than word if they
16202 have HARD_REGNO_NREGS of 1 for their mode.
16204 PR middle-end/80163
16205 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16206 conversions to integer types wider than word and pointer.
16209 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16210 (rtx_equal_for_cselib_p): Pass 0 to it.
16211 * cselib.c (cselib_hasher::equal): Likewise.
16212 (rtx_equal_for_cselib_1): Add depth argument. If depth
16213 is 128, don't look up VALUE locs and punt. Increment
16214 depth in recursive calls when walking VALUE locs.
16216 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16218 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16219 (make_gcov_file_name): Use the canonical path name for generating
16221 (read_line): Fix handling of files with ascii null bytes.
16223 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16225 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16226 to initialise a vector register instead
16227 of using a const_int.
16229 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16231 PR translation/80189
16232 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16233 diagnostic messages.
16235 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16238 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16239 (dfp_diex_<mode>): Update mode of operand 1.
16240 * doc/extend.texi (dxex, dxexq): Document change to return type.
16241 (diex, diexq): Document change to argument type.
16243 2017-03-30 Martin Jambor <mjambor@suse.cz>
16246 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16247 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16248 it reflects the signature changes performed at the callee side.
16249 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16250 to cgraph_build_function_type_skip_args.
16251 (build_function_decl_skip_args): Adjust call to the above function.
16253 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16256 * config/i386/sse.md
16257 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16258 register as dest whenever it is a MEM not rtx_equal_p to the
16259 corresponding dup operand, and when forcing into reg move the
16260 reg into the memory afterwards.
16261 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16262 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16263 for the force_reg mode.
16264 (avx512vl_vextractf128<mode>): Use register as dest either
16265 always when a MEM, or when it is a MEM not rtx_equal_p to the
16266 corresponding dup operand, or even not when it is a CONST_VECTOR
16267 depending on the mode and lo vs. hi.
16268 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16270 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16271 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16272 Likewise. Require that operands[2] is even.
16273 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16274 Remove extraneous parens. Require that operands[2] is a multiple
16276 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16277 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16278 disallow memory then.
16280 2017-03-30 Richard Biener <rguenther@suse.de>
16282 PR tree-optimization/77498
16283 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16284 to non-constants over backedges.
16286 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16288 PR rtl-optimization/80233
16289 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16290 as last_combined_insn. Do not test for BARRIER_P separately.
16292 2017-03-29 Andreas Schwab <schwab@suse.de>
16295 * calls.c (prepare_call_address): Convert funexp to Pmode before
16296 copying to temp reg.
16298 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16300 PR tree-optimization/80158
16301 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16302 Handle possible future case of more than one alternate
16304 (replace_rhs_if_not_dup): Likewise.
16305 (replace_one_candidate): Likewise.
16307 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16309 PR rtl-optimization/80193
16310 * ira.c (ira): Do not check allocation for LRA.
16312 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16314 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16315 (nvptx_output_simt_exit): Declare.
16316 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16317 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16318 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16319 (nvptx_declare_function_name): ...here. Emit declaration of local
16320 memory space buffer for omp_simt_enter insn.
16321 (nvptx_output_unisimt_switch): New.
16322 (nvptx_output_softstack_switch): New.
16323 (nvptx_output_simt_enter): New.
16324 (nvptx_output_simt_exit): New.
16325 * config/nvptx/nvptx.h (struct machine_function): New fields
16326 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16327 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16328 (UNSPECV_SIMT_EXIT): Ditto.
16329 (omp_simt_enter_insn): New insn.
16330 (omp_simt_enter): New expansion.
16331 (omp_simt_exit): New insn.
16332 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16334 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16335 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16336 (expand_GOMP_SIMT_EXIT): New.
16337 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16338 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16339 (GOMP_SIMT_EXIT): Ditto.
16340 * target-insns.def (omp_simt_enter): New insn.
16341 (omp_simt_exit): Ditto.
16342 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16344 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16345 (lower_rec_input_clauses): Likewise.
16346 (lower_lastprivate_clauses): Handle SIMT privatization.
16348 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16349 (ompdevlow_adjust_simt_enter): New.
16350 (find_simtpriv_var_op): New.
16351 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16352 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16354 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16355 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16356 (copy_decl_for_dup_finish): Ditto.
16358 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16360 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16363 * config/i386/i386.c (ix86_option_override_internal): Always
16364 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16366 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16368 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16370 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16372 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16373 mark new edge's irreducible flag accordign to it.
16374 (vect_do_peeling): Check loop preheader edge's irreducible flag
16375 and pass it to function slpeel_add_loop_guard.
16377 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16379 PR tree-optimization/80218
16380 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16381 Update block frequencies and counts.
16383 2017-03-28 Richard Biener <rguenther@suse.de>
16385 PR tree-optimization/78644
16386 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16387 of a simplification result we may not use it at all.
16389 2017-03-28 Richard Biener <rguenther@suse.de>
16392 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16393 without arguments, generate default definition of a SSA name.
16395 2017-03-28 Richard Biener <rguenther@suse.de>
16397 PR middle-end/80222
16398 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16399 TYPE_REF_CAN_ALIAS_ALL references.
16400 * fold-const.c (fold_indirect_ref_1): Likewise.
16402 2017-03-28 Martin Liska <mliska@suse.cz>
16405 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16406 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16408 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16409 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16411 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16412 (EXTRA_SPECS): Define.
16413 (SUBTARGET_EXTRA_SPECS): Likewise.
16414 (SUBTARGET_CPP_SPEC): Likewise.
16415 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16416 SUBTARGET_EXTRA_SPECS.
16417 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16419 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16421 * config/arc/simdext.md (vst64_insn): Update pattern.
16422 (vld32wh_insn): Likewise.
16423 (vld32wl_insn): Likewise.
16424 (vld64_insn): Likewise.
16425 (vld32_insn): Likewise.
16427 2017-03-28 Marek Polacek <polacek@redhat.com>
16430 * fold-const.c (fold_comparison): Use protected_set_expr_location
16431 instead of SET_EXPR_LOCATION.
16433 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16435 * tree.c (add_expr): Avoid name lookup warning.
16437 2017-03-27 Jeff Law <law@redhat.com>
16439 PR tree-optimization/80216
16440 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16441 function name. Limit recursion depth.
16442 (record_temporary_equivalences): Corresponding changes.
16444 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16446 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16449 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16452 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16454 * cfgcleanup.c (reg_note_cfa_p): New array.
16455 (insns_have_identical_cfa_notes): New function.
16456 (old_insns_match_p): Don't cross-jump in between /f
16457 and non-/f instructions. If both i1 and i2 are frame related,
16458 verify all CFA notes, their order and content.
16460 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16463 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16464 HImode and SImode with zero extend to DImode to one insn.
16465 (bswap<mode>2_extenddi): Likewise.
16466 (bswapsi2_extenddi): Likewise.
16467 (bswaphi2_extendsi): Likewise.
16468 (bswaphi2): Combine bswap HImode and SImode into one insn.
16469 Separate memory insns from swapping register.
16470 (bswapsi2): Likewise.
16471 (bswap<mode>2): Likewise.
16472 (bswaphi2_internal): Delete, no longer used.
16473 (bswapsi2_internal): Likewise.
16474 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16475 store, and gpr<-gpr swap insns.
16476 (bswap<mode>2_store): Likewise.
16477 (bswaphi2_reg): Register only splitter, combine with the splitter.
16478 (bswaphi2 splitter): Likewise.
16479 (bswapsi2_reg): Likewise.
16480 (bswapsi2 splitter): Likewise.
16481 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16482 the insns into load, store, and register/register insns.
16483 (bswapdi2_ldbrx): Likewise.
16484 (bswapdi2_load): Likewise.
16485 (bswapdi2_store): Likewise.
16486 (bswapdi2_reg): Likewise.
16488 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16490 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16491 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16493 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16496 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16498 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16499 special handling for target option conflicts between dform
16500 options (-mpower9-dform, -mpower9-dform-vector,
16501 -mpower9-dform-scalar) and -mno-direct-move.
16503 2017-03-27 Richard Biener <rguenther@suse.de>
16505 PR tree-optimization/80181
16506 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16508 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16510 * config/arc/predicates.md (move_double_src_operand): Replace the
16511 call to move_double_src_operand with a call to address_operand.
16513 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16515 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16516 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16517 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16519 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16521 * config/arc/predicates.md (long_immediate_loadstore_operand):
16522 Consider scaled addresses cases.
16524 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16526 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16527 restored when in interrupt.
16528 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16529 doesn't have delay slot.
16531 2017-03-27 Richard Biener <rguenther@suse.de>
16534 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16535 inlined thunk clones.
16537 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16540 * asan.c (instrument_derefs): Copy over last operand from
16541 original COMPONENT_REF to the new COMPONENT_REF with
16542 DECL_BIT_FIELD_REPRESENTATIVE.
16543 * ubsan.c (instrument_object_size): Likewise.
16545 2017-03-27 Richard Biener <rguenther@suse.de>
16547 PR tree-optimization/80170
16548 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16549 sure DR/SCEV didnt fold in constants we do not see when looking
16550 at the reference base alignment.
16552 2017-03-27 Richard Biener <rguenther@suse.de>
16554 PR middle-end/80171
16555 * gimple-fold.c (fold_ctor_reference): Properly guard against
16556 NULL return value from canonicalize_constructor_val.
16558 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16561 * config/i386/i386.c (ix86_expand_builtin)
16562 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16563 flags reg setting and flags reg using instructions.
16564 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16565 clobbering instructions to zero extend op2.
16567 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16569 * doc/install.texi (Configuration) <--with-aix-soname>:
16570 Update link to AIX ld.
16572 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16574 PR rtl-optimization/80160
16575 PR rtl-optimization/80159
16576 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16577 reg_alternate_class into account.
16579 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16582 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16583 to consider in curr_insn_transform.
16585 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16587 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16588 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16589 and emit_mode_inner.
16591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16593 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16594 argument to the overloaded builtin variants. Use the new flag to
16595 deprecate certain builtin variants.
16596 * config/s390/s390-builtin-types.def: Add new builtin types.
16597 * config/s390/s390-builtins.h: Support new flags field for
16598 overloaded builtins.
16599 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16600 (s390_macro_to_expand): Enable vector float data type.
16601 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16602 builtins by incrementing the __VEC__ version number.
16603 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16605 (s390_resolve_overloaded_builtin): Emit error messages depending
16606 on the builtin flags.
16607 * config/s390/s390.c (s390_expand_builtin): Support additional
16608 flags argument. Change error message to match the messages
16609 emitted in s390-c.c.
16610 * config/s390/s390.md: New UNSPEC_* constants.
16611 (op_type): Add new instruction types.
16612 * config/s390/vecintrin.h: Add new builtins and test data class
16614 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16615 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16616 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16617 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16618 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16619 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16621 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16622 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16623 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16624 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16626 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16627 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16628 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16629 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16631 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16632 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16633 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16634 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16635 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16636 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16637 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16639 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16640 ("vec_scatter_element<V_HW_4:mode>_DI")
16641 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16642 ("vec_fpint<mode>", "vflls")
16643 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16644 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16645 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16646 ("*vec_cmphe<mode>_cc"): ... these.
16648 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16649 mode constant instead of magic value.
16651 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16653 * config/s390/s390.c (s390_expand_vec_compare): Support other
16654 vector floating point modes than just V2DF.
16655 (s390_expand_vcond): Likewise.
16656 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16657 (s390_cannot_change_mode_class): Prevent mode changes between TF
16658 and V1TF in vector registers.
16659 * config/s390/s390.md (DF, SF): New mode attributes.
16660 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16661 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16662 SFmode support for VRs.
16663 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16665 (VFT, VF_HW): New mode iterators.
16666 (vw, sdx): New mode attributes.
16667 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16668 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16669 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16670 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16671 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16672 also the new vector floating point modes. Renaming to ...
16674 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16675 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16676 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16677 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16678 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16679 ("vec_unordered<mode>"): ... these.
16681 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16682 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16683 ("*vec_extendv2df"): New insn definitions.
16685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16687 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16688 ("mulditi3_2", "*muldi3_sign"): New patterns.
16689 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16690 rename the pattern definition.
16692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16694 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16696 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16700 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16701 instruction if possible.
16702 * config/s390/vector.md (vec_halfnumelts): New mode
16704 ("*vec_vllezlf<mode>"): New pattern.
16706 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16708 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16709 ("popcountv4si2", "popcountv2di2"): Rename to ...
16710 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16711 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16713 ("popcount<mode>2_vxe"): New pattern.
16715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16717 * common/config/s390/s390-common.c (processor_flags_table): Add
16719 * config.gcc: Add arch12.
16720 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16721 Default to arch12 for unknown CPU model numbers.
16722 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16723 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16724 PROCESSOR_max sanity check.
16725 * config/s390/s390-opts.h (enum processor_type): Add
16727 * config/s390/s390.c (processor_table): Add arch12.
16728 (s390_expand_builtin): Add check for B_VXE flag.
16729 (s390_issue_rate): Add PROCESSOR_ARCH12.
16730 (s390_get_sched_attrmask): Likewise.
16731 (s390_get_unit_mask): Likewise.
16732 (s390_sched_score): Enable z13 scheduling for arch12.
16733 (s390_sched_reorder): Likewise.
16734 (s390_sched_variable_issue): Likewise.
16735 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16737 (s390_tune_attr): Use z13 scheduling also for arch12.
16738 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16739 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16740 (TARGET_VXE_P): New macros.
16741 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16742 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16743 * config/s390/s390.opt: Add arch12 as processor_type.
16745 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16747 * config/s390/s390.md
16748 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16749 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16750 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16752 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16753 Rename expanders to ...
16755 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16756 ("fixuns_truncdddi2_emu"): ... these.
16758 ("fixuns_trunc<mode>si2_emu"): New expander.
16760 ("*fixuns_truncdfdi2_z13"): Rename to ...
16761 ("*fixuns_truncdfdi2_vx"): ... this.
16763 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16765 * config/s390/2964.md: Remove the single element vector compare
16766 instructions which are no longer used.
16767 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16769 (s390_canonicalize_comparison): Remove handling of DFmode
16771 (s390_expand_vec_compare_scalar): Remove function.
16772 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16773 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16775 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16777 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16779 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16781 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
16782 will anyway by matched by mov<mode>_64dfp.
16784 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16786 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
16787 vlef/vstef. Add missing operand to vleif.
16789 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16791 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
16792 pair for all vector types with 64 bit elements.
16793 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
16794 * config/s390/vector.md (V_HW_64): ... here.
16795 (V_128_NOSINGLE): New mode iterator.
16796 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
16797 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
16798 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
16799 ("*vec_load_pairv2di"): Change to ...
16800 ("*vec_load_pair<mode>"): ... this one.
16802 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16804 * config/s390/constraints.md: Add comments.
16805 (jKK): Reject element sizes > 8 bytes.
16806 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
16808 * config/s390/s390.md: Add the s_operand checks formerly in
16809 s390_split_ok_p to various splitters where they are still
16811 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
16812 for 128 bit vectors. Plus two splitters.
16814 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16816 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
16819 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16822 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
16823 error if the boundary argument is not constant.
16825 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16827 PR rtl-optimization/80112
16828 * loop-doloop.c (doloop_condition_get): Don't check condition
16829 if cmp isn't SET with IF_THEN_ELSE src.
16831 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16833 PR tree-optimization/80158
16834 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
16835 replacing a candidate statement, also replace it for the
16836 candidate's alternate interpretation.
16837 (replace_rhs_if_not_dup): Likewise.
16838 (replace_one_candidate): Likewise.
16840 2017-03-24 Richard Biener <rguenther@suse.de>
16842 PR tree-optimization/80167
16843 * graphite-isl-ast-to-gimple.c
16844 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
16846 (translate_isl_ast_to_gimple::get_rename): Likewise.
16848 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16850 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
16851 handling of certain combinations of target options, including the
16852 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
16853 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
16855 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16858 * config/arm/arm.md (*load_multiple): Add reload_completed to
16859 matching condition.
16861 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16862 Richard Biener <rguenth@suse.de>
16864 PR tree-optimization/79908
16865 PR tree-optimization/80136
16866 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16867 been cast away, gimplify_and_add suffices.
16869 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
16871 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
16873 2017-03-23 Richard Biener <rguenther@suse.de>
16875 PR tree-optimization/80032
16876 * gimplify.c (gimple_push_cleanup): Forced unconditional
16877 cleanups still have to go to the conditional_cleanups
16880 2017-03-22 Jakub Jelinek <jakub@redhat.com>
16882 PR tree-optimization/80072
16883 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
16885 (next_operand_entry_id): Change type to unsigned int.
16886 (sort_by_operand_rank): Make sure to return the right return value
16887 even if unsigned fields are bigger than INT_MAX.
16888 (struct oecount): Change cnt and id type to unsigned int.
16889 (oecount_hasher::equal): Formatting fix.
16890 (oecount_cmp): Make sure to return the right return value
16891 even if unsigned fields are bigger than INT_MAX.
16892 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
16895 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
16896 TREE_READONLY on result if writing it more than once.
16899 * doc/invoke.texi (-fsanitize=thread): Document that with
16900 -fnon-call-exceptions atomics are not able to throw
16904 * tsan.c: Include tree-eh.h.
16905 (instrument_builtin_call): Call maybe_clean_eh_stmt or
16906 maybe_clean_or_replace_eh_stmt where needed.
16907 (instrument_memory_accesses): Add cfg_changed argument.
16908 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
16909 if it returned true.
16910 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
16912 PR rtl-optimization/63191
16913 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
16914 wrapper function, moved the whole old content into ...
16915 (ix86_delegitimize_address_1): ... this. New inline function.
16916 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
16917 true as last argument instead of ix86_delegitimize_address.
16919 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16921 * config/aarch64/aarch64.c (generic_branch_cost): Copy
16922 cortexa57_branch_cost.
16924 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16926 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
16928 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16931 * doc/md.texi (Constraints): Document wA constraint.
16932 * config/rs6000/constraints.md (wA): New.
16933 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
16934 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
16935 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
16936 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
16938 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
16941 * gimplify.c (is_oacc_declared): New function.
16942 (oacc_default_clause): Use it to set default flags for acc declared
16943 variables inside parallel regions.
16944 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
16945 declared variables.
16946 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
16947 declare attribute to any decl as necessary.
16949 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16952 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
16953 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
16954 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
16955 (arm_arch_lpae): This.
16956 * config/arm/arm.c (arm_arch7ve): Rename into ...
16957 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
16958 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
16961 2017-03-22 Martin Liska <mliska@suse.cz>
16964 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
16965 error message instead of an ICE.
16967 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16969 * doc/extend.texi (6.11 Additional Floating Types): Revise.
16971 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
16973 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
16975 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16978 2017-03-21 Martin Sebor <msebor@redhat.com>
16980 * doc/extend.texi: Use "cannot" instead of "can't."
16981 * doc/hostconfig.texi: Same.
16982 * doc/install.texi: Same.
16983 * doc/invoke.texi: Same.
16984 * doc/loop.texi: Same.
16985 * doc/md.texi: Same.
16986 * doc/objc.texi: Same.
16987 * doc/rtl.texi: Same.
16988 * doc/tm.texi: Same.
16989 * doc/tm.texi.in: Same.
16990 * doc/trouble.texi: Same.
16992 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
16995 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
16996 (collect_checksum_attributes): Set it.
16997 (die_checksum_ordered): Use it.
16999 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17001 PR tree-optimization/79908
17002 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17003 change: For a VA_ARG whose LHS has been cast away, use
17004 force_gimple_operand to construct the side effects.
17006 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17008 PR translation/80001
17009 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17010 more amenable to translation.
17011 (oacc_loop_auto_partitions): Likewise.
17013 2017-03-21 Marek Polacek <polacek@redhat.com>
17014 Martin Sebor <msebor@redhat.com>
17016 PR tree-optimization/80109
17017 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17018 on INTEGRAL_TYPE_P.
17020 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17021 Segher Boessenkool <segher@kernel.crashing.org>
17024 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17025 check reg_used_between_p between insn and one of succ or succ2
17026 depending on if succ is artificial insn not inserted into insn
17029 2017-03-21 Martin Liska <mliska@suse.cz>
17031 PR gcov-profile/80081
17032 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17033 * doc/gcc.texi: Include gcov-dump stuff.
17034 * doc/gcov-dump.texi: New file.
17036 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17038 PR rtl-optimization/79150
17039 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17040 conditional jump, if the jump is the last insn of the loop.
17042 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17043 Richard Biener <rguenth@suse.de>
17045 PR tree-optimization/79908
17046 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17047 been cast away, use force_gimple_operand to construct the side
17050 2017-03-21 Martin Liska <mliska@suse.cz>
17052 PR libfortran/79956
17053 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17056 2017-03-21 Brad Spengler <spender@grsecurity.net>
17059 * plugin.c (htab_hash_plugin): New function.
17060 (add_new_plugin): Use it and adjust.
17061 (parse_plugin_arg_opt): Adjust.
17062 (init_one_plugin): Likewise.
17064 2017-03-21 Richard Biener <rguenther@suse.de>
17066 PR tree-optimization/80032
17067 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17068 if set force the cleanup to happen unconditionally.
17069 (gimplify_target_expr): Push inserted clobbers with force_uncond
17070 to avoid them being removed by control-dependent DCE.
17072 2017-03-21 Richard Biener <rguenther@suse.de>
17074 PR tree-optimization/80122
17075 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17076 va_arg_pack_len when the inlined call stmt requires pack
17078 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17080 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17083 * tsan.c (instrument_builtin_call): If the memory model argument
17084 is not a constant, assume it is valid.
17087 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17090 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17092 PR rtl-optimization/79910
17093 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17094 if its dest is used by an insn before I2 (other than the combined
17095 insns themselves, which are properly handled already).
17097 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17100 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17102 * combine.c (record_used_regs): New static function.
17103 (try_combine): Handle situations where there is an additional
17104 instruction between I2 and I3 which needs to have a LOG_LINK
17108 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17110 * combine.c (try_combine): Delete redundant i1 test. Call
17111 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17113 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17116 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17117 alternatives 13/14.
17119 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17121 PR tree-optimization/80054
17122 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17123 the optimization if a PHI or any of its arguments is not dominated
17124 by the candidate's basis. Use gphi* rather than gimple* as
17126 (replace_profitable_candidates): Clean up a gimple* variable that
17127 should be a gphi* variable.
17129 2017-03-20 Martin Sebor <msebor@redhat.com>
17132 * doc/extend.texi (attribute constructor): Document present limitation.
17134 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17137 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17138 __POWER9_VECTOR__ #ifdef control, change template definition to
17139 use Power9-specific built-in function.
17140 (vec_any_eq): Likewise.
17141 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17142 to control outcomes from this test.
17143 (vector_ae_<mode>p): For VEC_F modes, likewise.
17145 2017-03-20 Ian Lance Taylor <iant@google.com>
17147 * config/i386/i386.c (ix86_function_regparm): Save an extra
17148 register for -fsplit-stack with DECL_STATIC_CHAIN.
17150 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17153 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17154 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17156 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17158 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17160 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17163 2017-03-20 Marek Polacek <polacek@redhat.com>
17166 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17168 2017-03-20 Richard Biener <rguenther@suse.de>
17170 PR tree-optimization/80113
17171 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17172 allocate extra SSA name for PHI def.
17173 (add_close_phis_to_outer_loops): Likewise.
17174 (add_close_phis_to_merge_points): Likewise.
17175 (copy_loop_close_phi_args): Likewise.
17176 (copy_cond_phi_nodes): Likewise.
