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 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6 rather than a machine_mode.
7 (fixed_from_string): Likewise.
8 (fixed_convert): Likewise.
9 (fixed_convert_from_int): Likewise.
10 (fixed_convert_from_real): Likewise.
11 (real_convert_from_fixed): Likewise.
12 * fixed-value.c (fixed_from_double_int): Likewise.
13 (fixed_from_string): Likewise.
14 (fixed_convert): Likewise.
15 (fixed_convert_from_int): Likewise.
16 (fixed_convert_from_real): Likewise.
17 (real_convert_from_fixed): Likewise.
18 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
20 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
21 Alan Hayward <alan.hayward@arm.com>
22 David Sherwood <david.sherwood@arm.com>
24 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
25 of separate mode class checks. Do not allow vector modes here.
26 (immed_wide_int_const): Use as_a <scalar_mode>.
27 * explow.c (trunc_int_for_mode): Likewise.
28 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
30 (wi::min_value): Likewise.
31 (wi::max_value): Likewise.
32 * dwarf2out.c (loc_descriptor): Likewise.
33 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
36 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
37 Alan Hayward <alan.hayward@arm.com>
38 David Sherwood <david.sherwood@arm.com>
40 * tree.h (SCALAR_TYPE_MODE): New macro.
41 * expr.c (expand_expr_addr_expr_1): Use it.
42 (expand_expr_real_2): Likewise.
43 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
44 (fold_convert_const_fixed_from_int): Likewise.
45 (fold_convert_const_fixed_from_real): Likewise.
46 (native_encode_fixed): Likewise
47 (native_encode_complex): Likewise
48 (native_encode_vector): Likewise.
49 (native_interpret_fixed): Likewise.
50 (native_interpret_real): Likewise.
51 (native_interpret_complex): Likewise.
52 (native_interpret_vector): Likewise.
53 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
54 (simd_clone_adjust_argument_types): Likewise.
55 (simd_clone_init_simd_arrays): Likewise.
56 (simd_clone_adjust): Likewise.
57 * stor-layout.c (layout_type): Likewise.
58 * tree.c (build_minus_one_cst): Likewise.
59 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
60 * tree-inline.c (estimate_move_cost): Likewise.
61 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
62 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
63 (vectorizable_reduction): Likewise.
64 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
65 (vect_recog_mixed_size_cond_pattern): Likewise.
66 (check_bool_pattern): Likewise.
67 (adjust_bool_pattern): Likewise.
68 (search_type_for_mask_1): Likewise.
69 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
70 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
71 (vectorizable_load): Likewise.
72 (vectorizable_store): Likewise.
73 * ubsan.c (ubsan_encode_value): Likewise.
74 * varasm.c (output_constant): Likewise.
76 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
77 Alan Hayward <alan.hayward@arm.com>
78 David Sherwood <david.sherwood@arm.com>
80 * coretypes.h (scalar_mode): New class.
81 * machmode.h (scalar_mode): Likewise.
82 (scalar_mode::includes_p): New function.
83 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
84 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
85 * genmodes.c (get_mode_class): Handle remaining scalar modes.
86 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
87 * expmed.c (store_bit_field_1): Likewise.
88 (extract_bit_field_1): Likewise.
89 * expr.c (write_complex_part): Likewise.
90 (read_complex_part): Likewise.
91 (emit_move_complex_push): Likewise.
92 (expand_expr_real_2): Likewise.
93 * function.c (assign_parm_setup_reg): Likewise.
94 (assign_parms_unsplit_complex): Likewise.
95 * optabs.c (expand_binop): Likewise.
96 * rtlanal.c (subreg_get_info): Likewise.
97 * simplify-rtx.c (simplify_immed_subreg): Likewise.
98 * varasm.c (output_constant_pool_2): Likewise.
100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
101 Alan Hayward <alan.hayward@arm.com>
102 David Sherwood <david.sherwood@arm.com>
104 * expmed.c (extract_high_half): Use scalar_int_mode and remove
106 (expmed_mult_highpart_optab): Likewise.
107 (expmed_mult_highpart): Likewise.
109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
110 Alan Hayward <alan.hayward@arm.com>
111 David Sherwood <david.sherwood@arm.com>
113 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
114 instead of a machine_mode.
115 (builtin_memset_read_str): Likewise.
116 * builtins.c (c_readstr): Likewise.
117 (builtin_memcpy_read_str): Likewise.
118 (builtin_strncpy_read_str): Likewise.
119 (builtin_memset_read_str): Likewise.
120 (builtin_memset_gen_str): Likewise.
121 (expand_builtin_signbit): Use scalar_int_mode for local variables.
122 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
123 instead of a machine_mode.
124 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
126 (make_extraction): Likewise.
127 (try_widen_shift_mode): Take and return scalar_int_modes instead
129 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
130 a scalar_int_mode instead of a machine_mode.
131 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
132 (avr_addr_space_pointer_mode): Likewise.
133 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
134 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
135 (msp430_unwind_word_mode): Likewise.
136 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
137 (spu_addr_space_pointer_mode): Likewise.
138 (spu_addr_space_address_mode): Likewise.
139 (spu_libgcc_cmp_return_mode): Likewise.
140 (spu_libgcc_shift_count_mode): Likewise.
141 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
142 (rl78_addr_space_pointer_mode): Likewise.
143 (fl78_unwind_word_mode): Likewise.
144 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
146 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
147 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
148 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
149 (mips_valid_pointer_mode): Likewise.
150 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
151 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
152 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
154 (ft32_addr_space_address_mode): Likewise.
155 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
156 scalar_int_mode instead of a machine_mode.
157 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
159 (m32c_addr_space_address_mode): Likewise.
160 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
161 (rs6000_eh_return_filter_mode): Likewise.
162 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
163 (rs6000_eh_return_filter_mode): Likewise.
164 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
165 (s390_libgcc_shift_count_mode): Likewise.
166 (s390_unwind_word_mode): Likewise.
167 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
169 * target.def (mode_rep_extended): Likewise.
170 (valid_pointer_mode): Likewise.
171 (addr_space.valid_pointer_mode): Likewise.
172 (eh_return_filter_mode): Return a scalar_int_mode rather than
174 (libgcc_cmp_return_mode): Likewise.
175 (libgcc_shift_count_mode): Likewise.
176 (unwind_word_mode): Likewise.
177 (addr_space.pointer_mode): Likewise.
178 (addr_space.address_mode): Likewise.
179 * doc/tm.texi: Regenerate.
180 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
182 (do_jump): Use scalar_int_mode for local variables.
183 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
184 rather than a machine_mode.
185 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
186 (scompare_loc_descriptor_wide): Likewise.
187 (scompare_loc_descriptor_narrow): Likewise.
188 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
190 * except.c (sjlj_emit_dispatch_table): Likewise.
191 (expand_builtin_eh_copy_values): Likewise.
192 * explow.c (convert_memory_address_addr_space_1): Likewise.
193 Take a scalar_int_mode rather than a machine_mode.
194 (convert_memory_address_addr_space): Take a scalar_int_mode rather
196 (memory_address_addr_space): Use scalar_int_mode for local variables.
197 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
198 rather than a machine_mode.
199 * expmed.c (mask_rtx): Likewise.
200 (init_expmed_one_conv): Likewise.
201 (expand_mult_highpart_adjust): Likewise.
202 (extract_high_half): Likewise.
203 (expmed_mult_highpart_optab): Likewise.
204 (expmed_mult_highpart): Likewise.
205 (expand_smod_pow2): Likewise.
206 (expand_sdiv_pow2): Likewise.
207 (emit_store_flag_int): Likewise.
208 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
210 (extract_low_bits): Likewise.
211 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
213 * expr.c (pieces_addr::adjust): Likewise.
214 (can_store_by_pieces): Likewise.
215 (store_by_pieces): Likewise.
216 (clear_by_pieces_1): Likewise.
217 (expand_expr_addr_expr_1): Likewise.
218 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
219 (expand_expr_real_1): Likewise.
220 (try_casesi): Likewise.
221 * final.c (shorten_branches): Likewise.
222 * fold-const.c (fold_convert_const_int_from_fixed): Change the
223 type of "mode" to machine_mode.
224 * internal-fn.c (expand_arith_overflow_result_store): Take a
225 scalar_int_mode rather than a machine_mode.
226 (expand_mul_overflow): Use scalar_int_mode for local variables.
227 * loop-doloop.c (doloop_modify): Likewise.
228 (doloop_optimize): Likewise.
229 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
231 (expand_doubleword_shift_condmove): Likewise.
232 (expand_doubleword_shift): Likewise.
233 (expand_doubleword_clz): Likewise.
234 (expand_doubleword_popcount): Likewise.
235 (expand_doubleword_parity): Likewise.
236 (expand_absneg_bit): Use scalar_int_mode for local variables.
237 (prepare_float_lib_cmp): Likewise.
238 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
239 rather than a machine_mode.
240 (convert_memory_address_addr_space): Likewise.
241 (get_mode_bounds): Likewise.
242 (get_address_mode): Return a scalar_int_mode rather than a
244 * rtlanal.c (get_address_mode): Likewise.
245 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
247 * targhooks.c (default_mode_rep_extended): Likewise.
248 (default_valid_pointer_mode): Likewise.
249 (default_addr_space_valid_pointer_mode): Likewise.
250 (default_eh_return_filter_mode): Return a scalar_int_mode rather
252 (default_libgcc_cmp_return_mode): Likewise.
253 (default_libgcc_shift_count_mode): Likewise.
254 (default_unwind_word_mode): Likewise.
255 (default_addr_space_pointer_mode): Likewise.
256 (default_addr_space_address_mode): Likewise.
257 * targhooks.h (default_eh_return_filter_mode): Likewise.
258 (default_libgcc_cmp_return_mode): Likewise.
259 (default_libgcc_shift_count_mode): Likewise.
260 (default_unwind_word_mode): Likewise.
261 (default_addr_space_pointer_mode): Likewise.
262 (default_addr_space_address_mode): Likewise.
263 (default_mode_rep_extended): Take a scalar_int_mode rather than
265 (default_valid_pointer_mode): Likewise.
266 (default_addr_space_valid_pointer_mode): Likewise.
267 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
269 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
270 rather than a machine_mode.
271 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
273 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
274 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
278 Alan Hayward <alan.hayward@arm.com>
279 David Sherwood <david.sherwood@arm.com>
281 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
282 the mode argument to scalar_int_mode.
283 (do_jump_by_parts_zero_rtx): Likewise.
284 (do_jump_by_parts_equality_rtx): Likewise.
285 (do_jump_by_parts_greater): Take a mode argument.
286 (do_jump_by_parts_equality): Likewise.
287 (do_jump_1): Update calls accordingly.
289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
290 Alan Hayward <alan.hayward@arm.com>
291 David Sherwood <david.sherwood@arm.com>
293 * is-a.h (safe_dyn_cast): New function.
294 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
295 (jump_table_for_label): Likewise.
296 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
297 instead of an rtx_insn *.
298 (shorten_branches): Use dyn_cast instead of LABEL_P and
299 JUMP_TABLE_DATA_P. Use jump_table_for_label and
300 rtx_jump_table_data::get_data_mode.
301 (final_scan_insn): Likewise.
303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
304 Alan Hayward <alan.hayward@arm.com>
305 David Sherwood <david.sherwood@arm.com>
307 * combine.c (try_combine): Use is_a <scalar_int_mode> when
308 trying to combine a full-register integer set with a subreg
311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
316 that is always either address_mode or pointer_mode.
318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
319 Alan Hayward <alan.hayward@arm.com>
320 David Sherwood <david.sherwood@arm.com>
322 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
323 when the two are known to be equal.
325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
326 Alan Hayward <alan.hayward@arm.com>
327 David Sherwood <david.sherwood@arm.com>
329 * simplify-rtx.c (simplify_const_unary_operation): Use
330 is_a <scalar_int_mode> instead of checking for a nonzero
331 precision. Forcibly convert op_mode to a scalar_int_mode
332 in that case. More clearly differentiate the operand and
333 result modes and use the former when deciding what the value
334 of a count-bits operation should be. Use is_int_mode instead
335 of checking for a MODE_INT. Remove redundant check for whether
336 this mode has a zero precision.
338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * optabs.c (widen_leading): Change the type of the mode argument
343 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
344 (widen_bswap): Likewise.
345 (expand_parity): Likewise.
346 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
347 (expand_ffs): Likewise.
348 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
357 (expand_expr_real_1): Update call accordingly.
359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
360 Alan Hayward <alan.hayward@arm.com>
361 David Sherwood <david.sherwood@arm.com>
363 * expmed.c (store_bit_field_using_insv): Add op0_mode and
364 value_mode arguments. Use scalar_int_mode internally.
365 (store_bit_field_1): Rename the new integer mode from imode
366 to op0_mode and use it instead of GET_MODE (op0). Update calls
367 to store_split_bit_field, store_bit_field_using_insv and
368 store_fixed_bit_field.
369 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
370 Use scalar_int_mode internally. Use a bit count rather than a mode
371 when calculating the largest bit size for get_best_mode.
372 Update calls to store_split_bit_field and store_fixed_bit_field_1.
373 (store_fixed_bit_field_1): Add mode and value_mode arguments.
374 Remove assertion that OP0 has a scalar integer mode.
375 (store_split_bit_field): Add op0_mode and value_mode arguments.
376 Update calls to extract_fixed_bit_field.
377 (extract_bit_field_using_extv): Add an op0_mode argument.
378 Use scalar_int_mode internally.
379 (extract_bit_field_1): Rename the new integer mode from imode to
380 op0_mode and use it instead of GET_MODE (op0). Update calls to
381 extract_split_bit_field, extract_bit_field_using_extv and
382 extract_fixed_bit_field.
383 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
384 to extract_split_bit_field and extract_fixed_bit_field_1.
385 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
386 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
388 (extract_split_bit_field): Add an op0_mode argument. Update call
389 to extract_fixed_bit_field.
391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
392 Alan Hayward <alan.hayward@arm.com>
393 David Sherwood <david.sherwood@arm.com>
395 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
396 * explow.c (hard_function_value): Likewise.
397 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
398 convert_to_mode call outside the loop.
399 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
400 for the mode iterator. Require the mode specified by max_pieces
402 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
404 (copy_blkmode_to_reg): Likewise.
405 (set_storage_via_setmem): Likewise.
406 * optabs.c (prepare_cmp_insn): Likewise.
407 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
408 * stor-layout.c (finish_bitfield_representative): Likewise.
410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
411 Alan Hayward <alan.hayward@arm.com>
412 David Sherwood <david.sherwood@arm.com>
414 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
415 * expr.c (convert_move): Use them.
416 (convert_modes): Likewise.
417 (store_expr_with_bounds): Likewise.
419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
420 Alan Hayward <alan.hayward@arm.com>
421 David Sherwood <david.sherwood@arm.com>
423 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
424 parameter for the mode of "x". Remove the "known_x", "known_mode"
425 and "known_ret" arguments. Change the type of the mode argument
427 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
428 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
429 (reg_num_sign_bit_copies_for_combine): Likewise.
430 * rtlanal.c (nonzero_bits1): Likewise.
431 (num_sign_bit_copies1): Likewise.
432 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
433 (reg_num_sign_bit_copies_general): Likewise.
434 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
435 (reg_nonzero_bits_general): Likewise.
437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
438 Alan Hayward <alan.hayward@arm.com>
439 David Sherwood <david.sherwood@arm.com>
441 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
442 than in subroutines. Return 1 for non-integer modes.
443 (cached_num_sign_bit_copies): Change the type of the mode parameter
445 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
446 classes. Handle CONST_INT_P first and then check whether X also
447 has a scalar integer mode. Check the same thing for inner registers
448 of a SUBREG and for values that are being extended or truncated.
450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
451 Alan Hayward <alan.hayward@arm.com>
452 David Sherwood <david.sherwood@arm.com>
454 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
455 in subroutines. Return the mode mask for non-integer modes.
456 (cached_nonzero_bits): Change the type of the mode parameter
458 (nonzero_bits1): Likewise. Remove early exit for other mode
459 classes. Handle CONST_INT_P first and then check whether X
460 also has a scalar integer mode.
462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
463 Alan Hayward <alan.hayward@arm.com>
464 David Sherwood <david.sherwood@arm.com>
466 * combine.c (extract_left_shift): Add a mode argument and update
468 (make_compound_operation_int): Change the type of the mode parameter
469 to scalar_int_mode and update the call to extract_left_shift.
471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
472 Alan Hayward <alan.hayward@arm.com>
473 David Sherwood <david.sherwood@arm.com>
475 * combine.c (simplify_and_const_int): Change the type of the mode
476 parameter to scalar_int_mode.
477 (simplify_and_const_int_1): Likewise. Update recursive call.
479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
480 Alan Hayward <alan.hayward@arm.com>
481 David Sherwood <david.sherwood@arm.com>
483 * combine.c (simplify_compare_const): Check that the mode is a
484 scalar_int_mode (rather than VOIDmode) before testing its
486 (simplify_comparison): Move COMPARISON_P handling out of the
487 loop and restrict the latter part of the loop to scalar_int_modes.
488 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
489 and when considering SUBREG_REGs. Use is_int_mode instead of
490 checking GET_MODE_CLASS against MODE_INT.
492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
493 Alan Hayward <alan.hayward@arm.com>
494 David Sherwood <david.sherwood@arm.com>
496 * combine.c (try_widen_shift_mode): Move check for equal modes to...
497 (simplify_shift_const_1): ...here. Use scalar_int_mode for
498 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
505 (force_to_mode): ...here. Keep xmode up-to-date and use it
506 instead of GET_MODE (x).
508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
509 Alan Hayward <alan.hayward@arm.com>
510 David Sherwood <david.sherwood@arm.com>
512 * optabs-query.h (extraction_insn::struct_mode): Change type to
513 opt_scalar_int_mode and update comment.
514 (extraction_insn::field_mode): Change type to scalar_int_mode.
515 (extraction_insn::pos_mode): Likewise.
516 * combine.c (make_extraction): Update accordingly.
517 * optabs-query.c (get_traditional_extraction_insn): Likewise.
518 (get_optab_extraction_insn): Likewise.
519 * recog.c (simplify_while_replacing): Likewise.
520 * expmed.c (narrow_bit_field_mem): Change the type of the mode
521 parameter to opt_scalar_int_mode.
523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
524 Alan Hayward <alan.hayward@arm.com>
525 David Sherwood <david.sherwood@arm.com>
527 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
528 to a scalar_int_mode instead of a machine_mode.
529 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
530 (get_best_mode): Return a boolean and use a pointer argument to store
531 the selected mode. Replace the limit mode parameter with a bit limit.
532 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
533 for the values returned by bit_field_mode_iterator::next_mode.
534 (store_bit_field): Update call to get_best_mode.
535 (store_fixed_bit_field): Likewise.
536 (extract_fixed_bit_field): Likewise.
537 * expr.c (optimize_bitfield_assignment_op): Likewise.
538 * fold-const.c (optimize_bit_field_compare): Likewise.
539 (fold_truth_andor_1): Likewise.
540 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
541 Update for new type of m_mode.
542 (get_best_mode): As above.
544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
545 Alan Hayward <alan.hayward@arm.com>
546 David Sherwood <david.sherwood@arm.com>
548 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
549 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
550 (store_bit_field): Check is_a <scalar_int_mode> before calling
551 strict_volatile_bitfield_p.
552 (extract_bit_field): Likewise.
554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
555 Alan Hayward <alan.hayward@arm.com>
556 David Sherwood <david.sherwood@arm.com>
558 * target.def (cstore_mode): Return a scalar_int_mode.
559 * doc/tm.texi: Regenerate.
560 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
561 * targhooks.h (default_cstore_mode): Likewise.
562 * targhooks.c (default_cstore_mode): Likewise, using a forced
564 * expmed.c (emit_cstore): Expect the target of the cstore to be
567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
568 Alan Hayward <alan.hayward@arm.com>
569 David Sherwood <david.sherwood@arm.com>
571 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
572 * builtins.c (expand_builtin_signbit): Use it.
573 * cfgexpand.c (expand_debug_expr): Likewise.
574 * dojump.c (do_jump): Likewise.
575 (do_compare_and_jump): Likewise.
576 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
577 * expmed.c (make_tree): Likewise.
578 * expr.c (expand_expr_real_2): Likewise.
579 (expand_expr_real_1): Likewise.
580 (try_casesi): Likewise.
581 * fold-const-call.c (fold_const_call_ss): Likewise.
582 * fold-const.c (unextend): Likewise.
583 (extract_muldiv_1): Likewise.
584 (fold_single_bit_test): Likewise.
585 (native_encode_int): Likewise.
586 (native_encode_string): Likewise.
587 (native_interpret_int): Likewise.
588 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
589 * internal-fn.c (expand_addsub_overflow): Likewise.
590 (expand_neg_overflow): Likewise.
591 (expand_mul_overflow): Likewise.
592 (expand_arith_overflow): Likewise.
593 * match.pd: Likewise.
594 * stor-layout.c (layout_type): Likewise.
595 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
596 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
597 * tree-ssanames.c (get_range_info): Likewise.
598 * tree-switch-conversion.c (array_value_type) Likewise.
599 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
600 (vect_recog_divmod_pattern): Likewise.
601 (vect_recog_mixed_size_cond_pattern): Likewise.
602 * tree-vrp.c (extract_range_basic): Likewise.
603 (simplify_float_conversion_using_ranges): Likewise.
604 * tree.c (int_fits_type_p): Likewise.
605 * ubsan.c (instrument_bool_enum_load): Likewise.
606 * varasm.c (mergeable_string_section): Likewise.
607 (narrowing_initializer_constant_valid_p): Likewise.
608 (output_constant): Likewise.
610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
611 Alan Hayward <alan.hayward@arm.com>
612 David Sherwood <david.sherwood@arm.com>
614 * machmode.h (NARROWEST_INT_MODE): New macro.
615 * expr.c (alignment_for_piecewise_move): Use it instead of
616 GET_CLASS_NARROWEST_MODE (MODE_INT).
617 (push_block): Likewise.
618 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
620 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
623 Alan Hayward <alan.hayward@arm.com>
624 David Sherwood <david.sherwood@arm.com>
626 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
628 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
629 * optabs.c (expand_unop): Likewise.
631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
632 Alan Hayward <alan.hayward@arm.com>
633 David Sherwood <david.sherwood@arm.com>
635 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
636 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
637 (popcount_loc_descriptor): Likewise.
638 (bswap_loc_descriptor): Likewise.
639 (rotate_loc_descriptor): Likewise.
640 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
641 calling the functions above.
643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
644 Alan Hayward <alan.hayward@arm.com>
645 David Sherwood <david.sherwood@arm.com>
647 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
649 (try_combine): Likewise.
650 (simplify_if_then_else): Likewise.
651 * cse.c (cse_insn): Likewise.
652 * dwarf2out.c (mem_loc_descriptor): Likewise.
653 * emit-rtl.c (gen_lowpart_common): Likewise.
654 * simplify-rtx.c (simplify_truncation): Likewise.
655 (simplify_binary_operation_1): Likewise.
656 (simplify_const_relational_operation): Likewise.
657 (simplify_ternary_operation): Likewise.
658 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
661 Alan Hayward <alan.hayward@arm.com>
662 David Sherwood <david.sherwood@arm.com>
664 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
665 * reload.c (push_reload): Likewise.
666 (find_reloads): Likewise.
668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
669 Alan Hayward <alan.hayward@arm.com>
670 David Sherwood <david.sherwood@arm.com>
672 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
673 (make_compound_operation_int): Likewise.
674 (change_zero_ext): Likewise.
675 * expr.c (convert_move): Likewise.
676 (convert_modes): Likewise.
677 * fwprop.c (forward_propagate_subreg): Likewise.
678 * loop-iv.c (get_biv_step_1): Likewise.
679 * optabs.c (widen_operand): Likewise.
680 * postreload.c (move2add_valid_value_p): Likewise.
681 * recog.c (simplify_while_replacing): Likewise.
682 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
683 (simplify_binary_operation_1): Likewise. Remove redundant
686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
687 Alan Hayward <alan.hayward@arm.com>
688 David Sherwood <david.sherwood@arm.com>
690 * combine.c (combine_simplify_rtx): Add checks for
691 is_a <scalar_int_mode>.
692 (simplify_if_then_else): Likewise.
693 (make_field_assignment): Likewise.
694 (simplify_comparison): Likewise.
695 * ifcvt.c (noce_try_bitop): Likewise.
696 * loop-invariant.c (canonicalize_address_mult): Likewise.
697 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
700 Alan Hayward <alan.hayward@arm.com>
701 David Sherwood <david.sherwood@arm.com>
703 * simplify-rtx.c (simplify_binary_operation_1): Use
704 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
707 Alan Hayward <alan.hayward@arm.com>
708 David Sherwood <david.sherwood@arm.com>
710 * wide-int.h (int_traits<unsigned char>) New class.
711 (int_traits<unsigned short>) Likewise.
712 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
713 Use GET_MODE_UNIT_PRECISION and remove redundant test for
715 * combine.c (set_nonzero_bits_and_sign_copies): Use
716 is_a <scalar_int_mode>.
717 (find_split_point): Likewise.
718 (combine_simplify_rtx): Likewise.
719 (simplify_logical): Likewise.
720 (expand_compound_operation): Likewise.
721 (expand_field_assignment): Likewise.
722 (make_compound_operation): Likewise.
723 (extended_count): Likewise.
724 (change_zero_ext): Likewise.
725 (simplify_comparison): Likewise.
726 * dwarf2out.c (scompare_loc_descriptor): Likewise.
727 (ucompare_loc_descriptor): Likewise.
728 (minmax_loc_descriptor): Likewise.
729 (mem_loc_descriptor): Likewise.
730 (loc_descriptor): Likewise.
731 * expmed.c (init_expmed_one_mode): Likewise.
732 * lra-constraints.c (lra_constraint_offset): Likewise.