17178 2017-03-20 Martin Liska <mliska@suse.cz>
17180 PR middle-end/79753
17181 * tree-chkp.c (chkp_build_returned_bound): Do not build
17182 returned bounds for a LHS that's not a BOUNDED_P type.
17184 2017-03-20 Martin Liska <mliska@suse.cz>
17188 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17189 COMPLEX_CST and VECTOR_CST.
17191 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17194 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17195 target operand. A new splitter adds the clobber statement in case
17196 the target operand is dead anyway.
17198 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17200 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17201 to age-old versions of binutils and glibc.
17203 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17205 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17207 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17209 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17211 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17213 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17214 requirement for binutils 2.13.
17216 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17218 * combine.c (try_combine): Delete redundant i1 test. Call
17219 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17221 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17223 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17224 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17226 <riscv64-*-elf>: Re-arrange section
17227 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17228 <riscv32-*-linux>: Likewise.
17229 <riscv64-*-elf>: Likewise
17230 <riscv64-*-linux>: Likewise.
17232 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17235 * aarch64.opt(verbose-cost-dump): Fix typo.
17237 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17240 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17241 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17243 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17245 * reload.c (find_reloads): When reloading a nonoffsettable address,
17246 use RELOAD_OTHER for it and its address reloads.
17248 PR rtl-optimization/79910
17249 * combine.c (record_used_regs): New static function.
17250 (try_combine): Handle situations where there is an additional
17251 instruction between I2 and I3 which needs to have a LOG_LINK
17254 2017-03-17 Jeff Law <law@redhat.com>
17256 PR tree-optimization/71437
17257 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17258 conditional in the hash table first.
17259 (vrp_dom_walker::before_dom_children): Extract condition from
17260 ASSERT_EXPR. Record condition, its inverion and any implied
17261 conditions as well.
17263 2017-03-17 Marek Polacek <polacek@redhat.com>
17264 Markus Trippelsdorf <markus@trippelsdorf.de>
17266 PR tree-optimization/80079
17267 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17270 2017-03-17 Richard Biener <rguenther@suse.de>
17272 PR middle-end/80075
17273 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17274 Properly verify the LHS before the RHS possibly claims to be
17276 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17279 2017-03-17 Martin Jambor <mjambor@suse.cz>
17281 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17282 (List of -O2 options): Likewise.
17283 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17286 2017-03-17 Tom de Vries <tom@codesourcery.com>
17288 * gcov-dump.c (print_usage): Print bug_report_url.
17290 2017-03-17 Richard Biener <rguenther@suse.de>
17292 PR middle-end/80050
17293 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17294 (parser::peek): Likewise.
17296 2017-03-17 Richard Biener <rguenther@suse.de>
17298 PR tree-optimization/80048
17299 * sese.c (free_sese_info): Properly release rename_map and
17300 copied_bb_map elements.
17302 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17304 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17305 Add linked-list forward and backlinks. Insert on
17306 construction, remove on destruction.
17307 (class pass_store_merging): Add m_stores_head field.
17308 (pass_store_merging::terminate_and_process_all_chains):
17309 Iterate over m_stores_head list.
17310 (pass_store_merging::terminate_all_aliasing_chains):
17312 (pass_store_merging::execute): Check for debug stmts first.
17313 Push new chains onto the m_stores_head stack.
17315 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17318 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17319 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17320 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17322 2017-03-16 Jeff Law <law@redhat.com>
17324 PR tree-optimization/71437
17325 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17326 member function. Implementation moved into after_dom_children
17327 member function and into the threader's thread_outgoing_edges
17329 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17330 some code into new thread_outgoing_edges.
17331 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17332 definition. Simplify marker handling (do it here). Assume we always
17333 have the available expression and the const/copies tables.
17334 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17336 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17337 * tree-vrp.c (equiv_stack): No longer file scoped.
17338 (vrp_dom_walker): New class.
17339 (vrp_dom_walker::before_dom_children): New member function.
17340 (vrp_dom_walker::after_dom_children): Likewise.
17341 (identify_jump_threads): Setup domwalker. Use it rather than
17342 walking edges in a random order by hand. Simplify setup/finalization.
17343 (finalize_jump_threads): Remove.
17344 (vrp_finalize): Do not call identify_jump_threads here.
17345 (execute_vrp): Do it here instead and call thread_through_all_blocks
17348 PR tree-optimization/71437
17349 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17351 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17353 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17354 (dom_opt_dom_walker::thread_across_edge): Remove
17355 handle_dominating_asserts argument. All callers changed.
17356 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17357 changes. Remove calls to lhs_of_dominating_assert. Other
17358 uses of handle_dominating_asserts turn into unconditional code
17359 (simplify_control_stmt_condition_1): Likewise.
17360 (simplify_control_stmt_condition): Likewise.
17361 (thread_through_normal_block, thread_across_edge): Likewise.
17362 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17363 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17364 object if it is not an SSA_NAME.
17365 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17366 before calling into the VRP specific simplifiers.
17367 (identify_jump_threads): Remove handle_dominating_asserts
17370 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17373 * tree-diagnostic.c (default_tree_printer): No longer static.
17374 * tree-diagnostic.h (default_tree_printer): New prototype.
17376 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17378 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17379 Change ins into fmov.
17381 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17383 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17384 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17385 Use h_con constraint for operand 1.
17386 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17387 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17389 2017-03-15 Jeff Law <law@redhat.com>
17391 PR tree-optimization/71437
17392 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17393 (record_temporary_equivalences): Use it.
17395 PR tree-optimization/71437
17396 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17397 tree-ssa-scopedtables.
17398 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17399 (record_conditions, record_cond, vuse_eq): Likewise.
17400 (record_edge_info): Adjust to API tweak of record_conditions.
17401 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17402 (record_temporary_equivalences, optimize_stmt): Likewise.
17403 (eliminate_redundant_computations): Likewise.
17404 (record_equivalences_from_stmt): Likewise.
17405 * tree-ssa-scopedtables.c: Include options.h and params.h.
17406 (vuse_eq): New function, moved from tree-ssa-dom.c
17407 (build_and_record_new_cond): Likewise.
17408 (record_conditions): Likewise. Accept vector of conditions rather
17409 than edge_equivalence structure for first argument.
17410 for the first argument.
17411 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17412 from tree-ssa-dom.c.
17413 (avail_exprs_stack::record_cond): Likewise.
17414 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17415 from tree-ssa-dom.c.
17416 (avail_exprs_stack): Add new member functions lookup_avail_expr
17418 (record_conditions): Declare.
17420 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17423 * lra-constraints.c (process_alt_operands): Increase reject for
17424 reloading an input/output operand.
17426 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17429 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17430 insns to convert from signed/unsigned char/short to IEEE 128-bit
17432 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17434 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17437 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17438 subreg of inner mode for values already in registers.
17440 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17442 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17443 iteration reg is used after the loop.
17445 2017-03-14 Martin Sebor <msebor@redhat.com>
17447 PR tree-optimization/79800
17448 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17449 precision in negative-positive range.
17450 (format_floating): Call non-const overload with adjusted precision.
17452 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17455 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17458 2017-03-14 Martin Sebor <msebor@redhat.com>
17460 PR middle-end/80020
17461 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17462 * builtins.def (aligned_alloc): Use it.
17465 * Makefile.in (GTFILES): Add calls.c.
17466 * calls.c: Include "gt-calls.h".
17468 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17470 PR rtl-optimization/79728
17471 * regs.h (struct target_regs): New field
17472 x_contains_allocatable_regs_of_mode.
17473 (contains_allocatable_regs_of_mode): New macro.
17474 * reginfo.c (init_reg_sets_1): Initialize it, and change
17475 contains_reg_of_mode so it includes global regs as well.
17476 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17477 rather than contains_regs_of_mode.
17479 2017-03-14 Martin Liska <mliska@suse.cz>
17481 * doc/invoke.texi: Document options that can't be combined with
17482 -fcheck-pointer-bounds.
17484 2017-03-14 Martin Liska <mliska@suse.cz>
17486 PR middle-end/79831
17487 * doc/invoke.texi (-Wchkp): Document the option.
17489 2017-03-14 Martin Liska <mliska@suse.cz>
17491 * Makefile.in: Install gcov-dump.
17493 2017-03-14 Martin Liska <mliska@suse.cz>
17495 * multiple_target.c (expand_target_clones): Bail out for
17496 an invalid attribute.
17498 2017-03-14 Richard Biener <rguenther@suse.de>
17500 * alias.c (struct alias_set_entry): Pack properly.
17501 * cfgloop.h (struct loop): Likewise.
17502 * cse.c (struct set): Likewise.
17503 * ipa-utils.c (struct searchc_env): Likewise.
17504 * loop-invariant.c (struct invariant): Likewise.
17505 * lra-remat.c (struct cand): Likewise.
17506 * recog.c (struct change_t): Likewise.
17507 * rtl.h (struct address_info): Likewise.
17508 * symbol-summary.h (function_summary): Likewise.
17509 * tree-loop-distribution.c (struct partition): Likewise.
17510 * tree-object-size.c (struct object_size_info): Likewise.
17511 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17512 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17513 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17514 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17515 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17516 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17517 (struct _stmt_vec_info): Likewise.
17519 2017-03-14 Martin Liska <mliska@suse.cz>
17522 * multiple_target.c (create_dispatcher_calls): Check that
17523 a target can create a function dispatcher.
17525 2017-03-14 Martin Liska <mliska@suse.cz>
17528 * multiple_target.c (expand_target_clones): Drop local.local
17529 flag for default implementation.
17531 2017-03-14 Richard Biener <rguenther@suse.de>
17533 PR tree-optimization/80030
17534 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17536 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17538 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17539 gcc_fallthrough() instead of __attribute__((fallthrough));
17541 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17543 * doc/gcc.texi: Remove "up" link to (DIR).
17544 * doc/gccint.texi: Ditto.
17546 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17548 * doc/install.texi (Specific) <avr>: Remove reference to
17551 2017-03-13 Jeff Law <law@redhat.com>
17553 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17554 attribute rather than comments.
17556 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17557 match_scratch operand is highest.
17559 2017-03-13 Martin Liska <mliska@suse.cz>
17561 PR middle-end/78339
17562 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17563 is a CHKP clone, use original declaration.
17565 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17567 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17568 (arc_conditional_register_usage): Use a different allocation order
17569 when optimizing for size.
17570 * common/config/arc/arc-common.c (arc_option_optimization_table):
17571 Section anchors default on when optimizing for size.
17573 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17575 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17577 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17579 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17580 * config/arc/arc.md (cpu_facility): Add cd variant.
17581 (*movqi_insn): Add code density variant.
17582 (*movhi_insn): Likewise.
17583 (*movqi_insn): Likewise.
17584 (*addsi3_mixed): Likewise.
17585 (subsi3_insn): Likewise.
17587 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17589 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17591 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17593 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17594 expressions with MINUS and UNARY ops.
17596 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17599 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17601 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17602 between vec_select and vector argument.
17603 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17604 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17605 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17606 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17607 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17608 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17610 2017-03-13 Richard Biener <rguenther@suse.de>
17613 * params.def (vect-max-peeling-for-alignment): Fix typo.
17615 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17617 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17618 issue that only occurred with binutils below 2.18.
17620 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17622 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17623 refer to binutils 2.11/2.12 minimum.
17625 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17627 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17628 ftp.kernel.org and simplify binutils requirement.
17630 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17632 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17634 (Optimize Options): Ditto. Also remove redundancy.
17636 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17638 PR translation/79848
17639 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17641 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17642 to G_ to avoid double translation.
17644 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17646 PR translation/79923
17647 * auto-profile.c (get_combined_location): Convert leading
17648 character of diagnostics to lower case and remove trailing period.
17649 (read_profile): Likewise for various diagnostics.
17650 * config/arm/arm.c (arm_option_override): Remove trailing period
17651 from various diagnostics.
17652 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17653 (msp430_expand_delay_cycles): Likewise.
17655 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17658 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17659 full command-line argument, rather than just "str".
17660 (aarch64_validate_march): Likewise.
17661 (aarch64_validate_mtune): Likewise.
17663 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17665 PR rtl-optimization/78911
17666 * lra-assigns.c (must_not_spill_p): New function.
17667 (spill_for): Use it.
17669 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17671 PR tree-optimization/79981
17672 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17673 ATOMIC_COMPARE_EXCHANGE ifn result.
17674 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17675 IFN_ATOMIC_COMPARE_EXCHANGE.
17677 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17680 * opts.c (parse_sanitizer_options): Add missing question mark to
17681 "did you mean" message.
17683 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17685 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17687 (VMULEUH_UNS): Likewise.
17688 (VMULOUB_UNS): Likewise.
17689 (VMULOUH_UNS): Likewise.
17690 * config/rs6000/rs6000.c (builtin_function_type): Remove
17691 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17693 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17696 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17697 x with result of extra_parsing_for_operand_code_0.
17698 (function_reader::extra_parsing_for_operand_code_0): Convert
17699 return type from void to rtx, returning x. When reading
17700 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17701 larger size containing struct block_symbol.
17703 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17706 -mfloat128-hardware without -m64.
17708 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17711 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17712 entries to the case statement that marks unsigned arguments to
17713 overloaded functions.
17715 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17717 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17718 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17720 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17723 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17724 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17726 2017-03-10 Martin Liska <mliska@suse.cz>
17730 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17731 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17734 2017-03-10 Olivier Hainque <hainque@adacore.com>
17736 * tree-switch-conversion (array_value_type): Start by resetting
17737 candidate type to it's main variant.
17739 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17741 PR rtl-optimization/79909
17742 * combine.c (try_combine): Use simplify_replace_rtx on individual
17743 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17744 of the whole CALL_INSN_FUNCTION_USAGE.
17746 PR tree-optimization/79972
17747 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17748 get_range_info on SSA_NAMEs. Formatting fixes.
17750 2017-03-10 Richard Biener <rguenther@suse.de>
17751 Jakub Jelinek <jakub@redhat.com>
17753 PR tree-optimization/77975
17754 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17755 edge to be constant.
17756 (get_val_for): For constant x return it. Formatting fix.
17757 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17758 would use the same bases as the current one.
17760 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17762 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17763 instead of vec_select for V1TImode.
17764 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17766 (VSX_LE_128): Add V1TI to this mode iterator.
17767 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17768 (*vsx_le_perm_store_<mode>): Likewise.
17769 (pre-reload splitter for VSX stores): Likewise.
17770 (post-reload splitter for VSX stores): Likewise.
17771 (*vsx_xxpermdi2_le_<mode>): Likewise.
17772 (*vsx_lxvd2x2_le_<mode>): Likewise.
17773 (*vsx_stxvd2x2_le_<mode>): Likewise.
17775 2017-03-09 Michael Eager <eager@eagercon.com>
17777 Correct failures with --enable-checking=yes,rtl.
17779 * config/microblaze/microblaze.c (microblaze_expand_shift):
17780 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17781 test for const0_rtx.
17782 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
17783 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
17785 2017-03-09 Richard Biener <rguenther@suse.de>
17787 PR tree-optimization/79977
17788 * graphite-scop-detection.c (scop_detection::merge_sese):
17789 Handle the case of extra exits to blocks dominating the entry.
17791 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
17793 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17796 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
17798 PR rtl-optimization/79949
17799 * lra-constraints.c (process_alt_operands): Check memory when
17800 trying to predict a cycle. Print about the overall increase.
17802 2017-03-09 Richard Biener <rguenther@suse.de>
17804 PR middle-end/79971
17805 * gimple-expr.c (useless_type_conversion_p): Preserve
17806 TYPE_SATURATING for fixed-point types.
17808 2017-03-09 Richard Biener <rguenther@suse.de>
17811 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
17812 alignment of BLKmode params.
17814 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17817 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
17818 (VALL_NO_V2Q): Likewise.
17820 * config/aarch64/aarch64-simd.md
17821 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
17823 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
17824 VALL_NO_V2Q mode iterator.
17825 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
17827 2017-03-09 Martin Liska <mliska@suse.cz>
17829 PR tree-optimization/79631
17830 * tree-chkp-opt.c (chkp_is_constant_addr): Call
17831 tree_int_cst_sign_bit just for INTEGER constants.
17833 2017-03-09 Martin Liska <mliska@suse.cz>
17837 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
17840 2017-03-09 Marek Polacek <polacek@redhat.com>
17843 * tree.c (inchash::add_expr): Handle TREE_VEC.
17845 2017-03-09 Martin Liska <mliska@suse.cz>
17848 (chkp_narrow_size_and_offset): New function.
17849 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
17850 (void chkp_parse_bit_field_ref): New function.
17851 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
17852 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
17854 2017-03-09 Martin Liska <mliska@suse.cz>
17857 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
17858 (chkp_find_bounds_1): Remove gcc_unreachable.
17860 2017-03-09 Jakub Jelinek <jakub@redhat.com>
17863 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
17864 BUILT_IN_SYNC*, determine the access type from the size suffix and
17865 always build a MEM_REF with that type. Handle forgotten
17866 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
17869 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
17870 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
17871 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
17872 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
17873 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
17874 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
17875 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
17876 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
17877 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
17878 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
17879 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
17880 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
17881 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
17882 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
17883 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
17884 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
17885 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
17886 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
17887 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
17888 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
17889 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
17890 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
17891 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
17892 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
17893 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
17894 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
17895 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
17896 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
17897 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
17898 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
17899 definitions outside of __OPTIMIZE__ guarded section.
17902 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
17903 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
17904 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
17905 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
17908 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17910 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
17911 ("vfenez<mode>"): Add missing constraints.
17913 2017-03-08 Martin Sebor <msebor@redhat.com>
17916 * config/nds32/nds32.c (nds32_option_override):
17917 Fix misspelled diagnostic.
17919 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17922 * gimplify.c (gimplify_omp_for): Replace index var in outer
17923 taskloop statement with an artificial variable and add
17924 OMP_CLAUSE_PRIVATE clause for it.
17926 2017-03-08 Richard Biener <rguenther@suse.de>
17928 PR tree-optimization/79955
17929 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
17930 for accesses that are completely outside of the variable.
17932 2017-03-08 Andrew Haley <aph@redhat.com>
17934 PR tree-optimization/79943
17935 * tree-ssa-loop-split.c (compute_new_first_bound): When
17936 calculating the new upper bound, (END-BEG) should be added, not
17939 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17941 * config/avr/avr.md (setmemhi): Make sure match_dup
17942 operand number comes before match_scratch.
17944 2017-03-08 Richard Biener <rguenther@suse.de>
17946 PR tree-optimization/79920
17947 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
17948 with ncopies == 1 to ...
17949 (vect_transform_slp_perm_load): ... here. Properly compute
17950 all element loads by iterating VF times over the group. Do
17951 not handle ncopies (computed in a broken way) in
17952 vect_create_mask_and_perm.
17954 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17957 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
17958 is a uniform vector, use uniform_vector_p return value instead of
17959 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
17961 2017-03-07 Marek Polacek <polacek@redhat.com>
17963 PR middle-end/79809
17964 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
17965 (alloca_call_type): Likewise.
17967 2017-03-07 Martin Liska <mliska@suse.cz>
17969 * gcov.c (process_args): Put comment to correct location.
17971 2017-03-07 Martin Liska <mliska@suse.cz>
17973 PR middle-end/68270
17974 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
17975 Use array_at_struct_end_p instead of DECL_CHAIN (field).
17976 (chkp_narrow_bounds_for_field): Likewise.
17977 (chkp_parse_array_and_component_ref): Pass one more argument to
17980 2017-03-07 Richard Biener <rguenther@suse.de>
17982 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
17985 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17987 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
17988 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
17990 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17993 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
17994 to end of description.
17995 (PARAM_MAX_STORES_TO_MERGE): Likewise.
17997 2017-03-07 Jakub Jelinek <jakub@redhat.com>
17999 PR rtl-optimization/79901
18000 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18002 (*avx512f_<code><mode>3<mask_name>): ... this.
18003 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18004 iterator instead of VI8_AVX2_AVX512BW.
18006 PR rtl-optimization/79901
18007 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18008 min/max expander, expand it using expand_vec_cond_expr.
18011 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18014 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18017 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18018 to void * for PCH reasons.
18019 * dwarf2out.c (output_loc_operands, output_die): Cast
18020 v.val_vec.array to unsigned char *.
18022 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18025 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18028 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18030 PR rtl-optimization/79571
18031 * lra-constraints.c (process_alt_operands): Calculate static
18032 reject and subtract it from overall when only addresses will be
18035 2017-03-06 Julia Koval <julia.koval@intel.com>
18038 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18039 incoming stack boundary to 128 for 64-bit targets.
18041 2017-03-06 Richard Biener <rguenther@suse.de>
18043 PR tree-optimization/79894
18044 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18045 to NULL after folding it.
18047 2017-03-06 Richard Biener <rguenther@suse.de>
18049 PR tree-optimization/79824
18050 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18051 check disabling peeling for gaps.
18053 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18055 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18056 attributes): Document gettimeofday.
18058 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18060 * config/s390/s390.c (s390_option_override_internal): Set
18061 PARAM_MIN_VECT_LOOP_BOUND
18063 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18065 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18066 * config/s390/s390.md: Likewise.
18068 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18071 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18072 (<avx2_avx512>_perm<mode>): Rename to ...
18073 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18075 (<avx512>_perm<mode>_mask): Rename to ...
18076 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18078 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18079 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18080 instead of VI8F_256_512.
18081 (avx512f_perm<mode>): New define_expand.
18082 (avx512f_perm<mode>_mask): Likewise.
18083 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18084 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18086 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18088 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18089 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18091 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18093 2017-03-06 Martin Liska <mliska@suse.cz>
18096 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18097 when having a SSA NAME w/o VAR_DECL assigned to it.
18099 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18101 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18102 msa_dpsub_<su>_d): Fix MODE for vec_select.
18104 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18106 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18108 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18110 2017-03-06 Richard Biener <rguenther@suse.de>
18112 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18113 * plugin.c (register_plugin_info): Likewise.
18114 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18116 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18118 * config/i386/sse.md (sse_storehps, sse_storelps,
18119 avx_<castmode><avxsizesuffix>_<castmode>,
18120 avx512f_<castmode><avxsizesuffix>_<castmode>,
18121 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18122 in condition that at least one operand is not a MEM.
18124 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18126 PR middle-end/79805
18127 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18128 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18130 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18131 gimple_call_nothrow_p flag based on whether original builtin can throw.
18132 If it can, emit following stmts on the fallthrough edge.
18133 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18134 don't create new bb if inserting just debug stmts on the edge, try to
18135 insert them on the fallthru bb or just reset debug stmts.
18137 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18140 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18141 restore recog_data (including the operand rtxes inside it) around
18142 the call to get_insn_template.
18144 2017-03-03 Martin Sebor <msebor@redhat.com>
18146 PR tree-optimization/79699
18147 * context.c (context::~context): Free MPFR caches to avoid
18148 a memory leak on program exit.
18150 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18152 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18153 Use wide_int::ulow () instead of .elt (0).
18155 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18157 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18158 (*pushxf): Limit oF constraint to 32bit targets and add oC
18159 constraint for 64bit targets.
18160 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18161 (*pushdf): Change rmF constraint to rmC.
18163 2017-03-03 Martin Liska <mliska@suse.cz>
18165 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18166 Remove unused variable.
18168 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18171 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18172 is a memory operand, increase num_memory.
18173 (ix86_expand_args_builtin): Likewise.
18175 2017-03-03 Jan Hubicka <jh@suse.cz>
18178 * ipa-devirt.c (maybe_record_node): Properly handle
18179 __cxa_pure_virtual visibility.
18181 2017-03-03 Martin Liska <mliska@suse.cz>
18183 PR tree-optimization/79803
18184 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18186 (pass_loop_prefetch::execute): Disabled optimization if an
18187 assumption about L1 cache size is not met.
18189 2017-03-03 Martin Liska <mliska@suse.cz>
18191 PR rtl-optimization/79574
18192 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18193 (hash_scan_set): Likewise.
18194 (dump_hash_table): Likewise.
18195 (hoist_code): Likewise.
18197 2017-03-03 Richard Biener <rguenther@suse.de>
18199 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18200 in place of uhigh ().
18201 (fixed_convert_from_real): Likewise.
18203 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18206 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18208 2017-03-03 Richard Biener <rguenther@suse.de>
18210 PR middle-end/79818
18211 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18212 TYPE_OVERFLOW_UNDEFINED check.
18214 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18216 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18218 (vector_ae_<mode>_p): Likewise.
18219 (vector_nez_<mode>_p): Likewise.
18220 (vector_ne_v2di_p): Likewise.
18221 (vector_ae_v2di_p): Likewise.
18222 (vector_ne_<mode>_p): Likewise.
18223 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18225 (vsx_tsqrt<mode>2_fe): Likewise.
18227 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18230 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18232 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18234 PR rtl-optimization/79780
18235 * cprop.c (one_cprop_pass): When second and further conditional trap
18236 in a single basic block is turned into an unconditional trap, turn it
18237 into a deleted note to avoid RTL verification failures.
18239 2017-03-02 Richard Biener <rguenther@suse.de>
18241 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18243 2017-03-02 Richard Biener <rguenther@suse.de>
18245 PR tree-optimization/79345
18247 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18248 param and abort the walk, returning -1 if it is hit.
18249 (walk_aliased_vdefs): Take a limit param and pass it on.
18250 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18251 defaulting to 0 and return a signed int.
18252 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18253 (check_defs): New helper.
18254 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18255 about uninitialized memory.
18256 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18257 bogus uninitialized warning.
18258 (fixed_convert_from_real): Likewise.
18260 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18262 PR tree-optimization/66768
18263 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18264 iv_use if base object can't be determined.
18266 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18268 PR tree-optimization/79345
18269 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18270 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18271 (get_pattern_stats): Initialize it.
18272 * genemit.c (gen_expand): Verify match_scratch numbers come after
18273 match_operand/match_dup numbers.
18274 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18275 match_scratch numbers.
18276 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18278 * config/s390/s390.md (trunctdsd2): Likewise.
18280 2017-03-02 Richard Biener <rguenther@suse.de>
18282 * wide-int.h (wide_int_storage::operator=): Implement in terms
18285 2017-03-02 Richard Biener <rguenther@suse.de>
18287 PR tree-optimization/79777
18288 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18289 the to insert expression to sth existing.
18291 2017-03-01 Martin Sebor <msebor@redhat.com>
18293 PR middle-end/79692
18294 * gimple-ssa-sprintf.c
18295 (directive::known_width_and_precision): New function.
18296 (format_integer): Use it.
18297 (get_mpfr_format_length): Consider the full range of precision
18298 when computing %g output with the # flag. Set the likely byte
18299 count to 3 rather than 1 when precision is indeterminate.
18300 (format_floating): Correct the lower bound of precision.
18302 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18304 * doc/invoke.texi: Document default code model for 64-bit Linux.
18306 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18309 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18310 udiv rather than div since input pattern is unsigned.
18312 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18314 * config/i386/i386.c (print_reg): Warn for values of
18315 unsupported size in integer register.
18317 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18320 * config/rs6000/predicates.md (current_file_function_operand): Do
18321 not allow self calls to be local if the function is replaceable.
18323 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18326 * config/rs6000/altivec.h (vec_ctz and others): Change the
18327 preprocessor macro that controls conditional compilation from
18328 _ARCH_PWR9 to __POWER9_VECTOR__.
18329 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18330 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18331 control (instead of _ARCH_PWR9 control) so that template
18332 definition uses power9-specific function.
18333 (vec_any_eq): Likewise.
18334 (vec_all_ne): Change macro definition to use a power9-specific
18335 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18336 _ARCH_PWR9 control).
18337 (vec_any_eq) Likewise.
18338 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18339 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18340 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18341 support for xvcmpnedp instruction.
18342 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18343 macro expansion so that Power9 implementation of vec_all_ne does
18344 not use the AltiVec predicate framework.
18345 (VCMPNEH_P): Likewise.
18346 (VCMPNEW_P): Likewise.
18347 (VCMPNED_P): Likewise.
18348 (VCMPNEFP_P): Likewise.
18349 (VCMPNEDP_P): Likewise.
18350 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18351 implementation of vec_any_eq to not use AltiVec predicate
18353 (VCMPAEH_P): Likewise.
18354 (VCMPAEW_P): Likewise.
18355 (VCMPAED_P): Likewise.
18356 (VCMPAEFP_P): Likewise.
18357 (VCMPAEDP_P): Likewise.
18358 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18359 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18360 not use the AltiVec predicate framework.
18361 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18362 of vec_any_eq to not use AltiVec predicate framework.
18363 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18364 support for predefined __POWER9_VECTOR__ macro to indicate that
18365 Power9 instruction selection is enabled.
18366 (altivec_overloaded_builtins): Remove extraneous
18367 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18368 function argument types RS6000_BTI_bool_V16QI and
18369 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18370 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18371 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18372 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18373 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18374 Power9 for implementations of vec_cmpne. Change the signature for
18375 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18376 (representing vec_all_ne) to remove the previously described first
18377 argument of type RS6000_BTI_INTSI, as this was an artifact of
18378 reliance on the AltiVec predicate framework, which is no longer
18379 used in the implementation of these functions. Add
18380 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18381 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18382 since, unlike the AltiVec predicate framework implementation, we
18383 do not share function descriptors between vec_alle and vec_anyeq.
18384 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18385 set of modes that receive special treatment even when
18386 TARGET_P9_VECTOR is true. The special treatment emits code that
18387 does not depend on Power9 instructions.
18388 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18389 define_expand to not rely on AltiVec predicate framework.
18390 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18392 (vector_ne_v2di_p): Change this define_expand to not rely on
18393 AltiVec predicate framework.
18394 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18396 (vector_ne_<mode>_p): Change this define_expand to not rely on
18397 AltiVec predicate framework.
18398 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18400 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18401 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18402 define_insn pattern.
18403 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18404 define_insn pattern because the xvcmpne<VSs>. instruction is not
18406 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18407 instruction is not supported.
18409 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18411 * config/nvptx/nvptx.c: Include intl.h.
18413 2017-03-01 Martin Jambor <mjambor@suse.cz>
18416 * ipa-prop.h (ipa_bits): Removed field known.
18417 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18418 to pointers. Adjusted their comments to warn about their sharing.
18419 (ipcp_transformation_summary): Change bits to a vector of pointers.
18420 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18421 (ipa_get_ipa_bits_for_value): Declare.
18422 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18423 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18424 (ipa_bits_hash_table): Likewise.
18425 (ipa_vr_ggc_hash_traits): Likewise.
18426 (ipa_vr_hash_table): Likewise.
18427 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18428 being pointers and vr_known being removed.
18429 (ipa_set_jf_unknown): Likewise.
18430 (ipa_get_ipa_bits_for_value): New function.
18431 (ipa_set_jfunc_bits): Likewise.
18432 (ipa_get_value_range): New overloaded functions.
18433 (ipa_set_jfunc_vr): Likewise.
18434 (ipa_compute_jump_functions_for_edge): Use the above functions to
18435 construct bits and vr parts of jump functions.
18436 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18437 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18439 (ipcp_grow_transformations_if_necessary): Also allocate
18440 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18442 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18443 them. Fix too long lines.
18444 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18445 vr_known being removed.
18446 (ipa_read_jump_function): Use new setter functions to construct bits
18447 and vr parts of jump functions or set them to NULL.
18448 (write_ipcp_transformation_info): Adjust for bits being pointers.
18449 (read_ipcp_transformation_info): Likewise.
18450 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18452 Include gt-ipa-prop.h.
18453 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18455 (ipcp_store_bits_results): Likewise.
18456 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18457 Do not write to existing jump functions but use a temporary instead.
18459 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18462 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18463 attempt to use its first operand as BIT_FIELD_REF base.
18465 2017-03-01 Richard Biener <rguenther@suse.de>
18467 PR middle-end/79721
18468 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18469 interpolating formula in wrapping arithmetic.
18470 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18472 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18474 PR tree-optimization/79734
18475 * tree-vect-generic.c (expand_vector_condition): Optimize
18476 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18477 Handle VEC_COND_EXPR where comparison has different inner width from
18478 type's inner width.
18480 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18482 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18483 markup, and similar issues. Remove @opindex entries for things
18484 that aren't options. Add missing -mmpy-option entries.
18486 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18488 PR tree-optimization/79737
18489 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18490 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18491 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18492 instead of byte_size. Formatting fix.
18493 (shift_bytes_in_array_right): Formatting fix.
18495 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18498 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18499 condition on optimize for the leaf function test.
18501 2017-02-28 Martin Liska <mliska@suse.cz>
18504 * read-rtl-function.c (function_reader::handle_unknown_directive):
18505 Bail out when one uses -flto.
18507 2017-02-28 Martin Liska <mliska@suse.cz>
18509 * common.opt: Replace space with tabular for options of <number>
18511 * config/i386/i386.opt: Show <number> value for
18512 -mlarge-data-threshold.
18513 * opts.c (print_filtered_help): Do not display number in hexadecimal
18516 2017-02-28 Martin Liska <mliska@suse.cz>
18518 * common.opt: Fix --help=option -Q for options which are of
18521 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18523 * config/i386/i386.c (print_reg): Error out for values
18524 of 8-bit size in invalid integer register.
18526 2017-02-28 Martin Sebor <msebor@redhat.com>
18528 PR tree-optimization/79691
18529 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18531 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18534 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18535 gcc_unreachable with output_operand_lossage.
18537 2017-02-28 Richard Biener <rguenther@suse.de>
18539 PR tree-optimization/79740
18540 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18542 (visit_nary_op): Insert the nary into the hashtable if we
18543 pattern-matched sth.
18544 * tree-ssa-pre.c (eliminate_insert): Robustify.
18546 2017-02-28 Richard Biener <rguenther@suse.de>
18548 PR middle-end/79731
18549 * fold-const.c (decode_field_reference): Reject out-of-bound
18552 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18554 * config/i386/i386.c: Include intl.h.
18555 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18556 instead of just cond ? "..." : "...".
18557 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18558 * coverage.c (read_counts_file): Likewise.
18559 * omp-offload.c: Include intl.h.
18560 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18561 of just cond ? "..." : "...".
18562 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18563 of just cond ? "..." : "...".
18565 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18568 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18570 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18571 'tune for' CPU name.
18572 * config/arm/arm-cpu-data.h: Regenerated.
18574 2017-02-28 Richard Biener <rguenther@suse.de>
18576 PR tree-optimization/79732
18577 * tree-inline.c (expand_call_inline): Do not shadow var.
18579 2017-02-28 Richard Biener <rguenther@suse.de>
18581 PR tree-optimization/79723
18582 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18583 address-space properly.
18585 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18587 * doc/optinfo.texi (Optimization groups): Fix option used for
18589 * doc/invoke.texi (-fopt-info): Document "omp".
18590 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18591 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18592 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18593 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18594 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18596 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18598 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18599 OPTGROUP_OMP with "omp".
18601 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18604 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18605 for arithmetic shift of unsigned V2DI.
18607 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18609 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18610 arc/linux.h headers.
18611 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18612 (LINK_SPEC): Likewise.
18613 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18614 (EXTRA_SPECS): Likewise.
18615 (STARTFILE_SPEC): Likewise.
18616 (ENDFILE_SPEC): Likewise.
18617 (LIB_SPEC): Likewise.
18618 (TARGET_SDATA_DEFAULT): Likewise.
18619 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18620 (MULTILIB_DEFAULTS): Likewise.
18621 (DWARF2_UNWIND_INFO): Likewise.
18622 * config/arc/big.h: New file.
18623 * config/arc/elf.h: Likewise.
18624 * config/arc/linux.h: Likewise.
18625 * config/arc/t-uClibc: Remove.
18627 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18629 PR tree-optimization/77536
18630 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18631 (tree_transform_and_unroll_loop): Use above function to compute the
18632 estimated niter of unrolled loop and use it when scaling profile.
18633 Also use count info rather than frequency if it's non-zero.
18634 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18635 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18636 (vect_transform_loop): Call above function.
18638 2017-02-27 Richard Biener <rguenther@suse.de>
18640 PR tree-optimization/45397
18641 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18642 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18643 (visit_nary_op): Add pattern matching for CSEing sign-changed
18644 or truncated operations with wider ones.
18646 2017-02-27 Richard Biener <rguenther@suse.de>
18648 PR tree-optimization/79690
18649 * tree-vect-stmts.c (vectorizable_store): Use vector type
18650 built from the DR with address-space.
18652 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18654 * doc/invoke.texi (Optimize Options): Refine the description
18655 of asan-use-after-return.
18657 2017-02-25 Alan Modra <amodra@gmail.com>
18659 PR rtl-optimization/79584
18660 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18661 base, not ad->base_term, the reg within base. Remove assertion
18662 that ad->base == ad->base_term. Replace gen_int_mode using
18663 bogus mode with const0_rtx.
18665 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18667 PR middle-end/79396
18668 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18669 FMA_EXPR like tcc_binary or tcc_unary.
18671 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18674 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18676 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18677 (output_loc_operands): Handle DW_OP_call_ref and
18678 DW_OP_GNU_variable_value.
18679 (struct variable_value_struct): New type.
18680 (struct variable_value_hasher): Likewise.
18681 (variable_value_hash): New variable.
18682 (string_types): Remove.
18683 (copy_loc_descr): New function.
18684 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18685 (prepend_loc_descr_to_each): New function.
18686 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18687 instead of add_loc_descr_to_each if the first argument is single
18688 location list and the second has multiple.
18689 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18690 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18691 when looking for variable value which doesn't have other location info.
18692 (loc_list_from_tree): Formatting fix.
18693 (gen_array_type_die): Simplify DW_AT_string_length handling.
18694 (adjust_string_types): Remove.
18695 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18696 string_types. Call resolve_variable_values.
18697 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18698 (resolve_addr_in_expr): Likewise. Add A argument.
18699 (copy_deref_exprloc): Remove deref argument. Adjust for the
18700 original expression being DW_OP_GNU_variable_value with optionally
18701 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18702 optionally after it.
18703 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18704 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18705 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18706 (variable_value_hasher::hash, variable_value_hasher::equal): New
18708 (resolve_variable_value_in_expr, resolve_variable_value,
18709 resolve_variable_values, note_variable_value_in_expr,
18710 note_variable_value): New functions.
18711 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18714 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18717 * opts.h (handle_generated_option): Add GENERATED_P argument.
18718 * opts-common.c (handle_option): Adjust function comment.
18719 (handle_generated_option): Add GENERATED_P argument, pass it to
18721 (control_warning_option): Pass false to handle_generated_option
18723 * opts.c (maybe_default_option): Pass true to handle_generated_option
18725 * optc-gen.awk: Likewise.
18727 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18729 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18730 a REG, look at the REG it is a SUBREG of.
18731 (splitter for cmpeqsi_t): Ditto.
18733 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18735 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18736 the special USEs with the pattern of the insn, not the insn itself.
18738 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18741 * doc/invoke.texi: Document -mload-store-pairs.
18743 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18744 Sandra Loosemore <sandra@codesourcery.com>
18746 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18747 argument isn't a CONST_INT.
18748 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18749 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18750 (nios2_validate_compare): Bypass alternate compare logic if *op2
18751 is not a CONST_INT.
18752 (ldstwm_operation_p): Return false if first_base is not a REG or
18753 if first_offset is not a CONST_INT.
18755 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18757 * config/cris/cris.md: Use correct operand in a define_peephole2.
18759 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18761 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18763 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18765 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18766 this_insn if it is an INSN or JUMP_INSN.
18767 (force_offsettable): Look at base, not at addr.
18768 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18769 on things that aren't necessarily CONST_INTs.
18771 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18773 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18774 -mfpmath=sse is the default also for x86-32 targets with SSE2
18775 instruction set when @option{-ffast-math} is enabled
18777 2017-02-24 Jeff Law <law@redhat.com>
18779 PR rtl-optimizatoin/79286
18780 * ira.c (update_equiv_regs): Drop may_trap_p exception to
18783 2017-02-24 Richard Biener <rguenther@suse.de>
18785 PR tree-optimization/79389
18786 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
18789 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
18791 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
18792 function comment to reflect reality.
18793 (loop_exits_before_overflow): Fix typo in function description.
18795 2017-02-24 Richard Biener <rguenther@suse.de>
18797 PR tree-optimization/79389
18798 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
18799 properly that a threading opportunity exists. Detect conditional
18800 copy/constant propagation opportunities.
18802 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
18804 * config/visium/visium.md (type): Add trap.
18805 (b): New mode attribute.
18806 (*btst): Rename into...
18807 (*btst<mode>): ...this and adjust.
18808 (*cbranchsi4_btst_insn): Rename into...
18809 (*cbranch<mode>4_btst_insn): ...this and adjust.
18810 (trap): New define_insn.
18812 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18814 PR tree-optimization/79389
18815 * ifcvt.c (struct noce_if_info): Add rev_cond field.
18816 (noce_reversed_cond_code): New function.
18817 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
18818 reversed_comparison_code. Formatting fix.
18819 (noce_try_store_flag): Test rev_cond != NULL in addition to
18820 reversed_comparison_code.
18821 (noce_try_store_flag_constants): Likewise.
18822 (noce_try_store_flag_mask): Likewise.
18823 (noce_try_addcc): Use rev_cond if non-NULL instead of
18824 reversed_comparison_code.
18825 (noce_try_cmove_arith): Likewise. Formatting fixes.
18826 (noce_try_minmax, noce_try_abs): Clear rev_cond.
18827 (noce_find_if_block): Initialize rev_cond.
18828 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
18829 instead of false as last argument never attempt to reverse it
18832 2017-02-23 Bin Cheng <bin.cheng@arm.com>
18834 PR tree-optimization/79663
18835 * tree-predcom.c (combine_chains): Process refs in reverse order
18836 only for ZERO length chains, and add explaining comment.
18838 2017-02-23 Jeff Law <law@redhat.com>
18840 PR tree-optimization/79578
18841 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
18842 in call to operand_equal_p.
18844 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
18847 * config/i386/cpuid.h: Fix another undefined behavior.
18849 2017-02-23 Richard Biener <rguenther@suse.de>
18851 PR tree-optimization/79683
18852 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
18853 vector types for data-refs.
18855 2017-02-23 Martin Liska <mliska@suse.cz>
18857 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
18859 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18861 PR middle-end/79665
18862 * internal-fn.c (get_range_pos_neg): Moved to ...
18863 * tree.c (get_range_pos_neg): ... here. No longer static.
18864 * tree.h (get_range_pos_neg): New prototype.
18865 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
18866 are known to be in between 0 and signed maximum inclusive, try to
18867 expand both unsigned and signed divmod and use the cheaper one from
18870 2017-02-22 Jeff Law <law@redhat.com>
18872 PR tree-optimization/79578
18873 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
18874 to compare base operands.
18876 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
18879 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
18880 gpc_reg_operand instead of fpr_reg_operand.
18882 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
18884 * config/mips/mips.c (mips_return_in_memory): Force FP
18885 vector types to be returned in memory for o32 ABI.
18887 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18889 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
18890 instead of DW_TAG_member for static data member declarations and don't
18891 set no_linkage_name for static inline data members.
18892 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
18895 2017-02-22 Martin Liska <mliska@suse.cz>
18897 * doc/invoke.texi: Replace inequality signs with square brackets
18900 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18902 PR tree-optimization/68644
18903 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
18905 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
18908 * lra-constraints.c (simplify_operand_subreg): Handle
18909 WORD_REGISTER_OPERATIONS targets.
18911 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18914 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
18915 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
18916 elimination by swapping fld*.
18918 2017-02-22 Richard Biener <rguenther@suse.de>
18920 PR tree-optimization/79673
18921 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
18922 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
18923 irrelevant address-space qualifiers and avoiding a
18924 ADDR_SPACE_CONVERT_EXPR from fold_convert.
18926 2017-02-22 Richard Biener <rguenther@suse.de>
18928 PR tree-optimization/79666
18929 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
18930 to not symbolically negate if that may introduce undefined
18933 2017-02-22 Martin Liska <mliska@suse.cz>
18936 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
18937 * data-streamer-out.c (streamer_write_gcov_count_stream):
18939 * value-prof.c (stream_out_histogram_value): Make assert more
18940 precise based on type of counter.
18942 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
18945 * config/i386/i386.md (standard_x87sse_constant_load splitter):
18946 Use nonimmediate_operand instead of memory_operand for operand 1.
18947 (float-extend standard_x87sse_constant_load splitter): Ditto.
18949 2017-02-21 Jeff Law <law@redhat.com>
18951 PR tree-optimization/79621
18952 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
18953 blocks with edges to themselves.
18955 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18958 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
18959 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
18960 Use gimple_call_builtin_p.
18963 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
18964 on temporarily removed DEBUG_INSNs.
18966 PR tree-optimization/79649
18967 * tree-loop-distribution.c (classify_partition): Give up on
18968 non-generic address space loads/stores.
18970 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
18972 * doc/loop.texi (Loop manipulation): Remove nonexistent
18973 tree_ssa_loop_version from the documentation.
18974 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
18976 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18979 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
18980 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
18981 * config/rs6000/rs6000.c: Include except.h.
18982 (rs6000_expand_split_stack_prologue): Call
18983 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
18985 2017-02-21 Martin Jambor <mjambor@suse.cz>
18988 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
18989 have been analyzed.
18991 2017-02-21 Martin Jambor <mjambor@suse.cz>
18993 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
18994 for backward compatibility only.
18995 * doc/invoke.texi (Option Summary): Remove all references to
18996 -fipa-cp-alignment.
18998 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19002 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19004 * lra-constraints.c (curr_insn_transform): Handle
19005 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19007 2017-02-21 Martin Liska <mliska@suse.cz>
19009 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19011 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19013 PR translation/79638
19014 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19016 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19019 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19020 (arm_function_ok_for_sibcall): Return false for an indirect call by
19021 descriptor if all the argument registers are used.
19022 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19023 alignment of the function.
19025 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19027 PR tree-optimization/61441
19028 * simplify-rtx.c (simplify_const_unary_operation): For
19029 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19030 the sNaN unmodified.
19032 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19034 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19035 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19036 instead of SYSTEM_HEADER_DIR.
19038 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19039 Martin Liška <mliska@suse.cz>
19041 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19042 Fix typos and grammar, use active voice, and clarify.
19044 2017-02-20 Marek Polacek <polacek@redhat.com>
19046 PR middle-end/79537
19047 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19050 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19052 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19055 * config/i386/i386.c (ix86_expand_builtin): Handle
19056 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19057 ix86_builtins_isa[fcode].isa as a requirement of those
19058 flags and any other flag in the bitmask.
19059 (ix86_init_mmx_sse_builtins): Use 0 instead of
19060 ~OPTION_MASK_ISA_64BIT as mask.
19061 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19062 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19063 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19064 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19066 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19069 * lra-constraints.c (split_reg): Check requested split mode
19070 is supported by the register.
19072 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19074 * lra-constraints.c (simplify_operand_subreg): Remove early
19077 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19080 * lra-constraints.c (curr_insn_transform): Tighten condition
19081 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19083 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19086 * lra-constraints.c (curr_insn_transform): Handle
19087 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19089 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19092 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19094 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19096 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19099 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19102 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19104 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19107 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19109 * final.c (last_columnnum, override_columnnum): New variables.