733 * optabs.c (prepare_libcall_arg): Likewise.
734 * postreload.c (move2add_note_store): Likewise.
735 * reload.c (operands_match_p): Likewise.
736 * rtl.h (load_extend_op): Likewise.
737 * rtlhooks.c (gen_lowpart_general): Likewise.
738 * simplify-rtx.c (simplify_truncation): Likewise.
739 (simplify_unary_operation_1): Likewise.
740 (simplify_binary_operation_1): Likewise.
741 (simplify_const_binary_operation): Likewise.
742 (simplify_const_relational_operation): Likewise.
743 (simplify_subreg): Likewise.
744 * stor-layout.c (bitwise_mode_for_mode): Likewise.
745 * var-tracking.c (adjust_mems): Likewise.
746 (prepare_call_arguments): Likewise.
748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
749 Alan Hayward <alan.hayward@arm.com>
750 David Sherwood <david.sherwood@arm.com>
752 * machmode.h (is_int_mode): New fuction.
753 * combine.c (find_split_point): Use it.
754 (combine_simplify_rtx): Likewise.
755 (simplify_if_then_else): Likewise.
756 (simplify_set): Likewise.
757 (simplify_shift_const_1): Likewise.
758 (simplify_comparison): Likewise.
759 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
760 * cse.c (notreg_cost): Likewise.
761 (cse_insn): Likewise.
762 * cselib.c (cselib_lookup_1): Likewise.
763 * dojump.c (do_jump_1): Likewise.
764 (do_compare_rtx_and_jump): Likewise.
765 * dse.c (get_call_args): Likewise.
766 * dwarf2out.c (rtl_for_decl_init): Likewise.
767 (native_encode_initializer): Likewise.
768 * expmed.c (emit_store_flag_1): Likewise.
769 (emit_store_flag): Likewise.
770 * expr.c (convert_modes): Likewise.
771 (store_field): Likewise.
772 (expand_expr_real_1): Likewise.
773 * fold-const.c (fold_read_from_constant_string): Likewise.
774 * gimple-ssa-sprintf.c (get_format_string): Likewise.
775 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
776 * optabs.c (expand_binop): Likewise.
777 (expand_unop): Likewise.
778 (expand_abs_nojump): Likewise.
779 (expand_one_cmpl_abs_nojump): Likewise.
780 * simplify-rtx.c (mode_signbit_p): Likewise.
781 (val_signbit_p): Likewise.
782 (val_signbit_known_set_p): Likewise.
783 (val_signbit_known_clear_p): Likewise.
784 (simplify_relational_operation_1): Likewise.
785 * tree.c (vector_type_mode): Likewise.
787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
788 Alan Hayward <alan.hayward@arm.com>
789 David Sherwood <david.sherwood@arm.com>
791 * machmode.h (smallest_mode_for_size): Fix formatting.
792 (smallest_int_mode_for_size): New function.
793 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
794 instead of smallest_mode_for_size.
795 * combine.c (make_extraction): Likewise.
796 * config/arc/arc.c (arc_expand_movmem): Likewise.
797 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
798 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
799 * config/s390/s390.c (s390_expand_insv): Likewise.
800 * config/sparc/sparc.c (assign_int_registers): Likewise.
801 * config/spu/spu.c (spu_function_value): Likewise.
802 (spu_function_arg): Likewise.
803 * coverage.c (get_gcov_type): Likewise.
804 (get_gcov_unsigned_t): Likewise.
805 * dse.c (find_shift_sequence): Likewise.
806 * expmed.c (store_bit_field_1): Likewise.
807 * expr.c (convert_move): Likewise.
808 (store_field): Likewise.
809 * internal-fn.c (expand_arith_overflow): Likewise.
810 * optabs-query.c (get_best_extraction_insn): Likewise.
811 * optabs.c (expand_twoval_binop_libfunc): Likewise.
812 * stor-layout.c (layout_type): Likewise.
813 (initialize_sizetypes): Likewise.
814 * targhooks.c (default_get_mask_mode): Likewise.
815 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
818 Alan Hayward <alan.hayward@arm.com>
819 David Sherwood <david.sherwood@arm.com>
821 * machmode.h (opt_mode::else_blk): New function.
822 (int_mode_for_mode): Declare.
823 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
824 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
826 * cfgexpand.c (expand_debug_expr): Likewise.
827 * combine.c (gen_lowpart_or_truncate): Likewise.
828 (gen_lowpart_for_combine): Likewise.
829 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
830 * config/avr/avr.c (avr_to_int_mode): Likewise.
831 (avr_out_plus_1): Likewise.
832 (avr_out_plus): Likewise.
833 (avr_out_round): Likewise.
834 * config/i386/i386.c (ix86_split_to_parts): Likewise.
835 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
836 (s390_expand_vcond): Likewise.
837 * config/spu/spu.c (spu_split_immediate): Likewise.
838 (spu_expand_mov): Likewise.
839 * dse.c (get_stored_val): Likewise.
840 * expmed.c (store_bit_field_1): Likewise.
841 (convert_extracted_bit_field): Use int_mode_for_mode instead of
843 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
844 (extract_low_bits): Likewise.
845 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
846 handling rather than repeating the check.
847 (emit_group_store): Likewise.
848 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
849 * optabs.c (expand_absneg_bit): Likewise.
850 (expand_copysign_absneg): Likewise.
851 (expand_copysign_bit): Likewise.
852 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
853 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
854 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
855 * var-tracking.c (prepare_call_arguments): Likewise.
856 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
857 int_mode_for_mode instead of mode_for_size.
858 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
861 Alan Hayward <alan.hayward@arm.com>
862 David Sherwood <david.sherwood@arm.com>
864 * machmode.h (int_mode_for_size): New function.
865 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
866 instead of mode_for_size.
867 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
869 * combine.c (expand_field_assignment): Use int_mode_for_size
870 instead of mode_for_size.
871 (make_extraction): Likewise.
872 (simplify_shift_const_1): Likewise.
873 (simplify_comparison): Likewise.
874 * dojump.c (do_jump): Likewise.
875 * dwarf2out.c (mem_loc_descriptor): Likewise.
876 * emit-rtl.c (init_derived_machine_modes): Likewise.
877 * expmed.c (flip_storage_order): Likewise.
878 (convert_extracted_bit_field): Likewise.
879 * expr.c (copy_blkmode_from_reg): Likewise.
880 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
881 * internal-fn.c (expand_mul_overflow): Likewise.
882 * lower-subreg.c (simple_move): Likewise.
883 * optabs-libfuncs.c (init_optabs): Likewise.
884 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
885 * tree.c (vector_type_mode): Likewise.
886 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
887 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
888 * tree-vect-generic.c (expand_vector_parallel): Likewise.
889 * tree-vect-stmts.c (vectorizable_load): Likewise.
890 (vectorizable_store): Likewise.
892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
893 Alan Hayward <alan.hayward@arm.com>
894 David Sherwood <david.sherwood@arm.com>
896 * coretypes.h (pod_mode): New type.
897 (scalar_int_mode_pod): New typedef.
898 * machmode.h (pod_mode): New class.
899 (int_n_data_t::m): Change type to scalar_int_mode_pod.
900 * genmodes.c (emit_mode_int_n): Update accordingly.
901 * lower-subreg.h (target_lower_subreg): Change type to
903 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
907 Alan Hayward <alan.hayward@arm.com>
908 David Sherwood <david.sherwood@arm.com>
910 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
911 machine_mode to scalar_int_mode.
912 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
913 (rs6000_option_override_internal): Remove cast to int.
914 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
915 machine_mode to scalar_int_mode.
916 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
917 (rs6000_option_override_internal): Remove cast to int.
918 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
919 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
921 * config/s390/s390.c (s390_expand_builtin): Likewise.
922 * coretypes.h (scalar_int_mode): New type.
923 (opt_scalar_int_mode): New typedef.
924 * machmode.h (scalar_int_mode): New class.
925 (scalar_int_mode::includes_p): New function.
926 (byte_mode): Change type to scalar_int_mode.
927 (word_mode): Likewise.
928 (ptr_mode): Likewise.
929 * emit-rtl.c (byte_mode): Likewise.
930 (word_mode): Likewise.
931 (ptr_mode): Likewise.
932 (init_derived_machine_modes): Update accordingly.
933 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
934 and MODE_PARTIAL_INT.
935 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
939 Alan Hayward <alan.hayward@arm.com>
940 David Sherwood <david.sherwood@arm.com>
942 * target.def (libgcc_floating_mode_supported_p): Take a
944 * doc/tm.texi: Regenerate.
945 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
947 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
948 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
952 Alan Hayward <alan.hayward@arm.com>
953 David Sherwood <david.sherwood@arm.com>
955 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
956 * doc/tm.texi: Regenerate.
957 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
958 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
959 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
960 * targhooks.h (default_floatn_mode): Likewise.
961 * targhooks.c (default_floatn_mode): Likewise.
962 * tree.c (build_common_tree_nodes): Update accordingly.
964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
965 Alan Hayward <alan.hayward@arm.com>
966 David Sherwood <david.sherwood@arm.com>
968 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
969 (mode_iterator::iterate_p): Likewise.
970 (mode_iterator::get_wider): Likewise.
971 * expr.c (init_expr_target): Use opt_scalar_float_mode.
973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
974 Alan Hayward <alan.hayward@arm.com>
975 David Sherwood <david.sherwood@arm.com>
977 * coretypes.h (opt_scalar_float_mode): New typedef.
978 * machmode.h (float_mode_for_size): New function.
979 * emit-rtl.c (double_mode): Delete.
980 (init_emit_once): Use float_mode_for_size.
981 * stor-layout.c (layout_type): Likewise.
982 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
985 Alan Hayward <alan.hayward@arm.com>
986 David Sherwood <david.sherwood@arm.com>
988 * output.h (assemble_real): Take a scalar_float_mode.
989 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
990 * config/arm/arm.md (consttable_4): Likewise.
991 (consttable_8): Likewise.
992 (consttable_16): Likewise.
993 * config/mips/mips.md (consttable_float): Likewise.
994 * config/s390/s390.c (s390_output_pool_entry): Likewise.
995 * varasm.c (assemble_real): Take a scalar_float_mode.
996 (output_constant_pool_2): Update accordingly.
997 (output_constant): Likewise.
999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1000 Alan Hayward <alan.hayward@arm.com>
1001 David Sherwood <david.sherwood@arm.com>
1003 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1004 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1005 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1006 (native_encode_real): Likewise.
1007 (native_interpret_real): Likewise.
1008 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1009 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1012 Alan Hayward <alan.hayward@arm.com>
1013 David Sherwood <david.sherwood@arm.com>
1015 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1016 <scalar_float_mode>. Simplify.
1017 (gen_extend_conv_libfunc): Likewise.
1019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1020 Alan Hayward <alan.hayward@arm.com>
1021 David Sherwood <david.sherwood@arm.com>
1023 * coretypes.h (scalar_float_mode): New type.
1024 * machmode.h (mode_traits::from_int): Use machine_mode if
1025 USE_ENUM_MODES is defined.
1026 (is_a): New function.
1028 (dyn_cast): Likewise.
1029 (scalar_float_mode): New class.
1030 (scalar_float_mode::includes_p): New function.
1031 (is_float_mode): Likewise.
1032 * gdbhooks.py (MachineModePrinter): New class.
1033 (build_pretty_printer): Use it for scalar_float_mode.
1034 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1035 (format_helper::format_helper): Turn into a template.
1036 * genmodes.c (get_mode_class): New function.
1037 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1038 or machine_mode if none.
1039 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1040 as_a <scalar_float_mode>.
1041 * dwarf2out.c (mem_loc_descriptor): Likewise.
1042 (insert_float): Likewise.
1043 (add_const_value_attribute): Likewise.
1044 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1045 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1046 (expand_unop): Update accordingly.
1047 (expand_abs_nojump): Likewise.
1048 (expand_copysign_absneg): Take a scalar_float_mode.
1049 (expand_copysign_bit): Likewise.
1050 (expand_copysign): Update accordingly.
1052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1053 Alan Hayward <alan.hayward@arm.com>
1054 David Sherwood <david.sherwood@arm.com>
1056 * coretypes.h (opt_mode): New class.
1057 * machmode.h (opt_mode): Likewise.
1058 (opt_mode::else_void): New function.
1059 (opt_mode::require): Likewise.
1060 (opt_mode::exists): Likewise.
1061 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1062 (GET_MODE_2XWIDER_MODE): Likewise.
1063 (mode_iterator::get_wider): Update accordingly.
1064 (mode_iterator::get_2xwider): Likewise.
1065 (mode_iterator::get_known_wider): Likewise, turning into a template.
1066 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1067 forcing a wider mode to exist.
1068 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1069 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1070 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1071 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1072 * lower-subreg.c (init_lower_subreg): Likewise.
1073 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1074 on the final iteration.
1075 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1076 a wider mode exists before asking for a move pattern.
1077 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1078 forcing a wider mode to exist.
1079 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1080 returning false if no such mode exists.
1081 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1082 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1083 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1084 Avoid checking for a MODE_INT if we already know the mode is not a
1086 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1087 forcing a wider mode to exist.
1088 (expmed_mult_highpart_optab): Likewise.
1089 (expmed_mult_highpart): Likewise.
1090 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1092 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1093 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1094 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1095 * internal-fn.c (expand_mul_overflow): Update use of
1096 GET_MODE_2XWIDER_MODE.
1097 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1098 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1099 GET_MODE_WIDER_MODE.
1100 (convert_plusminus_to_widen): Likewise.
1101 * tree-switch-conversion.c (array_value_type): Likewise.
1102 * var-tracking.c (emit_note_insn_var_location): Likewise.
1103 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1104 Return false inside rather than outside the loop if no wider mode
1106 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1107 and GET_MODE_2XWIDER_MODE
1108 (can_compare_p): Use else_void.
1109 * gdbhooks.py (OptMachineModePrinter): New class.
1110 (build_pretty_printer): Use it for opt_mode.
1112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1113 Alan Hayward <alan.hayward@arm.com>
1114 David Sherwood <david.sherwood@arm.com>
1116 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1117 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1120 Alan Hayward <alan.hayward@arm.com>
1121 David Sherwood <david.sherwood@arm.com>
1123 * machmode.h (mode_traits): New structure.
1124 (get_narrowest_mode): New function.
1125 (mode_iterator::start): Likewise.
1126 (mode_iterator::iterate_p): Likewise.
1127 (mode_iterator::get_wider): Likewise.
1128 (mode_iterator::get_known_wider): Likewise.
1129 (mode_iterator::get_2xwider): Likewise.
1130 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1131 (FOR_EACH_MODE): Likewise.
1132 (FOR_EACH_MODE_FROM): Likewise.
1133 (FOR_EACH_MODE_UNTIL): Likewise.
1134 (FOR_EACH_WIDER_MODE): Likewise.
1135 (FOR_EACH_2XWIDER_MODE): Likewise.
1136 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1137 * combine.c (simplify_comparison): Likewise
1138 * config/i386/i386.c (type_natural_mode): Likewise.
1139 * cse.c (cse_insn): Likewise.
1140 * dse.c (find_shift_sequence): Likewise.
1141 * emit-rtl.c (init_derived_machine_modes): Likewise.
1142 (init_emit_once): Likewise.
1143 * explow.c (hard_function_value): Likewise.
1144 * expmed.c (extract_fixed_bit_field_1): Likewise.
1145 (extract_bit_field_1): Likewise.
1146 (expand_divmod): Likewise.
1147 (emit_store_flag_1): Likewise.
1148 * expr.c (init_expr_target): Likewise.
1149 (convert_move): Likewise.
1150 (alignment_for_piecewise_move): Likewise.
1151 (widest_int_mode_for_size): Likewise.
1152 (emit_block_move_via_movmem): Likewise.
1153 (copy_blkmode_to_reg): Likewise.
1154 (set_storage_via_setmem): Likewise.
1155 (compress_float_constant): Likewise.
1156 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1157 * optabs-query.c (get_best_extraction_insn): Likewise.
1158 * optabs.c (expand_binop): Likewise.
1159 (expand_twoval_unop): Likewise.
1160 (expand_twoval_binop): Likewise.
1161 (widen_leading): Likewise.
1162 (widen_bswap): Likewise.
1163 (expand_parity): Likewise.
1164 (expand_unop): Likewise.
1165 (prepare_cmp_insn): Likewise.
1166 (prepare_float_lib_cmp): Likewise.
1167 (expand_float): Likewise.
1168 (expand_fix): Likewise.
1169 (expand_sfix_optab): Likewise.
1170 * postreload.c (move2add_use_add2_insn): Likewise.
1171 * reg-stack.c (reg_to_stack): Likewise.
1172 * reginfo.c (choose_hard_reg_mode): Likewise.
1173 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1174 * stor-layout.c (mode_for_size): Likewise.
1175 (smallest_mode_for_size): Likewise.
1176 (mode_for_vector): Likewise.
1177 (finish_bitfield_representative): Likewise.
1178 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1179 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1180 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1181 * var-tracking.c (prepare_call_arguments): Likewise.
1183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1184 Alan Hayward <alan.hayward@arm.com>
1185 David Sherwood <david.sherwood@arm.com>
1187 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1188 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1189 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1190 * machmode.h (mode_size): Move earlier in file.
1191 (mode_precision): Likewise.
1192 (mode_inner): Likewise.
1193 (mode_nunits): Likewise.
1194 (mode_unit_size): Likewise.
1195 (unit_unit_precision): Likewise.
1196 (mode_wider): Likewise.
1197 (mode_2xwider): Likewise.
1198 (machine_mode): New class.
1199 (mode_to_bytes): New function.
1200 (mode_to_bits): Likewise.
1201 (mode_to_precision): Likewise.
1202 (mode_to_inner): Likewise.
1203 (mode_to_unit_size): Likewise.
1204 (mode_to_unit_precision): Likewise.
1205 (mode_to_nunits): Likewise.
1206 (GET_MODE_SIZE): Use mode_to_bytes.
1207 (GET_MODE_BITSIZE): Use mode_to_bits.
1208 (GET_MODE_PRECISION): Use mode_to_precision.
1209 (GET_MODE_INNER): Use mode_to_inner.
1210 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1211 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1212 (GET_MODE_NUNITS): Use mode_to_nunits.
1213 * system.h (ALWAYS_INLINE): New macro.
1214 * config/powerpcspe/powerpcspe-c.c
1215 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1216 int for arg1_mode and arg2_mode.
1218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1219 Alan Hayward <alan.hayward@arm.com>
1220 David Sherwood <david.sherwood@arm.com>
1222 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1223 Prefix mode names with E_ in case statements.
1224 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1225 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1226 (aarch64_split_simd_move): Likewise.
1227 (aarch64_gen_storewb_pair): Likewise.
1228 (aarch64_gen_loadwb_pair): Likewise.
1229 (aarch64_gen_store_pair): Likewise.
1230 (aarch64_gen_load_pair): Likewise.
1231 (aarch64_get_condition_code_1): Likewise.
1232 (aarch64_constant_pool_reload_icode): Likewise.
1233 (get_rsqrte_type): Likewise.
1234 (get_rsqrts_type): Likewise.
1235 (get_recpe_type): Likewise.
1236 (get_recps_type): Likewise.
1237 (aarch64_gimplify_va_arg_expr): Likewise.
1238 (aarch64_simd_container_mode): Likewise.
1239 (aarch64_emit_load_exclusive): Likewise.
1240 (aarch64_emit_store_exclusive): Likewise.
1241 (aarch64_expand_compare_and_swap): Likewise.
1242 (aarch64_gen_atomic_cas): Likewise.
1243 (aarch64_emit_bic): Likewise.
1244 (aarch64_emit_atomic_swap): Likewise.
1245 (aarch64_emit_atomic_load_op): Likewise.
1246 (aarch64_evpc_trn): Likewise.
1247 (aarch64_evpc_uzp): Likewise.
1248 (aarch64_evpc_zip): Likewise.
1249 (aarch64_evpc_ext): Likewise.
1250 (aarch64_evpc_rev): Likewise.
1251 (aarch64_evpc_dup): Likewise.
1252 (aarch64_gen_ccmp_first): Likewise.
1253 (aarch64_gen_ccmp_next): Likewise.
1254 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1255 (alpha_emit_xfloating_libcall): Likewise.
1256 (emit_insxl): Likewise.
1257 (alpha_arg_type): Likewise.
1258 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1259 (arc_preferred_simd_mode): Likewise.
1260 (arc_secondary_reload): Likewise.
1261 (get_arc_condition_code): Likewise.
1262 (arc_print_operand): Likewise.
1263 (arc_legitimate_constant_p): Likewise.
1264 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1265 * config/arc/arc.md (casesi_load): Likewise.
1266 (casesi_compact_jump): Likewise.
1267 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1268 (cc_use_register): Likewise.
1269 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1270 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1271 (arm_init_iwmmxt_builtins): Likewise.
1272 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1273 (neon_expand_vector_init): Likewise.
1274 (arm_attr_length_move_neon): Likewise.
1275 (maybe_get_arm_condition_code): Likewise.
1276 (arm_emit_vector_const): Likewise.
1277 (arm_preferred_simd_mode): Likewise.
1278 (arm_output_iwmmxt_tinsr): Likewise.
1279 (thumb1_output_casesi): Likewise.
1280 (thumb2_output_casesi): Likewise.
1281 (arm_emit_load_exclusive): Likewise.
1282 (arm_emit_store_exclusive): Likewise.
1283 (arm_expand_compare_and_swap): Likewise.
1284 (arm_evpc_neon_vuzp): Likewise.
1285 (arm_evpc_neon_vzip): Likewise.
1286 (arm_evpc_neon_vrev): Likewise.
1287 (arm_evpc_neon_vtrn): Likewise.
1288 (arm_evpc_neon_vext): Likewise.
1289 (arm_validize_comparison): Likewise.
1290 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1291 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1292 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1293 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1294 (c6x_preferred_simd_mode): Likewise.
1295 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1296 (epiphany_rtx_costs): Likewise.
1297 * config/epiphany/predicates.md (proper_comparison_operator):
1299 * config/frv/frv.c (condexec_memory_operand): Likewise.
1300 (frv_emit_move): Likewise.
1301 (output_move_single): Likewise.
1302 (output_condmove_single): Likewise.
1303 (frv_hard_regno_mode_ok): Likewise.
1304 (frv_matching_accg_mode): Likewise.
1305 * config/h8300/h8300.c (split_adds_subs): Likewise.
1306 (h8300_rtx_costs): Likewise.
1307 (h8300_print_operand): Likewise.
1308 (compute_mov_length): Likewise.
1309 (output_logical_op): Likewise.
1310 (compute_logical_op_length): Likewise.
1311 (compute_logical_op_cc): Likewise.
1312 (h8300_shift_needs_scratch_p): Likewise.
1313 (output_a_shift): Likewise.
1314 (compute_a_shift_length): Likewise.
1315 (compute_a_shift_cc): Likewise.
1316 (expand_a_rotate): Likewise.
1317 (output_a_rotate): Likewise.
1318 * config/i386/i386.c (classify_argument): Likewise.
1319 (function_arg_advance_32): Likewise.
1320 (function_arg_32): Likewise.
1321 (function_arg_64): Likewise.
1322 (function_value_64): Likewise.
1323 (ix86_gimplify_va_arg): Likewise.
1324 (ix86_legitimate_constant_p): Likewise.
1325 (put_condition_code): Likewise.
1326 (split_double_mode): Likewise.
1327 (ix86_avx256_split_vector_move_misalign): Likewise.
1328 (ix86_expand_vector_logical_operator): Likewise.
1329 (ix86_split_idivmod): Likewise.
1330 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1331 (ix86_build_const_vector): Likewise.
1332 (ix86_build_signbit_mask): Likewise.
1333 (ix86_match_ccmode): Likewise.
1334 (ix86_cc_modes_compatible): Likewise.
1335 (ix86_expand_branch): Likewise.
1336 (ix86_expand_sse_cmp): Likewise.
1337 (ix86_expand_sse_movcc): Likewise.
1338 (ix86_expand_int_sse_cmp): Likewise.
1339 (ix86_expand_vec_perm_vpermi2): Likewise.
1340 (ix86_expand_vec_perm): Likewise.
1341 (ix86_expand_sse_unpack): Likewise.
1342 (ix86_expand_int_addcc): Likewise.
1343 (ix86_split_to_parts): Likewise.
1344 (ix86_vectorize_builtin_gather): Likewise.
1345 (ix86_vectorize_builtin_scatter): Likewise.
1346 (avx_vpermilp_parallel): Likewise.
1347 (inline_memory_move_cost): Likewise.
1348 (ix86_tieable_integer_mode_p): Likewise.
1349 (x86_maybe_negate_const_int): Likewise.
1350 (ix86_expand_vector_init_duplicate): Likewise.
1351 (ix86_expand_vector_init_one_nonzero): Likewise.
1352 (ix86_expand_vector_init_one_var): Likewise.
1353 (ix86_expand_vector_init_concat): Likewise.
1354 (ix86_expand_vector_init_interleave): Likewise.
1355 (ix86_expand_vector_init_general): Likewise.
1356 (ix86_expand_vector_set): Likewise.
1357 (ix86_expand_vector_extract): Likewise.
1358 (emit_reduc_half): Likewise.
1359 (ix86_emit_i387_round): Likewise.
1360 (ix86_mangle_type): Likewise.
1361 (ix86_expand_round_sse4): Likewise.
1362 (expand_vec_perm_blend): Likewise.
1363 (canonicalize_vector_int_perm): Likewise.
1364 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1365 (expand_vec_perm_1): Likewise.
1366 (expand_vec_perm_interleave3): Likewise.
1367 (expand_vec_perm_even_odd_pack): Likewise.
1368 (expand_vec_perm_even_odd_1): Likewise.
1369 (expand_vec_perm_broadcast_1): Likewise.
1370 (ix86_vectorize_vec_perm_const_ok): Likewise.
1371 (ix86_expand_vecop_qihi): Likewise.
1372 (ix86_expand_mul_widen_hilo): Likewise.
1373 (ix86_expand_sse2_abs): Likewise.
1374 (ix86_expand_pextr): Likewise.
1375 (ix86_expand_pinsr): Likewise.
1376 (ix86_preferred_simd_mode): Likewise.
1377 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1378 * config/i386/sse.md (*andnot<mode>3): Likewise.
1379 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1380 (*<code><mode>3): Likewise.
1381 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1382 (ia64_expand_atomic_op): Likewise.
1383 (ia64_arg_type): Likewise.
1384 (ia64_mode_to_int): Likewise.
1385 (ia64_scalar_mode_supported_p): Likewise.
1386 (ia64_vector_mode_supported_p): Likewise.
1387 (expand_vec_perm_broadcast): Likewise.
1388 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1389 (iq2000_function_arg_advance): Likewise.
1390 (iq2000_function_arg): Likewise.
1391 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1392 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1393 (m68k_libcall_value): Likewise.
1394 (m68k_function_value): Likewise.
1395 (sched_attr_op_type): Likewise.
1396 * config/mcore/mcore.c (mcore_output_move): Likewise.
1397 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1399 (microblaze_function_arg): Likewise.
1400 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1401 (mips_print_operand): Likewise.
1402 (mips_mode_ok_for_mov_fmt_p): Likewise.
1403 (mips_vector_mode_supported_p): Likewise.
1404 (mips_preferred_simd_mode): Likewise.
1405 (mips_expand_vpc_loongson_even_odd): Likewise.
1406 (mips_expand_vec_unpack): Likewise.
1407 (mips_expand_vi_broadcast): Likewise.
1408 (mips_expand_vector_init): Likewise.
1409 (mips_expand_vec_reduc): Likewise.
1410 (mips_expand_msa_cmp): Likewise.
1411 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1412 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1413 (cc_flags_for_mode): Likewise.
1414 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1415 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1416 (nds32_output_casesi_pc_relative): Likewise.
1417 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1418 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1419 (nvptx_gen_unpack): Likewise.
1420 (nvptx_gen_pack): Likewise.
1421 (nvptx_gen_shuffle): Likewise.
1422 (nvptx_gen_wcast): Likewise.
1423 (nvptx_preferred_simd_mode): Likewise.
1424 * config/pa/pa.c (pa_secondary_reload): Likewise.
1425 * config/pa/predicates.md (base14_operand): Likewise.
1426 * config/powerpcspe/powerpcspe-c.c
1427 (altivec_resolve_overloaded_builtin): Likewise.
1428 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1430 (rs6000_preferred_simd_mode): Likewise.
1431 (output_vec_const_move): Likewise.
1432 (rs6000_expand_vector_extract): Likewise.
1433 (rs6000_split_vec_extract_var): Likewise.
1434 (reg_offset_addressing_ok_p): Likewise.
1435 (rs6000_legitimate_offset_address_p): Likewise.
1436 (rs6000_legitimize_address): Likewise.
1437 (rs6000_emit_set_const): Likewise.
1438 (rs6000_const_vec): Likewise.
1439 (rs6000_emit_move): Likewise.
1440 (spe_build_register_parallel): Likewise.
1441 (rs6000_darwin64_record_arg_recurse): Likewise.
1442 (swap_selector_for_mode): Likewise.
1443 (spe_init_builtins): Likewise.
1444 (paired_init_builtins): Likewise.
1445 (altivec_init_builtins): Likewise.
1446 (do_load_for_compare): Likewise.
1447 (rs6000_generate_compare): Likewise.
1448 (rs6000_expand_float128_convert): Likewise.
1449 (emit_load_locked): Likewise.
1450 (emit_store_conditional): Likewise.
1451 (rs6000_output_function_epilogue): Likewise.
1452 (rs6000_handle_altivec_attribute): Likewise.
1453 (rs6000_function_value): Likewise.
1454 (emit_fusion_gpr_load): Likewise.
1455 (emit_fusion_p9_load): Likewise.
1456 (emit_fusion_p9_store): Likewise.
1457 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1458 (fusion_gpr_mem_load): Likewise.
1459 (fusion_addis_mem_combo_load): Likewise.
1460 (fusion_addis_mem_combo_store): Likewise.
1461 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1462 (fusion_gpr_mem_load): Likewise.
1463 (fusion_addis_mem_combo_load): Likewise.
1464 (fusion_addis_mem_combo_store): Likewise.
1465 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1467 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1468 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1469 (rs6000_preferred_simd_mode): Likewise.
1470 (output_vec_const_move): Likewise.
1471 (rs6000_expand_vector_extract): Likewise.
1472 (rs6000_split_vec_extract_var): Likewise.
1473 (reg_offset_addressing_ok_p): Likewise.
1474 (rs6000_legitimate_offset_address_p): Likewise.
1475 (rs6000_legitimize_address): Likewise.
1476 (rs6000_emit_set_const): Likewise.
1477 (rs6000_const_vec): Likewise.
1478 (rs6000_emit_move): Likewise.
1479 (rs6000_darwin64_record_arg_recurse): Likewise.
1480 (swap_selector_for_mode): Likewise.
1481 (paired_init_builtins): Likewise.
1482 (altivec_init_builtins): Likewise.
1483 (rs6000_expand_float128_convert): Likewise.
1484 (emit_load_locked): Likewise.
1485 (emit_store_conditional): Likewise.
1486 (rs6000_output_function_epilogue): Likewise.
1487 (rs6000_handle_altivec_attribute): Likewise.
1488 (rs6000_function_value): Likewise.
1489 (emit_fusion_gpr_load): Likewise.
1490 (emit_fusion_p9_load): Likewise.
1491 (emit_fusion_p9_store): Likewise.
1492 * config/rx/rx.c (rx_gen_move_template): Likewise.
1493 (flags_from_mode): Likewise.
1494 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1495 (s390_slb_comparison): Likewise.
1496 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1497 (s390_vector_mode_supported_p): Likewise.
1498 (s390_cc_modes_compatible): Likewise.
1499 (s390_match_ccmode_set): Likewise.
1500 (s390_canonicalize_comparison): Likewise.
1501 (s390_emit_compare_and_swap): Likewise.
1502 (s390_branch_condition_mask): Likewise.
1503 (s390_rtx_costs): Likewise.
1504 (s390_secondary_reload): Likewise.
1505 (__SECONDARY_RELOAD_CASE): Likewise.
1506 (s390_expand_cs): Likewise.
1507 (s390_preferred_simd_mode): Likewise.
1508 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1509 * config/sh/sh.c (sh_print_operand): Likewise.
1510 (dump_table): Likewise.
1511 (sh_secondary_reload): Likewise.
1512 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1513 * config/sh/sh.md (casesi_worker_1): Likewise.
1514 (casesi_worker_2): Likewise.
1515 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1516 (fcc_comparison_operator): Likewise.
1517 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1518 (emit_soft_tfmode_cvt): Likewise.
1519 (sparc_preferred_simd_mode): Likewise.
1520 (output_cbranch): Likewise.
1521 (sparc_print_operand): Likewise.
1522 (sparc_expand_vec_perm_bmask): Likewise.
1523 (vector_init_bshuffle): Likewise.
1524 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1525 (spu_vector_mode_supported_p): Likewise.
1526 (spu_expand_insv): Likewise.
1527 (spu_emit_branch_or_set): Likewise.
1528 (spu_handle_vector_attribute): Likewise.
1529 (spu_builtin_splats): Likewise.
1530 (spu_builtin_extract): Likewise.
1531 (spu_builtin_promote): Likewise.
1532 (spu_expand_sign_extend): Likewise.
1533 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1534 (tilegx_simd_int): Likewise.
1535 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1536 (tilepro_simd_int): Likewise.
1537 * config/v850/v850.c (const_double_split): Likewise.
1538 (v850_print_operand): Likewise.
1539 (ep_memory_offset): Likewise.
1540 * config/vax/vax.c (vax_rtx_costs): Likewise.
1541 (vax_output_int_move): Likewise.
1542 (vax_output_int_add): Likewise.
1543 (vax_output_int_subtract): Likewise.
1544 * config/visium/predicates.md (visium_branch_operator): Likewise.
1545 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1546 (visium_print_operand_address): Likewise.
1547 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1548 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1549 (xtensa_expand_conditional_branch): Likewise.
1550 (xtensa_copy_incoming_a7): Likewise.
1551 (xtensa_output_literal): Likewise.
1552 * dfp.c (decimal_real_maxval): Likewise.
1553 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1556 Alan Hayward <alan.hayward@arm.com>
1557 David Sherwood <david.sherwood@arm.com>
1559 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1560 (mode_nunits_inline): Likewise.
1561 (mode_inner_inline): Likewise.
1562 (mode_unit_size_inline): Likewise.
1563 (mode_unit_precision_inline): Likewise.
1564 (emit_insn_modes_h): Likewise. Also emit a #define of the
1566 (emit_mode_wider): Add an E_ prefix to mode names.
1567 (emit_mode_complex): Likewise.
1568 (emit_mode_inner): Likewise.
1569 (emit_mode_adjustments): Likewise.
1570 (emit_mode_int_n): Likewise.
1571 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1572 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1573 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1574 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1575 (CRC32_BUILTIN, ENTRY): Likewise.
1576 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1577 (aarch64_pop_regs): Likewise.
1578 (aarch64_process_components): Likewise.
1579 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1580 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1581 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1582 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1583 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1584 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1586 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1587 (pcmpistr): Likewise.
1588 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1589 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1590 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1592 * config/rl78/rl78.c (mduc_regs): Likewise.
1593 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1594 (htm_expand_builtin): Likewise.
1595 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1596 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1597 * config/xtensa/xtensa.c (print_operand): Likewise.
1598 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1599 (NUM_MODE_VECTOR_INT): Likewise.
1600 * genoutput.c (null_operand): Likewise.
1601 (output_operand_data): Likewise.
1602 * genrecog.c (print_parameter_value): Likewise.
1603 * lra.c (debug_operand_data): Likewise.
1605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1607 * dwarf2out.c (scompare_loc_descriptor_wide)
1608 (scompare_loc_descriptor_narrow): New functions, split out from...
1609 (scompare_loc_descriptor): ...here.
1610 * expmed.c (emit_store_flag_int): New function, split out from...
1611 (emit_store_flag): ...here.
1613 2017-08-30 Richard Biener <rguenther@suse.de>
1615 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1616 (dwarf2out_early_finish): Move setting of AT_pubnames from
1617 early debug output to early finish.
1619 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1621 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1622 and -mdata-region to the assembler.
1624 2017-08-30 Richard Biener <rguenther@suse.de>
1626 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1628 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1629 (dwarf2out_early_global_decl): Only generate a DIE for the
1630 abstract origin if it doesn't already exist or is a declaration DIE.
1631 (resolve_addr): Do not add the linkage name twice when
1632 generating a stub DIE for the DW_TAG_GNU_call_site target.
1634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1636 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1637 Use machine_mode rather than int for arg1_mode.
1639 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1642 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1643 that the second argument of the built-in functions to unpack
1644 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1645 switch statement instead a lot of if statements.
1646 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1647 Allow 64-bit values to be in Altivec registers as well as
1648 traditional floating point registers.
1649 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1651 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1653 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1654 MAX_REGS_PER_ADDRESS == 1.
1656 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1658 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1659 * config/i386/i386.c (ix86_option_override_internal): Simplify
1660 setting of opts->x_flag_entry.
1662 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1663 Jakub Jelinek <jakub@redhat.com>
1664 Richard Biener <rguenther@suse.de>
1666 PR tree-optimization/81503
1667 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1668 folded constant fits in the target type; reorder tests for clarity.
1670 2017-08-29 Martin Liska <mliska@suse.cz>
1672 * passes.def: Include pass_lower_switch.
1673 * stmt.c (dump_case_nodes): Remove and move to
1674 tree-switch-conversion.
1675 (case_values_threshold): Likewise.
1676 (expand_switch_as_decision_tree_p): Likewise.
1677 (emit_case_decision_tree): Likewise.
1678 (expand_case): Likewise.
1679 (balance_case_nodes): Likewise.
1680 (node_has_low_bound): Likewise.
1681 (node_has_high_bound): Likewise.
1682 (node_is_bounded): Likewise.
1683 (emit_case_nodes): Likewise.
1684 (struct simple_case_node): New struct.
1685 (add_case_node): Remove.
1686 (emit_case_dispatch_table): Use vector instead of case_list.
1687 (reset_out_edges_aux): Remove.
1688 (compute_cases_per_edge): Likewise.
1689 (expand_case): Build list of simple_case_node.
1690 (expand_sjlj_dispatch_table): Use it.
1691 * tree-switch-conversion.c (struct case_node): Moved from
1692 stmt.c and adjusted.
1693 (emit_case_nodes): Likewise.
1694 (node_has_low_bound): Likewise.
1695 (node_has_high_bound): Likewise.
1696 (node_is_bounded): Likewise.
1697 (case_values_threshold): Likewise.
1698 (reset_out_edges_aux): Likewise.
1699 (compute_cases_per_edge): Likewise.
1700 (add_case_node): Likewise.
1701 (dump_case_nodes): Likewise.
1702 (balance_case_nodes): Likewise.
1703 (expand_switch_as_decision_tree_p): Likewise.
1704 (emit_jump): Likewise.
1705 (emit_case_decision_tree): Likewise.
1706 (try_switch_expansion): Likewise.
1707 (do_jump_if_equal): Likewise.
1708 (emit_cmp_and_jump_insns): Likewise.
1709 (fix_phi_operands_for_edge): New function.
1710 (record_phi_operand_mapping): Likewise.
1711 (class pass_lower_switch): New pass.
1712 (pass_lower_switch::execute): New function.
1713 (make_pass_lower_switch): Likewise.
1714 (conditional_probability):
1715 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1716 * tree-pass.h: Add make_pass_lower_switch.
1718 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1721 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1724 2017-08-29 Richard Biener <rguenther@suse.de>
1726 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1727 we do not add a DW_AT_inline attribute twice.
1728 (gen_subprogram_die): Remove code setting DW_AT_inline on
1729 DECL_ABSTRACT_P nodes.
1731 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1733 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1734 calls to internal functions.
1735 (gimplify_modify_expr): Likewise.
1736 * tree-call-cdce.c (use_internal_fn): Likewise.
1737 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1738 (convert_to_divmod): Set the nothrow flag.
1739 * tree-if-conv.c (predicate_mem_writes): Likewise.
1740 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1741 (vectorizable_call): Likewise.
1742 (vectorizable_store): Likewise.
1743 (vectorizable_load): Likewise.
1744 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1745 (vect_recog_mask_conversion_pattern): Likewise.
1747 2017-08-29 Martin Liska <mliska@suse.cz>
1750 * gcc.c (driver_handle_option): Add new argument.
1751 * opts-common.c (handle_option): Pass
1752 target_option_override_hook.
1753 * opts-global.c (lang_handle_option): Add new option.
1754 (set_default_handlers): Add new argument.
1755 (decode_options): Likewise.
1756 * opts.c (target_handle_option): Likewise.
1757 (common_handle_option): Call target_option_override_hook.
1758 * opts.h (struct cl_option_handler_func): Add hook for
1759 target option override.
1760 (struct cl_option_handlers): Likewise.
1761 (set_default_handlers): Add new argument.
1762 (decode_options): Likewise.
1763 (common_handle_option): Likewise.
1764 (target_handle_option): Likewise.
1765 * toplev.c (toplev::main): Pass targetm.target_option.override
1768 2017-08-29 Richard Biener <rguenther@suse.de>
1769 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1771 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1772 life to the active subtree.
1774 2017-08-28 Jeff Law <law@redhat.com>
1776 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1777 derive_equivalences.
1778 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1780 (edge_info::derive_equivalences): New private member function
1782 * tree-ssa-dom.c (class edge_info): Changed from a struct
1783 to a class. Add ctor/dtor, methods and data members.
1784 (edge_info::edge_info): Renamed from allocate_edge_info.
1785 Initialize additional members.
1786 (edge_info::~edge_info): New.
1787 (free_dom_edge_info): Delete the edge info.
1788 (record_edge_info): Use new class & associated member functions.
1789 Tighten forms for testing for edge equivalences.
1790 (record_temporary_equivalences): Iterate over the simple
1791 equivalences rather than assuming there's only one per edge.
1792 (cprop_into_successor_phis): Iterate over the simple
1793 equivalences rather than assuming there's only one per edge.
1794 (optimize_stmt): Use operand_equal_p rather than pointer
1795 equality for mini-DSE code.
1797 2017-08-28 Nathan Sidwell <nathan@acm.org>
1799 * gcc.c (execute): Fold SIGPIPE handling into switch
1800 statement. Adjust internal error message.
1802 2017-08-28 Richard Biener <rguenther@suse.de>
1805 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1806 Do nothing for removed DIEs.
1808 2017-08-28 Richard Biener <rguenther@suse.de>
1810 PR tree-optimization/81977
1811 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1814 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1817 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1818 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1819 using targetm.gen_mem_thread_fence.
1821 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1824 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1825 predicate to register_operand. Reorder operands.
1826 (*btr<mode>): Ditto.
1827 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1828 (*btr<mode>_mask): Ditto.
1830 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1832 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1833 * config/rs6000/xmmintrin.h: New file.
1834 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1836 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1839 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1840 parameter and_insn and return it.
1841 (recombine_lvx_pattern): Insert a copy to ensure availability of
1842 the base register of the copied masking operation at the point of
1843 the instruction replacement.
1844 (recombine_stvx_pattern): Likewise.
1846 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1848 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1849 undocumented switches.
1850 (-mpower9-dform-vector): Likewise.
1851 (-mpower9-dform): Likewise.
1852 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1853 comments to delete references to -mpower9-dform* switches.
1854 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1855 Delete reference to -mpower9-dform* switches, test for
1856 -mpower9-vector instead.
1857 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1858 (OTHER_P9_VECTOR_MASKS): Likewise.
1859 (POWERPC_MASKS): Likewise.
1860 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1861 tests against -mpower9-dform* to -mpower9-vector. Delete code
1862 that checked for -mpower9-dform* consistancy with other options.
1863 Add test for -mpower9-misc to enable other power9 switches.
1864 (rs6000_init_hard_regno_mode_ok): Likewise.
1865 (rs6000_option_override_internal): Likewise.
1866 (rs6000_emit_prologue): Likewise.
1867 (rs6000_emit_epilogue): Likewise.
1868 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1869 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1870 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1872 (emit_fusion_p9_store): Likewise.
1873 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1874 resetting these macros if the assembler does not support ISA 3.0
1876 (TARGET_P9_DFORM_VECTOR): Likewise.
1877 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1878 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1880 2017-08-25 Alan Modra <amodra@gmail.com>
1882 PR rtl-optimization/81747
1883 * cse.c (cse_extended_basic_block): Don't attempt to record
1884 equivalences for degenerate conditional branches that branch
1885 to their fall-through.
1887 2017-08-24 Martin Sebor <msebor@redhat.com>
1890 * gimple-fold.c (size_must_be_zero_p): New function.
1891 (gimple_fold_builtin_memory_op): Call it.
1893 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1895 * config/rs6000/mm_malloc.h: New file.
1897 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1899 PR tree-optimization/81913
1900 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1901 analysis when either IVs in condition can wrap.
1903 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1905 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1906 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1908 2017-08-24 Richard Biener <rguenther@suse.de>
1911 * targhooks.c (default_target_can_inline_p): Properly
1912 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1913 is present and always compare.
1914 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1915 infer -mfpmath=sse from TARGET_SSE_P.
1916 (ix86_can_inline_p): Properly use target_option_default_node when
1917 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1919 2017-08-24 Richard Biener <rguenther@suse.de>
1922 * dwarf2out.c (output_die): Handle flag_generate_offload like
1924 (output_comp_unit): Likewise.
1925 (gen_array_type_die): Likewise.
1926 (dwarf2out_early_finish): Likewise.
1927 (note_variable_value_in_expr): Likewise.
1928 (dwarf2out_finish): Likewise. Adjust assert.
1929 * cgraphunit.c (symbol_table::compile): Move setting of
1930 flag_generate_offload earlier ...
1931 (symbol_table::finalize_compilation_unit): ... here, before
1932 early debug finalization.
1934 2017-08-24 Richard Biener <rguenther@suse.de>
1936 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1937 and ipa-fnsummary.h.
1938 (ix86_can_inline_p): When ix86_fpmath flags do not match
1939 check whether the callee uses FP math at all.
1941 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1944 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1945 instead of TYPE_PRECISION.
1947 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1948 Alan Hayward <alan.hayward@arm.com>
1949 David Sherwood <david.sherwood@arm.com>
1951 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1953 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1955 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1957 (ix86_frame::stack_realign_allocate): New field.
1958 (struct machine_frame_state): Modify comments.
1959 (machine_frame_state::sp_realigned_fp_end): New field.
1960 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1962 (sp_valid_at): Add assertion to assure no attempt to access invalid
1963 offset of a realigned stack.
1964 (fp_valid_at): Likewise.
1965 (choose_baseaddr): Modify comments.
1966 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1967 ix86_expand_prologue.
1968 (ix86_expand_prologue): Modify stack realignment and allocation.
1969 (ix86_expand_epilogue): Modify comments.
1970 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1971 avx2_runtime, avx512f, and avx512f_runtime.
1973 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1975 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1976 (mstackrealign): Do not init to -1.
1977 * config/i386/i386.c (ix86_option_override_internal):
1978 Check opts_set, not opts when setting default value of
1979 opts->x_ix86_force_align_arg_pointer.
1981 2017-08-23 Richard Biener <rguenther@suse.de>
1983 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1984 lang_hooks.decl_printable_name.
1985 * print-rtl-function.c (print_rtx_function): Likewise.
1986 * tree-pretty-print.c (dump_function_header): Likewise.
1988 2017-08-23 Richard Biener <rguenther@suse.de>
1991 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1992 -g0 at compile-time.
1994 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1997 * doc/sourcebuild.texi (Other hardware attributes):
2000 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2003 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2004 Fix single-use check.
2006 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2008 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2010 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2012 * doc/install.texi: Modify to add more details on running selected
2015 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2017 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2018 is combined with -mabi=ms.
2019 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2022 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2024 PR tree-optimization/81488
2025 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2026 and cached_basis fields.
2027 (MAX_SPREAD): New constant.
2028 (alloc_cand_and_find_basis): Initialize new fields.
2029 (clear_visited): New function.
2030 (create_phi_basis_1): Rename from create_phi_basis, set visited
2031 and cached_basis fields.
2032 (create_phi_basis): New wrapper function.
2033 (phi_add_costs_1): Rename from phi_add_costs, add spread
2034 parameter, set visited field, short-circuit when limits reached.
2035 (phi_add_costs): New wrapper function.
2036 (record_phi_increments_1): Rename from record_phi_increments, set
2038 (record_phi_increments): New wrapper function.
2039 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2040 (phi_incr_cost): New wrapper function.
2041 (all_phi_incrs_profitable_1): Rename from
2042 all_phi_incrs_profitable, set visited field.
2043 (all_phi_incrs_profitable): New wrapper function.
2045 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2046 Alan Hayward <alan.hayward@arm.com>
2047 David Sherwood <david.sherwood@arm.com>
2049 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2050 that takes the outer and inner modes.
2051 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2052 comparison as the canonical test for a paradoxical subreg.
2053 * combine.c (simplify_set): Use paradoxical_subreg_p.
2054 (make_extraction): Likewise.
2055 (force_to_mode): Likewise.
2056 (rtx_equal_for_field_assignment_p): Likewise.
2057 (gen_lowpart_for_combine): Likewise.
2058 (simplify_comparison): Likewise.
2059 * cse.c (equiv_constant): Likewise.
2060 * expmed.c (store_bit_field_1): Likewise.
2061 * final.c (alter_subreg): Likewise.
2062 * fwprop.c (propagate_rtx): Likewise.
2063 (forward_propagate_subreg): Likewise.
2064 * ira-conflicts.c (ira_build_conflicts): Likewise.
2065 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2066 * lra-constraints.c (curr_insn_transform): Likewise.
2067 (split_reg): Likewise.
2068 * lra-eliminations.c (move_plus_up): Likewise.
2069 (lra_eliminate_regs_1): Likewise.
2070 * recog.c (general_operand): Likewise.
2071 * ree.c (combine_reaching_defs): Likewise.
2072 * reload.c (push_reload): Likewise.
2073 (find_reloads): Likewise.
2074 * reload1.c (elimination_effects): Likewise.
2075 (compute_reload_subreg_offset): Likewise.
2076 (choose_reload_regs): Likewise.
2077 * rtlanal.c (subreg_lsb_1): Likewise.
2078 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2079 (simplify_subreg): Likewise.
2080 * var-tracking.c (track_loc_p): Likewise.
2081 * emit-rtl.c (byte_lowpart_offset): Likewise.
2082 (paradoxical_subreg_p): Delete out-of-line definition.
2084 2017-08-22 Jeff Law <law@redhat.com>
2086 PR tree-optimization/81741
2087 PR tree-optimization/71947
2088 * tree-ssa-dom.c: Include tree-inline.h.
2089 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2090 equivalences if one is more expensive to compute than the other.
2091 * tree-ssa-scopedtables.h (class const_or_copies): Make
2092 record_const_or_copy_raw method private.
2093 (class avail_exprs_stack): New method simplify_binary_operation.
2094 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2095 avail_exprs_stack::simplify_binary_operation as needed.
2096 (avail_exprs_stack::simplify_binary_operation): New function.
2098 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2100 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2101 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2102 (DOT_SYMBOLS): Likewise.
2103 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2104 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2105 (RS6000_ABI_NAME): Likewise.
2106 (TARGET_CMODEL): Likewise.
2107 (TOC_SECTION_ASM_OP): Likewise.