19110 (final_start_function): Set last_columnnum, pass it to begin_prologue
19111 hook and pass 0 to dwarf2out_begin_prologue.
19112 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19113 to source_line debug hook.
19114 (notice_source_line): Compute last_columnnum and for debug_column_info
19115 return true on column changes.
19116 * debug.h (struct gcc_debug_hooks): Add column argument to
19117 source_line and begin_prologue hooks.
19118 (debug_nothing_int_charstar_int_bool): Remove prototype.
19119 (debug_nothing_int_int_charstar,
19120 debug_nothing_int_int_charstar_int_bool): New prototypes.
19121 (dwarf2out_begin_prologue): Add column argument.
19122 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19123 begin_prologue hooks.
19124 (debug_nothing_int_charstar_int_bool): Remove.
19125 (debug_nothing_int_int_charstar,
19126 debug_nothing_int_int_charstar_int_bool): New functions.
19127 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19128 through to dwarf2out_source_line.
19129 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19130 (dwarf2out_source_line): Add column argument, emit it if requested.
19131 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19133 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19134 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19135 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19136 through to dwarf2out_begin_prologue.
19137 (vmsdbgout_source_line): Add column argument, pass it through to
19138 dwarf2out_source_line.
19139 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19140 dbxout_source_line caller.
19141 (dbxout_source_line): Add column argument.
19143 * common.opt (gno-column-info, gcolumn-info): New options.
19144 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19145 (check_die): Also test for multiple DW_AT_decl_column attributes.
19146 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19147 DW_AT_decl_column if requested.
19148 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19150 (gen_variable_die): Likewise.
19151 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19152 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19155 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19156 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19157 (ix86_handle_option): Handle OPT_m3dnowa.
19158 * doc/invoke.texi (-m3dnowa): Document.
19159 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19160 -m3dnowa instead of -m3dnow -march=athlon.
19163 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19164 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19166 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19169 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19170 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19171 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19172 generator for vsx_xxpermdi_<mode>_be.
19173 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19174 force big-endian semantics.
19175 (vsx_xxpermdi_<mode>_be): New define_expand with same
19176 implementation as previous version of vsx_xxpermdi_<mode>.
19178 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19180 PR tree-optimization/79327
19181 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19182 variable, its initialization and use.
19184 2017-02-17 Julia Koval <julia.koval@intel.com>
19186 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19187 (OPTION_MASK_ISA_PKU_UNSET): New.
19188 (ix86_handle_option): Handle -mrdpid.
19189 * config/i386/cpuid.h (bit_RDPID): New.
19190 * config/i386/driver-i386.c (host_detect_local_cpu):
19191 Detect RDPID feature.
19192 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19193 * config/i386/i386-c.c (ix86_target_macros_internal):
19195 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19196 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19197 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19198 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19199 * config/i386/i386.md (define_insn "rdpid"): New.
19200 * config/i386/i386.opt Add -mrdpid.
19201 * config/i386/immintrin.h (_rdpid_u32): New.
19203 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19205 PR rtl-optimization/79541
19206 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19207 instead of transforming it into USE.
19209 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19211 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19212 If HONOR_SNANS (SFmode) force the input to a register.
19213 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19214 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19215 an frsp or similar insn.
19217 2017-02-17 Martin Liska <mliska@suse.cz>
19219 PR rtl-optimization/79577
19220 * params.def (selsched-max-sched-times): Increase minimum to 1.
19222 2017-02-17 Martin Liska <mliska@suse.cz>
19224 PR rtl-optimization/79574
19225 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19227 2017-02-17 Martin Liska <mliska@suse.cz>
19229 PR tree-optimization/79529
19230 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19231 ssa_defined_default_def_p to handle cases which are implicitly
19233 * tree-ssa.c (ssa_defined_default_def_p): New function.
19234 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19235 which are implicitly defined.
19236 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19238 2017-02-17 Richard Biener <rguenther@suse.de>
19240 PR middle-end/79576
19241 * params.def (max-ssa-name-query-depth): Limit to 10.
19243 2017-02-17 Richard Biener <rguenther@suse.de>
19245 PR tree-optimization/79552
19246 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19249 2017-02-17 Richard Biener <rguenther@suse.de>
19252 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19254 2017-02-17 Marek Polacek <polacek@redhat.com>
19256 PR middle-end/79536
19257 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19258 (fold_negate_expr): New wrapper.
19260 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19262 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19263 Correct terminology and de-emphasize pre-standard behavior.
19265 2017-02-16 Alan Modra <amodra@gmail.com>
19267 PR rtl-optimization/79286
19268 * ira.c (def_dominates_uses): New function.
19269 (update_equiv_regs): Don't create an equivalence for insns that
19270 may trap where the register def does not dominate the use.
19272 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19274 PR rtl-optimization/78127
19275 * lra.c (lra): Call lra_eliminate before finish the loop after
19278 2017-02-16 Richard Biener <rguenther@suse.de>
19280 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19282 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19283 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19284 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19285 (isl_val_int_from_wi): New function.
19286 (extract_affine_gmp): Rename to ...
19287 (extract_affine_wi): ... this, take a widest_int.
19288 (extract_affine_int): Just wrap extract_affine_wi.
19289 (add_param_constraints): Use isl_val_int_from_wi.
19290 (add_loop_constraints): Likewise, and extract_affine_wi.
19292 2017-02-15 Jeff Law <law@redhat.com>
19294 PR middle-end/79521
19295 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19296 ira_init_register_move_cost_if_necessary.
19298 2017-02-15 Martin Sebor <msebor@redhat.com>
19300 PR middle-end/32003
19301 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19302 removed in a prior commit.
19304 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19306 PR tree-optimization/79347
19307 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19308 counters during peeling.
19310 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19312 * Makefile.in (site.exp): Remove "set ISLVER".
19314 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19317 * real.c (real_from_integer): Call real_convert even for decimal.
19319 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19322 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19324 2017-02-14 Andrew Pinski <apinski@cavium.com>
19326 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19327 cores and change the partno/implementer to be correct.
19328 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19329 the 'B" as the implementer.
19330 * config/aarch64/aarch64-tune.md: Regenerate.
19332 2017-02-14 Carl Love <cel@us.ibm.com>
19334 * config/rs6000/rs6000.c: Add case statement entry to make the
19335 xvcvuxdsp built-in argument unsigned.
19336 * config/rs6000/vsx.md: Fix the source and return operand types so they
19337 match the instruction definitions from the ISA document. Fix typo
19338 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19341 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19344 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19345 member early_clobber_alts.
19346 * lra-lives.c (reg_early_clobber_p): New.
19347 (process_bb_lives): Use it.
19348 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19349 (debug_operand_data): Initialize early_clobber_alts.
19350 (setup_operand_alternative): Set up early_clobber_alts.
19351 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19352 alternatives to new_insn_reg.
19353 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19355 (lra_update_insn_regno_info): Pass the new arg.
19357 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19359 PR middle-end/79505
19360 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19361 (new_oacc_loop_raw): Don't clear already cleared fields.
19364 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19365 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19366 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19368 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19371 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19373 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19376 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19377 the extra instruction to the right place to store 128-bit constant
19380 2017-02-14 Martin Sebor <msebor@redhat.com>
19382 PR middle-end/79448
19383 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19384 warning for strings of unknown length.
19386 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19388 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19390 2017-02-14 Jeff Law <law@redhat.com>
19393 * ira-costs.c (scan_one_insn): Initialize register move costs
19394 for pseudos seen in USE/CLOBBER insns.
19396 PR tree-optimization/79095
19397 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19398 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19399 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19400 if the operands are known to be not equal, then the resulting range
19402 (intersect_ranges): If the new range is ~[0,0] and the old range is
19403 wide, then prefer ~[0,0].
19404 * tree-vrp.c (overflow_comparison_p_1): New function.
19405 (overflow_comparison_p): New function.
19406 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19407 if NAME is used in an overflow test.
19408 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19409 overflow check that can be expressed as an equality test, then adjust
19410 ops to be that equality test.
19412 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19414 * config/s390/s390-builtin-types.def: Remove flags argument.
19415 * config/s390/s390.c (s390_init_builtins): Likewise.
19417 2017-02-14 Martin Liska <mliska@suse.cz>
19419 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19420 vector. Fix trailing white spaces.
19422 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19424 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19427 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19429 PR rtl-optimization/68664
19430 * config/arm/arm.c (arm_sched_can_speculate_insn):
19431 New function. Declare prototype.
19432 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19434 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19436 PR rtl-optimization/68664
19437 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19439 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19441 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19443 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19444 max skip bytes for function, loop and jump.
19446 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19448 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19449 ABS_EXPR for gimple dump.
19451 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19454 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19456 PR tree-optimization/79408
19457 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19458 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19459 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19460 also if rhs1 is INTEGER_CST.
19462 2017-02-14 Richard Biener <rguenther@suse.de>
19464 PR middle-end/79432
19465 * tree-into-ssa.c (insert_phi_nodes): When the function can
19466 have abnormal edges rewrite SSA names with broken use-def
19467 dominance out of SSA and register them for PHI insertion.
19469 2017-02-13 Martin Sebor <msebor@redhat.com>
19471 PR middle-end/79496
19472 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19473 clearing info.nowrite flag when snprintf size argument is a range.
19475 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19477 * cprop.c (cprop_jump): Add missing space in string literal.
19478 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19479 (get_constraint_for_component_ref): Likewise.
19480 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19481 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19482 * lra-constraints.c (process_alt_operands): Likewise.
19483 * ipa-inline.c (inline_small_functions): Likewise.
19484 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19485 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19486 * trans-mem.c (diagnose_tm_1_op): Likewise.
19487 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19488 (grid_parallel_clauses_gridifiable): Likewise.
19490 * config/nvptx/mkoffload.c (process): Add space in between
19493 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19494 "MOD4_SSE_REGS" and "ALL_REGS".
19496 * spellcheck.c (test_data): Add , in between "foo" and "food".
19498 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19501 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19502 boundary crossing check and subsequent code generation agree.
19504 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19506 * config/aarch64/aarch64.c (has_memory_op): Delete.
19507 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19510 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19512 PR rtl-optimization/79388
19513 PR rtl-optimization/79450
19514 * combine.c (distribute_notes): When removing TEM_INSN for which
19515 corresponding dest has last value recorded, invalidate that last
19518 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19520 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19521 of explicit '@'. Add missing assembly comment marker on branch costs
19524 2017-02-13 Nathan Sidwell <nathan@acm.org>
19526 * gengtype-lex.l (<in_struct>): Add '/'.
19528 2017-02-13 Martin Liska <mliska@suse.cz>
19531 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19533 2017-02-13 Richard Biener <rguenther@suse.de>
19535 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19537 * configure: Re-generate.
19538 * config.in: Likewise.
19539 * graphite-dependences.c: Simplify as if
19540 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19541 * graphite-isl-ast-to-gimple.c: Likewise.
19542 * graphite-optimize-isl.c: Likewise.
19543 * graphite-poly.c: Likewise.
19544 * graphite-sese-to-poly.c: Likewise.
19545 * graphite.h: Likewise.
19546 * toplev.c: Include isl/version.h and use isl_version () for
19547 printing the ISL version.
19548 * doc/install.texi: Update ISL requirement.
19550 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19552 * doc/standards.texi (Standards): Update reference to
19555 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19557 * doc/extend.texi (Named Address Spaces): sourceware.org now
19559 * doc/install.texi (Binaries): Ditto.
19562 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19564 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19565 standard terminology "stringize"/"stringizing" throughout.
19566 * doc/cppinternals.texi: Likewise.
19568 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19570 * doc/extend.texi: Fix some spelling mistakes and typos.
19571 * doc/invoke.texi: Likewise.
19573 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19576 * params.def (inline-min-speedup) Change from 10 to 8.
19578 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19580 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19583 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19586 * ipa-inline-analysis.c (get_minimal_bb): New function.
19587 (record_modified): Use it.
19588 (remap_edge_change_prob): Handle also ancestor functions.
19590 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19592 * doc/contrib.texi (Contributors): Remove broken link into
19593 the Mauve CVS repository.
19595 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19597 PR middle-end/79454
19598 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19599 result computation whenever lhs doesn't have vector mode, not
19600 just when it has BLKmode.
19602 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19604 * doc/makefile.texi (profiledbootstrap): Refer to the
19605 installation instructions only in textual form.
19607 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19610 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19612 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19614 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19615 (Specific): Update mingw-w64 reference.
19617 (Specific): Remove broken link to Renesas RX processor.
19619 2017-02-10 Richard Biener <rguenther@suse.de>
19621 * toplev.c (process_options): Do not mention obsolete graphite
19622 options when printing sorry message about missing graphite support.
19623 Mention -floop-nest-optimize.
19625 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19627 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19628 (vtst_p16): Likewise.
19629 (vtstq_p8): Likewise.
19630 (vtstq_p16): Likewise.
19632 (vtstq_p64): Likewise.
19633 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19634 (vset_lane_p64): New.
19635 (vsetq_lane_p64): New.
19637 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19639 PR tree-optimization/79411
19640 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19641 stmt operands are SSA_NAMEs used in abnormal phis.
19642 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19645 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19648 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19651 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19653 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19655 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19657 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19661 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19662 not arbitrary TREE_CONSTANT.
19665 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19666 "omp declare target link" attribute unless is_global_var.
19667 * omp-offload.c (find_link_var_op): Likewise.
19669 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19670 Chung-Lin Tang <cltang@codesourcery.com>
19672 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19674 (gimplify_adjust_omp_clauses): Don't delete TILE.
19675 (gimplify_omp_for): Deal with TILE.
19676 * internal-fn.c (expand_GOACC_TILE): New function.
19677 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19679 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19680 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19682 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19683 avoid DIV for outermost collapse var.
19684 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19685 Remove out of date comments, fix whitespace.
19686 * omp-general.c (omp_extract_for_data): Deal with tiling.
19687 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19688 adjust OLF_DIM_BASE value.
19689 (struct omp_for_data): Add tiling field.
19690 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19691 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19692 for auto loops. Remove default auto determining, moved to
19693 oacc_loop_fixed_partitions.
19694 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19695 stmts, add e_mask field.
19696 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19697 (oacc_thread_numbers): Use oacc_dim_call.
19698 (oacc_xform_tile): New.
19699 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19700 (finish_oacc_loop): Adjust for ifns vector.
19701 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19702 add case for GOACC_TILE fns.
19703 (oacc_loop_xform_loop): Delete.
19704 (oacc_loop_process): Iterate over call list directly, and add
19705 handling for GOACC_TILE fns.
19706 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19708 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19709 vector partitioning to outer loops. Assign 2 partitions to loops
19710 when available. Add TILE handling.
19711 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19712 (execite_oacc_device_lower): Process GOACC_TILE fns,
19713 ignore unknown specs.
19714 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19715 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19716 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19718 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19720 * configure.ac (ACX_BUGURL): Update.
19721 * configure: Regenerate.
19723 2017-02-09 Richard Biener <rguenther@suse.de>
19725 PR tree-optimization/69823
19726 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19727 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19729 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19731 * config/arc/arc-c.def: Add __NPS400__ definition.
19732 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19733 (TARGET_NPS400): Define.
19735 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19737 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19739 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19740 pointer, arch_info.
19741 (arc_cpu_types): Fill the arch_info field with a pointer into the
19742 arc_arch_types table.
19743 (arc_selected_cpu): Declare.
19744 * config/arc/arc.c (arc_selected_cpu): Make global.
19745 (arc_selected_arch): Delete.
19746 (arc_base_cpu): Delete.
19747 (arc_override_options): Remove references to deleted variables,
19748 update access to arch information.
19749 (ARC_OPT): Update access to arch information.
19750 (ARC_OPTX): Likewise.
19751 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19752 (TARGET_ARC600): Update access to arch information.
19753 (TARGET_ARC601): Likewise.
19754 (TARGET_ARC700): Likewise.
19755 (TARGET_EM): Likewise.
19756 (TARGET_HS): Likewise.
19757 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19760 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19763 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19764 condition/operands for integer GE/LE/GEU/LEU operations.
19766 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19768 PR translation/79397
19769 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19772 2017-02-08 Martin Jambor <mjambor@suse.cz>
19775 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19776 whether allocation happened.
19777 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19778 nothing was allocated.
19780 2017-02-08 Jakub Jelinek <jakub@redhat.com>
19782 PR tree-optimization/79408
19783 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
19784 constant, but SSA_NAME with a known integer range, use the minimum
19785 of that range instead of op1 to determine if modulo can be replaced
19786 with its first operand.
19788 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19790 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
19792 2017-02-08 Richard Biener <rguenther@suse.de>
19794 PR tree-optimization/71824
19795 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19796 Check all loops contained in the merged region.
19798 2017-02-07 Andrew Pinski <apinski@cavium.com>
19800 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
19802 2017-02-07 Andrew Pinski <apinski@cavium.com>
19804 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
19805 (thunderxt88): Likewise.
19806 (thunderxt81): Disable LSE and change v8.1 to v8.
19807 (thunderxt83): Likewise.
19809 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19810 Richard Biener <rguenther@suse.de>
19812 PR middle-end/79399
19813 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
19814 type from int to size_t.
19815 * ira-costs.c (struct_costs_size): Change type from int to size_t.
19817 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19819 PR rtl-optimization/79386
19820 * cprop.c (bypass_conditional_jumps): Initialize
19821 bypass_last_basic_block already before splitting bbs after
19822 unconditional traps...
19823 (bypass_conditional_jumps): ... rather than here.
19826 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
19827 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
19828 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
19829 fix -masm=intel patterns.
19831 2017-02-07 Richard Biener <rguenther@suse.de>
19833 PR tree-optimization/79256
19834 PR middle-end/79278
19835 * builtins.c (get_object_alignment_2): Use min_align_of_type
19836 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
19837 and ADJUST_FIELD_ALIGN.
19839 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
19841 * doc/tm.texi: Regenerate.
19842 * stor-layout.c (layout_decl): Adjust.
19843 (update_alignment_for_field): Likewise.
19844 (place_field): Likewise.
19845 (min_align_of_type): Likewise.
19846 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
19847 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
19848 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
19849 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
19850 * config/frv/frv.c (frv_adjust_field_align): Likewise.
19851 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
19852 * config/i386/i386.c (x86_field_alignment): Likewise.
19853 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
19854 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
19855 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
19856 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
19857 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
19858 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
19862 2017-01-30 Richard Biener <rguenther@suse.de>
19864 PR tree-optimization/79256
19865 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
19866 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
19869 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
19871 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
19872 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
19873 builtins to SImode and emit a zero-extend, if necessary.
19875 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19877 * docs/invoke.texi (RISC-V Options): Alphabetize.
19879 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19881 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
19884 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19886 * config/riscv/riscv.c: New file.
19887 * common/config/riscv/riscv-common.c: Likewise.
19888 * config.gcc: Likewise.
19889 * config/riscv/constraints.md: Likewise.
19890 * config/riscv/elf.h: Likewise.
19891 * config/riscv/generic.md: Likewise.
19892 * config/riscv/linux.h: Likewise.
19893 * config/riscv/multilib-generator: Likewise.
19894 * config/riscv/peephole.md: Likewise.
19895 * config/riscv/pic.md: Likewise.
19896 * config/riscv/predicates.md: Likewise.
19897 * config/riscv/riscv-builtins.c: Likewise.
19898 * config/riscv/riscv-c.c: Likewise.
19899 * config/riscv/riscv-ftypes.def: Likewise.
19900 * config/riscv/riscv-modes.def: Likewise.
19901 * config/riscv/riscv-opts.h: Likewise.
19902 * config/riscv/riscv-protos.h: Likewise.
19903 * config/riscv/riscv.h: Likewise.
19904 * config/riscv/riscv.md: Likewise.
19905 * config/riscv/riscv.opt: Likewise.
19906 * config/riscv/sync.md: Likewise.
19907 * config/riscv/t-elf-multilib: Likewise.
19908 * config/riscv/t-linux: Likewise.
19909 * config/riscv/t-linux-multilib: Likewise.
19910 * config/riscv/t-riscv: Likewise.
19911 * configure.ac: Likewise.
19912 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
19913 Waterman as RISC-V maintainers.
19914 * doc/install.texi: Add RISC-V entries.
19915 * doc/invoke.texi: Add RISC-V options section.
19916 * doc/md.texi: Add RISC-V constraints section.
19917 * configure: Regenerated.
19919 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19922 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
19923 false values to be constant vectors with all 0 or all 1 bits set.
19924 (vcondu<mode><mode>): Likewise.
19925 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
19927 (fpmask_comparison_operator): Update comment.
19928 (vecint_comparison_operator): New predicate.
19929 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
19930 vector conditionals when the true and false values are constant
19931 vectors with all 0 bits or all 1 bits set.
19933 2017-02-06 Martin Sebor <msebor@redhat.com>
19935 PR tree-optimization/79376
19936 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
19938 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
19940 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
19941 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
19942 to simplify split condition.
19944 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19946 * omp-expand.c (oxpand_omp_atomic_fetch_op,
19947 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
19950 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
19952 PR rtl-optimization/68664
19953 * target.def (can_speculate_insn): New hook.
19954 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
19955 * doc/tm.texi: Regenerate.
19956 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
19957 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
19958 (rs6000_sched_can_speculate_insn): New function.
19960 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19962 PR tree-optimization/79284
19963 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
19964 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
19965 vectorizable_mask_load_store, vectorizable_operation,
19966 vect_is_simple_cond, get_same_sized_vectype): Use it instead
19967 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
19968 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
19969 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
19970 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
19971 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19972 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
19973 is_gimple_assign (stmt). Replace another such test with
19974 is_gimple_assign (stmt).
19976 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19979 * config/avr/avr.c (rtl-iter.h): Include it.
19980 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
19981 (avr_legitimate_combined_insn): ...and implementation.
19983 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
19985 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
19986 * config/s390/s390.c (s390_const_operand_ok)
19987 (s390_canonicalize_comparison, s390_extract_part)
19988 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
19989 (s390_contiguous_bitmask_p, s390_rtx_costs)
19990 (legitimize_pic_address): Likewise.
19991 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
19992 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
19993 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
19994 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
19995 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
19997 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19999 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20000 REGNO($0) == REGNO($1).
20002 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20004 * config/s390/linux.h(SIZE_TYPE): Add comment.
20006 2017-02-06 Julian Brown <julian@codesourcery.com>
20007 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20008 Virendra Pathak <virendra.pathak@broadcom.com>
20010 * config/aarch64/aarch64-cores.def: Change the scheduler
20012 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20013 * config/aarch64/thunderx2t99.md: New file.
20015 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20017 * doc/standards.texi (Go Language): Update link to language
20020 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20022 * tree-eh.c (lower_resx): Sanitize profile.
20023 (cleanup_empty_eh_move_lp): Likewise.
20025 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20028 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20030 * cfgloopmanip.h (loop_version): Update prototype.
20031 * modulo-sched.c (sms_schedule): Update call of loop_version.
20032 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20033 * tree-parloops.c (gen_parallel_loop): Likewise.
20034 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20035 * tree-ssa-loop-split.c (split_loop): Likewise.
20036 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20037 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20039 2017-02-05 Martin Liska <mliska@suse.cz>
20042 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20044 (print_operand_address): Initialize a struct to zero.
20046 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20048 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20049 garbage collector only in textual form.
20051 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20053 * doc/extend.texi (x86 specific memory model extensions for
20054 transactional memory): Simplify a phrase.
20056 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20059 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20060 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20061 (atomic_storedi_1): Likewise.