2108 (SET_CMODEL): New macro.
2109 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2111 2017-08-22 Richard Biener <rguenther@suse.de>
2113 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2114 to allow for free-lang-data replacements similar to verify_type_variant.
2116 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2117 Alan Hayward <alan.hayward@arm.com>
2118 David Sherwood <david.sherwood@arm.com>
2120 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2121 VOIDmode for the LABEL_REF.
2123 2017-08-22 Richard Biener <rguenther@suse.de>
2125 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2127 2017-08-22 Richard Biener <rguenther@suse.de>
2129 * common.opt (feliminate-dwarf2-dups): Ignore.
2130 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2131 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2132 same_die_p_wrap, compute_section_prefix,
2133 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2134 (comdat_symbol_id, comdat_symbol_number): Likewise.
2135 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2137 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2138 (output_die): Mark unreachable path unreachable.
2139 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2140 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2141 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2142 (dwarf2out_early_finish): Likewise.
2144 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2146 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2148 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2151 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2152 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2153 (avr_attribute_table) <io, io_low, address>: Initialize
2154 .decl_required with true.
2156 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2158 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2159 undocumented debugging options.
2160 (-mvsx-scalar-double): Likewise.
2161 (-mallow-df-permute): Likewise.
2162 (-mvectorize-builtins): Likewise.
2163 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2164 (rs6000_builtin_vectorized_function): Likewise.
2165 (rs6000_builtin_md_vectorized_function): Likewise.
2166 (rs6000_opt_vars): Likewise.
2168 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2171 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2172 (*btrq_imm): Rename from *btrq.
2173 (*btcq_imm): Rename from *btcq.
2174 (btsc): New code attribute.
2175 (*<btsc><mode>): New insn pattern.
2176 (*btr<mode>): Ditto.
2177 (*<btsc><mode>_mask): New insn_and_split pattern.
2178 (*btr<mode>_mask): Ditto.
2180 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2181 Alan Hayward <alan.hayward@arm.com>
2182 David Sherwood <david.sherwood@arm.com>
2184 * function.c (pad_below): Simplify padding calculation.
2186 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2187 Alan Hayward <alan.hayward@arm.com>
2188 David Sherwood <david.sherwood@arm.com>
2190 * target.def (function_prologue): Remove frame size argument.
2191 (function_epilogue): Likewise.
2192 * doc/tm.texi: Regenerate.
2193 * final.c (final_start_function): Update call to function_prologue.
2194 (final_end_function): Update call to function_epilogue.
2195 (default_function_pro_epilogue): Remove frame size argument.
2196 * output.h (default_function_pro_epilogue): Likewise.
2197 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2198 (arm_output_function_prologue): Likewise.
2199 * config/frv/frv.c (frv_function_prologue): Likewise.
2200 (frv_function_epilogue): Likewise.
2201 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2202 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2203 (ia64_output_function_epilogue): Likewise.
2204 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2205 (m32r_output_function_epilogue): Likewise.
2206 * config/microblaze/microblaze.c (microblaze_function_prologue)
2207 (microblaze_function_epilogue): Likewise.
2208 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2209 (mips_output_function_epilogue): Likewise.
2210 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2211 (mmix_target_asm_function_epilogue): Likewise.
2212 * config/msp430/msp430.c (msp430_start_function): Likewise.
2213 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2214 (nds32_asm_function_epilogue): Likewise.
2215 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2216 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2217 (pa_output_function_epilogue): Likewise.
2218 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2219 (rs6000_output_function_epilogue): Likewise.
2220 * config/rl78/rl78.c (rl78_start_function): Likewise.
2221 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2222 (rs6000_output_function_epilogue): Likewise.
2223 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2224 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2225 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2226 (sparc_asm_function_epilogue): Likewise.
2228 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2230 * tree.h (type_has_mode_precision_p): New function.
2231 * convert.c (convert_to_integer_1): Use it.
2232 * expr.c (expand_expr_real_2): Likewise.
2233 (expand_expr_real_1): Likewise.
2234 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2235 * match.pd: Likewise.
2236 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2237 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2238 * tree-tailcall.c (process_assignment): Likewise.
2239 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2240 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2241 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2242 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2243 (vectorizable_assignment): Likewise.
2244 (vectorizable_shift): Likewise.
2245 (vectorizable_operation): Likewise.
2246 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2248 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2250 * match.pd: Add pow (C, x) simplification.
2252 2017-08-21 Richard Biener <rguenther@suse.de>
2254 PR tree-optimization/81900
2255 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2256 for blocks with abnormal predecessors.
2257 (compute_antic): Do not set visited flag prematurely.
2259 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2262 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2264 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2266 * stor-layout.h (vector_type_mode): Move to...
2267 * tree.h (vector_type_mode): ...here.
2268 * stor-layout.c (vector_type_mode): Move to...
2269 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2271 2017-08-21 Richard Biener <rguenther@suse.de>
2273 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2274 register_external_die hooks.
2275 (debug_false_tree_charstarstar_uhwistar): Declare.
2276 (debug_nothing_tree_charstar_uhwi): Likewise.
2277 * debug.c (do_nothing_debug_hooks): Adjust.
2278 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2279 (debug_nothing_tree_charstar_uhwi): Likewise.
2280 * dbxout.c (dbx_debug_hooks): Adjust.
2281 (xcoff_debug_hooks): Likewise.
2282 * sdbout.c (sdb_debug_hooks): Likewise.
2283 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2284 * dwarf2out.c (macinfo_label_base): New global.
2285 (dwarf2out_register_external_die): New function for the
2286 register_external_die hook.
2287 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2288 (dwarf2_debug_hooks): Use them.
2289 (dwarf2_lineno_debug_hooks): Adjust.
2290 (struct die_struct): Add with_offset flag.
2291 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2292 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2293 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2294 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2295 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2296 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2297 defining section names for the early LTO debug variants.
2298 (reset_indirect_string): New helper.
2299 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2300 (print_dw_val): Add support for offsetted symbol references.
2301 (get_ultimate_context): Split out from is_cxx.
2302 (is_cxx): Use get_ultimate_context.
2303 (is_fortran): Add decl overload.
2304 (compute_comp_unit_symbol): Split out worker from
2305 compute_section_prefix.
2306 (compute_section_prefix): Call compute_comp_unit_symbol and
2307 set comdat_type_p here.
2308 (output_die): Skip DIE symbol output for the LTO added one.
2309 Handle DIE symbol references with offset.
2310 (output_comp_unit): Guard section name mangling properly.
2311 For LTO debug sections emit a symbol at the section beginning
2312 which we use to refer to its DIEs.
2313 (add_abstract_origin_attribute): For DIEs registered via
2314 dwarf2out_register_external_die directly refer to the early
2315 DIE rather than indirectly through the shadow one we created.
2316 Remove obsolete call to dwarf2out_abstract_function for
2317 non-function/block origins.
2318 (gen_array_type_die): When generating early LTO debug do
2319 not emit DW_AT_string_length.
2320 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2321 late when in LTO. As suggested place a gcc_unreachable for
2322 the DECL_ABSTRACT_P case.
2323 (gen_subprogram_die): Avoid another specification DIE
2324 for early built declarations/definitions for the late LTO case.
2325 (gen_variable_die): Add type references for late duplicated VLA dies
2327 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2328 we have the abstract instance already.
2329 (process_scope_var): Adjust decl DIE contexts in LTO which
2330 first puts them in limbo.
2331 (gen_decl_die): Do not generate type DIEs late apart from
2332 types for VLAs or for decls we do not yet have a DIE. Do not
2333 call dwarf2out_abstract_function late.
2334 (dwarf2out_early_global_decl): Make sure to create DIEs
2335 for abstract instances of a decl first.
2336 (dwarf2out_late_global_decl): Adjust comment.
2337 (output_macinfo_op): With multiple macro sections use
2338 macinfo_label_base to distinguish labels.
2339 (output_macinfo): Likewise. Update macinfo_label_base.
2340 Pass in the line info label.
2341 (note_variable_value_in_expr): When generating LTO resolve
2342 all variable values here by generating DIEs as needed.
2343 (init_sections_and_labels): Add early LTO debug flag parameter
2344 and generate different sections and names if set. Add generation
2345 counter for the labels so we can have multiple of them.
2346 (reset_dies): Helper to allow DIEs to be output multiple times.
2347 (dwarf2out_finish): When outputting DIEs to the fat part of an
2348 LTO object first reset DIEs.
2349 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2350 (modified_type_die): Check for decl_ultimate_origin being self
2352 (gen_type_die_with_usage): Likewise.
2353 (gen_typedef_die): Allow decl_ultimate_origin being self.
2354 (set_decl_abstract_flags): Remove.
2355 (set_block_abstract_flags): Likewise.
2356 (dwarf2out_abstract_function): Treat the early generated DIEs
2357 as the abstract copy and only add DW_AT_inline and
2358 DW_AT_artificial here and call set_decl_origin_self.
2359 If the DIE has an abstract origin don't do anything.
2360 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2361 if we have none yet (Go fails to build one, PR78628).
2362 (variably_modified_type_p): Prevent endless recursion for Ada
2363 cyclic pointer types.
2364 * lto-streamer-in.c: Include debug.h.
2365 (dref_queue): New global.
2366 (lto_read_tree_1): Stream in DIE references.
2367 (lto_input_tree): Register DIE references.
2368 (input_function): Stream DECL_DEBUG_ARGS.
2369 * lto-streamer-out.c: Include debug.h.
2370 (lto_write_tree_1): Output DIE references.
2371 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2372 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2373 (output_function): Stream DECL_DEBUG_ARGS.
2374 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2375 Stream DECL_ABSTRACT_ORIGIN.
2376 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2377 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2378 DECL_CONTEXT for file-scope decls.
2379 * lto-streamer.h (struct dref_entry): Declare.
2380 (dref_queue): Likewise.
2381 * cfgexpand.c (pass_expand::execute): Do not call the
2382 outlining_inline_function hook here.
2383 * lto-wrapper.c (debug_obj): New global.
2384 (tool_cleanup): Unlink it if required.
2385 (debug_objcopy): New function.
2386 (run_gcc): Handle early debug sections in the IL files by
2387 extracting them to separate files, partially linkin them and
2388 feeding the result back as result to the linker.
2389 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2390 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2391 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2392 sections into a separate segment.
2393 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2395 (darwin_asm_dwarf_section): Likewise.
2396 (darwin_asm_output_dwarf_offset): Likewise.
2397 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2399 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2400 Alan Hayward <alan.hayward@arm.com>
2401 David Sherwood <david.sherwood@arm.com>
2403 * read-md.h (md_reader::record_potential_iterator_use): Replace
2404 pointer argument with an rtx and an index.
2405 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2406 (apply_mode_iterator): Likewise.
2407 (apply_code_iterator): Likewise.
2408 (apply_int_iterator): Likewise.
2409 (apply_subst_iterator): Likewise.
2410 (record_iterator_use): Likewise.
2411 (record_attribute_use): Likewise.
2412 (md_reader::record_potential_iterator_use): Likewise. Update calls
2413 to record_iterator_use and apply_iterator.
2414 (iterator_use): Replace ptr with x and index.
2415 (attribute_use): Likewise.
2416 (apply_attribute_uses): Update calls to apply_iterator.
2417 (apply_iterators): Likewise. Update initialization of iterator_use.
2418 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2419 and record_potential_iterator_use.
2420 (rtx_reader::read_rtx_operand): Likewise.
2422 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2423 Alan Hayward <alan.hayward@arm.com>
2424 David Sherwood <david.sherwood@arm.com>
2426 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2429 2017-08-21 Richard Biener <rguenther@suse.de>
2432 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2433 at struct end conservatively when comparing common bases.
2435 2017-08-21 Richard Biener <rguenther@suse.de>
2437 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2438 (mem_ref_in_stmt): Remove.
2439 (determine_max_movement): Use ref index to get at the reference.
2440 (invariantness_dom_walker::before_dom_children): Deal with
2441 lim data already initialized.
2442 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2444 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2446 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2447 (TARGET_ISA_ROUND): Ditto.
2448 (TARGET_ROUND): Ditto.
2449 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2450 * config/i386/i386.md: Ditto.
2451 * config/i386/sse.md: Ditto.
2452 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2453 with OPTION_MASK_ISA_SSE4_1.
2455 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2458 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2459 __builtin_ia32_lzcnt_u16.
2461 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2464 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2465 (rs6000_set_current_function): Rewrite function to use it.
2467 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2470 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2471 and TYPE_WARN_IF_NOT_ALIGN.
2472 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2473 (handle_warn_if_not_align): New.
2474 (place_union_field): Call handle_warn_if_not_align.
2475 (place_field): Call handle_warn_if_not_align.
2476 Copy TYPE_WARN_IF_NOT_ALIGN.
2477 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2478 (layout_type): Likewise.
2479 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2481 (tree_decl_common): Add warn_if_not_align.
2482 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2483 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2484 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2485 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2486 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2487 * doc/extend.texi: Document warn_if_not_aligned attribute.
2488 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2490 2017-08-17 Martin Liska <mliska@suse.cz>
2493 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2494 (get_data_dependence): Use it as pointer type.
2495 (distribute_loop): Likewise.
2497 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2499 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2500 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2501 (p8_vmrgow_<mode>_direct): New define_insn.
2502 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2503 handle endianness for vmrgew and vmrgow permute patterns.
2505 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2507 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2508 * config/rs6000/rs6000-cpus.def: Remove comment.
2509 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2510 (POWERPC_MASKS): Likewise.
2511 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2512 use of TARGET_VSX_TIMODE.
2513 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2514 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2515 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2516 (rs6000_option_override_internal): Remove dead code.
2517 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2518 (rs6000_legitimize_reload_address): Likewise.
2519 (rs6000_legitimate_address_p): Likewise.
2520 (rs6000_opt_masks): Delete "vsx-timode".
2521 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2522 from function comment.
2523 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2524 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2525 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2527 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2528 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2529 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2532 (define_peephole2): Likewise.
2534 2017-08-17 Martin Sebor <msebor@redhat.com>
2537 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2538 past the end of an array.
2539 (test_pp_format): Add test cases.
2541 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2543 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2544 missing ECF_NOTHROW flags.
2546 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2549 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2550 operands residing in integer registers.
2551 (*vsx_le_perm_load_<mode>): Likewise.
2552 (*vsx_le_perm_store_<mode>): Likewise.
2553 (define_peephole2): Add peepholes to optimize the above.
2555 2017-08-17 Marek Polacek <polacek@redhat.com>
2558 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2559 to mimic what shorten_compare did. Change the return type to bool.
2560 (fold_cond_expr_with_comparison): Update call to
2561 operand_equal_for_comparison_p.
2562 (fold_ternary_loc): Likewise.
2564 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2566 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2568 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2570 2017-08-17 Richard Biener <rguenther@suse.de>
2572 * tree-ssa-structalias.c (solve_graph): When propagating
2573 to successors update the graphs succ edges and avoid duplicate work.
2575 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2578 * config/i386/i386.c (ix86_option_override_internal): Save target
2579 specific options after ix86_stack_protector_guard_reg was changed.
2581 2017-08-17 Richard Biener <rguenther@suse.de>
2583 PR tree-optimization/81827
2584 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2585 (new_var_info): Initialize it conservatively.
2586 (get_call_vi): Mark register vars.
2587 (new_scalar_tmp_constraint_exp): Likewise.
2588 (handle_rhs_call): Likewise.
2589 (handle_const_call): Likewise.
2590 (create_function_info_for): Likewise.
2591 (solve_constraints): Sort varinfos to separate register from
2592 non-register vars to pack points-to solution bitmaps during
2595 2017-08-17 Marek Polacek <polacek@redhat.com>
2597 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2599 2017-08-17 Richard Biener <rguenther@suse.de>
2601 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2602 to true when overflow is undefined and we saturated the result.
2604 2017-08-17 Alan Modra <amodra@gmail.com>
2607 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2608 Don't use store multiple if only one reg needs saving.
2609 (interesting_frame_related_regno): New function.
2610 (rs6000_frame_related): Don't emit frame info for regs that
2612 (rs6000_emit_epilogue): Likewise.
2614 2017-08-16 Nathan Sidwell <nathan@acm.org>
2616 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2617 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2618 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2619 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2620 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2621 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2622 (verify_type): Adjust for TYPE_BINFO move.
2623 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2624 process TYPE_BINFO directly.
2625 (hash_tree): Likewise.
2626 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2628 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2631 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2633 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2635 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2638 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2639 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2640 Add ix86_binary_operator_ok to insn constraint.
2641 (*iordi_1_bts): Ditto.
2642 (*xordi_1_btc): Ditto.
2643 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2644 Update corresponding peephole2 pattern.
2648 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2650 PR tree-optimization/81832
2651 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2652 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2654 2017-08-16 Marek Polacek <polacek@redhat.com>
2657 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2658 regarding size_zero_node.
2660 2017-08-16 Martin Liska <mliska@suse.cz>
2663 * config.gcc: Respect previously set extra_objs in case
2666 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2668 PR tree-optimization/81835
2669 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2670 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2671 not depend on the phi.
2673 2017-08-16 Alan Modra <amodra@gmail.com>
2675 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2678 2017-08-16 Alan Modra <amodra@gmail.com>
2680 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2681 (save_reg_p): ..into this. Update all callers.
2682 (first_reg_to_save): Simplify.
2684 2017-08-16 Alan Modra <amodra@gmail.com>
2686 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2689 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2693 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2694 constant count if that count is less than 32.
2696 2017-08-15 Nathan Sidwell <nathan@acm.org>
2698 * gcc.c (execute): Emit friendlier message if inferior is killed
2699 by an external cause.
2701 2017-08-15 Richard Biener <rguenther@suse.de>
2703 PR tree-optimization/81790
2704 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2705 CONSTRUCTORs from simplifying and VN.
2707 2017-08-14 Martin Sebor <msebor@redhat.com>
2709 * builtin-attrs.def: Add comments.
2711 2017-08-14 Martin Sebor <msebor@redhat.com>
2714 * doc/extend.texi (attribute nonstring): Document new attribute.
2716 2017-08-14 Martin Sebor <msebor@redhat.com>
2719 * tree-diagnostic.c (default_tree_printer): Handle %G.
2720 * gimple-pretty-print.h (percent_G_format): Declare new function.
2721 * gimple-pretty-print.c (percent_G_format): Define.
2722 * tree-pretty-print.c (percent_K_format): Add argument.
2724 2017-08-14 Martin Sebor <msebor@redhat.com>
2726 PR translation/79998
2727 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2728 Remove a stray space.
2730 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2733 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2734 (*iordi_1_bts): Ditto.
2735 (*xordi_1_btc): Ditto.
2737 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2740 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2741 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2743 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2744 quoted strings, and make more translator-friendly.
2745 (darwin_rs6000_override_options): Likewise.
2746 (rs6000_option_override_internal): Likewise.
2747 (rs6000_return_in_memory): Fix overlong line.
2748 (init_cmulative_args): Use quoted strings, and make more
2749 translator-friendly.
2750 (rs6000_pass_by_reference): Fix overlong line.
2751 (def_builtin): Use quoted strings.
2752 (altivec_expand_predicate_builtin): Use quoted strings, and make
2753 more translator-friendly.
2754 (htm_expand_builtin): Use quoted strings.
2755 (cpu_expand_builtin): Use quoted strings, and make more
2756 translator-friendly.
2757 (altivec_expand_builtin): Likewise.
2758 (paired_expand_predicate_builtin): Likewise.
2759 (rs6000_invalid_builtin): Likewise.
2760 (builtin_function_type): Use quoted strings.
2761 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2762 more translator-friendly.
2763 (rs6000_trampoline_init): Likewise.
2764 (rs6000_handle_altivec_attribute): Likewise.
2765 (rs6000_inner_target_options): Use quoted strings.
2766 (rs6000_disable_incompatible_switches): Likewise.
2767 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2768 strings, and make more translator-friendly.
2769 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2771 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2773 PR tree-optimization/81799
2774 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2775 cond_expr to simple gimple operand.
2777 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2780 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2782 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2786 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2787 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2788 TARGET_GASISR_PROLOGUES.
2789 * config/avr/avr.c (avr_option_override): Same.
2790 (avr_pass_pre_proep::execute): Same.
2792 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2795 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2796 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2798 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2800 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2801 define_insn_and_split. Split to a memory load from 0 in
2802 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2803 using PTR mode iterator.
2804 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2805 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2806 (*add_tp_<mode>): Redefine as define_insn_and_split.
2807 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2808 address space. Merge with *add_tp_x32 using PTR mode iterator.
2809 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2810 Split to an add with a memory load from 0 in
2811 DEFAULT_TLS_SEG_REG address space.
2813 2017-08-12 Andrew Pinski <apinski@cavium.com>
2815 * config/aarch64/aarch64-option-extensions.def (rdma):
2816 Fix feature string to what Linux prints out in /proc/cpuinfo.
2818 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2821 * dwarf2out.c (modified_type_die): For C typedef types that have
2822 an ultimate origin, process the ultimate origin instead of the
2824 (gen_typedef_die): Assert that input DECLs have no ultimate
2826 (gen_type_die_with_usage): For typedef variants that have an
2827 ultimate origin, just call gen_decl_die on the original DECL.
2828 (process_scope_var): Avoid creating DIEs for local typedefs and
2829 concrete static variables.
2831 2017-08-12 Alan Modra <amodra@gmail.com>
2835 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2836 match gnu-user.h startfile.
2837 (ENDFILE_LINUX_SPEC): Similarly.
2839 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2842 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2844 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2846 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2847 * config/aarch64/aarch64.md (mov<mode>): Change.
2848 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2849 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2850 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2852 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2854 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2855 for storage order barriers.
2857 2017-08-11 Martin Liska <mliska@suse.cz>
2860 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2861 variables of void type.
2863 2017-08-11 Martin Liska <mliska@suse.cz>
2865 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2866 TARGET_SUPPORTS_ALIASES.
2867 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2868 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2869 (optimize_weakref): Likewise.
2870 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2871 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2872 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2874 2017-08-11 Martin Liska <mliska@suse.cz>
2877 * config/i386/i386.c (make_resolver_func): Do complete
2878 refactoring of the function.
2880 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2883 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2884 * config/i386/i386.c (ix86_stack_protect_guard): Use
2885 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2886 * doc/invoke.texi (x86 Options): Document
2887 -mstack-protector-guard-symbol= option.
2889 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2891 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2892 * config/i386/i386.c (ix86_split_stack_guard): New function.
2893 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2894 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2895 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2896 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2897 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2898 (split_stack_space_check): Call ix86_split_stack_guard.
2900 2017-08-10 Martin Sebor <msebor@redhat.com>
2902 * print-tree.c (print_node): Print location using the established
2904 Replace spaces with colons.
2905 (debug_raw, debug): Ditto.
2907 2017-08-10 Martin Sebor <msebor@redhat.com>
2910 * pretty-print.c (pp_format): Correct the handling of %s precision.
2912 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2915 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2917 (ix86_finalize_stack_frame_flags): This. Also clear
2918 frame_pointer_needed if -fno-omit-frame-pointer is used without
2920 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2921 with ix86_finalize_stack_frame_flags.
2922 (ix86_expand_epilogue): Likewise.
2923 (ix86_expand_split_stack_prologue): Likewise.
2924 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2926 2017-08-10 Martin Liska <mliska@suse.cz>
2929 * c-attribs.c (handle_target_attribute):
2930 Report warning for an empty string argument of target attribute.
2932 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2935 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2937 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2938 or DECL_NONLOCAL labels.
2939 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2940 or DECL_NONLOCAL labels here.
2942 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2944 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2945 to indicate when early gimple folding has been disabled.
2946 (rs6000_gimple_fold_builtin): Add debug content.
2947 (rs6000_invalid_builtin): Fix whitespace.
2948 (rs6000_expand_builtin): Fix whitespace.
2949 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2951 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2954 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2955 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2957 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2959 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2960 (qdf24xx): Likewise.
2961 * config/aarch64/aarch64.md: Include falkor.md.
2962 * config/aarch64/falkor.md: New.
2964 2017-08-09 Marek Polacek <polacek@redhat.com>
2967 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2968 * diagnostic.c (emit_diagnostic): Add a comment.
2969 (emit_diagnostic_valist): New function.
2971 2017-08-09 Marek Polacek <polacek@redhat.com>
2974 * input.c (make_location): New overload.
2975 * input.h (make_location): Declare.
2977 2017-08-08 Alan Modra <amodra@gmail.com>
2978 H.J. Lu <hongjiu.lu@intel.com>
2981 * gcc.c (NO_PIE_SPEC): Delete.
2982 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2984 (LINK_PIE_SPEC): ..to here.
2985 (LINK_COMMAND_SPEC): Support -no-pie.
2986 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2987 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2989 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2990 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2991 (ENDFILE_CRTEND_SPEC): Similarly.
2993 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2996 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2997 (mstack-protector-guard-offset=): Ditto.
2998 * config/i386/i386.c (ix86_option_override): Handle
2999 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3001 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3002 ix86_stack_protect_guard_offset variables.
3003 (TARGET_STACK_PROTECT_GUARD): Always define.
3004 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3005 and -mstack-protector-guard-offset= options.
3007 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3009 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3010 boundary case for the last candidate.
3012 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3014 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3015 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3017 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3020 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3021 * config/aarch64/aarch64-builtins.c
3022 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3023 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3024 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3026 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3027 Andrew Pinski <pinskia@gmail.com>
3030 * internal-fn.def (XORSIGN): New.
3031 * optabs.def (xorsign_optab): New.
3032 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3033 (convert_expand_mult_copysign): New.
3034 (pass_optimize_widening_mul::execute): Call
3035 convert_expand_mult_copysign.
3037 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3039 PR tree-optimization/81354
3040 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3041 Insert on edges rather than explicitly creating landing pads.
3042 (analyze_candidates_and_replace): Commit edge inserts.