20063 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20065 PR tree-optimization/79338
20066 * tree-parloops.c (gather_scalar_reductions): Don't call
20067 vect_analyze_loop_form for loop->inner before destroying loop's
20070 2017-02-03 Martin Sebor <msebor@redhat.com>
20072 PR tree-optimization/79327
20073 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20074 when precision has resulted in leading zeros.
20075 (format_integer): Adjust the likely counter to assume an unknown
20076 argument that may be zero is non-zero.
20078 2017-02-03 Jason Merrill <jason@redhat.com>
20081 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20082 avoid copying non-taken branch.
20084 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20086 PR tree-optimization/79340
20087 * tree-vect-loop.c (vectorizable_reduction): Release
20088 vec_defs elements after safe_splicing them into other vectors.
20091 PR tree-optimization/79327
20092 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20093 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20095 (format_integer): Use wide_int_to_tree instead of build_int_cst
20096 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20097 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20098 of shortest and longest sequence.
20100 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20102 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20103 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20105 2017-02-03 Walter Lee <walt@tilera.com>
20108 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20109 after initial stackframe link reg save.
20110 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20112 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20115 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20116 wu for stxssp alternative.
20118 2017-02-03 Martin Sebor <msebor@redhat.com>
20120 PR tree-optimization/79352
20121 * gimple-fold.c (get_range_strlen): Add argument.
20122 (get_range_strlen): Change return type to bool.
20123 (get_maxval_strlen): Pass in a dummy argument.
20124 * gimple-fold.h (get_range_strlen): Change return type to bool.
20125 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20126 * tree.h (array_at_struct_end_p): Add argument.
20127 * tree.c (array_at_struct_end_p): Handle it.
20129 2017-02-03 Martin Liska <mliska@suse.cz>
20132 * multiple_target.c (create_dispatcher_calls): Redirect edge
20133 from a caller of a dispatcher.
20134 (expand_target_clones): Make the clones local.
20135 (ipa_target_clone): Do both target clones and resolvers.
20136 (ipa_dispatcher_calls): Remove the pass.
20137 (pass_dispatcher_calls::gate): Likewise.
20138 (make_pass_dispatcher_calls): Likewise.
20139 * passes.def (pass_target_clone): Put as very first IPA early
20142 2017-02-03 Martin Liska <mliska@suse.cz>
20144 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20145 in case of a function with ifunc attribute.
20147 2017-02-03 Martin Liska <mliska@suse.cz>
20149 * cgraph.c (cgraph_node::dump): Dump function version info.
20150 * symtab.c (symtab_node::dump_base): Add missing new line.
20152 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20154 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20155 (ifcombine_ifandif): Use it.
20157 2017-02-03 Martin Liska <mliska@suse.cz>
20159 * doc/invoke.texi: Document default value for
20160 use-after-scope-direct-emission-threshold.
20162 2017-02-03 Martin Liska <mliska@suse.cz>
20164 PR tree-optimization/79339
20165 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20166 (format_floating): Likewise.
20168 2017-02-03 Martin Liska <mliska@suse.cz>
20171 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20173 (ipa_node_params_t::remove): Likewise.
20174 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20175 initialization from removed ipa_node_params_t::insert.
20176 (ipa_node_params::~ipa_node_params): Move from removed
20177 ipa_node_params_t::release.
20178 * symbol-summary.h (symbol_summary::m_released): New member.
20179 Do not release a summary twice. Do not allow to call finalizer
20180 for types of a summary that live in GGC memory.
20182 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20184 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20187 2017-02-02 Martin Sebor <msebor@redhat.com>
20189 PR middle-end/79275
20190 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20191 (format_string): Tighten up the range of output for non-constant
20192 strings and correct the expected range for wide non-constant strings.
20194 2017-02-02 Martin Sebor <msebor@redhat.com>
20196 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20198 PR middle-end/32003
20199 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20201 (-fdump-tree-@var): Add to index and document how to come up
20202 with pass-specific option and dump file names.
20203 (-fdump-passes): Clarify where to look for output.
20205 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20207 PR middle-end/77445
20208 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20209 statistics of the analyzed path; allow threading for speed when
20210 any of BBs along the path are optimized for speed.
20212 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20214 PR middle-end/78468
20215 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20216 settings of the virtual registers.
20219 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20221 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20222 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20225 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20227 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20228 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20230 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20232 * config/s390/s390.md: Add missing comments with the expanded
20234 * config/s390/vector.md: Likewise.
20235 * config/s390/vx-builtins.md: Likewise.
20237 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20240 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20241 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20242 conditions on a single line.
20244 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20246 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20247 __S390_VX__ to __VX__.
20249 2017-02-01 Andrew Pinski <apinski@cavium.com>
20251 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20252 stmt_info to record_stmt_cost.
20253 (vect_get_known_peeling_cost): Pass stmt_info if known to
20255 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20256 cpu_vector_cost field into
20257 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20258 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20259 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20260 splitting of scalar_stmt_cost and vec_stmt_cost.
20261 (thunderx_vector_cost): Likewise.
20262 (cortexa57_vector_cost): LIkewise.
20263 (exynosm1_vector_cost): Likewise.
20264 (xgene1_vector_cost): Likewise.
20265 (thunderx2t99_vector_cost): Improve after the splitting of the two
20267 (aarch64_builtin_vectorization_cost): Update for the splitting of
20268 scalar_stmt_cost and vec_stmt_cost.
20270 2017-02-01 Torvald Riegel <triegel@redhat.com>
20271 Richard Henderson <rth@redhat.com>
20273 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20274 conditional on existance of a fast atomic load.
20275 * optabs-query.c (can_atomic_load_p): New function.
20276 * optabs-query.h (can_atomic_load_p): Declare it.
20277 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20278 no fast atomic load is available for the particular size of access.
20279 (expand_atomic_compare_and_swap): Likewise.
20280 (expand_atomic_load): Likewise.
20281 (expand_atomic_store): Likewise.
20282 (expand_atomic_fetch_op): Likewise.
20283 * testsuite/lib/target-supports.exp
20284 (check_effective_target_sync_int_128): Remove x86 because it provides
20285 no fast atomic load.
20286 (check_effective_target_sync_int_128_runtime): Likewise.
20288 2017-02-01 Richard Biener <rguenther@suse.de>
20290 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20291 (graphite_transform_loops): Provide opt-info for optimized nests.
20292 * tree-parloop.c (parallelize_loops): Provide opt-info for
20293 parallelized loops.
20295 2017-02-01 Richard Biener <rguenther@suse.de>
20297 PR middle-end/79315
20298 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20299 was not set before.
20301 2017-02-01 Richard Biener <rguenther@suse.de>
20303 PR tree-optimization/71824
20304 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20305 Verify the loops are valid in the merged SESE region.
20306 (scop_detection::can_represent_loop_1): Check analyzing the
20307 evolution of the number of iterations in the region succeeds.
20309 2017-01-31 Ian Lance Taylor <iant@golang.org>
20311 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20312 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20314 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20316 PR preprocessor/79210
20317 * input.c (get_substring_ranges_for_loc): Replace line_width
20318 assertion with error-handling.
20320 2017-01-31 Richard Biener <rguenther@suse.de>
20322 PR tree-optimization/77318
20323 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20324 (create_pw_aff_from_tree): Take loop parameter.
20325 (add_condition_to_pbb): Pass loop of the condition to
20326 create_pw_aff_from_tree.
20328 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20330 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20331 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20333 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20337 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20339 (convert_int_to_float128): Likewise.
20340 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20341 (convert_int_to_float128): Likewise.
20342 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20343 (UNSPEC_IEEE128_CONVERT): Likewise.
20344 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20345 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20346 Use local variables for IBM extended format.
20347 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20348 (fix_trunc<mode>si2_fprs): Likewise.
20349 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20350 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20351 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20352 to know that we can now have integers of all sizes in vector
20354 (fix<uns>_<mode>di2_hw): Likewise.
20355 (float<uns>_<mode>si2_hw): Likewise.
20356 (fix_<mode>si2_hw): Likewise.
20357 (fixuns_<mode>si2_hw): Likewise.
20358 (float<uns>_<mode>di2_hw): Likewise.
20359 (float_<mode>di2_hw): Likewise.
20360 (float_<mode>si2_hw): Likewise.
20361 (floatuns_<mode>di2_hw): Likewise.
20362 (floatuns_<mode>si2_hw): Likewise.
20363 (xscvqp<su>wz_<mode>): Delete, no longer used.
20364 (xscvqp<su>dz_<mode>): Likewise.
20365 (xscv<su>dqp_<mode>): Likewise.
20366 (ieee128_mfvsrd_64bit): Likewise.
20367 (ieee128_mfvsrd_32bit): Likewise.
20368 (ieee128_mfvsrwz): Likewise.
20369 (ieee128_mtvsrw): Likewise.
20370 (ieee128_mtvsrd_64bit): Likewise.
20371 (ieee128_mtvsrd_32bit): Likewise.
20373 2017-01-31 Martin Liska <mliska@suse.cz>
20376 * ipa-prop.c (ipa_free_all_node_params): Call release method
20377 instead of ~sumbol_summary to not to trigger double times
20380 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20382 PR tree-optimization/71691
20383 * bitmap.h (class auto_bitmap): New.
20384 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20385 is_maybe_undefined instead of ssa_undefined_value_p.
20387 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20389 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20390 __S390_ARCH_LEVEL__ to __ARCH__.
20392 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20394 PR tree-optimization/79267
20395 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20396 if should_remove_lhs_p is true.
20398 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20401 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20402 (add_alignment_attribute): New.
20403 (base_type_die): Add alignment attribute.
20404 (subrange_type_die): Likewise.
20405 (modified_type_die): Likewise.
20406 (gen_array_type_die): Likewise.
20407 (gen_descr_array_type_die: Likewise.
20408 (gen_enumeration_type_die): Likewise.
20409 (gen_subprogram_die): Likewise.
20410 (gen_variable_die): Likewise.
20411 (gen_field_die): Likewise.
20412 (gen_ptr_to_mbr_type_die): Likewise.
20413 (gen_struct_or_union_type_die): Likewise.
20414 (gen_subroutine_type_die): Likewise.
20415 (gen_typedef_die): Likewise.
20416 (base_type_cmp): Compare alignment attribute.
20418 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20421 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20422 (setb_unsigned) New pattern for setb with CCUNS.
20423 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20424 subfc./subfe sequence to avoid overflow problems. Generate a
20425 shorter sequence with cmpld/setb for power9.
20426 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20427 for generating subfc. instruction.
20428 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20429 now uses this instruction.
20431 2017-01-30 Ian Lance Taylor <iant@google.com>
20434 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20435 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20437 2017-01-30 Martin Sebor <msebor@redhat.com>
20439 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20440 Move constant to the right of a relational operator.
20441 (get_mpfr_format_length, format_character, format_string): Ditto.
20442 (should_warn_p, maybe_warn): Same.
20444 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20446 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20449 * asan.c (get_translation_unit_decl): Remove function.
20450 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20452 2017-01-30 Martin Liska <mliska@suse.cz>
20454 PR gcov-profile/79259
20455 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20456 -fprofile-generate.
20458 2017-01-30 Martin Liska <mliska@suse.cz>
20461 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20462 Initialize variables with NULL value.
20464 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20467 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20469 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20471 2017-01-30 Richard Biener <rguenther@suse.de>
20473 PR tree-optimization/79276
20474 * tree-vrp.c (process_assert_insertions): Properly adjust common
20475 when removing a duplicate.
20477 * gcc.dg/torture/pr79276.c: New testcase.
20479 2017-01-30 Richard Biener <rguenther@suse.de>
20481 PR tree-optimization/79256
20482 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20483 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20485 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20487 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20490 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20491 ("*r<noxa>sbg_<mode>_sll_bitmask")
20492 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20493 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20494 Use contiguous_bitmask_nowrap_operand.
20496 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20499 * config/rs6000/altivec.h (vec_xl): Revise #define.
20500 (vec_xst): Likewise.
20502 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20504 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20506 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20508 PR rtl-optimization/79194
20509 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20510 traps before call to bypass_conditional_jumps.
20512 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20514 PR tree-optimization/71374
20515 * lra-constraints.c (check_conflict_input_operands): New.
20516 (match_reload): Use it.
20518 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20521 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20522 account to calculate conflict_set.
20524 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20526 PR rtl-optimization/78559
20527 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20528 other_insn in combine.
20530 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20532 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20533 uint16_type_node for BT_UINT16.
20535 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20537 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20538 "RTL Tests" to menu.
20539 (GIMPLE Tests): New node.
20540 (RTL Tests): New node.
20542 2017-01-27 Richard Biener <rguenther@suse.de>
20544 PR tree-optimization/79245
20545 * tree-loop-distribution.c (distribute_loop): Apply cost
20546 modeling also to detected patterns.
20548 2017-01-27 Richard Biener <rguenther@suse.de>
20550 PR tree-optimization/71433
20551 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20552 (compare_assert_loc): New function.
20553 (process_assert_insertions): Sort and optimize assert locations
20554 to remove duplicates and push down identical assertions on
20555 edges to their destination block.
20557 2017-01-27 Richard Biener <rguenther@suse.de>
20559 PR tree-optimization/79244
20560 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20561 out SSA names even if abnormal.
20563 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20565 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20566 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20567 instead of MPFR_RNDN.
20569 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20572 * arm.c (arm_option_override): Don't call build_target_option_node
20573 until after doing all option overrides.
20574 (arm_valid_target_attribute_tree): Likewise.
20576 2017-01-27 Martin Liska <mliska@suse.cz>
20578 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20579 for {cd}tors and C++ {cd}tors.
20581 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20583 * config/s390/s390.md ("*setmem_long_and")
20584 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20586 2017-01-26 Martin Sebor <msebor@redhat.com>
20588 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20591 2017-01-26 Martin Sebor <msebor@redhat.com>
20593 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20594 HAVE_DFmode before using XFmode or DFmode.
20595 (parse_directive): Avoid using the z length modifier to avoid
20596 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20598 PR middle-end/78703
20599 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20600 to accept adjustment as an array.
20601 (get_int_range): New function.
20602 (struct directive): Make width and prec arrays.
20603 (directive::set_width, directive::set_precision): Call get_int_range.
20604 (format_integer, format_floating): Handle width and precision ranges.
20605 (format_string, parse_directive): Same.
20607 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20610 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20611 comdat_type_p set, just clone them, but keep the children in the
20615 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20616 which have direct callers with -fvar-tracking-assignments enabled
20618 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20619 inside of type units.
20621 2017-01-26 Martin Sebor <msebor@redhat.com>
20623 PR middle-end/78703
20624 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20626 (struct format_result): Replace number_chars, number_chars_min,
20627 and number_chars_max with a single member of struct result_range.
20629 (format_result::operator+=): Adjust.
20630 (struct fmtresult): Remove bounded. Handle likely and unlikely
20632 (fmtresult::adjust_for_width_or_precision): New function.
20633 (fmtresult:type_max_digits): New function.
20634 (bytes_remaining): Handle likely and unlikely counters.
20635 (min_bytes_remaining): Remove.
20636 (format_percent): Simplify.
20637 (format_integer, format_floating): Set likely and unlikely counters.
20638 (get_string_length, format_character, format_string): Same.
20639 (format_plain, should_warn_p): New function.
20640 (maybe_warn): Call should_warn_p. Update diagnostic messages
20641 and handle those for all directives, including plain strings.
20642 (format_directive): Handle likely and unlikely counters.
20643 Remove unnecessary quoting from diagnostics. Add an informational
20645 (add_bytes): Remove.
20646 (pass_sprintf_length::compute_format_length): Simplify.
20647 (try_substitute_return_value): Handle likely and unlikely counters.
20649 2017-01-26 Carl Love <cel@us.ibm.com>
20651 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20652 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20654 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20657 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20658 endianess for subregs into account.
20659 * lra-constraints.c (lra_constraints): Do risky transformations
20660 always on the first iteration.
20661 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20662 last_call_used_reg_set.
20663 (process_bb_lives): Define and use last_call_used_reg_set.
20664 * lra.c (lra): Always continue after lra_constraints on the first
20667 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20669 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20671 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20673 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20675 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20676 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20677 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20678 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20679 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20680 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20681 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20682 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20683 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20685 2017-01-26 Marek Polacek <polacek@redhat.com>
20688 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20689 for the third operand.
20691 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20693 PR middle-end/79236
20694 * omp-low.c (struct omp_context): Add simt_stmt field.
20695 (scan_omp_for): Return omp_context *.
20696 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20697 context to the _simt_ SIMD stmt.
20698 (lower_omp_for): For combined SIMD with sibling _simt_
20699 SIMD, make sure to use the same decls in _looptemp_
20700 clauses as in the sibling.
20702 2017-01-26 David Sherwood <david.sherwood@arm.com>
20704 PR middle-end/79212
20705 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20708 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20711 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20712 emit fld b; fld a; if possible.
20714 * brig-builtins.def: Update copyright years.
20715 * config/arm/arm_acle_builtins.def: Update copyright years.
20717 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20720 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20721 constraint instead of o for the stxsd instruction.
20723 2017-01-25 Carl Love <cel@us.ibm.com>
20725 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20726 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20728 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20730 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20732 2017-01-25 Richard Biener <rguenther@suse.de>
20734 PR tree-optimization/69264
20735 * target.def (vector_alignment_reachable): Improve documentation.
20736 * doc/tm.texi: Regenerate.
20737 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20739 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20740 earlier changes with respect to TYPE_USER_ALIGN.
20741 (vector_alignment_reachable_p): Likewise. Improve dumping.
20743 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20746 * config/arm/arm.md (xordi3): Force constant operand into a register
20749 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20751 * doc/invoke.texi (-fstore-merging): Correct default optimization
20752 levels at which it is enabled.
20753 (-O): Move -fstore-merging from list to...
20756 2017-01-25 Richard Biener <rguenther@suse.de>
20759 * omp-expand.c: Include debug.h.
20760 (expand_omp_taskreg): Make sure to generate early debug before
20761 outlining anything from a function.
20762 (expand_omp_target): Likewise.
20763 (grid_expand_target_grid_body): Likewise.
20765 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20768 * asan.c (get_translation_unit_decl): New function.
20769 (asan_add_global): Extract modules file name from globals
20770 TRANSLATION_UNIT_DECL name.
20772 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20775 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20776 for long calls with APCS frame and VFP.
20778 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20780 * cfg.c (original_copy_tables_initialized_p): New function.
20781 * cfg.h (original_copy_tables_initialized_p): New decl.
20782 * cfgrtl.c (relink_block_chain): Guard the call to
20783 free_original_copy_tables with a call to
20784 original_copy_tables_initialized_p.
20785 * cgraph.h (symtab_node::native_rtl_p): New decl.
20786 * cgraphunit.c (symtab_node::native_rtl_p): New function.
20787 (symtab_node::needed_p): Don't assert for early assembly output
20788 for __RTL functions.
20789 (cgraph_node::finalize_function): Set "force_output" for __RTL
20791 (cgraph_node::analyze): Bail out early for __RTL functions.
20792 (analyze_functions): Update assertion to support __RTL functions.
20793 (cgraph_node::expand): Bail out early for __RTL functions.
20794 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
20796 * function.h (struct function): Update comment for field
20798 * gimple-expr.c: Include "tree-pass.h".
20799 (gimple_has_body_p): Return false for __RTL functions.
20800 * Makefile.in (OBJS): Add run-rtl-passes.o.
20801 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
20803 (gcc::pass_manager::get_clean_slate): New accessor.
20804 * passes.c: Include "insn-addr.h".
20805 (should_skip_pass_p): Add logging. Update logic for running
20806 "expand" to be compatible with both __GIMPLE and __RTL. Guard
20807 property-provider override so it is only done for gimple passes.
20809 (skip_pass): New function.
20810 (execute_one_pass): Call skip_pass when skipping passes.
20811 * read-md.c (md_reader::read_char): Support filtering
20812 the input to a subset of line numbers.
20813 (md_reader::md_reader): Initialize fields
20814 m_first_line and m_last_line.
20815 (md_reader::read_file_fragment): New function.
20816 * read-md.h (md_reader::read_file_fragment): New decl.
20817 (md_reader::m_first_line): New field.
20818 (md_reader::m_last_line): New field.
20819 * read-rtl-function.c (function_reader::create_function): Only
20820 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
20821 curr_properties. Set DECL_INITIAL to a dummy block.
20822 (read_rtl_function_body_from_file_range): New function.
20823 * read-rtl-function.h (read_rtl_function_body_from_file_range):
20825 * run-rtl-passes.c: New file.
20826 * run-rtl-passes.h: New file.
20828 2017-01-24 Jeff Law <law@redhat.com>
20830 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
20833 2017-01-24 Bin Cheng <bin.cheng@arm.com>
20835 PR tree-optimization/79159
20836 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
20837 (record_nonwrapping_iv): Improve boundary using above function if no
20838 value range information.
20840 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
20841 Martin Jambor <mjambor@suse.cz>
20843 * brig-builtins.def: New file.
20844 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
20845 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
20846 (DEF_HSAIL_SAT_BUILTIN): Likewise.
20847 (DEF_HSAIL_INTR_BUILTIN): Likewise.
20848 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
20849 * builtin-types.def (BT_INT8): New.
20850 (BT_INT16): Likewise.
20851 (BT_UINT8): Likewise.
20852 (BT_UINT16): Likewise.
20853 (BT_FN_ULONG): Likewise.
20854 (BT_FN_UINT_INT): Likewise.
20855 (BT_FN_UINT_ULONG): Likewise.
20856 (BT_FN_UINT_LONG): Likewise.
20857 (BT_FN_UINT_PTR): Likewise.
20858 (BT_FN_ULONG_PTR): Likewise.
20859 (BT_FN_INT8_FLOAT): Likewise.
20860 (BT_FN_INT16_FLOAT): Likewise.
20861 (BT_FN_UINT32_FLOAT): Likewise.
20862 (BT_FN_UINT16_FLOAT): Likewise.
20863 (BT_FN_UINT8_FLOAT): Likewise.
20864 (BT_FN_UINT64_FLOAT): Likewise.
20865 (BT_FN_UINT16_UINT32): Likewise.
20866 (BT_FN_UINT32_UINT16): Likewise.
20867 (BT_FN_UINT16_UINT16_UINT16): Likewise.
20868 (BT_FN_INT_PTR_INT): Likewise.
20869 (BT_FN_UINT_PTR_UINT): Likewise.
20870 (BT_FN_LONG_PTR_LONG): Likewise.
20871 (BT_FN_ULONG_PTR_ULONG): Likewise.
20872 (BT_FN_VOID_UINT64_UINT64): Likewise.
20873 (BT_FN_UINT8_UINT8_UINT8): Likewise.
20874 (BT_FN_INT8_INT8_INT8): Likewise.
20875 (BT_FN_INT16_INT16_INT16): Likewise.
20876 (BT_FN_INT_INT_INT): Likewise.
20877 (BT_FN_UINT_FLOAT_UINT): Likewise.
20878 (BT_FN_FLOAT_UINT_UINT): Likewise.
20879 (BT_FN_ULONG_UINT_UINT): Likewise.
20880 (BT_FN_ULONG_UINT_PTR): Likewise.
20881 (BT_FN_ULONG_ULONG_ULONG): Likewise.
20882 (BT_FN_UINT_UINT_UINT): Likewise.
20883 (BT_FN_VOID_UINT_PTR): Likewise.
20884 (BT_FN_UINT_UINT_PTR: Likewise.
20885 (BT_FN_UINT32_UINT64_PTR): Likewise.
20886 (BT_FN_INT_INT_UINT_UINT): Likewise.
20887 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
20888 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
20889 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
20890 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
20891 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
20892 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
20893 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
20894 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
20895 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
20896 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
20897 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
20898 * doc/frontends.texi: List BRIG FE.
20899 * doc/install.texi (Testing): Add BRIG tesring requirements.
20900 * doc/invoke.texi (Overall Options): Mention BRIG.
20901 * doc/standards.texi (Standards): Doucment BRIG HSA version.
20903 2017-01-24 Richard Biener <rguenther@suse.de>
20905 PR translation/79208
20906 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
20908 2017-01-24 Martin Jambor <mjambor@suse.cz>
20911 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
20912 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
20913 and known_contexts.
20915 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
20917 PR middle-end/79123
20918 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
20919 casts from signed to unsigned really don't have a range.
20921 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
20923 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
20924 GMP_RNDx for compatiblity.
20926 2017-01-24 Martin Liska <mliska@suse.cz>
20929 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
20930 that would prevent us to call alloca with -1 as argument.
20932 2017-01-24 Jakub Jelinek <jakub@redhat.com>
20934 * dwarf2out.c (output_compilation_unit_header, output_file_names):
20935 Avoid -Wformat-security warning.
20937 2017-01-23 Andrew Pinski <apinski@cavium.com>
20939 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
20942 2017-01-23 Martin Sebor <msebor@redhat.com>
20944 PR middle-end/78703
20945 * gimple-ssa-sprintf.c (warn_level): New global.
20946 (format_integer): Use it here and throughout the rest of the file.
20947 Use the same switch to compute sign as base.
20948 (maybe_warn): New function.
20949 (format_directive): Factor out warnings into maybe_warn.
20950 Add debugging output. Use warn_level.
20951 (add_bytes): Use warn_level.
20952 (pass_sprintf_length::compute_format_length): Add debugging output.
20953 (try_substitute_return_value): Same.
20954 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
20956 PR middle-end/78703
20957 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
20958 (struct fmtresult, format_integer, format_floating): Adjust.
20959 (fmtresult::fmtresult): Set max correctly in two argument ctor.
20960 (get_string_length, format_string,format_directive): Same.
20961 (pass_sprintf_length::compute_format_length): Same.
20962 (try_substitute_return_value): Simplify slightly.
20964 PR middle-end/78703
20965 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
20966 (fmtresult::operator+=): Outlined.
20967 (struct fmtresult): Add ctors.
20968 (struct conversion_spec): Rename...
20969 (struct directive): ...to this. Add and remove data members.
20970 (directive::set_width, directive::set_precision): New functions.
20971 (format_percent): Use fmtresult ctor.
20972 (get_width_and_precision): Remove.
20973 (format_integer): Make naming changes. Avoid computing width and
20975 (format_floating): Same. Adjust indentation.
20976 (format_character, format_none): New functions.
20977 (format_string): Moved character handling to format_character.
20978 (format_directive): Remove arguments, change return type.
20979 (parse_directive): New function.
20980 (pass_sprintf_length::compute_format_length): Move directive
20981 parsing to parse_directive.
20983 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20985 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
20986 (assign_assembler_name_if_needed): ... this.
20987 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
20988 (assign_assembler_name_if_needed): ... this.
20989 (free_lang_data_in_cgraph): Adjust callers.
20990 * cgraphunit.c (cgraph_node::analyze): Likewise.
20991 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
20994 2017-01-23 Richard Biener <rguenther@suse.de>
20996 PR tree-optimization/79088
20997 PR tree-optimization/79188
20998 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
20999 resetting loop bounds after last path deletion. Reset loop
21000 bounds of the target loop, make code match the comments.
21001 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21002 Make sure loops need no fixups.
21004 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21006 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21007 exponent support with double type for first argument.
21008 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21009 type returned by __builtin_vec_extract_sig,
21010 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21011 functions from "vector int" to "vector unsigned int" or from
21012 "vector long long int" to "vector unsigned long long int".
21013 Changed type returned by __builtin_vec_extract_exp,
21014 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21015 functions from "vector int" to "vector unsigned int" or from
21016 "vector long long int" to "vector unsigned long long int".
21017 Changed return type of __builtin_vec_test_data_class,
21018 __builtin_vec_test_data_class_sp, and
21019 __builtin_vec_test_data_class_dp from "vector int" to
21020 "vector bool int" or from "vector long long int" to "vector bool
21021 long long int" and changed second argument type from "unsigned
21022 int" to "int". Added new overloaded function forms "vector float
21023 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21024 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21025 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21026 double, vector unsigned long long int)" and "vector double
21027 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21028 long int)". Changed return type of
21029 __builtin_scalar_test_data_class and
21030 __builtin_scalar_test_data_class_sp and
21031 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21032 int" and changed second argument from "unsigned int" to "int".
21033 Changed type returned by __builtin_scalar_test_neg,
21034 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21035 from "int" to "bool int". Added new overloaded function form
21036 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21037 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21038 exponent double-precision with floating point first argument.
21039 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21040 documentation of scalar_test_data_class, scalar_test_neg,
21041 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21042 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21043 vec_test_data_class built-in functions to reflect refinements in
21044 their type signatures.
21046 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21048 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21050 (aarch64_elf_asm_destructor): Likewise.
21052 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21054 PR rtl-optimization/78634
21055 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21056 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21057 * ifcvt.c (noce_try_cmove): Add missing cost check.
21059 PR rtl-optimization/71724
21060 * combine.c (if_then_else_cond): Look for situations where it is
21061 beneficial to undo the work of one of the recursive calls.
21063 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21065 PR tree-optimization/70754
21066 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21067 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21068 combined stmt before it if not NULL.
21069 (combine_chains): Process refs reversely and compute dominance point
21072 2017-01-23 Martin Liska <mliska@suse.cz>
21074 PR tree-optimization/79196
21075 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21076 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21078 (strlen_optimize_stmt): Call the renamed function.
21080 2017-01-23 Michael Matz <matz@suse.de>
21082 PR tree-optimization/78384
21083 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21085 2017-01-23 Richard Biener <rguenther@suse.de>
21087 PR tree-optimization/79186
21088 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21089 both incoming edges before moving an assert.
21091 2017-01-23 Martin Jambor <mjambor@suse.cz>
21093 * ipa-prop.c (load_from_param_1): Removed.
21094 (load_from_unmodified_param): Bits from load_from_param_1 put back
21096 (load_from_param): Removed.
21097 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21098 with stmt. Reverted back to use of load_from_unmodified_param.
21100 2017-01-23 Martin Jambor <mjambor@suse.cz>
21103 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21104 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21105 field a pointer to garbage collected vector, mark lattices and
21106 ipcp_orig_node with GTY((skip)).
21107 (ipa_get_param_count): Adjust to descriptors being a pointer.
21108 (ipa_get_param): Likewise.
21109 (ipa_get_type): Likewise.
21110 (ipa_get_param_move_cost): Likewise.
21111 (ipa_set_param_used): Likewise.
21112 (ipa_get_controlled_uses): Likewise.
21113 (ipa_set_controlled_uses): Likewise.
21114 (ipa_is_param_used): Likewise.
21115 (ipa_node_params_t): Move into garbage collector. New methods insert
21117 (ipa_node_params_sum): Annotate wth GTY(()).
21118 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21120 (ipa_load_from_parm_agg): Adjust declaration.
21121 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21122 * ipa-profile.c (ipa_profile): Likewise.
21123 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21124 (ipa_populate_param_decls): Make descriptors parameter garbage
21126 (ipa_dump_param): Adjust to descriptors being a pointer.
21127 (ipa_alloc_node_params): Likewise.
21128 (ipa_initialize_node_params): Likewise.
21129 (load_from_param_1): Make descriptors parameter garbage collected.
21130 (load_from_unmodified_param): Likewise.
21131 (load_from_param): Likewise.
21132 (ipa_load_from_parm_agg): Likewise.
21133 (ipa_node_params::~ipa_node_params): Removed.
21134 (ipa_free_all_node_params): Remove call to delete operator.
21135 (ipa_node_params_t::insert): New.
21136 (ipa_node_params_t::remove): Likewise.
21137 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21138 copy known_csts and known_contexts vectors.
21139 (ipa_read_node_info): Adjust to descriptors being a pointer.
21140 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21142 (ipcp_transform_function): Make descriptors variable garbage
21145 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21147 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21148 * config/i386/avx512dqintrin.h: Ditto.
21149 * config/i386/avx512fintrin.h: Ditto.
21150 * config/i386/i386.c: Handle new builtins.
21151 * config/i386/i386-builtin.def: Add new builtins.
21152 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21153 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21155 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21156 Martin Liska <mliska@suse.cz>
21158 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21159 * asan.c (asan_expand_poison_ifn): Support stores and use
21160 appropriate ASAN report function.
21161 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21162 * internal-fn.def (ASAN_POISON_USE): Declare.
21163 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21164 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21165 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21166 ASAN_POISON calls w/o LHS.
21167 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21168 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21169 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21170 * gimplify.c (asan_poison_variables): Add attribute
21171 use_after_scope_memory to variables that really needs to live
21173 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21174 having the attribute.
21176 2017-01-23 Martin Liska <mliska@suse.cz>
21178 * asan.c (create_asan_shadow_var): New function.
21179 (asan_expand_poison_ifn): Likewise.
21180 * asan.h (asan_expand_poison_ifn): New declaration.
21181 * internal-fn.c (expand_ASAN_POISON): Likewise.
21182 * internal-fn.def (ASAN_POISON): New builtin.
21183 * sanopt.c (pass_sanopt::execute): Expand
21184 asan_expand_poison_ifn.
21185 * tree-inline.c (copy_decl_for_dup_finish): Make function
21187 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21188 * tree-ssa.c (is_asan_mark_p): New function.
21189 (execute_update_addresses_taken): Rewrite local variables
21190 (identified just by use-after-scope as addressable) into SSA.
21192 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21194 * doc/install.texi (Specific): opensource.apple.com uses https
21195 now. Remove trailing slash.
21197 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21199 * README.Portability: Remove note on an Irix compatibility issue.
21201 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21203 * gcov.c (INCLUDE_ALGORITHM): Define.
21204 (INCLUDE_VECTOR): Define.
21205 No longer include <vector> and <algorithm> directly.
21207 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21209 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21211 * doc/invoke.texi (Code Gen Options): Ditto.
21213 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21216 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21218 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21220 rtl-optimization/79125
21221 * cprop.c (local_cprop_pass): Handle cases where we make an
21222 unconditional trap.
21224 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21228 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21229 read from, for big endian.
21231 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21233 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21234 register pauth builtins for LP64 only.
21236 2017-01-20 Marek Polacek <polacek@redhat.com>
21239 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21242 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21244 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21246 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21247 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21248 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21250 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21253 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21254 in big-endian mode when they are not a single duplicated value.
21256 2017-01-20 Richard Biener <rguenther@suse.de>
21258 * BASE-VER: Bump to 7.0.1.
21260 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21262 * omp-low.c (omplow_simd_context): New struct. Use it...
21263 (lower_rec_simd_input_clauses): ...here and...
21264 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21265 references to idx, lane, max_vf, is_simt.
21267 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21269 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21272 2017-01-20 Martin Jambor <mjambor@suse.cz>
21274 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21275 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21277 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21278 (GTFILES): Rename hsa.c to hsa-common.c.
21279 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21280 * hsa-dump.c: Likewise.
21281 * hsa-gen.c: Likewise.
21282 * hsa-regalloc.c: Likewise.
21283 * ipa-hsa.c: Likewise.
21284 * omp-expand.c: Likewise.
21285 * omp-low.c: Likewise.
21286 * toplev.c: Likewise.
21288 2017-01-20 Marek Polacek <polacek@redhat.com>
21291 * doc/invoke.texi: Document -Wduplicated-branches.
21292 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21293 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21294 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21295 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21296 return 0 only when not OEP_LEXICOGRAPHIC.
21297 (fold_build_cleanup_point_expr): Use the expression
21298 location when building CLEANUP_POINT_EXPR.
21299 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21300 * tree.c (add_expr): Handle error_mark_node.
21302 2017-01-20 Martin Liska <mliska@suse.cz>
21305 * tree-profile.c (init_ic_make_global_vars): Do not call
21307 (gimple_init_gcov_profiler): Likewise.
21309 2017-01-20 Martin Liska <mliska@suse.cz>
21312 * cgraph.h (maybe_create_reference): Remove argument and
21314 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21316 * ipa-cp.c (create_specialized_node): Likewise.
21317 * symtab.c (symtab_node::maybe_create_reference): Handle
21318 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21320 2017-01-20 Martin Liska <mliska@suse.cz>
21322 * read-rtl-function.c (function_reader::create_function): Use
21323 build_decl instread of build_decl_stat.
21325 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21327 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21328 * config/i386/avx512dqintrin.h: Ditto.
21329 * config/i386/avx512fintrin.h: Ditto.
21330 * config/i386/i386-builtin-types.def: Add new types.
21331 * config/i386/i386.c: Handle new types.
21332 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21333 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21334 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21335 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21336 (__builtin_ia32_kshiftridi): New.
21337 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21339 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21343 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21344 define to rs6000_init_stack_protect_guard.
21345 (rs6000_init_stack_protect_guard): New function.
21347 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21348 Yunqiang Su <yunqiang.su@imgtec.com>
21350 * config.gcc (supported_defaults): Add madd4.
21351 (with_madd4): Add validation.
21352 (all_defaults): Add madd4.
21353 * config/mips/mips.opt (mmadd4): New option.
21354 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21356 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21358 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21359 (ISA_HAS_FUSED_MADD4): Likewise.
21360 * doc/invoke.texi (-mmadd4): Document the new option.
21361 * doc/install.texi (--with-madd4): Document the new option.
21363 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21365 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21366 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21367 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21368 (aarch64_init_pauth_hint_builtins): New.
21369 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21370 (aarch64_expand_builtin): Expand new builtins.
21372 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21374 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21375 * combine-stack-adj.c (no_unhandled_cfa): Handle
21376 REG_CFA_TOGGLE_RA_MANGLE.
21377 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21378 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21379 info for return address signing.
21380 (aarch64_expand_epilogue): Likewise.
21382 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21384 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21385 * config/aarch64/aarch64-protos.h
21386 (aarch64_return_address_signing_enabled): New declaration.
21387 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21389 (aarch64_expand_prologue): Sign return address before it's pushed onto
21391 (aarch64_expand_epilogue): Authenticate return address fetched from
21393 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21394 (aarch64_attributes): New function attributes for "sign-return-address".
21395 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21396 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21397 ("*do_return"): Generate combined instructions according to key index.
21398 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21399 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21401 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21402 * config/aarch64/aarch64.opt (msign-return-address=): New.
21403 * doc/extend.texi (AArch64 Function Attributes): Documents
21404 "sign-return-address=".
21405 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21407 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21409 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21410 overall option summary.
21412 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21414 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21415 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21416 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21417 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21419 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21421 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21422 -mpower9-minmax by default for -mcpu=power9.
21423 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21424 128-bit floating point.
21426 2017-01-20 Alan Modra <amodra@gmail.com>
21428 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21429 optimizing for size.
21431 2017-01-20 Alan Modra <amodra@gmail.com>
21434 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21435 for strcmp and strncmp from corresponding builtin decl.
21437 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21439 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21440 instead of i386/rtems-64.h.
21441 * config/i386/rtems-64.h: Remove.
21443 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21447 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21449 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21451 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21453 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21454 Change int to HOST_WIDE_INT.
21455 * config/aarch64/aarch64-protos.h
21456 (aarch64_simd_gen_const_vector_dup): Likewise.
21457 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21459 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21461 * langhooks-def.h (lhd_type_for_size): New decl.
21462 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21463 * langhooks.c (lhd_type_for_size): New function, taken from
21466 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21468 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21469 define_bypass for CR latency.
21470 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21471 (power9-alu2): Add define_bypass for CR latency.
21473 (power9-mul): Update insn latency.
21474 (power9-mul-compare): Update insn latency, bypass latency and remove
21477 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21479 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21481 * config/aarch64/aarch64.md
21482 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21483 aarch64_nopcrelative_literal_loads.
21484 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21486 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21488 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21489 TARGET_LOONGSON_3A.
21490 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21492 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21495 * config.gcc (supported_defaults): Add lxc1-sxc1.
21496 (with_lxc1_sxc1): Add validation.
21497 (all_defaults): Add lxc1-sxc1.
21498 * config/mips/mips.opt (mlxc1-sxc1): New option.
21499 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21501 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21502 __mips_no_lxc1_sxc1.
21503 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21504 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21505 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21507 2017-01-19 Richard Biener <rguenther@suse.de>
21509 PR tree-optimization/72488
21510 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21511 sure to restore SSA info.
21512 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21514 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21516 PR rtl-optimization/79121
21517 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21518 of the inner type when shifting an extended value.
21520 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21523 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21524 interposable aliases.
21526 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21529 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21530 Use the evmergelohi instruction.
21531 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21532 (mov_sitf_e500_subreg8_2_be): Likewise.
21533 (mov_sitf_e500_subreg12_2_le): Likewise.
21534 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21535 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21536 (mov_sitf_e500_subreg8_2_le): Likewise.
21537 (mov_sitf_e500_subreg12_2_be): Likewise.
21539 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21541 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21542 attribute from vecsimple to vecperm.
21543 (altivec_vbpermq2): Likewise.
21545 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21548 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21550 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21551 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21552 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21553 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21554 case where N arg is SIZE_MAX.
21555 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21556 (cmpstrsi): Add pattern.
21558 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21560 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21561 __builtin_vec_revb builtins.
21562 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21563 built-in functions to support generation of the ISA 3.0 XXBR<x>
21564 vector byte reverse instructions.
21565 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21566 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21567 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21568 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21569 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21570 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21571 (P9V_BUILTIN_VEC_REVB): Likewise.
21572 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21573 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21574 (p9_xxbrq_v16qi): Likewise.
21575 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21576 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21577 (p9_xxbrh_v8hi): Likewise.
21578 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21579 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21580 vec_revb built-in functions.
21582 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21584 PR rtl-optimization/78952
21585 * config/i386/i386.md (any_extract): New code iterator.
21586 (*insvqi_2): Use any_extract for source operand.
21587 (*insvqi_3): Use any_shiftrt for source operand.
21589 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21591 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21593 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21595 2017-01-18 Matthias Klose <doko@ubuntu.com>
21597 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21599 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21601 * config/rs6000/altivec.h (vec_bperm): Change #define.
21602 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21603 (altivec_vbpermq2): New define_insn.
21604 (altivec_vbpermd): Likewise.
21605 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21606 function interface.
21607 (VBPERMD): Likewise.
21608 (VBPERM): New polymorphic function interface.
21609 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21610 Add entries for P9V_BUILTIN_VEC_VBPERM.
21611 * doc/extend.texi: Add interfaces for vec_bperm.
21613 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21615 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21616 first letter of error messages.
21617 (s390_resolve_overloaded_builtin): Likewise.
21618 * config/s390/s390.c (s390_expand_builtin): Likewise.
21619 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21620 (s390_valid_target_attribute_inner_p): Likewise.
21621 * config/s390/s390.md ("tabort"): Likewise.
21623 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21625 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21626 (ISA_AVOID_DIV_HILO): New macro.
21627 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21628 (ISA_HAS_DDIV): Likewise.
21630 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21632 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21634 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21636 * doc/invoke.texi (fabi-version): Spelling fix.
21638 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21641 * doc/invoke.texi (fabi-version): Mention mangling fix for
21644 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21647 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21649 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21652 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21655 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21657 (stack_protect_test): Ditto.
21658 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21659 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21661 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21662 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21663 -mstack-protector-guard-offset=.
21664 (RS/6000 and PowerPC Options): Ditto.
21666 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21668 * config/i386/i386.h (MASK_CLASS_P): New define.
21669 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21670 there are no registers from different register sets also when
21671 mask registers are used. Update function comment.
21672 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21673 to (*k/*r) and (*k/*km) alternatives.
21675 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21677 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21678 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21679 (EH_RETURN_HANDLER_RTX): New define.
21680 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21681 Force frame pointer in EH return functions.
21682 (aarch64_expand_epilogue): Add barrier for eh_return.
21683 (aarch64_final_eh_return_addr): Remove.
21684 (aarch64_eh_return_handler_rtx): New function.
21685 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21687 (aarch64_eh_return_handler_rtx): New prototype.
21689 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21691 * config/rs6000/altivec.h (vec_rlmi): New #define.
21692 (vec_vrlnm): Likewise.
21693 (vec_rlnm): Likewise.
21694 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21695 (UNSPEC_VRLNM): Likewise.
21696 (VIlong): New mode iterator.
21697 (altivec_vrl<VI_char>mi): New define_insn.
21698 (altivec_vrl<VI_char>nm): Likewise.
21699 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21701 (VRLDNM): Likewise.
21702 (RLNM): New polymorphic function entry.
21703 (VRLWMI): New monomorphic function entry.
21704 (VRLDMI): Likewise.
21705 (RLMI): New polymorphic function entry.
21706 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21707 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21708 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21711 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21714 * dwarf2out.c (field_byte_offset): Restore the
21715 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21716 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21717 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21720 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21723 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21725 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21728 * dwarf2out.c (add_data_member_location_attribute): For constant
21729 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21730 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21731 DW_AT_byte_size attributes.
21733 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21735 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21736 after forcing to constant memory when the code model is medium.
21738 2017-01-17 Julia Koval <julia.koval@intel.com>
21741 * config/i386/avx512fintrin.h
21742 (_mm512_i32gather_ps): Change __addr type to void const*.
21743 (_mm512_mask_i32gather_ps): Ditto.
21744 (_mm512_i32gather_pd): Ditto.
21745 (_mm512_mask_i32gather_pd): Ditto.
21746 (_mm512_i64gather_ps): Ditto.
21747 (_mm512_mask_i64gather_ps): Ditto.
21748 (_mm512_i64gather_pd): Ditto.
21749 (_mm512_mask_i64gather_pd): Ditto.
21750 (_mm512_i32gather_epi32): Ditto.
21751 (_mm512_mask_i32gather_epi32): Ditto.
21752 (_mm512_i32gather_epi64): Ditto.
21753 (_mm512_mask_i32gather_epi64): Ditto.
21754 (_mm512_i64gather_epi32): Ditto.
21755 (_mm512_mask_i64gather_epi32): Ditto.
21756 (_mm512_i64gather_epi64): Ditto.
21757 (_mm512_mask_i64gather_epi64): Ditto.
21758 (_mm512_i32scatter_ps): Change __addr type to void*.
21759 (_mm512_mask_i32scatter_ps): Ditto.
21760 (_mm512_i32scatter_pd): Ditto.
21761 (_mm512_mask_i32scatter_pd): Ditto.
21762 (_mm512_i64scatter_ps): Ditto.
21763 (_mm512_mask_i64scatter_ps): Ditto.
21764 (_mm512_i64scatter_pd): Ditto.
21765 (_mm512_mask_i64scatter_pd): Ditto.
21766 (_mm512_i32scatter_epi32): Ditto.
21767 (_mm512_mask_i32scatter_epi32): Ditto.
21768 (_mm512_i32scatter_epi64): Ditto.
21769 (_mm512_mask_i32scatter_epi64): Ditto.
21770 (_mm512_i64scatter_epi32): Ditto.
21771 (_mm512_mask_i64scatter_epi32): Ditto.
21772 (_mm512_i64scatter_epi64): Ditto.
21773 (_mm512_mask_i64scatter_epi64): Ditto.
21774 * config/i386/avx512pfintrin.h
21775 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21776 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21777 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21778 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21779 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21780 (_mm512_prefetch_i32scatter_ps): Ditto.