3044 2017-08-08 Richard Biener <rguenther@suse.de>
3047 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3048 (expand_simple_operations): Also look through ADDR_EXPRs with
3049 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3051 2017-08-08 Richard Biener <rguenther@suse.de>
3053 PR tree-optimization/81723
3054 * tree-vect-slp.c (struct bst_traits): New hash traits.
3055 (bst_fail): New global.
3056 (vect_build_slp_tree_2): New worker, split out from ...
3057 (vect_build_slp_tree): ... this now wrapping it with using
3058 bst_fail set to cache SLP tree build fails. Properly handle
3060 (vect_analyze_slp_instance): Allocate and free bst_fail.
3062 2017-08-08 Martin Liska <mliska@suse.cz>
3065 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3066 LABEL_DECLs that can be from a different function.
3068 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3070 PR tree-optimization/81744
3071 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3072 loop's number of iterations.
3074 2017-08-08 Martin Liska <mliska@suse.cz>
3076 * asan.c: Include header files.
3077 * attribs.c (build_decl_attribute_variant): New function moved
3079 (build_type_attribute_qual_variant): Likewise.
3080 (cmp_attrib_identifiers): Likewise.
3081 (simple_cst_list_equal): Likewise.
3082 (omp_declare_simd_clauses_equal): Likewise.
3083 (attribute_value_equal): Likewise.
3084 (comp_type_attributes): Likewise.
3085 (build_type_attribute_variant): Likewise.
3086 (lookup_ident_attribute): Likewise.
3087 (remove_attribute): Likewise.
3088 (merge_attributes): Likewise.
3089 (merge_type_attributes): Likewise.
3090 (merge_decl_attributes): Likewise.
3091 (merge_dllimport_decl_attributes): Likewise.
3092 (handle_dll_attribute): Likewise.
3093 (attribute_list_equal): Likewise.
3094 (attribute_list_contained): Likewise.
3095 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3096 (lookup_attribute_by_prefix): Likewise.
3097 * bb-reorder.c: Include header files.
3098 * builtins.c: Likewise.
3099 * calls.c: Likewise.
3100 * cfgexpand.c: Likewise.
3101 * cgraph.c: Likewise.
3102 * cgraphunit.c: Likewise.
3103 * convert.c: Likewise.
3104 * dwarf2out.c: Likewise.
3105 * final.c: Likewise.
3106 * fold-const.c: Likewise.
3107 * function.c: Likewise.
3108 * gimple-expr.c: Likewise.
3109 * gimple-fold.c: Likewise.
3110 * gimple-pretty-print.c: Likewise.
3111 * gimple.c: Likewise.
3112 * gimplify.c: Likewise.
3113 * hsa-common.c: Likewise.
3114 * hsa-gen.c: Likewise.
3115 * internal-fn.c: Likewise.
3116 * ipa-chkp.c: Likewise.
3117 * ipa-cp.c: Likewise.
3118 * ipa-devirt.c: Likewise.
3119 * ipa-fnsummary.c: Likewise.
3120 * ipa-inline.c: Likewise.
3121 * ipa-visibility.c: Likewise.
3123 * lto-cgraph.c: Likewise.
3124 * omp-expand.c: Likewise.
3125 * omp-general.c: Likewise.
3126 * omp-low.c: Likewise.
3127 * omp-offload.c: Likewise.
3128 * omp-simd-clone.c: Likewise.
3129 * opts-global.c: Likewise.
3130 * passes.c: Likewise.
3131 * predict.c: Likewise.
3132 * sancov.c: Likewise.
3133 * sanopt.c: Likewise.
3134 * symtab.c: Likewise.
3135 * toplev.c: Likewise.
3136 * trans-mem.c: Likewise.
3137 * tree-chkp.c: Likewise.
3138 * tree-eh.c: Likewise.
3139 * tree-into-ssa.c: Likewise.
3140 * tree-object-size.c: Likewise.
3141 * tree-parloops.c: Likewise.
3142 * tree-profile.c: Likewise.
3143 * tree-ssa-ccp.c: Likewise.
3144 * tree-ssa-live.c: Likewise.
3145 * tree-ssa-loop.c: Likewise.
3146 * tree-ssa-sccvn.c: Likewise.
3147 * tree-ssa-structalias.c: Likewise.
3148 * tree-ssa.c: Likewise.
3149 * tree-streamer-in.c: Likewise.
3150 * tree-vectorizer.c: Likewise.
3151 * tree-vrp.c: Likewise.
3153 * ubsan.c: Likewise.
3154 * varasm.c: Likewise.
3155 * varpool.c: Likewise.
3156 * tree.c: Remove functions moved to attribs.[ch].
3158 * config/aarch64/aarch64.c: Add attrs.h header file.
3159 * config/alpha/alpha.c: Likewise.
3160 * config/arc/arc.c: Likewise.
3161 * config/arm/arm.c: Likewise.
3162 * config/avr/avr.c: Likewise.
3163 * config/bfin/bfin.c: Likewise.
3164 * config/c6x/c6x.c: Likewise.
3165 * config/cr16/cr16.c: Likewise.
3166 * config/cris/cris.c: Likewise.
3167 * config/darwin.c: Likewise.
3168 * config/epiphany/epiphany.c: Likewise.
3169 * config/fr30/fr30.c: Likewise.
3170 * config/frv/frv.c: Likewise.
3171 * config/ft32/ft32.c: Likewise.
3172 * config/h8300/h8300.c: Likewise.
3173 * config/i386/winnt.c: Likewise.
3174 * config/ia64/ia64.c: Likewise.
3175 * config/iq2000/iq2000.c: Likewise.
3176 * config/lm32/lm32.c: Likewise.
3177 * config/m32c/m32c.c: Likewise.
3178 * config/m32r/m32r.c: Likewise.
3179 * config/m68k/m68k.c: Likewise.
3180 * config/mcore/mcore.c: Likewise.
3181 * config/microblaze/microblaze.c: Likewise.
3182 * config/mips/mips.c: Likewise.
3183 * config/mmix/mmix.c: Likewise.
3184 * config/mn10300/mn10300.c: Likewise.
3185 * config/moxie/moxie.c: Likewise.
3186 * config/msp430/msp430.c: Likewise.
3187 * config/nds32/nds32-isr.c: Likewise.
3188 * config/nds32/nds32.c: Likewise.
3189 * config/nios2/nios2.c: Likewise.
3190 * config/nvptx/nvptx.c: Likewise.
3191 * config/pa/pa.c: Likewise.
3192 * config/pdp11/pdp11.c: Likewise.
3193 * config/powerpcspe/powerpcspe.c: Likewise.
3194 * config/riscv/riscv.c: Likewise.
3195 * config/rl78/rl78.c: Likewise.
3196 * config/rx/rx.c: Likewise.
3197 * config/s390/s390.c: Likewise.
3198 * config/sh/sh.c: Likewise.
3199 * config/sol2.c: Likewise.
3200 * config/sparc/sparc.c: Likewise.
3201 * config/spu/spu.c: Likewise.
3202 * config/stormy16/stormy16.c: Likewise.
3203 * config/tilegx/tilegx.c: Likewise.
3204 * config/tilepro/tilepro.c: Likewise.
3205 * config/v850/v850.c: Likewise.
3206 * config/vax/vax.c: Likewise.
3207 * config/visium/visium.c: Likewise.
3208 * config/xtensa/xtensa.c: Likewise.
3210 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3213 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3214 constraints since the -mupper-regs-* switches have been
3216 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3217 into a vector from a double word element that was extracted from
3218 another vector, and eliminate extra XXPERMDI instructions.
3219 (vsx_concat_<mode>_2): Likewise.
3220 (vsx_concat_<mode>_3): Likewise.
3221 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3222 concat to allow optimizing inserts from previous extracts.
3224 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3226 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3227 memory reference to a SSP offset in TLS address space.
3228 (ix86_print_operand) <case '@'>: Remove.
3229 (ix86_print_operand_punct_valid_p): Remove '@' code.
3230 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3232 (stack_tls_protect_set_<mode>): Remove.
3233 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3234 (stack_tls_protect_test_<mode>): Remove.
3235 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3237 2017-08-07 Olivier Hainque <hainque@adacore.com>
3240 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3242 2017-08-07 Douglas Rupp <rupp@adacore.com>
3244 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3245 variable was referenced as multidir in command.
3247 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3250 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3253 2017-08-07 Martin Liska <mliska@suse.cz>
3255 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3256 * config/rl78/rl78.c: Add include of attribs.h.
3257 * config/sh/sh.c: Likewise.
3258 * config/v850/v850.c: Likewise.
3260 2017-08-07 Tom de Vries <tom@codesourcery.com>
3263 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3265 2017-08-07 Martin Liska <mliska@suse.cz>
3267 * config/mips/mips.c: Include attribs.h.
3269 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3272 * doc/invoke.texi: Document change in behvaior for -Ofast for
3275 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3277 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3279 (aarch64_pop_regs): Likewise.
3280 (aarch64_gen_load_pair): Likewise.
3281 (aarch64_save_callee_saves): Likewise.
3282 (aarch64_restore_callee_saves): Likewise.
3284 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3286 * config/i386/i386.c: Revert the last change.
3288 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3291 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3293 (ix86_finalize_stack_frame_flags): This. Also clear
3294 frame_pointer_needed if -fno-omit-frame-pointer is used without
3296 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3297 with ix86_finalize_stack_frame_flags.
3298 (ix86_expand_epilogue): Likewise.
3299 (ix86_expand_split_stack_prologue): Likewise.
3301 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3304 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3305 to P_AES for Westmere.
3307 2017-08-07 Jonathan Yong <10walls@gmail.com>
3309 * config/i386/mingw.opt (fset-stack-executable): Removed.
3310 * config/i386/cygming.opt (fset-stack-executable): Moved
3312 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3314 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3316 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3318 2017-08-07 Marek Polacek <polacek@redhat.com>
3321 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3323 2017-08-07 Martin Liska <mliska@suse.cz>
3325 * attribs.h (canonicalize_attr_name): New function.
3326 (cmp_attribs): Move from c-format.c and adjusted.
3327 (is_attribute_p): Moved from tree.h.
3328 * tree-inline.c: Add new includes.
3329 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3330 (private_is_attribute_p): Remove.
3331 (private_lookup_attribute): Likewise.
3332 (private_lookup_attribute_by_prefix): Simplify.
3333 (remove_attribute): Use is_attribute_p.
3334 * tree.h: Remove removed declarations.
3336 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3339 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3340 instead of computing it in the function. Formatting fix.
3341 (expand_case): Don't rely on default_edge being the first edge,
3342 clear it if removing it, pass default_edge to
3343 emit_case_dispatch_table.
3344 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3347 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3349 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3350 insn in the function, emit NOP after the insn.
3352 2017-08-06 Tom de Vries <tom@codesourcery.com>
3354 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3357 2017-08-06 Tom de Vries <tom@codesourcery.com>
3359 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3362 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3364 PR tree-optimization/57371
3365 * match.pd: New pattern.
3367 2017-08-04 Marek Polacek <polacek@redhat.com>
3370 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3371 perform the computation in offset_int.
3373 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3375 PR tree-optimization/81136
3376 * tree-vectorizer.h: Include tree-hash-traits.h.
3377 (vec_base_alignments): New typedef.
3378 (vec_info): Add a base_alignments field.
3379 (vect_record_base_alignments): Declare.
3380 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3382 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3383 (create_data_ref): Add an is_conditional_in_stmt argument.
3384 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3385 the is_conditional_in_stmt field.
3386 (data_ref_loc): Add an is_conditional_in_stmt field.
3387 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3388 (find_data_references_in_stmt): Update call to create_data_ref.
3389 (graphite_find_data_references_in_stmt): Likewise.
3390 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3391 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3392 (vect_record_base_alignment): New function.
3393 (vect_record_base_alignments): Likewise.
3394 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3395 for nested statements even if we fail to compute a misalignment.
3396 Use pooled base alignments for unconditional references.
3397 (vect_find_same_alignment_drs): Compare base addresses instead
3399 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3400 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3402 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3404 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3405 Add an explicit name for the enum. Use auto_vec for slp_instances
3407 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3409 (_bb_vec_info): Add a constructor and destructor.
3410 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3411 (destroy_loop_vec_info): Delete.
3412 (vect_destroy_datarefs): Likewise.
3413 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3414 (vec_info::vec_info): New function.
3415 (vec_info::~vec_info): Likewise.
3416 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3417 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3418 destroy_loop_vec_info.
3419 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3420 (_loop_vec_info::_loop_vec_info): ...this.
3421 (destroy_loop_vec_info): Replace with...
3422 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3423 the stmt_vec_infos. Leave handling of vec_info information to its
3424 destructor. Remove explicit vector releases.
3425 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3426 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3427 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3428 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3429 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3430 (destroy_bb_vec_info): Replace with...
3431 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3432 information to its destructor.
3433 (vect_slp_analyze_bb_1): Use new and delete instead of
3434 new_bb_vec_info and destroy_bb_vec_info.
3435 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3438 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3440 * tree-data-ref.h (subscript): Add access_fn field.
3441 (data_dependence_relation): Add could_be_independent_p.
3442 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3443 (same_access_functions): Move to tree-data-ref.c.
3444 * tree-data-ref.c (ref_contains_union_access_p): New function.
3445 (access_fn_component_p): Likewise.
3446 (access_fn_components_comparable_p): Likewise.
3447 (dr_analyze_indices): Add a reference to access_fn_component_p.
3448 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3450 (constant_access_functions): Likewise.
3451 (add_other_self_distances): Likewise.
3452 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3453 (initialize_data_dependence_relation): Use XCNEW and remove
3454 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3455 of access functions that have the same type. Allow the
3456 subsequence to end with different bases in some circumstances.
3457 Record the chosen access functions in SUB_ACCESS_FN.
3458 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3459 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3460 (subscript_dependence_tester_1): Likewise dra and drb.
3461 (build_classic_dist_vector): Update calls accordingly.
3462 (subscript_dependence_tester): Likewise.
3463 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3464 DDR_COULD_BE_INDEPENDENT_P.
3465 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3466 comp_alias_ddrs instead of may_alias_ddrs.
3467 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3469 (vect_analyze_data_ref_dependence): Use it if
3470 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3471 distance vectors if that fails.
3472 (dependence_distance_ge_vf): New function.
3473 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3474 LOOP_VINFO_MAY_ALIAS_DDRS.
3476 2017-08-04 Richard Biener <rguenther@suse.de>
3479 * fold-const.c (fold_binary_loc): Properly restrict
3480 minus_var0 && minus_var1 case when associating undefined overflow
3483 2017-08-04 Tom de Vries <tom@codesourcery.com>
3485 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3487 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3489 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3490 Don't start diagnostic messages with a capital letter.
3491 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3493 (rs6000_invalid_builtin): Likewise.
3494 (rs6000_trampoline_init): Likewise.
3496 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3499 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3500 after setting changeable df flags.
3502 2017-08-03 Richard Biener <rguenther@suse.de>
3504 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3505 up if the use is in USE - X.
3507 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3509 * toplev.c (dumpfile.h): New include.
3510 (internal_error_reentered): New static function. Use it...
3511 (internal_error_function): ...here to handle reentered internal_error.
3513 2017-08-03 Richard Biener <rguenther@suse.de>
3516 * fold-const.c (split_tree): Add minus_var and minus_con
3517 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3518 here but always use minus_*.
3519 (associate_trees): Assert we never associate with MINUS_EXPR
3520 and NULL first operand. Do not recurse for PLUS_EXPR operands
3521 when associating as MINUS_EXPR either.
3522 (fold_binary_loc): Track minus_var and minus_con.
3524 2017-08-03 Tom de Vries <tom@codesourcery.com>
3527 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3528 ACCEL_COMPILER, apply finish_options on
3529 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3531 2017-08-03 Tom de Vries <tom@codesourcery.com>
3534 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3535 function_entry_patch_area_size > 0.
3537 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3540 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3541 instead of 10??LU, perform unit multiplication in wide_int,
3542 don't change alloc_object_size_limit if the limit is larger
3545 PR tree-optimization/81655
3546 PR tree-optimization/81588
3547 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3548 the case when ranges[i].low and high are 1 for unsigned type with
3552 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3553 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3555 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3557 * tree-vrp.h: Add include guard.
3559 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3562 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3563 (ud2): New insn pattern.
3564 * config/i386/i386.c (ix86_expand_epilogue):
3565 For naked functions, generate ud2 instead of trap insn.
3567 2017-08-02 Marek Polacek <polacek@redhat.com>
3570 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3572 2017-08-02 Tom de Vries <tom@codesourcery.com>
3573 Cesar Philippidis <cesar@codesourcery.com>
3575 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3576 Add missing edge probabilities.
3578 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3580 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3583 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3586 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3587 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3588 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3589 prologue_seq sequences - if any.
3591 2017-08-02 Richard Biener <rguenther@suse.de>
3593 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3594 via vectors if supported, integer extracts via punning if supported
3595 or otherwise vector extracts.
3597 2017-08-02 Richard Biener <rguenther@suse.de>
3599 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3601 (bitmap_insert_into_set): ... this.
3603 2017-08-02 Richard Biener <rguenther@suse.de>
3605 PR tree-optimization/81633
3607 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3609 PR tree-optimization/71752
3610 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3612 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3614 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3615 (machine_function::call_ms2sysv_pad_out): Remove field.
3616 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3617 (ix86_compute_frame_layout): Likewise.
3619 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3622 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3623 attribute with interrupt attribute.
3625 2017-08-01 Andrew Pinski <apinski@cavium.com>
3627 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3628 BIT_INSERT_EXPR's operand 1
3629 to see if the types precision matches.
3631 2017-08-01 Martin Liska <mliska@suse.cz>
3634 * builtins.c (expand_builtin_memcpy_args): Remove.
3635 (expand_builtin_memcpy): Call newly added function
3636 expand_builtin_memory_copy_args.
3637 (expand_builtin_memcpy_with_bounds): Likewise.
3638 (expand_builtin_mempcpy): Remove last argument.
3639 (expand_builtin_mempcpy_with_bounds): Likewise.
3640 (expand_builtin_memory_copy_args): New function created from
3641 expand_builtin_mempcpy_args with small modifications.
3642 (expand_builtin_mempcpy_args): Remove.
3643 (expand_builtin_stpcpy): Remove unused argument.
3644 (expand_builtin): Likewise.
3645 (expand_builtin_with_bounds): Likewise.
3647 2017-08-01 Martin Liska <mliska@suse.cz>
3650 Make mempcpy more optimal (PR middle-end/70140).
3652 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3655 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3656 __builtin_vec_cmpne verify both arguments are compatible vectors
3657 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3658 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3659 move computation of aligned to after checking the argument types.
3663 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3664 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3667 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3668 Doug Rupp <rupp@adacore.com>
3669 Olivier Hainque <hainque@adacore.com>
3671 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3672 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3674 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3675 for TARGET_OS_CPP_BUILTIN.
3676 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3677 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3679 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3680 passing required abi options to the assembler for EABI configurations.
3681 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3682 of .text.hot and .text.unlikely sections for kernel modules when
3683 using ARM style exceptions.
3684 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3685 options. Add EXTRA_CC1_SPEC.
3686 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3688 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3690 (ARM_TARGET2_DWARF_FORMAT): Define.
3691 * config/arm/t-vxworks: Adjust multilib control to removal of the
3692 Diab command line options.
3694 2017-08-01 Martin Liska <mliska@suse.cz>
3696 PR gcov-profile/81561
3697 * gcov.c (unblock): Make unblocking safe as we need to preserve
3698 index correspondence of blocks and block_lists.
3700 2017-08-01 Richard Biener <rguenther@suse.de>
3702 PR tree-optimization/81181
3703 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3704 (compute_antic): ... end of iteration here.
3706 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3708 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3709 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3710 (ftree-slp-vectorize): Likewise.
3711 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3712 can no longer be set independent of flag_tree_loop_vectorize.
3713 * omp-general.c (emp_max_vf): Likewise.
3714 * opts.c (enable_fdo_optimizations): Remove references to
3715 flag_tree_vectorize, these are now implicit.
3716 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3717 and leave it for the options machinery.
3719 2017-08-01 Martin Liska <mliska@suse.cz>
3722 * builtins.c (expand_builtin_memcpy_args): Remove.
3723 (expand_builtin_memcpy): Call newly added function
3724 expand_builtin_memory_copy_args.
3725 (expand_builtin_memcpy_with_bounds): Likewise.
3726 (expand_builtin_mempcpy): Remove last argument.
3727 (expand_builtin_mempcpy_with_bounds): Likewise.
3728 (expand_builtin_memory_copy_args): New function created from
3729 expand_builtin_mempcpy_args with small modifications.
3730 (expand_builtin_mempcpy_args): Remove.
3731 (expand_builtin_stpcpy): Remove unused argument.
3732 (expand_builtin): Likewise.
3733 (expand_builtin_with_bounds): Likewise.
3735 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3738 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3739 print "ds:" only for immediates in generic address space.
3741 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3744 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3745 (ix86_function_ok_for_sibcall): Return false for naked functions.
3747 2017-08-01 Richard Biener <rguenther@suse.de>
3749 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3750 (compute_antic): Seed worklist with exit block predecessors.
3751 * cfganal.c (dfs_find_deadend): For a cycle return the source
3752 of the edge closing it.
3754 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3756 * config/aarch64/aarch64.c
3757 (aarch64_can_const_movi_rtx_p): Move 0 check.
3759 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3761 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3762 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3764 * match.pd: Ditto in address comparison pattern.
3766 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3768 PR tree-optimization/81627
3769 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3770 closed ssa form for store-store chain.
3772 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3774 PR tree-optimization/81620
3775 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3776 for store-store chain.
3778 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3780 PR tree-optimization/81588
3781 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3782 ranges[i].in_p, invert comparison code ccode. For >/>=,
3783 swap rhs1 and rhs2 and comparison code unconditionally,
3784 for </<= don't do that. Don't swap rhs1/rhs2 again if
3785 ranges[i].in_p, instead invert comparison code ccode if
3786 opcode or oe->rank is BIT_IOR_EXPR.
3789 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3790 a direct optab to conversion optab.
3791 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3792 with GET_MODE_INNER as last argument instead of optab_handler.
3793 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3794 vector extraction if possible and optab is available.
3795 * expr.c (store_constructor): Use convert_optab_handler instead
3796 of optab_handler. Use vector initialization from smaller
3797 vectors if possible and optab is available.
3798 * tree-vect-stmts.c (vectorizable_load): Likewise.
3799 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3801 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3802 of vec_init from half-sized vectors with the same element mode.
3803 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3804 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3805 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3806 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3807 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3808 after mode in gen_vec_extract* calls.
3809 (vec_extract<mode>): Renamed to ...
3810 (vec_extract<mode><ssescalarmodelower>): ... this.
3811 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3812 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3813 element mode after mode in gen_vec_init* calls.
3814 (VEC_INIT_HALF_MODE): New mode iterator.
3815 (vec_init<mode>): Renamed to ...
3816 (vec_init<mode><ssescalarmodelower>): ... this.
3817 (vec_init<mode><ssehalfvecmodelower>): New expander.
3818 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3819 (vec_extractv2sfsf): ... this.
3820 (vec_initv2sf): Renamed to ...
3821 (vec_initv2sfsf): ... this.
3822 (vec_extractv2si): Renamed to ...
3823 (vec_extractv2sisi): ... this.
3824 (vec_initv2si): Renamed to ...
3825 (vec_initv2sisi): ... this.
3826 (vec_extractv4hi): Renamed to ...
3827 (vec_extractv4hihi): ... this.
3828 (vec_initv4hi): Renamed to ...
3829 (vec_initv4hihi): ... this.
3830 (vec_extractv8qi): Renamed to ...
3831 (vec_extractv8qiqi): ... this.
3832 (vec_initv8qi): Renamed to ...
3833 (vec_initv8qiqi): ... this.
3834 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3835 (vec_init<mode>): Renamed to ...
3836 (vec_init<mode><VEC_base_l>): ... this.
3837 (vec_extract<mode>): Renamed to ...
3838 (vec_extract<mode><VEC_base_l>): ... this.
3839 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3840 (vec_initv2sfsf): ... this.
3841 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3842 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3843 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3844 element mode after mode in gen_vec_init* calls.
3845 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3846 (vec_init<mode><Vel>): ... this.
3847 (vec_extract<mode>): Renamed to ...
3848 (vec_extract<mode><Vel>): ... this.
3849 * config/aarch64/iterators.md (Vel): New mode attribute.
3850 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3851 Add element mode after mode in gen_vec_extract* calls.
3852 * config/s390/vector.md (non_vec_l): New mode attribute.
3853 (vec_extract<mode>): Renamed to ...
3854 (vec_extract<mode><non_vec_l>): ... this.
3855 (vec_init<mode>): Renamed to ...
3856 (vec_init<mode><non_vec_l>): ... this.
3857 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3858 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3860 * config/arm/iterators.md (V_elem_l): New mode attribute.
3861 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3862 (vec_extract<mode><V_elem_l>): ... this.
3863 (vec_extractv2di): Renamed to ...
3864 (vec_extractv2didi): ... this.
3865 (vec_init<mode>): Renamed to ...
3866 (vec_init<mode><V_elem_l>): ... this.
3867 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3868 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3869 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3870 Add element mode after gen_vec_extract* calls.
3871 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3872 (vec_init<mode><unitmode>): ... this.
3873 (vec_extract<mode>): Renamed to ...
3874 (vec_extract<mode><unitmode>): ... this.
3875 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3876 (vec_init<mode><unitmode>): ... this.
3877 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3878 (vec_initv2sfsf): ... this.
3879 (vec_extractv2sf): Renamed to ...
3880 (vec_extractv2sfsf): ... this.
3881 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3882 Add element mode after gen_vec_extract* calls.
3883 * config/mips/mips.md (unitmode): New mode iterator.
3884 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3885 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3886 * config/spu/spu.md (inner_l): New mode attribute.
3887 (vec_init<mode>): Renamed to ...
3888 (vec_init<mode><inner_l>): ... this.
3889 (vec_extract<mode>): Renamed to ...
3890 (vec_extract<mode><inner_l>): ... this.
3891 * config/sparc/sparc.md (veltmode): New mode iterator.
3892 (vec_init<VMALL:mode>): Renamed to ...
3893 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3894 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3895 (vec_initv2sisi): ... this.
3896 (vec_initv2sf): Renamed to ...
3897 (vec_initv2sfsf): ... this.
3898 (vec_extractv2sf): Renamed to ...
3899 (vec_extractv2sfsf): ... this.
3900 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3901 (vec_init<mode>): Renamed to ...
3902 (vec_init<mode><VEC_base_l>): ... this.
3903 (vec_extract<mode>): Renamed to ...
3904 (vec_extract<mode><VEC_base_l>): ... this.
3905 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3906 (vec_initv2sfsf): ... this.
3907 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3908 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3909 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3910 gen_vec_init* calls.
3912 2017-08-01 Richard Biener <rguenther@suse.de>
3914 PR tree-optimization/81297
3915 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3916 TREE_OVERFLOW from INTEGER_CSTs.
3918 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3920 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3922 2017-07-31 Carl Love <cel@us.ibm.com>
3924 * config/rs6000/rs6000-c: Add support for built-in functions
3925 vector signed char vec_xl_be (signed long long, signed char *);
3926 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3927 vector signed int vec_xl_be (signed long long, signed int *);
3928 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3929 vector signed long long vec_xl_be (signed long long, signed long long *);
3930 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3931 vector signed short vec_xl_be (signed long long, signed short *);
3932 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3933 vector double vec_xl_be (signed long long, double *);
3934 vector float vec_xl_be (signed long long, float *);
3935 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3936 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3937 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3939 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3940 (altivec_expand_builtin): Add switch statement to call
3941 altivec_expand_xl_be for each builtin.
3942 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3943 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3944 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3945 __builtin_vsx_le_be_v16qi.
3946 * doc/extend.texi: Update the built-in documentation file for the
3947 new built-in functions.
3949 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3952 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3954 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3956 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3958 * config.gcc: Add z14.
3959 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3960 CPU model numbers for z13s and z14.
3961 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3963 * config/s390/s390-opts.h (enum processor_type): Rename
3964 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3965 * config/s390/s390.c (processor_table): Add field for CPU name to
3966 be passed to Binutils.
3967 (s390_asm_output_machine_for_arch): Use the new field in
3968 processor_table for Binutils.
3969 (s390_expand_builtin): Replace arch12 with z14.
3970 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3971 (s390_get_sched_attrmask): Likewise.
3972 (s390_get_unit_mask): Likewise.
3973 * config/s390/s390.opt: Add z14 to processor_type enum.
3975 2017-07-31 Martin Jambor <mjambor@suse.cz>
3978 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3979 regardless of optimization level.
3981 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3982 Martin Liska <mliska@suse.cz>
3984 * predict.def: Remove old comment and adjust probability.
3985 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3988 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3991 * config/i386/i386.c (ix86_function_naked): New function.
3992 (ix86_can_use_return_insn_p): Return false for naked functions.
3993 (ix86_expand_prologue): Skip prologue for naked functions.
3994 (ix86_expand_epilogue): Skip epilogue for naked functions
3995 and emit trap instruction.
3996 (ix86_warn_func_return): New function.
3997 (ix86_attribute_table): Add "naked" attribute specification.
3998 (TARGET_WARN_FUNC_RETURN): Define.
3999 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4001 2017-07-31 Martin Liska <mliska@suse.cz>
4003 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4004 (dump_gimple_bb_header): Always dump BB info.
4005 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4007 2017-07-31 Martin Liska <mliska@suse.cz>
4010 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4011 also with current_function_decl non-null equality.
4013 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4016 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4017 change type to the element type, instead add eltype variable and
4018 use it where we are interested in the element type.
4020 PR tree-optimization/81603
4021 * ipa-polymorphic-call.c
4022 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4023 offset arithmetic in offset_int, bail out if the resulting bit offset
4024 doesn't fit into shwi.
4026 2017-07-31 Martin Liska <mliska@suse.cz>
4028 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4029 (gimplify_save_expr): Fix comment.
4031 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4034 * config/i386/i386.c (ix86_function_arg): Update arguments for
4036 (ix86_compute_frame_layout): Set the initial stack offset to
4037 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4038 INCOMING_FRAME_SP_OFFSET.
4039 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4040 stack before exception handler returns.
4041 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4042 the 'ERROR_CODE' for exception handler.
4044 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4046 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4047 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4048 (ASM_OUTPUT_REG_POP): Ditto.
4049 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4050 instead of asm_fprintf to output pure string.
4052 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4054 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4055 to imported_module_or_decl hook.
4056 (debug_nothing_tree_tree_tree_bool): Remove.
4057 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4058 * debug.c (do_nothing_debug_hooks): Use
4059 debug_nothing_tree_tree_tree_bool_bool instead of
4060 debug_nothing_tree_tree_tree_bool.
4061 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4062 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4063 * sdbout.c (sdb_debug_hooks): Likewise.
4064 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4065 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4067 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4068 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4069 attribute, don't add anything.
4071 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4073 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4074 (fold_build2_stat_loc): Likewise.
4075 (fold_build3_stat_loc): Likewise.
4076 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4077 (fold_build1_loc): Remove macro.
4078 (fold_build2_loc): Likewise.
4079 (fold_build3_loc): Likewise.
4081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4083 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4084 (gimple_build_debug_bind_source_stat): Likewise.
4085 * gimple.h (gimple_build_debug_bind): Remove macro.
4086 (gimple_build_debug_bind_source): Likewise.
4088 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4090 * bitmap.c (bitmap_alloc): Adjust.
4091 (bitmap_gc_alloc): Likewise.
4092 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4094 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4096 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4097 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4098 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4099 (bitmap_gc_alloc_stat): Likewise.
4100 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4104 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4105 * rtl.h (shallow_copy_rtx): Remove macro.
4107 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4109 * emit-rtl.c (gen_raw_REG): Adjust.
4110 * gengenrtl.c (gendef): Likewise.
4111 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4112 * rtl.h (rtx_alloc): Remove macro.
4114 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4116 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4117 (build_tree_list_stat): Likewise.
4118 * tree.h (build_tree_list): Remove macro.
4119 (build_tree_list_vec): Likewise.
4121 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4123 * tree.c (make_vector_stat): Remove _stat from name.
4124 (build_vector_stat): Likewise.
4125 * tree.h (make_vector_stat): Remove macro.
4126 (build_vector_stat): Likewise.
4128 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4130 * tree.h (build_var_debug_value): Remove prototype.
4132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4134 * tree.c (tree_cons_stat): Remove _stat from name.
4135 * tree.h (tree_cons): Remove macro.
4137 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4139 * tree.c (build_vl_exp_stat): Remove _stat from name.
4140 * tree.h (build_vl_exp): Remove macro.
4142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4144 * tree.c (build_decl_stat): Remove _stat from name.
4145 * tree.h (build_decl): Remove macro.
4147 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4149 * gimple.c (gimple_build_with_ops_stat): Adjust.
4150 (gimple_alloc_stat): Remove _stat from name.
4151 * gimple.h (gimple_alloc): Remove macro.
4153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4155 * tree.c (make_tree_vec_stat): Remove _stat from name.
4156 (grow_tree_vec_stat): Likewise.
4157 * tree.h (make_tree_vec_stat): Adjust prototype.
4158 (grow_tree_vec_stat): Likewise.
4159 (make_tree_vec): Remove macro.
4160 (grow_tree_vec): Likewise.
4162 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4164 * fold-const.c (fold_build1_stat_loc): Adjust.
4165 (fold_build2_stat_loc): Likewise.
4166 (fold_build3_stat_loc): Likewise.
4167 * tree.c (build0_stat): Remove _stat from name.
4168 (build1_stat): Likewise.
4169 (build2_stat): Likewise.
4170 (build3_stat): Likewise.
4171 (build4_stat): Likewise.
4172 (build5_stat): Likewise.
4173 * tree.h (build1_loc): Remove macro, and rename _stat function
4175 (build2_loc): Likewise.
4176 (build3_loc): Likewise.
4177 (build4_loc): Likewise.
4178 (build5_loc): Likewise.
4180 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4182 * tree.c (make_int_cst_stat): Remove _stat from name.
4183 * tree.h (make_int_cst_stat): Adjust prototype.
4184 (make_int_cst): Remove macro.
4186 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4188 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4189 * tree.h (make_tree_binfo_stat): Adjust prototype.
4190 (make_tree_binfo): Remove.
4192 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4194 * tree.c (copy_node_stat): Rename to copy_node.
4195 (build_distinct_type_copy): Adjust.
4196 * tree.h (copy_node_stat): Adjust prototype.
4197 (copy_node): Remove macro.
4199 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4201 * tree.c (make_node_stat): rename to make_node.
4202 (build_tree_list_stat): Adjust.
4203 (build0_stat): Likewise.
4204 (build2_stat): Likewise.
4205 (build3_stat): Likewise.
4206 (build4_stat): Likewise.
4207 (build5_stat): Likewise.
4208 (build_decl_stat): Likewise.
4209 * tree.h (make_node_stat): Adjust prototype.
4210 (make_node): remove macro.
4212 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4214 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4215 (PPC_FEATURE2_SCV): Likewise.
4216 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4218 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4220 * config/aarch64/aarch64.c
4221 (aarch64_internal_mov_immediate): Add new special pattern.
4222 * config/aarch64/aarch64.md (*movdi_aarch64):
4223 Add reg/32bit const mov case.
4225 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4226 Richard Sandiford <richard.sandiford@linaro.org>
4228 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4229 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4230 Add integer and movi cases.
4231 (movi-split-hf-df-sf split, fp16): New.
4232 (enabled): Added TARGET_FP_F16INST.
4233 * config/aarch64/iterators.md (GPF_HF): New.
4234 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4236 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4238 * config/aarch64/aarch64.c
4239 (aarch64_simd_container_mode): Add prototype.
4240 (aarch64_expand_mov_immediate): Add HI support.
4241 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4242 (aarch64_can_const_movi_rtx_p): New.
4243 (aarch64_preferred_reload_class):
4244 Remove restrictions of using FP registers for certain SIMD operations.
4245 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4246 (aarch64_valid_floating_const): Add integer move validation.
4247 (aarch64_simd_imm_scalar_p): Remove.
4248 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4249 (aarch64_legitimate_constant_p): Expand list of supported cases.
4250 * config/aarch64/aarch64-protos.h
4251 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4252 (aarch64_reinterpret_float_as_int): New.
4253 (aarch64_simd_imm_scalar_p): Remove.
4254 * config/aarch64/constraints.md (Uvi): New.
4255 (Dd): Split into Ds and new Dd.
4256 * config/aarch64/aarch64.md (*movsi_aarch64):
4258 (*movdi_aarch64): Add SIMD mov case.
4260 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4262 * tree-predcom.c: (struct chain): Handle store-store chain in which
4263 stores for elimination only store loop invariant values.
4264 (execute_pred_commoning_chain): Ditto.
4265 (prepare_initializers_chain_store_elim): Ditto.
4266 (prepare_finalizers): Ditto.
4267 (is_inv_store_elimination_chain): New function.
4268 (initialize_root_vars_store_elim_1): New function.
4270 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4272 * tree-predcom.c: Revise general description of the pass.
4273 (enum chain_type): New enum type for store elimination.
4274 (struct chain): New field supporting store elimination.
4275 (struct component): Ditto.
4276 (dump_chain): Dump store-stores chain.
4277 (release_chain): Release resources.
4278 (split_data_refs_to_components): Compute and create component
4279 contains only stores for elimination.
4280 (get_chain_last_ref_at): New function.
4281 (make_invariant_chain): Initialization.
4282 (make_rooted_chain): Specify chain type in parameter and record it.
4283 (add_looparound_copies): Skip for store-stores chain.
4284 (determine_roots_comp): Compute type of chain and pass it to
4286 (initialize_root_vars_store_elim_2): New function.
4287 (finalize_eliminated_stores): New function.
4288 (remove_stmt): Handle store for elimination.
4289 (execute_pred_commoning_chain): Execute predictive commoning on
4291 (determine_unroll_factor): Skip unroll for store-stores chain.
4292 (prepare_initializers_chain_store_elim): New function.
4293 (prepare_initializers_chain): Hanlde store-store chain.
4294 (prepare_finalizers_chain, prepare_finalizers): New function.
4295 (tree_predictive_commoning_loop): Return integer value indicating
4296 if loop is unrolled or lcssa form is corrupted.
4297 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4299 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4301 * tree-predcom.c (initialize_root): Delete.
4302 (execute_pred_commoning_chain): Initialize root vars and replace
4303 reference of non-combined chain directly, rather than call above
4306 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4308 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4309 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4311 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4313 * tree-predcom.c (struct chain): New field init_seq.
4314 (release_chain): Release init_seq.
4315 (prepare_initializers_chain): Record intialization stmts in above
4317 (insert_init_seqs): New function.
4318 (tree_predictive_commoning_loop): Call insert_init_seqs.
4320 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4322 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4324 2017-07-28 Richard Biener <rguenther@suse.de>
4326 * match.pd: Remove superfluous :c.
4327 * genmatch.c (simplify::id): Add member.
4328 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4330 (current_id): New global.
4331 (dt_node::parent): Move from ...
4332 (dt_operand::parent): ... here. Add for_id member.
4333 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4334 (decision_tree::find_node): Relax order requirement when
4335 merging DT_TRUE nodes to ones inbetween the current simplify
4336 and the one we try to merge with. Add diagnostic whenever
4337 we need to enforce pattern order by not merging.
4338 (decision_tree::insert): Set current_id.
4339 (decision_tree::print_node): Dump parent node and for_id.
4340 (parser::last_id): Add member.
4341 (parser::push_simplify): Assign unique id.
4342 (parser::parser): Initialize last_id.
4344 2017-07-28 Martin Liska <mliska@suse.cz>
4347 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4348 gimple_build_debug_bind.
4350 2017-07-28 Richard Biener <rguenther@suse.de>
4352 PR tree-optimization/81502
4353 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4355 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4357 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4358 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4359 for BIT_FIELD_REF args.
4360 * fold-const.c (make_bit_field_ref): Likewise.
4361 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4363 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4366 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4367 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4368 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4369 Or it into SANITIZER_UNDEFINED.
4370 * ubsan.c: Include gimple-fold.h and varasm.h.
4371 (ubsan_expand_ptr_ifn): New function.
4372 (instrument_pointer_overflow): New function.
4373 (maybe_instrument_pointer_overflow): New function.
4374 (instrument_object_size): Formatting fix.
4375 (pass_ubsan::execute): Call instrument_pointer_overflow
4376 and maybe_instrument_pointer_overflow.
4377 * internal-fn.c (expand_UBSAN_PTR): New function.
4378 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4379 * sanitizer.def (__ubsan_handle_pointer_overflow,
4380 __ubsan_handle_pointer_overflow_abort): New builtins.
4381 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4382 * internal-fn.def (UBSAN_PTR): New internal function.
4383 * opts.c (sanitizer_opts): Add pointer-overflow.
4384 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4385 * fold-const.c (build_range_check): Compute pointer range check in
4386 integral type if pointer arithmetics would be needed. Formatting
4389 2017-07-28 Martin Liska <mliska@suse.cz>
4392 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4393 parameters that are of a variable-length.
4395 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4397 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4399 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4400 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4401 (CRT_CALL_STATIC_FUNCTION): Likewise.
4402 (ASM_DEFAULT_SPEC): New define.
4403 (ASM_SPEC32): Likewise.
4404 (ASM_SPEC64): Likewise.
4405 (ASM_SPEC_COMMON): Likewise.
4406 (ASM_SPEC): Likewise.
4407 (INVALID_64BIT): Likewise.
4408 (LINK_OS_DEFAULT_SPEC): Likewise.
4409 (LINK_OS_SPEC32): Likewise.
4410 (LINK_OS_SPEC64): Likewise.
4411 (POWERPC_LINUX): Likewise.
4412 (PTRDIFF_TYPE): Likewise.
4413 (RESTORE_FP_PREFIX): Likewise.
4414 (RESTORE_FP_SUFFIX): Likewise.
4415 (SAVE_FP_PREFIX): Likewise.
4416 (SAVE_FP_SUFFIX): Likewise.
4417 (SIZE_TYPE): Likewise.
4418 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4419 (TARGET_64BIT): Likewise.
4420 (TARGET_64BIT): Likewise.
4421 (TARGET_AIX): Likewise.
4422 (WCHAR_TYPE_SIZE): Likewise.
4423 (WCHAR_TYPE): Undefine.
4424 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4425 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4426 (CPP_OS_RTEMS_SPEC): Delete.
4427 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4428 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4430 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4432 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4434 PR tree-optimization/81578
4435 * tree-parloops.c (build_new_reduction): Bail out if
4436 reduction_code isn't one of the standard OpenMP reductions.
4437 Move the details printing after that decision.
4439 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4441 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4442 related to reload_in_progress.
4443 (splat_input_operand): Likewise.
4444 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4446 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4448 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4449 (TARGET_INSTANTIATE_DECLS): Likewise.
4450 (legitimate_indexed_address_p): Delete reload_in_progress code.
4451 (rs6000_debug_legitimate_address_p): Likewise.
4452 (rs6000_eliminate_indexed_memrefs): Likewise.
4453 (rs6000_emit_le_vsx_store): Likewise.
4454 (rs6000_emit_move_si_sf_subreg): Likewise.
4455 (rs6000_emit_move): Likewise.
4456 (register_to_reg_type): Likewise.
4457 (rs6000_pre_atomic_barrier): Likewise.
4458 (rs6000_machopic_legitimize_pic_address): Likewise.
4459 (rs6000_allocate_stack_temp): Likewise.
4460 (rs6000_address_for_fpconvert): Likewise.
4461 (rs6000_address_for_altivec): Likewise.
4462 (rs6000_secondary_memory_needed_rtx): Delete function.
4463 (rs6000_check_sdmode): Likewise.
4464 (rs6000_alloc_sdmode_stack_slot): Likewise.
4465 (rs6000_instantiate_decls): Likewise.
4466 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4467 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4468 Delete reload_in_progress.
4469 (*vec_reload_and_plus_<mptrsize>): Likewise.
4470 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4471 (vsx_div_v2di): Likewise.
4472 (vsx_udiv_v2di): Likewise.
4474 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4476 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4477 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4478 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4479 (rs6000_debug_reg_global): Delete print of LRA status.
4480 (rs6000_option_override_internal): Delete dead LRA related code.
4481 (rs6000_lra_p): Delete function.
4482 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4484 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4486 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4487 * config/riscv/rtems.h: New file.
4489 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4490 Sudakshina Das <sudi.das@arm.com>
4492 * config/aarch64/aarch64.md
4493 (define_split for and<mode>3nr_compare): Move
4494 non aarch64_logical_operand to a register.
4495 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4496 register immediate operand to a register.
4497 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4499 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4502 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4504 2017-07-27 Richard Biener <rguenther@suse.de>
4506 PR tree-optimization/81573
4507 PR tree-optimization/81494
4508 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4509 multi defuse cycle case.
4511 2017-07-27 Richard Biener <rguenther@suse.de>
4513 PR tree-optimization/81571
4514 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4517 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4519 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4520 earlier and only if MASK_FPU is set. Adjust formatting.
4522 2017-07-27 Martin Liska <mliska@suse.cz>
4524 * opt-functions.awk: Add validation of value of Init.
4525 * optc-gen.awk: Pass new argument.
4527 2017-07-27 Martin Liska <mliska@suse.cz>
4529 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4530 Fix wrong condition.
4532 2017-07-27 Martin Liska <mliska@suse.cz>
4534 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4535 BBs and edges seen by autoFDO.
4537 2017-07-27 Richard Biener <rguenther@suse.de>
4539 PR tree-optimization/81502
4540 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4541 with incompatible but same sized type.
4542 (execute_update_addresses_taken): Likewise.
4544 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4546 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4547 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4549 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4552 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4553 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4554 Change s_operand to memory_operand.
4556 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4558 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4559 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4560 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4561 Emit instructions rather than returning an expression. Handle TFmode
4562 and KFmode by casting to TImode.
4563 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4564 (rs6000_emit_le_vsx_store): Likewise.
4565 * config/rs6000/vsx.md (VSX_TI): New iterator.
4566 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4567 (*vsx_le_undo_permute_<mode>): Likewise.
4568 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4569 emit the split sequence.
4570 (*vsx_le_perm_store_<mode>): Likewise.
4572 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4574 PR tree-optimization/81555
4575 PR tree-optimization/81556
4576 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4577 if true, force CHANGED for the recursive invocation.
4578 (reassociate_bb): Remember original length of ops array, pass
4579 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4581 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4582 attributes for noipa attribute. For naked attribute use
4583 lookup_attribute first before lookup_attribute_spec.
4584 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4586 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4587 for functions with noipa attribute.
4588 (cgraph_externally_visible_p): Return true for functions with noipa
4590 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4591 for functions with noipa attribute.
4592 * doc/extend.texi: Document noipa function attribute.
4593 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4594 also for functions with noipa attribute.
4595 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4597 2017-07-26 Andrew Pinski <apinski@cavium.com>
4599 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4600 vec_unalign_load_cost and vec_unalign_store_cost.
4602 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4604 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4605 -mvsx-small-integer option.
4606 (ISA_3_0_MASKS_IEEE): Likewise.
4607 (OTHER_VSX_VECTOR_MASKS): Likewise.
4608 (POWERPC_MASKS): Likewise.
4609 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4610 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4611 code, only testing for DImode being allowed in non-VSX floating
4613 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4614 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4616 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4617 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4618 TARGET_P8_VECTOR test.
4619 (rs6000_secondary_reload_simple_move): Likewise.
4620 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4621 since TARGET_P9_VECTOR was already tested.
4622 (rs6000_opt_masks): Remove -mvsx-small-integer.
4623 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4624 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4626 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4627 test for TARGET_VEXTRACTUB was used, and that uses
4629 (p9 extract splitter): Likewise.
4630 (vsx_extract_<mode>_di_p9): Likewise.
4631 (vsx_extract_<mode>_store_p9): Likewise.
4632 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4633 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4634 the elimination of TARGET_VSX_SMALL_INTEGER.
4635 (vsx_extract_<mode>_p8): Likewise.
4636 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4637 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4638 (vsx_set_<mode>_p9): Likewise.
4639 (vsx_set_v4sf_p9): Likewise.
4640 (vsx_set_v4sf_p9_zero): Likewise.
4641 (vsx_insert_extract_v4sf_p9): Likewise.
4642 (vsx_insert_extract_v4sf_p9_2): Likewise.
4643 * config/rs6000/rs6000.md (sign extend splitter): Change
4644 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4645 (floatsi<mode>2_lfiwax_mem): Likewise.
4646 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4647 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4648 since a test for TARGET_P9_VECTOR was used.
4649 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4650 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4651 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4652 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4653 TARGET_P8_VECTOR test.
4654 (fix_trunc<mode>si2_stfiwx): Likewise.
4655 (fix_trunc<mode>si2_internal): Likewise.
4656 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4657 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4659 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4660 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4661 TARGET_P8_VECTOR test.
4662 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4663 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4664 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4666 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4667 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4668 since a test for TARGET_P9_VECTOR was used.
4669 (splitter for loading small constants): Likewise.
4671 2017-07-26 Andrew Pinski <apinski@cavium.com>
4673 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4676 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4679 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4680 (fp_valid_at): Likewise.
4682 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4684 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4685 (qdf24xx_addrcost_table): Likewise.
4686 (cortexa57_tunings): Update to use generic_branch_cost.
4687 (cortexa72_tunings): Likewise.
4688 (cortexa73_tunings): Likewise.
4689 (qdf24xx_tunings): Likewise.
4691 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4693 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4694 (thunderx2t99_branch_cost): Likewise.
4695 (cortexa35_tunings): Update to use generic_branch_cost.
4696 (cortexa53_tunings): Likewise.
4697 (cortexa57_tunings): Likewise.
4698 (cortexa72_tunings): Likewise.
4699 (cortexa73_tunings): Likewise.
4700 (thunderx2t99_tunings): Likewise.
4702 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4704 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4705 (sparc_option_override): Honour MASK_FSMULD.
4706 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4707 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4708 * config/sparc/sparc.opt (mfsmuld): New option.
4709 * doc/invoke.texi (mfsmuld): Document option.
4711 2017-07-26 Marek Polacek <polacek@redhat.com>
4714 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4716 2017-07-26 Richard Biener <rguenther@suse.de>
4718 * gimple-match-head.c (do_valueize): Return OP if valueize
4720 (get_def): New helper to get at the def stmt of a SSA name
4722 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4723 do_valueize to get at the def stmt.
4724 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4726 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4729 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4731 2017-07-26 Martin Liska <mliska@suse.cz>
4734 * function.c (expand_function_start): Make expansion of
4735 nonlocal_goto_save_area after parm_birth_insn.
4737 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4739 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4740 from all CPU target flags enable members.
4742 2017-07-26 Richard Biener <rguenther@suse.de>
4744 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4745 (decision_tree::gen): Make 'type' const.
4746 (write_predicate): Likewise.
4748 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4750 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4751 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4752 (rs6000_option_override_internal): Likewise.
4753 (rs6000_expand_vector_set): Likewise.
4754 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4755 (TARGET_UPPER_REGS_SF): Likewise.
4756 (TARGET_UPPER_REGS_DI): Likewise.
4757 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4758 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4759 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4760 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4761 (Splitters for DI constants in Altivec registers): Likewise.
4762 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4763 (vsx_set_v4sf_p9): Likewise.
4764 (vsx_set_v4sf_p9_zero): Likewise.