21781 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
21782 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
21783 (_mm512_prefetch_i64scatter_pd): Ditto.
21784 (_mm512_prefetch_i64scatter_ps): Ditto.
21785 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
21786 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
21787 * config/i386/avx512vlintrin.h
21788 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
21789 (_mm_mmask_i32gather_ps): Ditto.
21790 (_mm256_mmask_i32gather_pd): Ditto.
21791 (_mm_mmask_i32gather_pd): Ditto.
21792 (_mm256_mmask_i64gather_ps): Ditto.
21793 (_mm_mmask_i64gather_ps): Ditto.
21794 (_mm256_mmask_i64gather_pd): Ditto.
21795 (_mm_mmask_i64gather_pd): Ditto.
21796 (_mm256_mmask_i32gather_epi32): Ditto.
21797 (_mm_mmask_i32gather_epi32): Ditto.
21798 (_mm256_mmask_i32gather_epi64): Ditto.
21799 (_mm_mmask_i32gather_epi64): Ditto.
21800 (_mm256_mmask_i64gather_epi32): Ditto.
21801 (_mm_mmask_i64gather_epi32): Ditto.
21802 (_mm256_mmask_i64gather_epi64): Ditto.
21803 (_mm_mmask_i64gather_epi64): Ditto.
21804 (_mm256_i32scatter_ps): Change __addr type to void*.
21805 (_mm256_mask_i32scatter_ps): Ditto.
21806 (_mm_i32scatter_ps): Ditto.
21807 (_mm_mask_i32scatter_ps): Ditto.
21808 (_mm256_i32scatter_pd): Ditto.
21809 (_mm256_mask_i32scatter_pd): Ditto.
21810 (_mm_i32scatter_pd): Ditto.
21811 (_mm_mask_i32scatter_pd): Ditto.
21812 (_mm256_i64scatter_ps): Ditto.
21813 (_mm256_mask_i64scatter_ps): Ditto.
21814 (_mm_i64scatter_ps): Ditto.
21815 (_mm_mask_i64scatter_ps): Ditto.
21816 (_mm256_i64scatter_pd): Ditto.
21817 (_mm256_mask_i64scatter_pd): Ditto.
21818 (_mm_i64scatter_pd): Ditto.
21819 (_mm_mask_i64scatter_pd): Ditto.
21820 (_mm256_i32scatter_epi32): Ditto.
21821 (_mm256_mask_i32scatter_epi32): Ditto.
21822 (_mm_i32scatter_epi32): Ditto.
21823 (_mm_mask_i32scatter_epi32): Ditto.
21824 (_mm256_i32scatter_epi64): Ditto.
21825 (_mm256_mask_i32scatter_epi64): Ditto.
21826 (_mm_i32scatter_epi64): Ditto.
21827 (_mm_mask_i32scatter_epi64): Ditto.
21828 (_mm256_i64scatter_epi32): Ditto.
21829 (_mm256_mask_i64scatter_epi32): Ditto.
21830 (_mm_i64scatter_epi32): Ditto.
21831 (_mm_mask_i64scatter_epi32): Ditto.
21832 (_mm256_i64scatter_epi64): Ditto.
21833 (_mm256_mask_i64scatter_epi64): Ditto.
21834 (_mm_i64scatter_epi64): Ditto.
21835 (_mm_mask_i64scatter_epi64): Ditto.
21836 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
21837 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
21838 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
21839 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
21840 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
21841 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
21842 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
21843 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
21844 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
21845 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
21846 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
21847 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
21848 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
21849 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
21850 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
21851 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
21852 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
21853 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
21854 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
21855 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
21856 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
21857 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
21858 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
21859 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
21860 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
21861 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
21862 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
21863 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
21864 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
21865 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
21866 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
21867 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
21868 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
21869 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
21870 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
21871 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
21872 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
21873 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
21874 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
21875 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
21876 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
21877 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
21878 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
21879 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
21880 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
21881 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
21882 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
21883 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
21884 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
21885 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
21886 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
21887 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
21888 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
21889 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
21890 definitions accordingly.
21892 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
21893 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21896 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
21899 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
21902 * ira-conflicts.c (ira_build_conflicts): Update total conflict
21903 hard regs for inner regno.
21905 2017-01-17 Martin Liska <mliska@suse.cz>
21908 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
21909 assumption and add comment.
21911 2017-01-17 Nathan Sidwell <nathan@acm.org>
21913 * ipa-visibility.c (localize_node): New function, broken out of ...
21914 (function_and_variable_visibility): ... here. Call it.
21916 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21918 PR middle-end/77445
21919 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
21920 correctly set frequency of oudgoing edge.
21921 (duplicate_thread_path): Fix profile updating.
21923 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21926 * configure.ac: Add GCC_BASE_VER.
21927 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
21928 version from BASE-VER file.
21929 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
21930 (gcc.o): Depend on $(BASEVER).
21931 * common.opt (dumpfullversion): New option.
21932 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
21933 * doc/invoke.texi: Document -dumpfullversion.
21934 * doc/install.texi: Document --with-gcc-major-version-only.
21935 * configure: Regenerated.
21937 2017-01-17 Richard Biener <rguenther@suse.de>
21939 PR tree-optimization/71433
21940 * tree-vrp.c (register_new_assert_for): Merge same asserts
21941 on all incoming edges.
21942 (process_assert_insertions_for): Handle insertions at the
21945 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
21947 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
21948 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
21950 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
21953 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
21956 2017-01-17 Alan Modra <amodra@gmail.com>
21959 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
21960 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
21961 symbolic stack limit when pic.
21963 2017-01-16 Martin Sebor <msebor@redhat.com>
21965 PR tree-optimization/78608
21966 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
21968 2017-01-16 Jeff Law <law@redhat.com>
21971 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
21972 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
21973 for several include directories that may be relative to sysroot.
21974 * config/i386/x-mingw32 (gplus_includedir): Define.
21975 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
21976 (native_system_includedir): Likewise.
21977 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
21978 override if TARGET_SYSTEM_ROOT is defined.
21979 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
21981 PR tree-optimization/79090
21982 PR tree-optimization/33562
21983 PR tree-optimization/61912
21984 PR tree-optimization/77485
21985 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
21986 and computed trims into the dump file.
21988 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21990 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
21992 2017-01-16 Jakub Jelinek <jakub@redhat.com>
21995 * gimplify.c (gimplify_init_constructor): If want_value and
21996 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22000 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22001 sequence. Formatting fixes.
22002 (doloop_optimize): Formatting fixes.
22005 * gcc.c (debug_level_greater_than_spec_func): New function.
22006 (static_spec_functions): Add debug-level-gt spec function.
22007 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22009 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22010 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22011 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22012 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22013 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22014 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22016 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22018 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22019 QImode fixups to general and mask registers only.
22021 2017-01-16 Carl Love <cel@us.ibm.com>
22023 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22024 for built-in functions
22025 vector signed char vec_nabs (vector signed char)
22026 vector signed short vec_nabs (vector signed short)
22027 vector signed int vec_nabs (vector signed int)
22028 vector signed long long vec_nabs (vector signed long long)
22029 vector float vec_nabs (vector float)
22030 vector double vec_nabs (vector double)
22031 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22033 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22034 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22035 * doc/extend.texi: Update the documentation file for the new built-in
22038 2017-01-16 Martin Sebor <msebor@redhat.com>
22040 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22043 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22045 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22046 UNSPEC_VSX__XXSPLTD to require special splat handling.
22048 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22051 * system.h: Poison strndup.
22053 2017-01-16 Alan Modra <amodra@gmail.com>
22056 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22059 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22061 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22063 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22065 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22066 call recog here. Assert that INSN_CODE (insn) is non-negative.
22068 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22071 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22073 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22074 in the currently scheduled RTL fragment.
22076 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22078 PR rtl-optimization/78751
22079 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22082 2017-01-14 Jeff Law <law@redhat.com>
22084 PR tree-optimization/79090
22085 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22086 variable length stores.
22087 (compute_trims): Delete dead assignment to *trim_tail.
22088 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22091 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22093 PR rtl-optimization/78626
22094 PR rtl-optimization/78727
22095 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22096 of a block, and split such blocks after everything else is finished.
22098 2017-01-14 Alan Modra <amodra@gmail.com>
22101 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22102 target legitimate_combined_insn.
22103 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22104 (rs6000_legitimate_combined_insn): New function.
22105 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22107 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22108 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22109 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22111 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22113 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22115 2017-01-13 Jeff Law <law@redhat.com>
22117 PR tree-optimization/33562
22118 PR tree-optimization/61912
22119 PR tree-optimization/77485
22120 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22122 (delete_dead_assignment): Likewise.
22123 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22124 statement to delete_dead_call and delete_dead_assignment.
22126 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22129 * substring-locations.c (format_warning_va): Strengthen case 1 so
22130 that both endpoints of the substring must be within the format
22131 range for just the substring to be printed.
22133 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22135 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22136 * config/i386/i386.c (ix86_target_string): Add missing options
22137 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22138 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22139 flags_other and ix86_target_other to flags2_other. Display unknown
22141 (ix86_valid_target_attribute_inner_p): Add missing options and
22142 reorder options by implied ISAs, as in ix86_target_string.
22144 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22146 * hash-table.h (hash_table::too_empty_p): New function.
22147 (hash_table::expand): Use it.
22148 (hash_table::traverse): Likewise.
22149 (hash_table::empty_slot): Use sizeof (value_type) instead of
22150 sizeof (PTR) to convert bytes to elements. Shrink the table
22151 if the current size is excessive for the current number of
22154 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22156 * ira-costs.c (record_reg_classes): Break from the inner loop
22157 early once alt_fail is known to be true. Update outer loop
22158 handling accordingly.
22160 2017-01-13 Jeff Law <law@redhat.com>
22162 * tree-ssa-dse.c (decrement_count): New function.
22163 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22164 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22165 when we know the partially dead statement is a mem* function.
22167 PR tree-optimization/61912
22168 PR tree-optimization/77485
22169 * tree-ssa-dse.c: Include expr.h.
22170 (maybe_trim_constructor_store): New function.
22171 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22173 PR tree-optimization/33562
22174 PR tree-optimization/61912
22175 PR tree-optimization/77485
22176 * doc/invoke.texi: Document new dse-max-object-size param.
22177 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22178 * tree-ssa-dse.c: Include params.h.
22179 (dse_store_status): New enum.
22180 (initialize_ao_ref_for_dse): New, partially extracted from
22182 (valid_ao_ref_for_dse, normalize_ref): New.
22183 (setup_live_bytes_from_ref, compute_trims): Likewise.
22184 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22185 (maybe_trim_partially_dead_store): Likewise.
22186 (maybe_trim_complex_store): Likewise.
22187 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22188 Track what bytes live from the original store. Return tri-state
22189 for dead, partially dead or live.
22190 (dse_dom_walker): Add constructor, destructor and new private members.
22191 (delete_dead_call, delete_dead_assignment): New extracted from
22193 (dse_optimize_stmt): Make a member of dse_dom_walker.
22194 Use initialize_ao_ref_for_dse.
22196 PR tree-optimization/33562
22197 PR tree-optimization/61912
22198 PR tree-optimization/77485
22199 * sbitmap.h (bitmap_count_bits): Prototype.
22200 (bitmap_clear_range, bitmap_set_range): Likewise.
22201 * sbitmap.c (bitmap_clear_range): New function.
22202 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22204 2017-01-13 Martin Liska <mliska@suse.cz>
22207 * function.c (set_cfun): Add new argument force.
22208 * function.h (set_cfun): Likewise.
22209 * ipa-inline-transform.c (inline_call): Use the function when
22210 strict alising from is dropped for function we inline to.
22212 2017-01-13 Richard Biener <rguenther@suse.de>
22214 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22215 for dumping GIMPLE INTEGER_CSTs.
22217 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22219 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22220 to 201112L since C++17.
22222 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22225 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22226 if -fsanitize=kernel-address is present.
22228 2017-01-13 Richard Biener <rguenther@suse.de>
22230 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22231 as _Literal ( type ) number in case usual suffixes do not
22232 preserve all information.
22234 2017-01-13 Richard Biener <rguenther@suse.de>
22236 PR tree-optimization/77283
22237 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22238 and ssa-iterators.h.
22239 (is_feasible_trace): Implement a cost model based on joiner
22242 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22245 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22246 char or short to __float128/_Float128 directly.
22248 2017-01-12 Martin Sebor <msebor@redhat.com>
22250 to -Wformat-overflow.
22251 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22252 (min_bytes_remaining): Same.
22253 (get_string_length): Same.
22254 (format_string): Same.
22255 (format_directive): Same.
22257 (pass_sprintf_length::handle_gimple_call): Same.
22259 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22261 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22262 info.nowrite calls with no lhs that can't throw. Return bool
22263 whether gsi_remove has been called or not.
22264 (pass_sprintf_length::handle_gimple_call): Return bool whether
22265 try_substitute_return_value called gsi_remove. Formatting fix.
22266 (pass_sprintf_length::execute): Don't use gsi_remove if
22267 handle_gimple_call returned true.
22270 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22271 be removed due to side-effects, don't remove following barrier nor
22272 turn the successor edge into fallthru edge.
22274 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22277 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22278 element-reversing loads and stores as not swappable.
22280 2017-01-12 Nathan Sidwell <nathan@acm.org>
22281 Nicolai Stange <nicstange@gmail.com>
22283 * combine.c (try_combine): Don't ignore result of overlap checking
22284 loop. Combine overlap & asm check into single loop.
22286 2017-01-12 Richard Biener <rguenther@suse.de>
22288 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22289 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22291 2017-01-12 Richard Biener <rguenther@suse.de>
22293 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22294 and TS_TARGET_OPTION directly derive from TS_BASE.
22295 * tree-core.h (tree_optimization_option): Derive from tree_base.
22296 (tree_target_option): Likewise.
22298 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22300 * config/i386/i386.c (memory_address_length): Increase len
22301 only when rip_relative_addr_p returns false.
22303 2017-01-11 Julia Koval <julia.koval@intel.com>
22305 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22306 (OPTION_MASK_ISA_SGX_SET): New.
22307 (ix86_handle_option): Handle OPT_msgx.
22308 * config.gcc: Added sgxintrin.h.
22309 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22310 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22311 * config/i386/i386.c (ix86_target_string): Add -msgx.
22313 (ix86_option_override_internal): Handle new options.
22314 (ix86_valid_target_attribute_inner_p): Add sgx.
22315 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22316 * config/i386/i386.opt: Add msgx.
22317 * config/i386/sgxintrin.h: New file.
22318 * config/i386/x86intrin.h: Add sgxintrin.h.
22320 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22323 * fold-const.c (maybe_nonzero_address): Return 1 for function
22325 (tree_single_nonzero_warnv_p): Don't handle function local objects
22329 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22332 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22335 * opts.c: Include "spellcheck.h"
22336 (struct string_fragment): New struct.
22337 (struct edit_distance_traits<const string_fragment &>): New
22339 (get_closest_sanitizer_option): New function.
22340 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22342 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22344 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22346 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22347 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22348 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22349 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22350 for initial die_offset if dwarf_split_debug_info.
22351 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22352 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22354 (output_skeleton_debug_sections): Formatting fix. Use
22355 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22356 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22358 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22360 * config/arm/cortex-a53.md: Add bypasses for
22361 cortex_a53_r2f_cvt.
22362 (cortex_a53_r2f): Only use for transfers.
22363 (cortex_a53_f2r): Likewise.
22364 (cortex_a53_r2f_cvt): Add reservation for conversions.
22365 (cortex_a53_f2r_cvt): Likewise.
22367 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22369 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22370 to all inlined functions, change static to extern.
22372 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22375 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22377 (arm_assemble_integer): Likewise.
22379 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22381 * config.gcc: Use new awk script to check CPU, FPU and architecture
22382 parameters for --with-... options.
22383 * config/arm/parsecpu.awk: New file
22384 * config/arm/arm-cpus.in: New file.
22385 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22387 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22389 * config/arm/t-arm: Update dependency rules.
22390 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22391 of processing .def files.
22392 * config/arm/genopt.sh: Deleted.
22393 * config/arm/gentune.sh: Deleted.
22394 * config/arm/arm-cores.def: Deleted.
22395 * config/arm/arm-arches.def: Deleted.
22396 * config/arm/arm-fpus.def: Deleted.
22397 * config/arm/arm-tune.md: Regenerated.
22398 * config/arm/arm-tables.opt: Regenerated.
22399 * config/arm/arm-cpu.h: New generated file.
22400 * config/arm/arm-cpu-data.h: New generated file.
22401 * config/arm/arm-cpu-cdata.h: New generated file.
22403 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22406 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22408 (input_varpool_node): Unpack dynamically_initialized bit.
22410 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22412 PR rtl-optimization/79032
22413 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22414 the alignment of the adjusted memory reference against that of MODE,
22415 instead of the alignment of the original memory reference.
22417 2017-01-11 Martin Jambor <mjambor@suse.cz>
22419 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22421 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22422 decorated functions.
22424 2017-01-11 Richard Biener <rguenther@suse.de>
22426 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22427 set range/nonnull info for PHI results. Do not set it on
22428 stmts marked for removal.
22430 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22432 * expr.c (store_field): In the bitfield case, fetch the return value
22433 from the registers before applying a single big-endian adjustment.
22434 Always do a final load for a BLKmode value not larger than a word.
22436 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22439 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22440 that we correctly handle column numbers greater than
22441 LINE_MAP_MAX_COLUMN_NUMBER.
22443 2017-01-10 Martin Sebor <msebor@redhat.com>
22445 PR middle-end/78245
22446 * gimple-ssa-sprintf.c (get_destination_size): Call
22447 {init,fini}object_sizes.
22448 * tree-object-size.c (addr_object_size): Adjust.
22449 (pass_through_call): Adjust.
22450 (pass_object_sizes::execute): Adjust.
22451 * tree-object-size.h (fini_object_sizes): Declare.
22453 2017-01-10 Martin Sebor <msebor@redhat.com>
22455 PR tree-optimization/78775
22456 * builtins.c (get_size_range): Move...
22457 * calls.c: ...to here.
22458 (alloc_max_size): Accept zero argument.
22459 (operand_signed_p): Remove.
22460 (maybe_warn_alloc_args_overflow): Call get_size_range.
22461 * calls.h (get_size_range): Declare.
22463 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22465 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22466 from TI's devices.csv file as of September 2016.
22467 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22469 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22471 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22472 * doc/invoke.texi: Likewise.
22473 * doc/md.texi: Likewise.
22474 * doc/objc.texi: Likewise.
22476 2017-01-10 Joshua Conner <joshconner@google.com>
22478 * config/arm/fuchsia-elf.h: New file.
22479 * config/fuchsia.h: New file.
22480 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22481 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22483 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22485 2016-01-10 Richard Biener <rguenther@suse.de>
22487 PR tree-optimization/79034
22488 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22489 Propagate out degenerate PHIs in the joiner.
22491 2017-01-10 Martin Liska <mliska@suse.cz>
22493 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22494 (sort_congruence_classes_by_decl_uid): Likewise.
22495 (sort_congruence_class_groups_by_decl_uid): Likewise.
22496 (sem_item_optimizer::merge_classes): Sort class, groups in these
22497 classes and members in the groups by DECL_UID of declarations.
22498 This would make merge operations stable.
22500 2017-01-10 Martin Liska <mliska@suse.cz>
22502 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22503 usage of m_classes_vec.
22504 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22505 (sem_item_optimizer::get_group_by_hash): Likewise.
22506 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22507 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22508 (sem_item_optimizer::verify_classes): Likewise.
22509 (sem_item_optimizer::process_cong_reduction): Likewise.
22510 (sem_item_optimizer::dump_cong_classes): Likewise.
22511 (sem_item_optimizer::merge_classes): Likewise.
22512 * ipa-icf.h (congruence_class_hash): Rename from
22513 congruence_class_group_hash. Remove declaration of m_classes_vec.
22515 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22517 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22518 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22519 * config.gcc: Add avx512vpopcntdqintrin.h.
22520 * config/i386/avx512vpopcntdqintrin.h: New.
22521 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22522 * config/i386/i386-builtin-types.def: Add new types.
22523 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22524 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22525 __builtin_ia32_vpopcountq_v8di_mask): New.
22526 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22527 __AVX512VPOPCNTDQ__.
22528 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22529 (PTA_AVX512VPOPCNTDQ): Define.
22530 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22531 TARGET_AVX512VPOPCNTDQ_P): Define.
22532 * config/i386/i386.opt: Add mavx512vpopcntdq.
22533 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22534 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22536 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22538 PR middle-end/77484
22539 * predict.def (PRED_CALL): Set to 67.
22541 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22543 * expr.c (store_field): In the bitfield case, if the value comes from
22544 a function call and is of an aggregate type returned in registers, do
22545 not modify the field mode; extract the value in all cases if the mode
22546 is BLKmode and the size is not larger than a word.
22548 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22551 * config/i386/cpuid.h: Fix undefined behavior.
22553 2017-01-04 Jeff Law <law@redhat.com>
22555 PR tree-optimization/79007
22556 PR tree-optimization/67955
22557 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22558 conservative for pt.null when flag_non_call_exceptions is on.
22560 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22562 PR translation/79019
22563 PR translation/79020
22564 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22565 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22566 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22568 * config/avr/avr.opt (maccumulate-args): Likewise.
22569 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22570 * common.opt (freport-bug): Likewise.
22571 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22572 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22573 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22574 translatable string.
22575 * config/i386/i386.c (function_value_32): Likewise.
22576 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22577 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22579 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22580 * common/config/msp430/msp430-common.c (msp430_handle_option):
22582 * symtab.c (symtab_node::verify_base): Likewise.
22583 * opts.c (set_debug_level): Likewise.
22584 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22585 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22586 missing whitespace to translatable strings.
22587 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22588 * config/sh/superh.h: Likewise.
22589 * config/i386/xopintrin.h: Likewise.
22590 * config/i386/znver1.md: Likewise.
22591 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22592 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22593 * double-int.h (struct double_int): Likewise.
22594 * double-int.c (div_and_round_double): Likewise.
22595 * wide-int.cc: Likewise.
22596 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22597 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22598 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22599 (crossjumps_occurred): ... this.
22600 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22603 PR tree-optimization/78899
22604 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22605 returning bool return struct loop *, NULL for failure and the new
22607 (versionable_outer_loop_p): Don't version outer loop if it has
22608 dont_vectorized bit set.
22609 (tree_if_conversion): When versioning outer loop, ensure
22610 tree_if_conversion is performed also on the inner loop of the
22611 non-vectorizable outer loop copy.
22612 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22613 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22614 prevent vectorization of it.
22615 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22616 the outer loop vectorization of the non-scalar version is attempted
22617 before vectorization of the inner loop in scalar version. If
22618 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22619 vectorization of its inner loop.
22620 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22621 has 2 inner loops, rename also on edges from bb whose single pred
22622 is outer_loop->header. Fix typo in function comment.
22624 2017-01-09 Martin Sebor <msebor@redhat.com>
22627 * asan.c (asan_emit_stack_protection): Increase local buffer size
22628 to avoid snprintf truncation warning.
22630 2017-01-09 Andrew Pinski <apinski@cavium.com>
22632 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22633 to reference thunderx2t99 for the tuning structure
22634 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22636 (thunderx2t99_extra_costs): This.
22637 * config/aarch64/aarch64-tune.md: Regenerate.
22638 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22639 (vulcan_addrcost_table): This.
22640 (vulcan_regmove_cost): Rename to ...
22641 (thunderx2t99_regmove_cost): This.