4765 (vsx_insert_extract_v4sf_p9): Likewise.
4766 (vsx_insert_extract_v4sf_p9_2): Likewise.
4768 2017-07-25 Carl Love <cel@us.ibm.com>
4770 * doc/extend.texi: Update the built-in documentation file for the
4771 existing built-in functions
4772 vector signed char vec_cnttz (vector signed char);
4773 vector unsigned char vec_cnttz (vector unsigned char);
4774 vector signed short vec_cnttz (vector signed short);
4775 vector unsigned short vec_cnttz (vector unsigned short);
4776 vector signed int vec_cnttz (vector signed int);
4777 vector unsigned int vec_cnttz (vector unsigned int);
4778 vector signed long long vec_cnttz (vector signed long long);
4779 vector unsigned long long vec_cnttz (vector unsigned long long);
4781 2017-07-25 Andrew Pinski <apinski@cavium.com>
4783 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4784 accesses where the use is for the first operand of a BIT_INSERT.
4786 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4789 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4790 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4792 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4794 * config/i386/gstabs.h: Delete.
4795 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4797 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4799 * config/i386/i386.c (ix86_decompose_address): Do not check for
4800 register RTX when looking at index_reg or base_reg.
4801 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4803 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4805 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4806 to update EH info here.
4808 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4810 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4812 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4814 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4816 2017-07-25 Torsten Duwe <duwe@suse.de>
4818 * common.opt: Introduce -fpatchable-function-entry
4819 command line option, and its variables function_entry_patch_area_size
4820 and function_entry_patch_area_start.
4821 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4822 including a two-value parser.
4823 * target.def (print_patchable_function_entry): New target hook.
4824 * targhooks.h (default_print_patchable_function_entry): New function.
4825 * targhooks.c (default_print_patchable_function_entry): Likewise.
4826 * toplev.c (process_options): Switch off IPA-RA if
4827 patchable function entries are being generated.
4828 * varasm.c (assemble_start_function): Look at the
4829 patchable-function-entry command line switch and current
4830 function attributes and maybe generate NOP instructions by
4831 calling the print_patchable_function_entry hook.
4832 * doc/extend.texi: Document patchable_function_entry attribute.
4833 * doc/invoke.texi: Document -fpatchable_function_entry
4834 command line option.
4835 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4837 * doc/tm.texi: Re-generate.
4839 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4842 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4843 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4845 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4847 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4849 2017-07-25 Richard Biener <rguenther@suse.de>
4851 PR tree-optimization/81455
4852 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4853 not walk in cycles when looking for guards.
4855 2017-07-25 Richard Biener <rguenther@suse.de>
4857 PR tree-optimization/81529
4858 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4859 when optimizing backedge uses.
4861 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4863 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4865 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4866 to dl_section_ref. On AIX, append an expression to subtract
4867 the size of the section length to dl_section_ref.
4869 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4871 * configure.ac: If any of the config.* scripts fail, exit 1.
4872 * configure: Regenerate.
4874 2017-07-25 Richard Biener <rguenther@suse.de>
4877 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4878 of immediate uses, be more verbose on errors.
4880 2017-07-25 Richard Biener <rguenther@suse.de>
4882 PR tree-optimization/81510
4883 * tree-vect-loop.c (vect_is_simple_reduction): When the
4884 reduction stmt is not inside the loop bail out.
4886 2017-07-25 Richard Biener <rguenther@suse.de>
4888 PR tree-optimization/81303
4889 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4890 profitability check against LOOP_VINFO_NITERSM1.
4892 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4894 * domwalk.c (cmp_bb_postorder): Simplify.
4895 (sort_bbs_postorder): New function. Use it...
4896 (dom_walker::walk): ...here to optimize common cases.
4898 2017-07-25 Martin Liska <mliska@suse.cz>
4901 * ipa-visibility.c (function_and_variable_visibility): Make the
4902 redirection just on target that supports aliasing.
4903 Fix GNU coding style.
4905 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4908 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4910 * config/arm/rtems.h: Likewise.
4911 * config/bfin/rtems.h: Likewise.
4912 * config/i386/rtemself.h: Likewise.
4913 * config/lm32/rtems.h: Likewise.
4914 * config/m32c/rtems.h: Likewise.
4915 * config/m68k/rtemself.h: Likewise.
4916 * config/microblaze/rtems.h: Likewise.
4917 * config/mips/rtems.h: Likewise.
4918 * config/moxie/rtems.h: Likewise.
4919 * config/nios2/rtems.h: Likewise.
4920 * config/powerpcspe/rtems.h: Likewise.
4921 * config/rs6000/rtems.h: Likewise.
4922 * config/rtems.h: Likewise.
4923 * config/sh/rtems.h: Likewise.
4924 * config/sh/rtemself.h: Likewise.
4925 * config/sparc/rtemself.h: Likewise.
4927 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4930 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4931 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4932 * tree-ssa-structalias.c (alias_get_name): Same.
4934 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4937 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4938 instructions if no du chain is found.
4940 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4942 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4944 2017-07-25 Richard Biener <rguenther@suse.de>
4947 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4950 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4952 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4954 (ISA_2_7_MASKS_SERVER): Likewise.
4955 (ISA_3_0_MASKS_IEEE): Likewise.
4956 (OTHER_P8_VECTOR_MASKS): Likewise.
4957 (OTHER_VSX_VECTOR_MASKS): Likewise.
4958 (POWERPC_MASKS): Likewise.
4959 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4960 duplicate list of options.
4961 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4962 explicit -mupper-regs options.
4963 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4964 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4965 alias for -mupper-regs-df.
4966 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4967 (rs6000_init_hard_regno_mode_ok): Likewise.
4968 (rs6000_option_override_internal): Likewise.
4969 (rs6000_opt_masks): Likewise.
4970 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4971 options in terms of whether -mvsx or -mpower8-vector was used.
4972 (TARGET_UPPER_REGS_DI): Likewise.
4973 (TARGET_UPPER_REGS_SF): Likewise.
4974 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4975 -mupper-regs-* options.
4977 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4979 * passes.c (emergency_dump_function): Print some empty lines and a
4980 header before the RTL dump.
4982 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4984 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4986 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4989 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4990 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4992 2017-07-24 Carl Love <cel@us.ibm.com>
4994 * config/rs6000/rs6000-c.c: Add support for built-in functions
4995 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4996 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4997 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4998 vec_extract_fp_from_shortl): Add defines for the two builtins.
4999 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5000 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5002 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5003 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5004 * doc/extend.texi: Update the built-in documentation file for the
5005 new built-in function.
5007 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5010 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5012 * doc/generic.texi: Likewise.
5013 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5014 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5016 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5018 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5019 (aarch64_mls_elt_merge<mode>): Likewise.
5021 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5023 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5024 having __cxa_atexit.
5026 2017-07-23 Michael Collison <michael.collison@arm.com>
5028 * config/arm/arm.c (arm_option_override): Deprecate
5029 use of -mstructure-size-boundary.
5030 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5031 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5033 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5036 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5037 Reduce cost estimate for direct moves.
5039 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5042 * config/i386/i386.c (ix86_option_override_internal): Disable
5043 BMI, BMI2 and TBM instructions for -m16.
5045 2017-07-21 Carl Love <cel@us.ibm.com>
5047 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5048 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5049 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5050 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5051 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5052 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5053 VMULOSW): New enum "unspec" values.
5054 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5055 altivec_vmulosw): New patterns.
5056 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5057 VMULOSW): Add definitions.
5059 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5061 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5062 (qdf24xx): Likewise.
5063 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5064 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5065 (AARCH64_FL_V8_1): Renumber.
5066 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5067 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5068 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5069 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5070 rdma to feature modifiers list.
5072 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5075 * ipa-visibility (function_and_variable_visibility): Convert
5076 recursive PLT call to direct call if appropriate.
5078 2017-07-21 Andrew Pinski <apinski@cavium.com>
5080 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5081 operand 1 to see if the types precision matches.
5082 * fold-const.c (operand_equal_p): Likewise.
5084 2017-07-21 Richard Biener <rguenther@suse.de>
5086 PR tree-optimization/81303
5087 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5088 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5089 (vect_peeling_hash_get_lowest_cost): Adjust.
5090 (vect_enhance_data_refs_alignment): Likewise. Use
5091 vect_get_peeling_costs_all_drs to compute the penalty for no
5092 peeling to match up costs.
5094 2017-07-21 Richard Biener <rguenther@suse.de>
5096 PR tree-optimization/81500
5097 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5098 we didn't identify a reduction path.
5100 2017-07-21 Tom de Vries <tom@codesourcery.com>
5101 Cesar Philippidis <cesar@codesourcery.com>
5103 PR gcov-profile/81442
5104 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5107 2017-07-21 Tom de Vries <tom@codesourcery.com>
5110 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5112 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5113 nvptx_override_options_after_change.
5115 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5117 * dwarf2out.c (output_file_names): Avoid double testing for
5120 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5122 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5124 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5126 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5128 (try_crossjump_to_edge): Do not punt on partitioned functions.
5130 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5132 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5133 Put all BBs reachable only via paths crossing cold region to cold
5135 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5137 2016-07-21 Richard Biener <rguenther@suse.de>
5139 PR tree-optimization/81303
5140 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5141 into account prologue and epilogue iterations when raising
5142 min_profitable_iters to sth at least covering one vector iteration.
5144 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5146 * config/arm/arm.c (arm_test_cpu_arch_dat):
5149 2017-07-20 Nathan Sidwell <nathan@acm.org>
5151 Remove TYPE_METHODS.
5152 * tree.h (TYPE_METHODS): Delete.
5153 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5154 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5155 (dbxout_type_methods): Scan TYPE_FIELDS.
5156 (dbxout_type): Don't check TYPE_METHODS here.
5157 * function.c (use_register_for_decl): Always ignore register for
5158 class types when not optimizing.
5159 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5160 * tree.c (free_lang_data_in_type): Stitch out member functions and
5161 templates from TYPE_FIELDS.
5162 (build_distinct_type_copy, verify_type_variant,
5163 verify_type): Member fns are on TYPE_FIELDS.
5164 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5165 * tree-pretty-print.c (dump_generic_node): Likewise.
5167 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5170 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5171 V2TImode and V4TImode.
5172 (ix86_expand_vector_extract): Likewise.
5173 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5174 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5175 (ssescalarmode): Handle V4TImode and V2TImode.
5176 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5177 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5178 (VEXTRACTI128_MODE): New mode iterator.
5179 (splitter for *vec_extractv?ti first element): New.
5180 (VEC_INIT_MODE): New mode iterator.
5181 (vec_init<mode>): Consolidate 3 expanders into one using
5182 VEC_INIT_MODE mode iterator.
5184 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5186 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5187 non_spilled_static_chain_regno_p.
5189 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5191 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5193 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5195 * bb-reorder.c (connect_traces): Allow copying of blocks within
5198 2017-07-20 Richard Biener <rguenther@suse.de>
5200 * gimple.h (gimple_phi_result): Add gphi * overload.
5201 (gimple_phi_result_ptr): Likewise.
5202 (gimple_phi_arg): Likewise. Adjust index assert to only
5203 allow actual argument accesses rather than all slots available
5205 (gimple_phi_arg_def): Add gphi * overload.
5206 * tree-phinodes.c (make_phi_node): Initialize only actual
5208 (resize_phi_node): Clear memory not covered by old node,
5209 do not initialize excess argument slots.
5210 (reserve_phi_args_for_new_edge): Initialize new argument slot
5213 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5215 PR tree-optimization/81388
5217 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5219 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5220 by removing computation of may_be_zero.
5222 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5223 Tom de Vries <tom@codesourcery.com>
5226 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5227 when gimple level profile disagrees with what RTL expander did.
5229 2017-07-20 Richard Biener <rguenther@suse.de>
5231 PR tree-optimization/61171
5232 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5233 (vect_analyze_stmt): Add slp instance parameter.
5234 (vectorizable_reduction): Likewise.
5235 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5236 (vect_is_simple_reduction): Deal with chains not detected
5237 as SLP reduction chain, specifically not properly associated
5238 chains containing a mix of plus/minus.
5239 (get_reduction_op): Remove.
5240 (get_initial_defs_for_reduction): Simplify, pass in whether
5241 this is a reduction chain, pass in the SLP node for the PHIs.
5242 (vect_create_epilog_for_reduction): Get the SLP instance as
5244 (vectorizable_reduction): Get the SLP instance as arg.
5245 During analysis remember the SLP node with the PHIs in the
5246 instance. Simplify getting at the vectorized reduction PHIs.
5247 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5248 through SLP instance.
5249 (vect_slp_analyze_operations): Likewise.
5250 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5251 (vect_transform_stmt): Likewise.
5253 2017-07-20 Tom de Vries <tom@codesourcery.com>
5255 PR tree-optimization/81489
5256 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5257 read of phi arg location to before loop that modifies phi.
5259 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5261 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5264 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5267 * except.c (execute): Fix ordering issue.
5269 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5271 PR rtl-optimization/81423
5272 * combine.c (make_compound_operation_int): Don't try to optimize
5273 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5275 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5277 PR rtl-optimization/81423
5278 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5279 with a constant that is -1 in the truncated to mode.
5281 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5283 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5284 (determine_unlikely_bbs): ... here.
5285 * predict.h (propagate_unlikely_bbs_forward): Declare.
5286 * cfgexpand.c (pass_expand::execute): Use it.
5287 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5289 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5290 propagate_unlikely_bbs_forward.
5292 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5295 * except.c (maybe_add_nop_after_section_switch): New function.
5298 2017-07-19 Tom de Vries <tom@codesourcery.com>
5300 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5302 2017-07-19 Tom de Vries <tom@codesourcery.com>
5304 * gimple.h (gimple_phi_arg): Make assert more strict.
5306 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5308 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5309 * config/rs6000/mmintrin.h: New file.
5310 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5312 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5314 PR tree-optimization/81346
5315 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5317 2017-07-19 Tom de Vries <tom@codesourcery.com>
5319 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5321 2017-07-19 Tom de Vries <tom@codesourcery.com>
5323 * config/nvptx/nvptx-modes.def: Add V2DImode.
5324 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5325 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5326 (nvptx_output_mov_insn): Handle lack of mov.b128.
5327 (nvptx_print_operand): Handle 'H' and 'L' codes.
5328 (nvptx_vector_mode_supported): Allow V2DImode.
5329 (nvptx_preferred_simd_mode): New function.
5330 (nvptx_data_alignment): New function.
5331 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5332 nvptx_preferred_simd_mode.
5333 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5335 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5337 2017-07-19 Tom de Vries <tom@codesourcery.com>
5339 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5340 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5341 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5342 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5343 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5344 (mov<VECIM>_insn): New define_insn.
5345 (define_expand "mov<VECIM>): New define_expand.
5347 2017-07-19 Tom de Vries <tom@codesourcery.com>
5349 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5351 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5353 PR tree-optimization/81346
5354 * fold-const.h (fold_div_compare, range_check_type): Declare.
5355 * fold-const.c (range_check_type): New function.
5356 (build_range_check): Use range_check_type.
5357 (fold_div_compare): No longer static, rewritten into
5358 a match.pd helper function.
5359 (fold_comparison): Don't call fold_div_compare here.
5360 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5363 2017-07-19 Nathan Sidwell <nathan@acm.org>
5365 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5366 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5367 * tree.c (find_decls_types_r, verify_type): Use
5368 TYPE_{MIN,MAX}_VALUE_RAW.
5369 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5370 (hash_tree): Likewise.
5371 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5373 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5376 2017-07-18 Tom de Vries <tom@codesourcery.com>
5379 * omp-expand.c (expand_omp_for_static_chunk): Handle
5380 equal-argument loop exit phi.
5382 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5385 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5386 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5387 operand 2 predicate.
5388 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5389 operand 2 predicate.
5390 (ror,rol -> rorx splitters): Use const_int_operand as
5391 operand 2 predicate.
5393 2017-06-18 Richard Biener <rguenther@suse.de>
5395 PR tree-optimization/81410
5396 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5397 the gap in the ! slp_perm SLP case after each group.
5399 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5402 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5405 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5408 * predict.c (set_even_probabilities): Cleanup; do not affect
5409 probabilities that are already known.
5410 (combine_predictions_for_bb): Call even when count is set.
5412 2017-07-18 Nathan Sidwell <nathan@acm.org>
5414 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5417 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5420 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5421 optimization for loop niter analysis.
5423 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5426 * config/avr/avr.c (avr_optimize_casesi): Don't use
5427 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5429 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5431 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5432 body_cost_vec from _vect_peel_extended_info.
5433 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5434 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5437 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5439 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5441 2017-07-18 Richard Biener <rguenther@suse.de>
5443 PR tree-optimization/80620
5444 PR tree-optimization/81403
5445 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5446 info when re-using a VN table entry.
5448 2017-07-18 Richard Biener <rguenther@suse.de>
5450 PR tree-optimization/81418
5451 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5452 vectype_in. Verify that with lane-reducing reduction operations
5453 we have a single def-use cycle.
5455 2017-07-17 Carl Love <cel@us.ibm.com>
5457 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5459 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5460 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5461 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5462 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5463 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5464 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5465 VMULOSW): New enum "unspec" values.
5466 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5467 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5468 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5469 altivec_vmulosw): New patterns.
5470 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5471 VMULOSW): Add definitions.
5472 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5474 * config/alpha/alpha.c: Include predict.h.
5476 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5478 * tree-vrp.c (compare_assert_loc): Fix comparison function
5479 to return predictable results.
5481 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5483 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5486 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5487 * doc/invoke.texi (mexpand-adddi): Update text.
5489 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5491 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5492 that also clobbers the CC register. The old expand code is moved
5494 (*arc_clzsi2): ... here.
5495 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5496 the CC register. The old expand code is moved to ...
5497 (arc_ctzsi2): ... here.
5499 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5501 * config/arc/arc.opt (mindexed-loads): Use initial value
5502 TARGET_INDEXED_LOADS_DEFAULT.
5503 (mauto-modify-reg): Use initial value
5504 TARGET_AUTO_MODIFY_REG_DEFAULT.
5505 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5506 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5507 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5508 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5510 2017-07-17 Martin Liska <mliska@suse.cz>
5513 * opts.c (finish_options): Do not allow -fgnu-tm
5514 w/ -fsanitize={kernel-,}address. Say sorry.
5516 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5519 * tree-loop-distribution.c (classify_partition): Only assert on
5520 numer of iterations.
5521 (merge_dep_scc_partitions): Delete prameter. Update function call.
5522 (distribute_loop): Remove code handling loop with unknown niters.
5523 (pass_loop_distribution::execute): Skip loop with unknown niters.
5525 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5528 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5529 function sort_partitions_by_post_order.
5531 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5533 PR tree-optimization/81374
5534 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5535 the max index of basic blocks, rather than number of basic blocks.
5537 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5539 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5541 (arc_legitimate_pic_operand_p): Likewise.
5542 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5544 (arc_needs_pcl_p): Likewise.
5545 (arc_legitimate_pc_offset_p): Likewise.
5546 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5547 function is also used in constrains.md.
5548 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5549 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5550 PLUS. Only return true/false in known cases, otherwise assert.
5551 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5552 is already called in arc_legitimate_constant_p.
5553 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5555 (LEGITIMATE_PIC_OPERAND_P): Use
5556 arc_raw_symbolic_reference_mentioned_p function.
5557 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5562 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5563 Andrew Burgess <andrew.burgess@embecosm.com>
5565 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5566 (arc_return_address_register): New function.
5567 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5568 (arc_handle_fndecl_attribute): Add naked attribute.
5569 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5570 (TARGET_WARN_FUNC_RETURN): Likewise.
5571 (arc_allocate_stack_slots_for_args): New function.
5572 (arc_warn_func_return): Likewise.
5573 (machine_function): Change type fn_type.
5574 (arc_compute_function_type): Consider new naked function type,
5575 change function return type.
5576 (arc_must_save_register): Adapt to handle new
5577 arc_compute_function_type's return type.
5578 (arc_expand_prologue): Likewise.
5579 (arc_expand_epilogue): Likewise.
5580 (arc_return_address_regs): Delete.
5581 (arc_return_address_register): New function.
5582 (arc_epilogue_uses): Use above function.
5583 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5584 (arc_function_type): Change encoding, add naked type.
5585 (ARC_INTERRUPT_P): Change to handle the new encoding.
5586 (ARC_FAST_INTERRUPT_P): Likewise.
5587 (ARC_NORMAL_P): Define.
5588 (ARC_NAKED_P): Likewise.
5589 (arc_compute_function_type): Delete prototype.
5590 * config/arc/arc.md (in_ret_delay_slot): Use
5591 arc_return_address_register function.
5592 (simple_return): Likewise.
5593 (p_return_i): Likewise.
5595 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5597 PR tree-optimization/81428
5598 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5599 can't be built for those types.
5601 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5603 Remove stuff dead since r239246.
5605 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5606 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5607 (avr_inform_devices): Remove dead stuff.
5609 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5611 * config/arm/arm_neon.h: Fix softp typo.
5613 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5615 PR tree-optimization/81365
5616 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5617 aggregate moves onto bb predecessor edges, make sure there are no
5618 loads that could alias the lhs in between the start of bb and the
5621 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5624 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5625 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5626 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5628 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5630 PR tree-optimization/81396
5631 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5632 (init_symbolic_number): Initialize it to 1.
5633 (perform_symbolic_merge): Add n_ops from both operands into the new
5635 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5636 without base_addr as useless if they need more than one operation.
5637 (bswap_replace): Handle !bswap case for NULL base_addr.
5639 2017-07-17 Tom de Vries <tom@codesourcery.com>
5642 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5645 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5647 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5648 conditional builtin define __FIX_LEON3FT_B2BST.
5650 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5652 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5653 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5656 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5658 PR rtl-optimization/81424
5659 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5660 to remove potential trapping from operands if -fnon-call-exceptions.
5662 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5664 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5665 profile_proability for scalling.
5666 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5668 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5670 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5672 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5674 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5675 fixpoint arithmetics.
5677 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5679 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5680 fixpoint arithmetics.
5682 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5684 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5685 fixpoint arithmetics.
5687 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5689 * profile-count.h (profile_probability::from_reg_br_prob_note,
5690 profile_probability::to_reg_br_prob_note): New functions.
5691 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5692 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5693 * predict.c (probability_reliable_p): Update.
5694 (edge_probability_reliable_p): Update.
5695 (br_prob_note_reliable_p): Update.
5696 (invert_br_probabilities): Update.
5697 (add_reg_br_prob_note): New function.
5698 (combine_predictions_for_insn): Update.
5699 * asan.c (asan_clear_shadow): Update.
5700 * cfgbuild.c (compute_outgoing_frequencies): Update.
5701 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5702 (update_br_prob_note): Update.
5703 (rtl_verify_edges): Update.
5704 (purge_dead_edges): Update.
5705 (fixup_reorder_chain): Update.
5706 * emit-rtl.c (try_split): Update.
5707 * ifcvt.c (cond_exec_process_insns): Update.
5708 (cond_exec_process_if_block): Update.
5709 (dead_or_predicable): Update.
5710 * internal-fn.c (expand_addsub_overflow): Update.
5711 (expand_neg_overflow): Update.
5712 (expand_mul_overflow): Update.
5713 * loop-doloop.c (doloop_modify): Update.
5714 * loop-unroll.c (compare_and_jump_seq): Update.
5715 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5716 * predict.h: Update.
5717 * reorg.c (mostly_true_jump): Update.
5719 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5720 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5721 * config/arc/arc.c: (emit_unlikely_jump): Update.
5722 * config/arm/arm.c: (emit_unlikely_jump): Update.
5723 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5724 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5725 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5726 (ix86_print_operand): Update.
5727 (ix86_split_fp_branch): Update.
5728 (predict_jump): Update.
5729 * config/ia64/ia64.c (ia64_print_operand): Update.
5730 * config/mmix/mmix.c (mmix_print_operand): Update.
5731 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5732 (rs6000_expand_split_stack_prologue): Update.
5733 * config/rs6000/rs6000.c: Update.
5734 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5735 (s390_expand_vec_movstr): Update.
5736 (s390_expand_cs_tdsi): Update.
5737 (s390_expand_split_stack_prologue): Update.
5738 * config/sh/sh.c (sh_print_operand): Update.
5739 (expand_cbranchsi4): Update.
5740 (expand_cbranchdi4): Update.
5741 * config/sparc/sparc.c (output_v9branch): Update.
5742 * config/spu/spu.c (get_branch_target): Update.
5743 (ea_load_store_inline): Update.
5744 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5745 * config/tilepro/tilepro.c: Update.
5747 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5749 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5750 (gimplify_save_expr): Likewise.
5752 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5754 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5756 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5758 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5760 * timevar.def (TV_IPA_FNSUMMARY): Define.
5762 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5764 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5765 to back store errata sensitive sequence from being generated.
5766 (sqrtdf2_fix): Likewise.
5768 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5770 * tree-ssa-threadupdate.c (compute_path_counts,
5771 update_joiner_offpath_counts): Use profile_probability.
5773 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5776 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5778 * config/arm/arm-c.c (arm_cpu_builtins): Define
5779 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5781 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5783 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5784 array entries to represent __ieee128 versions of the
5785 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5786 scalar_extract_sig, and scalar_insert_exp built-in functions.
5787 (altivec_resolve_overloaded_builtin): Add special case handling
5788 for the __builtin_scalar_insert_exp function, as represented by
5789 the P9V_BUILTIN_VEC_VSIEDP constant.
5790 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5791 exponent support for __ieee128 argument.
5792 (VSESQP): Add scalar extract signature support for __ieee128
5794 (VSTDCNQP): Add scalar test negative support for __ieee128
5796 (VSIEQP): Add scalar insert exponent support for __int128 argument
5797 with __ieee128 result.
5798 (VSIEQPF): Add scalar insert exponent support for __ieee128
5799 argument with __ieee128 result.