22642 (vulcan_vector_cost): Rename to ...
22643 (thunderx2t99_vector_cost): this.
22644 (vulcan_branch_cost): Rename to ...
22645 (thunderx2t99_branch_cost): This.
22646 (vulcan_tunings): Rename to ...
22647 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22648 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22650 2017-01-09 Martin Jambor <mjambor@suse.cz>
22654 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22655 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22656 (propagate_vr_accross_jump_function): Use the above function for all
22657 value range computations for pass-through jump functions and type
22658 converasion from explicit value range values.
22659 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22660 parameters from TYPE_ARG_TYPES.
22661 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22662 (ipa_write_node_info): Stream type of the actual argument.
22663 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22665 2017-01-09 Martin Liska <mliska@suse.cz>
22668 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22669 (lookup_compiler): Do not show error message with have_E.
22671 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22673 PR tree-optimization/78938
22674 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22675 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22676 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22677 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22680 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22682 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22685 2017-01-09 Richard Biener <rguenther@suse.de>
22687 PR tree-optimization/78997
22688 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22689 name condition properly.
22691 2017-01-09 Richard Biener <rguenther@suse.de>
22694 * dwarf2out.c (is_cxx): New overload with context.
22695 (is_naming_typedef_decl): Use it.
22697 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22699 * invoke.texi (Option Summary): Correct spacing in option lists
22700 and add line breaks to fix over-long lines.
22702 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22704 PR middle-end/17660
22706 * extend.texi (Common Variable Attributes): Add xref to GCC
22707 Internals manual to explain mode attribute keywords.
22709 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22712 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22713 and Preprocessor Options.
22714 (Options for Linking): Document -pthread here....
22715 (RS/6000 and PowerPC Options): ...not here.
22716 (Solaris 2 Options): ...or here.
22717 * doc/cppopts.texi: Document -pthread.
22719 2017-01-08 Martin Sebor <msebor@redhat.com>
22721 PR middle-end/77708
22722 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22723 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22724 New member functions.
22725 (format_directive): Used them.
22727 (pass_sprintf_length::handle_gimple_call): Same.
22728 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22729 to avoid truncation for any argument.
22730 (extract_affine_mul): Same.
22731 * tree.c (get_file_function_name): Same.
22733 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22735 PR middle-end/77484
22736 * predict.def (PRED_INDIR_CALL): Set to 86.
22738 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22740 PR preprocessor/54124
22741 * doc/cppopts.texi: Reformat -d subtable to list the full name
22742 of the options. Add cross-reference to the docs for the general
22743 compiler -d options.
22744 * doc/invoke.texi (Developer Options): Add cross-reference to the
22745 preprocessor-specific -d option documentation.
22747 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22749 PR preprocessor/13498
22750 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22751 redudant material, and reflect new command-line options.
22752 (System Headers): Likewise.
22754 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22756 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22757 -isystem, and -idirafter. Copy-edit.
22758 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22759 default for -ftrack-macro-expansion. Delete obsolete and
22760 badly-formatted implementation details about -fdebug-cpp output.
22761 * doc/cppwarnopts.texi: Copy-edit.
22763 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22766 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22767 that the transition from a max line width >= 1<<10 to narrower
22768 lines works correctly.
22770 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22772 * doc/options.texi (PerFunction): New.
22773 * opt-functions.awk (switch_flags): Map both Optimization and
22774 PerFunction to CL_OPTIMIZATION.
22775 * opth-gen.awk: Test for PerFunction flag along with
22777 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22778 it only when the latter is present. Skip those that don't in
22779 the hash function generator.
22780 * common.opt (fvar-tracking): Mark as PerFunction instead of
22782 (fvar-tracking-assignments): Likewise.
22783 (fvar-tracking-assignments-toggle): Likewise.
22784 (fvar-tracking-uninit): Likewise.
22786 2017-01-07 Jakub Jelinek <jakub@redhat.com>
22788 PR translation/79018
22789 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
22792 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
22795 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
22796 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
22797 TARGET_LONG_JUMP_TABLE_OFFSETS.
22798 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
22799 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
22800 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
22801 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
22802 * config/m68k/m68k.md (tablejump expander): Likewise.
22803 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
22804 TARGET_LONG_JUMP_TABLE_OFFSETS.
22805 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
22806 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
22808 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
22809 David Holsgrove <david.holsgrove@xilinx.com>
22811 * common/config/microblaze/microblaze-common.c
22812 (TARGET_EXCEPT_UNWIND_INFO): Remove.
22813 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
22815 * config/microblaze/microblaze.c (microblaze_must_save_register)
22816 (microblaze_expand_epilogue, microblaze_return_addr): Handle
22818 (microblaze_eh_return): New function.
22819 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
22820 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
22821 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
22822 * config/microblaze/microblaze.md (eh_return): New pattern.
22824 2017-01-06 Jakub Jelinek <jakub@redhat.com>
22826 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
22827 GCC_DIAGNOSTIC_STRINGIFY): Define.
22829 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
22831 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22833 * config/arm/arm.md (<mcrr>): New.
22835 * config/arm/arm.c (arm_arch5te): New.
22836 (arm_option_override): Set arm_arch5te.
22837 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
22839 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
22840 (arm_mcrr_qualifiers): ... this. New.
22841 (MRRC_QUALIFIERS): Define to...
22842 (arm_mrrc_qualifiers): ... this. New.
22843 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
22845 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
22846 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
22847 (MRRCI, mrrc, MRRC): New.
22848 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
22849 VUNSPEC_MRRC2): New.
22851 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22853 * config/arm/arm.md (<mcr>): New.
22855 * config/arm/arm.c (arm_coproc_builtin_available): Add
22856 support for mcr, mrc, mcr2 and mrc2.
22857 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
22858 (arm_mcr_qualifiers): ... this. New.
22859 (MRC_QUALIFIERS): Define to ...
22860 (arm_mrc_qualifiers): ... this. New.
22861 (MCR_QUALIFIERS): Define to ...
22862 (arm_mcr_qualifiers): ... this. New.
22863 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
22865 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
22866 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
22867 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
22868 VUNSPEC_MRC2): New.
22870 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22872 * config/arm/arm.md (*ldc): New.
22876 * config/arm/arm.c (arm_coproc_builtin_available): Add
22877 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
22878 (arm_coproc_ldc_stc_legitimate_address): New.
22879 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
22880 'qualifier_const_pointer'.
22881 (LDC_QUALIFIERS): Define to...
22882 (arm_ldc_qualifiers): ... this. New.
22883 (STC_QUALIFIERS): Define to...
22884 (arm_stc_qualifiers): ... this. New.
22885 * config/arm/arm-protos.h
22886 (arm_coproc_ldc_stc_legitimate_address): New.
22887 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
22888 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
22889 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
22890 stc2, stcl, stc2l): New.
22891 * config/arm/constraints.md (Uz): New.
22892 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
22893 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
22894 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
22895 VUNSPEC_STC2L): New.
22897 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22899 * config/arm/arm.md (<cdp>): New.
22900 * config/arm/arm.c (neon_const_bounds): Rename this ...
22901 (arm_const_bounds): ... this.
22902 (arm_coproc_builtin_available): New.
22903 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
22904 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
22905 (CDP_QUALIFIERS): Define to...
22906 (arm_cdp_qualifiers): ... this. New.
22908 (arm_expand_builtin_args): Add case for 6 arguments.
22909 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
22910 (arm_const_bounds): ... this.
22911 (arm_coproc_builtin_available): New.
22912 * config/arm/arm_acle.h (__arm_cdp): New.
22914 * config/arm/arm_acle_builtins.def (cdp): New.
22916 * config/arm/iterators.md (CDPI,CDP,cdp): New.
22917 * config/arm/neon.md: Rename all 'neon_const_bounds' to
22918 'arm_const_bounds'.
22919 * config/arm/types.md (coproc): New.
22920 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
22921 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
22922 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
22923 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
22925 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22927 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
22928 (UBINOP_QUALIFIERS): New.
22930 (acle_builtin_data): New. Change comment.
22931 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
22932 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
22933 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
22934 arm_acle_builtins.def.
22935 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
22936 (arm_init_acle_builtins): New.
22937 (CRC32_BUILTIN): Remove.
22938 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
22939 crc32cb, crc32ch and crc32cw.
22940 (arm_init_crc32_builtins): Remove.
22941 (arm_init_builtins): Use arm_init_acle_builtins rather
22942 than arm_init_crc32_builtins.
22943 (arm_expand_acle_builtin): New.
22944 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
22945 * config/arm/arm_acle_builtins.def: New.
22947 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22949 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
22950 (arm_builtin_datum): ... this.
22951 (arm_init_neon_builtin): Rename to ...
22952 (arm_init_builtin): ... this. Add a new parameters PREFIX
22953 and USE_SIG_IN_NAME.
22954 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
22955 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
22956 'arm_builtin_datum'.
22957 (arm_init_vfp_builtins): Likewise.
22958 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
22959 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
22960 (arm_expand_neon_args): Rename to ...
22961 (arm_expand_builtin_args): ... this. Rename builtin_arg
22962 enum values and differentiate between ARG_BUILTIN_MEMORY
22963 and ARG_BUILTIN_NEON_MEMORY.
22964 (arm_expand_neon_builtin_1): Rename to ...
22965 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
22966 values, arm_expand_builtin_args and add bool parameter NEON.
22967 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
22968 (arm_expand_vfp_builtin): Likewise.
22969 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
22971 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22973 PR middle-end/77484
22974 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
22975 * predict.c (tree_estimate_probability_bb): Reverse direction of
22976 polymorphic call predictor.
22978 2017-01-06 David Malcolm <dmalcolm@redhat.com>
22980 * passes.c (execute_one_pass): Split out pass-skipping logic into...
22981 (determine_pass_name_match): ...this new function and...
22982 (should_skip_pass_p): ...this new function.
22984 2017-01-06 Nathan Sidwell <nathan@acm.org>
22986 * ipa-visibility.c (function_and_variable_visibility): Reformat
22987 comments and long lines. Remove extrneous if.
22988 * symtab.c (symtab_node::make_decl_local): Fix code format.
22989 (symtab_node::set_section_for_node): Fix comment typo.
22991 2017-01-06 Martin Liska <mliska@suse.cz>
22994 * lra-constraints.c: Rename invariant to lra_invariant.
22995 * predict.c (set_even_probabilities): Initialize e to NULL.
22997 2017-01-05 Martin Sebor <msebor@redhat.com>
22999 PR tree-optimization/78910
23000 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23001 (format_integer): Correct off-by-one error in the handling
23002 of precision with negative numbers in signed conversions..
23004 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23006 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23008 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23010 PR tree-optimization/71016
23011 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23012 factor_out_conditional_conversion. Formatting fix.
23013 (factor_out_conditional_conversion): Add cond_stmt argument.
23014 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23015 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23018 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23020 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23021 read-rtl-function.o, and selftest-rtl.o.
23022 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23023 (selftest::aarch64_test_loading_full_dump): New function.
23024 (selftest::aarch64_run_selftests): New function.
23025 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23026 selftest::aarch64_run_selftests.
23027 * config/i386/i386.c
23028 (selftest::ix86_test_loading_dump_fragment_1): New function.
23029 (selftest::ix86_test_loading_call_insn): New function.
23030 (selftest::ix86_test_loading_full_dump): New function.
23031 (selftest::ix86_test_loading_unspec): New function.
23032 (selftest::ix86_run_selftests): Call the new functions.
23033 * emit-rtl.c (maybe_set_max_label_num): New function.
23034 * emit-rtl.h (maybe_set_max_label_num): New decl.
23035 * function.c (instantiate_decls): Guard call to
23036 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23037 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23039 * gensupport.c (gen_reader::gen_reader): Pass "false"
23040 for new "compact" param of rtx_reader.
23041 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23042 rather than an empty string for NULL strings.
23043 * read-md.c: Potentially include config.h rather than bconfig.h.
23044 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23045 (have_error): New global, copied from errors.c.
23046 (md_reader::read_name): Rename to...
23047 (md_reader::read_name_1): ...this, adding "out_loc" param,
23048 and converting "missing name or number" to returning false, rather
23050 (md_reader::read_name): Reimplement in terms of read_name_1.
23051 (md_reader::read_name_or_nil): New function.
23052 (md_reader::read_string): Handle "(nil)" by returning NULL.
23053 (md_reader::md_reader): Add new param "compact".
23054 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23055 (md_reader::read_file): New method.
23056 * read-md.h (md_reader::md_reader): Add new param "compact".
23057 (md_reader::read_file): New method.
23058 (md_reader::is_compact): New accessor.
23059 (md_reader::read_name): Convert return type from void to file_location.
23060 (md_reader::read_name_or_nil): New decl.
23061 (md_reader::read_name_1): New decl.
23062 (md_reader::m_compact): New field.
23063 (noop_reader::noop_reader): Pass "false" for new "compact" param
23065 (rtx_reader::rtx_reader): Add new "compact" param.
23066 (rtx_reader::read_rtx_operand): Make virtual and convert return
23067 type from void to rtx.
23068 (rtx_reader::read_until): New decl.
23069 (rtx_reader::handle_any_trailing_information): New virtual function.
23070 (rtx_reader::postprocess): New virtual function.
23071 (rtx_reader::finalize_string): New virtual function.
23072 (rtx_reader::m_in_call_function_usage): New field.
23073 (rtx_reader::m_reuse_rtx_by_id): New field.
23074 * read-rtl-function.c: New file.
23075 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23076 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23077 (selftest::verify_three_block_rtl_cfg): New decl.
23078 * read-rtl-function.h: New file.
23079 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23080 For host, include function.h, memmodel.h, and emit-rtl.h.
23081 (one_time_initialization): New function.
23082 (struct compact_insn_name): New struct.
23083 (compact_insn_names): New array.
23084 (find_code): Handle insn codes in compact dumps.
23085 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23086 (bind_subst_iter_and_attr): Likewise.
23087 (add_condition_to_string): Likewise.
23088 (add_condition_to_rtx): Likewise.
23089 (apply_attribute_uses): Likewise.
23090 (add_current_iterators): Likewise.
23091 (apply_iterators): Likewise.
23092 (initialize_iterators): Guard usage of apply_subst_iterator with
23093 #ifdef GENERATOR_FILE.
23094 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23095 (md_reader::read_mapping): Likewise.
23096 (add_define_attr_for_define_subst): Likewise.
23097 (add_define_subst_attr): Likewise.
23098 (read_subst_mapping): Likewise.
23099 (check_code_iterator): Likewise.
23100 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23102 (one_time_initialization): New function.
23103 (rtx_reader::read_until): New method.
23104 (read_flags): New function.
23105 (parse_reg_note_name): New function.
23106 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23107 Handle reuse_rtx ids.
23108 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23109 Add parsing support for RTL dumps, mirroring the special-cases in
23110 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23111 values, and calling handle_any_trailing_information.
23112 (rtx_reader::read_rtx_operand): Convert return type from void
23113 to rtx, returning return_rtx. Handle case 'e'. Call
23114 finalize_string on XSTR and XTMPL fields.
23115 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23116 "(nil)" values were omitted. Call the postprocess vfunc on the
23118 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23119 class ctor. Initialize m_in_call_function_usage. Call
23120 one_time_initialization.
23121 * rtl-tests.c (selftest::test_uncond_jump): Call
23122 set_new_first_and_last_insn.
23123 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23124 * selftest-rtl.c: New file.
23125 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23126 (selftest::get_insn_by_uid): New decl.
23127 * selftest-run-tests.c (selftest::run_tests): Call
23128 read_rtl_function_c_tests.
23129 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23130 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23133 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23135 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23136 operands in a special way. Assert that pos+len <= mode precision.
23138 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23140 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23141 3 argument Alias with unlimited for the negative form.
23142 (fno-vect-cost-model): Removed.
23144 2017-01-05 Martin Liska <mliska@suse.cz>
23146 * hsa-gen.c (gen_hsa_divmod): New function.
23147 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23149 2017-01-05 Martin Liska <mliska@suse.cz>
23152 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23155 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23157 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23158 small constant length operands.
23160 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23162 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23163 between loop iterations.
23165 2017-01-05 Martin Liska <mliska@suse.cz>
23168 * gimplify.c (gimplify_decl_expr): Compare to
23169 asan_poisoned_variables instread of checking flags.
23170 (gimplify_target_expr): Likewise.
23171 (gimplify_expr): Likewise.
23172 (gimplify_function_tree): Conditionally initialize
23173 asan_poisoned_variables.
23175 2017-01-04 Jeff Law <law@redhat.com>
23177 PR tree-optimizatin/78812
23178 * rtl.h (contains_mem_rtx_p): Prototype.
23179 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23180 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23181 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23182 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23183 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23185 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23187 * input.c (assert_char_at_range): Default-initialize actual_range.
23189 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23191 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23192 to match the caller.
23194 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23196 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23197 insns after final jump in test to emit dummy move.
23199 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23201 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23202 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23204 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23206 * multiple_target.c (create_dispatcher_calls): Init e_next.
23207 * tree-ssa-loop-split.c (split_loop): Init border.
23208 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23211 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23216 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23217 (altivec_register_operand): Do not return true if the operand
23218 contains a SUBREG mixing SImode and SFmode.
23219 (vsx_register_operand): Likewise.
23220 (vsx_reg_sfsubreg_ok): New predicate.
23221 (vfloat_operand): Do not return true if the operand contains a
23222 SUBREG mixing SImode and SFmode.
23223 (vint_operand): Likewise.
23224 (vlogical_operand): Likewise.
23225 (gpc_reg_operand): Likewise.
23226 (int_reg_operand): Likewise.
23227 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23228 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23229 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23231 (rs6000_emit_move_si_sf_subreg): New helper function.
23232 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23233 fixup SUBREGs involving SImode and SFmode.
23234 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23235 numbers for the new peephole2 optimization.
23236 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23237 the GLIBC math library that do AND/IOR/XOR operations on single
23238 precision floating point.
23239 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23240 target macros to say whether we need to avoid SUBREGs mixing
23242 (TARGET_ALLOW_SF_SUBREG): Likewise.
23243 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23244 (UNSPEC_SI_FROM_SF): Likewise.
23245 (iorxor): Change spacing.
23246 (and_ior_xor): New iterator for AND, IOR, and XOR.
23247 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23248 (movdi_from_sf_zero_ext): Likewise.
23249 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23250 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23251 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23252 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23253 (fms<mode>4): Likewise.
23254 (fnma<mode>4): Likewise.
23255 (fnms<mode>4): Likewise.
23256 (nfma<mode>4): Likewise.
23257 (nfms<mode>4): Likewise.
23259 2017-01-04 Marek Polacek <polacek@redhat.com>
23262 * doc/invoke.texi: Document -Wpointer-compare.
23264 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23266 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23269 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23270 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23273 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23276 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23277 documentation of the powerpc_popcntb_ok attribute.
23278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23279 code to issue warning messages if a requested CPU configuration is
23280 not supported by the binary (assembler and loader) toolchain.
23281 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23282 made to define a built-in function that has been disabled.
23283 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23284 made to define a built-in function that has been disabled.
23285 (altivec_init_builtins): Add comment explaining why definition
23286 of the DST built-in functions is not preceded by an assertion
23287 check. Add assertions to prevent ICE if attempts are made to
23288 define an altivec predicate or an abs* built-in function that has
23290 (htm_init_builtins): Add comment explaining why definition of the
23291 htm built-in functions is not preceded by an assertion check.
23293 2017-01-04 Jeff Law <law@redhat.com>
23295 PR tree-optimizatin/67955
23296 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23297 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23298 the points-to solution does not include pt_null. Use DECL_PT_UID
23301 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23303 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23304 Use gen_int_mode instead of gen_lopwart for const_int operands.
23306 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23308 PR tree-optimization/71563
23309 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23310 out of range value - has low bits known to be zero.
23312 2017-01-04 Alan Modra <amodra@gmail.com>
23314 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23315 * configure: Regenerate.
23316 * config.in: Regenerate.
23318 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23321 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23322 a substring of the message, but strcmp with the whole message. Ifdef
23323 ENABLE_NLS, translate the message first using dgettext.
23325 2017-01-03 Jeff Law <law@redhat.com>
23327 PR tree-optimizatin/78856
23328 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23329 (mark_threaded_blocks): Remove code to truncate thread paths that
23330 cross multiple loop headers. Instead invalidate the cached loop
23331 iteration information and handle case of a thread path walking
23332 into an irreducible region.
23334 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23337 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23338 assertions. Add support for doing the signbit if the IEEE 128-bit
23339 floating point value is in a GPR.
23340 * config/rs6000/rs6000.md (Fsignbit): Delete.
23341 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23342 Update the length attribute if the value is in a GPR.
23343 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23344 the sign or zero extension instruction, since the value is always 0/1.
23345 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23348 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23349 extracting SImode to a GPR register so that we can generate a
23350 store, limit the vector to be in a traditional Altivec register
23351 for the vextuwrx instruction.
23353 2017-01-03 Ian Lance Taylor <iant@google.com>
23355 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23357 2017-01-03 Martin Sebor <msebor@redhat.com>
23359 PR tree-optimization/78696
23360 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23361 precision. Use MPFR for %f for greater fidelity. Correct handling
23363 (pass_sprintf_length::compute_format_length): Set width and precision
23364 specified by asrerisk to void_node for vararg functions.
23365 (try_substitute_return_value): Adjust dump output.
23367 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23369 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23371 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23373 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23374 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23375 -mlra/-mno-lra was passed to the compiler.
23377 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23379 PR rtl-optimization/65618
23380 * emit-rtl.c (try_split): Move initialization of "before" and
23381 "after" to just before the call to emit_insn_after_setloc.
23383 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23385 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23387 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23389 * dwarf2out.c (gen_enumeration_type_die): When
23390 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23392 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23394 PR tree-optimization/78965
23395 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23396 Change first argument from const call_info & to call_info &. For %n
23397 set info.nowrite to false.
23399 PR middle-end/78901
23400 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23401 possibly throwing calls.
23403 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23404 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23405 and fns handling, rather than in a separate case SSA_NAME.
23407 2017-01-02 Jeff Law <law@redhat.com>
23409 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23410 fixes for first_period and second_period variables.
23412 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23415 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23416 (*insvqi_1): New insn pattern.
23417 (*insvqi_1_mem_rex64): Ditto.
23418 (*insvqi_2): Ditto.
23419 (*insvqi_3): Rename from *insvqi.
23421 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23423 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23425 * doc/cfg.texi (Edges): Remove reference to Java.
23426 (Maintaining the CFG): Ditto.
23428 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23430 PR middle-end/77674
23431 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23432 transparent aliases.
23434 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23436 PR middle-end/77484
23437 * predict.def (PRED_CALL): Update hitrate.
23438 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23439 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23440 into direct/indirect/polymorphic variants.
23442 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23444 Update copyright years.
23446 * gcc.c (process_command): Update copyright notice dates.
23447 * gcov-dump.c (print_version): Ditto.
23448 * gcov.c (print_version): Ditto.
23449 * gcov-tool.c (print_version): Ditto.
23450 * gengtype.c (create_file): Ditto.
23451 * doc/cpp.texi: Bump @copying's copyright year.
23452 * doc/cppinternals.texi: Ditto.
23453 * doc/gcc.texi: Ditto.
23454 * doc/gccint.texi: Ditto.
23455 * doc/gcov.texi: Ditto.
23456 * doc/install.texi: Ditto.
23457 * doc/invoke.texi: Ditto.
23459 Copyright (C) 2017 Free Software Foundation, Inc.
23461 Copying and distribution of this file, with or without modification,
23462 are permitted in any medium without royalty provided the copyright
23463 notice and this notice are preserved.