5800 (VSTDCQP): Add scalar test data class support for __ieee128
5802 (VSTDCNQP): Add overload support for scalar test negative with
5804 (VSTDCQP): Add overload support for scalar test data class
5806 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5808 (UNSPEC_VSX_SIEXPQP): New constant.
5809 (xsxexpqp): New insn for VSX scalar extract exponent quad
5811 (xsxsigqp): New insn for VSX scalar extract significand quad
5813 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5814 precision with floating point argument.
5815 (xststdcqp): New expand for VSX scalar test data class quad
5817 (xststdcnegqp): New expand for VSX scalar test negative quad
5819 (xststdcqp): New insn to match expansions for VSX scalar test data
5820 class quad precision and VSX scalar test negative quad precision.
5821 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5822 special case operand checking to enforce that second operand of
5823 VSX scalar test data class with quad precision argument is a 7-bit
5825 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5826 prototypes and descriptions of __ieee128 versions of
5827 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5828 scalar_test_data_class, and scalar_test_neg built-in functions.
5830 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5832 PR tree-optimization/81162
5833 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5834 replace a negate with an add.
5836 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5838 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5840 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5842 * config/arm/arm-c.c (arm_cpu_builtins): Define
5843 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5845 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5847 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5848 (armv8-r): Set ARM Cortex-R52 as default CPU.
5849 * config/arm/arm-tables.opt: Regenerate.
5850 * config/arm/arm-tune.md: Regenerate.
5851 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5853 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5854 extension for -mcpu=cortex-r52.
5856 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5858 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5859 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5860 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5861 (fp-armv8): Define it as FP_ARMv8 only.
5862 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5863 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5865 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5866 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5867 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5868 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5869 than TARGET_FPU_ARMV8.
5870 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5871 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5872 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5874 * config/arm/neon.md (neon_vrint): Likewise.
5875 (neon_vcvt): Likewise.
5876 (neon_<fmaxmin_op><mode>): Likewise.
5877 (<fmaxmin><mode>3): Likewise.
5878 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5879 * config/arm/predicates.md (arm_cond_move_operator): Check against
5880 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5882 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5884 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5887 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5889 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5890 loop in comment with memset.
5892 2017-07-14 Martin Liska <mliska@suse.cz>
5894 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5895 * dwarf2out.c (is_java): Remove the function.
5896 (output_pubname): Remove usage of the function.
5897 (lower_bound_default): Remove usage of DW_LANG_Java.
5898 (gen_compile_unit_die): Likewise.
5899 * gcc.c: Remove compiler defaults for .java and .zip files.
5900 * gimple-expr.c (remove_suffix): Change as there's no longer
5901 extension than 4-letter one.
5902 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5903 (gimplify_save_expr): Likewise.
5904 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5905 as it's possible even for other languages than Java.
5906 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5907 * lto-opts.c (lto_write_options): Remove reference to Java.
5908 * opts.c (strip_off_ending): Update file extension handling.
5909 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5910 * tree-eh.c (lower_resx): Likewise.
5911 * tree.c (free_lang_data_in_type): Remove dead code.
5912 (find_decls_types_r): Likewise.
5913 (build_common_builtin_nodes): Remove Java from a comment.
5914 (verify_type): Remove dead code.
5915 * varasm.c (assemble_external): Remove Java from a comment.
5917 2017-07-14 Martin Liska <mliska@suse.cz>
5919 * opts.c (finish_options): Add quotes.
5920 (common_handle_option): Likewise.
5922 2017-07-14 Martin Liska <mliska@suse.cz>
5924 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5925 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5927 * dwarf2out.c (lower_bound_default): Do not handle
5929 (gen_compile_unit_die): Likewise.
5930 * gcc.c: Remove default extension binding for GNU Pascal.
5931 * stmt.c: Remove Pascal language from a comment.
5932 * xcoffout.c: Likewise.
5934 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5937 * diagnostic-show-locus.c (fixit_cmp): New function.
5938 (layout::layout): Sort m_fixit_hints.
5939 (column_range::column_range): Assert that the values are valid.
5940 (struct char_span): New struct.
5941 (correction::overwrite): New method.
5942 (struct source_line): New struct.
5943 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5944 calls in terms of classes source_line and char_span, and
5945 correction::overwrite.
5946 (selftest::test_overlapped_fixit_printing_2): New function.
5947 (selftest::diagnostic_show_locus_c_tests): Call it.
5949 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5951 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5952 early if there is no lhs.
5954 2017-07-13 Martin Liska <mliska@suse.cz>
5956 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5957 (gen_reference_type_die): Likewise.
5958 * stor-layout.c: Remove Pascal-related comment.
5960 2017-07-13 Martin Liska <mliska@suse.cz>
5962 * opts.c (finish_options): Add quotes to error messages.
5963 (parse_sanitizer_options): Likewise.
5965 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5967 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5969 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5971 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5973 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5975 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5977 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5979 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5982 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5983 provides the hardware capability bits, define the macro
5984 __BUILTIN_CPU_SUPPORTS__.
5985 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5986 if GLIBC does not provide the hardware capability bits. Add a
5987 gcc_unreachable call if the built-in cpu function is neither
5988 __builtin_cpu_is nor __builtin_cpu_supports.
5989 (rs6000_get_function_versions_dispatcher): Change the warning
5990 that an old GLIBC is used which does not export the capability
5991 bits to be an error.
5992 * doc/extend.texi (target_clones attribute): Document the
5993 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5994 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5995 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5996 the macros defined by GCC if the newer GLIBC is available.
5998 2017-07-12 Jeff Law <law@redhat.com>
6000 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6001 remaining includes slightly.
6002 * config/riscv/riscv-builtins.c: Include profile-count.h.
6004 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6007 * config/avr/avr.c (avr_set_current_function): In diagnostic
6008 messages: Quote keywords and (parts of) identifiers.
6009 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6012 2017-07-12 Carl Love <cel@us.ibm.com>
6014 * config/rs6000/rs6000-c.c: Add support for built-in functions
6015 vector bool char vec_revb (vector bool char);
6016 vector bool short vec_revb (vector short char);
6017 vector bool int vec_revb (vector bool int);
6018 vector bool long long vec_revb (vector bool long long);
6019 * doc/extend.texi: Update the built-in documentation file for the
6020 new built-in functions.
6022 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6024 * config/s390/s390.md: Remove movcc splitter.
6026 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6028 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6029 load/store on condition.
6031 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6034 * config/avr/avr.c (avr_encode_section_info)
6035 [progmem && !TREE_READONLY]: Error if progmem object needs
6038 2017-07-11 Michael Collison <michael.collison@arm.com>
6040 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6043 2017-07-11 Carl Love <cel@us.ibm.com>
6045 * config/rs6000/rs6000-c.c: Add support for builtins
6046 vector unsigned int vec_parity_lsbb (vector signed int);
6047 vector unsigned int vec_parity_lsbb (vector unsigned int);
6048 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6049 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6050 vector unsigned long long vec_parity_lsbb (vector signed long long);
6051 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6052 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6053 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6054 * doc/extend.texi: Update the built-in documentation file for the
6055 new built-in functions.
6057 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6059 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6060 (layout::m_primary_loc): New field.
6061 (layout::layout): Initialize new field. Move location filtering
6062 logic from here to...
6063 (layout::maybe_add_location_range): ...this new method. Add
6064 support for filtering to just the lines already specified by other
6066 (layout::will_show_line_p): New method.
6067 (gcc_rich_location::add_location_if_nearby): New method.
6068 (selftest::test_add_location_if_nearby): New test function.
6069 (selftest::diagnostic_show_locus_c_tests): Call it.
6070 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6073 2017-07-11 Tom de Vries <tom@codesourcery.com>
6075 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6076 (bb_first_real_insn): New function.
6077 (nvptx_single): Add extra initialization of broadcasted condition
6080 2017-07-11 Nathan Sidwell <nathan@acm.org>
6082 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6084 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6086 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6087 Binutils doc as TEXI will mess them up.
6088 * doc/invoke.texi (AVR Options): Same here.
6090 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6092 * config/sparc/sparc.opt (mfix-ut700): New option.
6093 (mfix-gr712rc): Likewise.
6094 (sparc_fix_b2bst): New variable.
6095 * doc/invoke.texi (SPARC options): Document them.
6096 (ARM options): Fix warnings.
6097 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6098 instructions to prevent sequences that can trigger the store-store
6099 errata for certain LEON3FT processors.
6100 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6101 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6102 * config/sparc/sparc.md (fix_b2bst): New attribute.
6103 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6105 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6108 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6110 (*rsqrtsf2_sse): Ditto.
6112 (div<mode>3): Macroize insn from divdf3 and divsf3
6113 using MODEF mode iterator.
6115 2017-07-10 Martin Sebor <msebor@redhat.com>
6117 PR tree-optimization/80397
6118 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6119 instead of testing for equality to INTEGER_TYPE.
6121 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6123 * config.gcc: Remove uclibc from arc target spec.
6125 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6127 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6129 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6132 * lto-wrapper.c (remove_option): New function.
6133 (merge_and_complain): Merge PIC/PIE options more realistically.
6135 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6137 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6141 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6142 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6143 * config.in: Regenerate.
6144 * configure: Regenerate.
6145 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6146 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6147 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6148 (TARGET_GASISR_PROLOGUES): ...target mask.
6149 * common/config/avr/avr-common.c
6150 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6151 Set -mgas-isr-prologues.
6152 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6153 INSERT_PASS_BEFORE for it.
6154 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6155 * config/avr/avr.c (avr_option_override)
6156 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6157 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6158 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6159 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6160 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6162 (make_avr_pass_pre_proep): New function.
6163 (emit_push_sfr) <treg>: Add argument to function and use it
6165 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6166 and set machine->gasisr.yes.
6167 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6168 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6169 __gcc_isr.n_pushed to .L__stack_usage.
6170 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6171 (avr_asm_final_postscan_insn): ...this new static function.
6172 * config/avr/avr.h (machine_function)
6173 <is_no_gccisr, use_L__stack_usage>: New fields.
6174 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6175 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6176 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6177 (gasisr, *gasisr): New expander and insn.
6178 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6179 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6180 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6182 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6184 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6187 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6189 Move jump-tables out of .text again.
6192 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6193 (ASM_OUTPUT_ADDR_VEC): New function.
6194 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6195 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6196 INSN_ADDRESSes as asm comment.
6197 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6198 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6199 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6200 * config/avr/avr.md (*tablejump): Adjust comment.
6201 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6202 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6204 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6205 (avr_output_addr_vec): New proto.
6206 (avr_log_t) <insn_addresses>: New field.
6208 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6211 * config/i386/i386.c (ix86_function_arg_advance): Set
6212 outgoing_args_on_stack to true if there are outgoing arguments
6214 (ix86_function_arg): Likewise.
6215 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6216 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6217 * config/i386/i386.h (machine_function): Add
6218 outgoing_args_on_stack.
6220 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6222 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6224 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6226 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6228 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6229 (REAL_H): Remove $(MACHMODE_H).
6230 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6232 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6233 $(MACHMODE_H) and double-int.h.
6234 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6236 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6239 2017-07-07 Andrew Pinski <apinski@cavium.com>
6241 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6242 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6244 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6246 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6247 Add warning if GCC was not configured to link against a GLIBC that
6248 exports the hardware capability bits.
6249 (make_resolver_func): Make resolver function private and not a
6250 COMDAT function. Create the name with clone_function_name instead
6251 of make_unique_name.
6254 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6255 correct operand in doing the split.
6257 2017-07-07 Carl Love <cel@us.ibm.com>
6259 * config/rs6000/rs6000-c: Add support for built-in function
6260 vector unsigned short vec_pack_to_short_fp32 (vector float,
6262 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6263 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6264 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6265 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6266 (convert_4f32_8i16): Add define_expand.
6267 * doc/extend.texi: Update the built-in documentation file for the
6268 new built-in function.
6270 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6272 * config/sparc/m8.md: New file.
6273 * config/sparc/sparc.md: Include m8.md.
6275 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6277 * config/sparc/sparc.opt: New option -mvis4b.
6278 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6279 (sparc_option_override): Handle VIS4B.
6280 (enum sparc_builtins): Define
6281 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6282 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6283 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6284 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6285 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6286 (check_constant_argument): New function.
6287 (sparc_vis_init_builtins): Define builtins
6288 __builtin_vis_dictunpack{8,16,32},
6289 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6290 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6291 __builtin_vis_fpcmpde{8,16,32}shl and
6292 __builtin_vis_fpcmpur{8,16,32}shl.
6293 (sparc_expand_builtin): Check that the constant operands to
6294 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6295 constant and in range.
6296 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6298 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6299 (SPARC_IMM5_P): Likewise.
6300 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6301 (enabled): Handle vis4b.
6302 (UNSPEC_DICTUNPACK): New unspec.
6303 (UNSPEC_FPCMPSHL): Likewise.
6304 (UNSPEC_FPUCMPSHL): Likewise.
6305 (UNSPEC_FPCMPDESHL): Likewise.
6306 (UNSPEC_FPCMPURSHL): Likewise.
6307 (cpu_feature): New CPU feature `vis4b'.
6308 (dictunpack{8,16,32}): New insns.
6309 (FPCSMODE): New mode iterator.
6310 (fpcscond): New code iterator.
6311 (fpcsucond): Likewise.
6312 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6313 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6314 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6315 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6316 * config/sparc/constraints.md: Define constraints `q' for unsigned
6317 2-bit integer constants and `t' for unsigned 5-bit integer
6319 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6321 (imm5_operand_dictunpack16): Likewise.
6322 (imm5_operand_dictunpack32): Likewise.
6323 (imm2_operand): Likewise.
6324 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6325 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6326 ditunpack* and fpcmp*shl builtins.
6328 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6330 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6331 * config.in: Add HAVE_AS_SPARC6 define.
6332 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6334 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6336 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6337 (CPP_CPU_SPEC): Handle m8.
6338 (ASM_CPU_SPEC): Likewise.
6339 * config/sparc/sparc-opts.h (enum processor_type): Add
6341 * config/sparc/sparc.c (m8_costs): New struct.
6342 (sparc_option_override): Handle TARGET_CPU_m8.
6343 (sparc32_initialize_trampoline): Likewise.
6344 (sparc64_initialize_trampoline): Likewise.
6345 (sparc_issue_rate): Likewise.
6346 (sparc_register_move_cost): Likewise.
6347 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6348 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6349 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6350 (CPP_CPU_SPEC): Handle M8.
6351 (ASM_CPU_SPEC): Likewise.
6352 (AS_M8_FLAG): Define.
6353 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6354 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6355 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6357 * configure: Regenerate.
6358 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6361 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6363 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6365 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6366 ("*movdi_insn_sp32"): Do not set v3pipe.
6367 ("*movsi_insn"): Likewise.
6368 ("*movdi_insn_sp64"): Likewise.
6369 ("*movsf_insn"): Likewise.
6370 ("*movdf_insn_sp32"): Likewise.
6371 ("*movdf_insn_sp64"): Likewise.
6372 ("*zero_extendsidi2_insn_sp64"): Likewise.
6373 ("*sign_extendsidi2_insn"): Likewise.
6374 ("*mov<VM32:mode>_insn"): Likewise.
6375 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6376 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6377 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6378 ("<vlop:code><VL:mode>3"): Likewise.
6379 ("*not_<vlop:code><VL:mode>3"): Likewise.
6380 ("*nand<VL:mode>_vis"): Likewise.
6381 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6382 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6383 ("one_cmpl<VL:mode>2"): Likewise.
6384 ("faligndata<VM64:mode>_vis"): Likewise.
6385 ("alignaddrsi_vis"): Likewise.
6386 ("alignaddrdi_vis"): Likweise.
6387 ("alignaddrlsi_vis"): Likewise.
6388 ("alignaddrldi_vis"): Likewise.
6389 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6390 ("bmaskdi_vis"): Likewise.
6391 ("bmasksi_vis"): Likewise.
6392 ("bshuffle<VM64:mode>_vis"): Likewise.
6393 ("cmask8<P:mode>_vis"): Likewise.
6394 ("cmask16<P:mode>_vis"): Likewise.
6395 ("cmask32<P:mode>_vis"): Likewise.
6396 ("pdistn<P:mode>_vis"): Likewise.
6397 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6399 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6401 * config/sparc/sparc.md ("subtype"): New insn attribute.
6402 ("*wrgsr_sp64"): Set insn subtype.
6403 ("*rdgsr_sp64"): Likewise.
6404 ("alignaddrsi_vis"): Likewise.
6405 ("alignaddrdi_vis"): Likewise.
6406 ("alignaddrlsi_vis"): Likewise.
6407 ("alignaddrldi_vis"): Likewise.
6408 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6409 ("fexpand_vis"): Likewise.
6410 ("fpmerge_vis"): Likewise.
6411 ("faligndata<VM64:mode>_vis"): Likewise.
6412 ("bshuffle<VM64:mode>_vis"): Likewise.
6413 ("cmask8<P:mode>_vis"): Likewise.
6414 ("cmask16<P:mode>_vis"): Likewise.
6415 ("cmask32<P:mode>_vis"): Likewise.
6416 ("fchksm16_vis"): Likewise.
6417 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6418 ("fmean16_vis"): Likewise.
6419 ("fp<plusminus_insn>64_vis"): Likewise.
6420 ("<plusminus_insn>v8qi3"): Likewise.
6421 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6422 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6423 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6424 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6425 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6426 ("*movqi_insn"): Likewise.
6427 ("*movhi_insn"): Likewise.
6428 ("*movsi_insn"): Likewise.
6429 ("movsi_pic_gotdata_op"): Likewise.
6430 ("*movdi_insn_sp32"): Likewise.
6431 ("*movdi_insn_sp64"): Likewise.
6432 ("movdi_pic_gotdata_op"): Likewise.
6433 ("*movsf_insn"): Likewise.
6434 ("*movdf_insn_sp32"): Likewise.
6435 ("*movdf_insn_sp64"): Likewise.
6436 ("*zero_extendhisi2_insn"): Likewise.
6437 ("*zero_extendqihi2_insn"): Likewise.
6438 ("*zero_extendqisi2_insn"): Likewise.
6439 ("*zero_extendqidi2_insn"): Likewise.
6440 ("*zero_extendhidi2_insn"): Likewise.
6441 ("*zero_extendsidi2_insn_sp64"): Likewise.
6442 ("ldfsr"): Likewise.
6443 ("prefetch_64"): Likewise.
6444 ("prefetch_32"): Likewise.
6445 ("tie_ld32"): Likewise.
6446 ("tie_ld64"): Likewise.
6447 ("*tldo_ldub_sp32"): Likewise.
6448 ("*tldo_ldub1_sp32"): Likewise.
6449 ("*tldo_ldub2_sp32"): Likewise.
6450 ("*tldo_ldub_sp64"): Likewise.
6451 ("*tldo_ldub1_sp64"): Likewise.
6452 ("*tldo_ldub2_sp64"): Likewise.
6453 ("*tldo_ldub3_sp64"): Likewise.
6454 ("*tldo_lduh_sp32"): Likewise.
6455 ("*tldo_lduh1_sp32"): Likewise.
6456 ("*tldo_lduh_sp64"): Likewise.
6457 ("*tldo_lduh1_sp64"): Likewise.
6458 ("*tldo_lduh2_sp64"): Likewise.
6459 ("*tldo_lduw_sp32"): Likewise.
6460 ("*tldo_lduw_sp64"): Likewise.
6461 ("*tldo_lduw1_sp64"): Likewise.
6462 ("*tldo_ldx_sp64"): Likewise.
6463 ("*mov<VM32:mode>_insn"): Likewise.
6464 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6465 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6467 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6469 * config/sparc/sparc.md ("type"): New insn type viscmp.
6470 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6472 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6473 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6474 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6475 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6477 ("n7_vis_logical_11cycle"): Likewise.
6478 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6479 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6480 * config/sparc/niagara.md ("niag_vis"): Likewise.
6481 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6482 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6484 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6486 * config/sparc/sparc.md: New instruction type `bmask'.
6487 (bmaskdi_vis): Use the `bmask' type.
6488 (bmasksi_vis): Likewise.
6489 * config/sparc/ultra3.md (us3_array): Likewise.
6490 * config/sparc/niagara7.md (n7_array): Likewise.
6491 * config/sparc/niagara4.md (n4_array): Likewise.
6492 * config/sparc/niagara2.md (niag2_vis): Likewise.
6493 (niag3_vis): Likewise.
6494 * config/sparc/niagara.md (niag_vis): Likewise.
6496 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6498 * ipa-comdats.c: Remove optimize check from gate.
6499 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6500 for functions not optimized.
6501 (ipa_fn_summary_read): Skip optimize check.
6502 (ipa_fn_summary_write): Likewise.
6503 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6505 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6507 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6508 (check_callers): Give up on caller which is not optimized.
6509 (inline_small_functions): Likewise.
6510 (ipa_inline): Do not give up when not optimizing.
6511 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6512 away unoptimizes cdtors.
6513 (whole_program_function_and_variable_visibility): Do
6514 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6515 * ipa.c (process_references): Do not check optimize.
6516 (symbol_table::remove_unreachable_nodes): Update optimize check.
6517 (set_writeonly_bit): Update optimize check.
6518 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6519 (pass_ipa_single_use::gate): Remove.
6521 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6523 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6524 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6525 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6526 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6527 permute_load, permute_store, adjust_extract, adjust_splat,
6528 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6529 replace_swap_with_copy, dump_swap_insn_table,
6530 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6531 recombine_lvx_pattern, recombine_stvx_pattern,
6532 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6533 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6534 to file rs6000-p8swap.c.
6535 * config/rs6000/rs6000-p8swap.c: New file.
6536 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6537 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6538 and rs6000*-*-* targets.
6540 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6542 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6544 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6546 * lto-wrapper.c (merge_and_complain): Do not merge
6547 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6548 fsigned_zeros, ftrapping_math, fwrapv.
6549 (append_compiler_options): Do not track these options.
6550 (append_linker_options): Likewie
6552 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6554 * cgraphunit.c (cgraph_node::finalize_function): When
6555 !flag_toplevel_reorde set no_reorder flag.
6556 (varpool_node::finalize_decl): Likewise.
6557 (symbol_table::compile): Drop no toplevel reorder path.
6559 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6561 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6562 edges; zero probability is not better than uninitialized.
6564 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6566 * asan.h (asan_sanitize_allocas_p): Declare.
6567 * asan.c (asan_sanitize_allocas_p): New function.
6568 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6569 (handle_builtin_alloca): Likewise.
6570 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6571 if !asan_sanitize_allocas_p.
6572 * params.def (asan-instrument-allocas): Add new option.
6573 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6574 * opts.c (common_handle_option): Disable allocas sanitization for
6577 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6579 * asan.c: Include gimple-fold.h.
6580 (get_last_alloca_addr): New function.
6581 (handle_builtin_stackrestore): Likewise.
6582 (handle_builtin_alloca): Likewise.
6583 (asan_emit_allocas_unpoison): Likewise.
6584 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6585 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6586 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6587 (instrument_builtin_call): Pass gimple iterator to
6588 get_mem_refs_of_builtin_call.
6589 (last_alloca_addr): New global.
6590 * asan.h (asan_emit_allocas_unpoison): Declare.
6591 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6592 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6593 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6594 if function calls alloca.
6595 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6596 * gimple-fold.h (replace_call_with_value): Declare.
6597 * internal-fn.c: Include asan.h.
6598 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6599 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6601 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6603 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6604 (C_SELFTEST_FLAGS): New.
6605 (CPP_SELFTEST_FLAGS): New.
6606 (SELFTEST_DEPS): New, from deps of s-selftest.
6607 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6608 (CPP_SELFTEST_DEPS): New.
6609 (selftest): Add dependency on s-selftest-c++.
6610 (s-selftest): Rename to...
6611 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6612 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6613 than SELFTEST_FLAGS.
6614 (selftest-gdb): Rename to...
6615 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6617 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6618 (selftest-valgrind): Rename to...
6619 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6621 (selftest-valgrind): Reintroduce as an alias for
6622 selftest-c-valgrind.
6623 (s-selftest-c++): New.
6624 (selftest-c++-gdb): New.
6625 (selftest-c++-valgrind): New.
6627 2017-07-06 Olivier Hainque <hainque@adacore.com>
6629 * gcc.c (process_command): When deciding if undefined variables
6630 should be ignored when processing specs, accept "gcc -v" as well.
6632 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6634 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6635 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6637 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6639 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6640 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6641 * config/arm/arm-tables.opt: Regenerate.
6642 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6644 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6646 2017-07-06 Carl Love <cel@us.ibm.com>
6648 * ChangeLog: Clean up from mid air collision
6650 2017-07-06 Carl Love <cel@us.ibm.com>
6652 * config/rs6000/rs6000-c.c: Add support for built-in functions
6653 vector signed int vec_subc (vector signed int, vector signed int);
6654 vector signed __int128 vec_subc (vector signed __int128,
6655 vector signed __int128);
6656 vector unsigned __int128 vec_subc (vector unsigned __int128,
6657 vector unsigned __int128);
6658 vector signed int vec_sube (vector signed int, vector signed int,
6660 vector unsigned int vec_sube (vector unsigned int,
6661 vector unsigned int,
6662 vector unsigned int);
6663 vector signed __int128 vec_sube (vector signed __int128,
6664 vector signed __int128,
6665 vector signed__int128);
6666 vector unsigned __int128 vec_sube (vector unsigned __int128,
6667 vector unsigned __int128,
6668 vector unsigned __int128);
6669 vector signed int vec_subec (vector signed int, vector signed int,
6671 vector unsigned int vec_subec (vector unsigned int,
6672 vector unsigned int,
6673 vector unsigned int);
6674 vector signed __int128 vec_subec (vector signed __int128,
6675 vector signed __int128,
6676 vector signed__int128);
6677 vector unsigned __int128 vec_subec (vector unsigned __int128,
6678 vector unsigned __int128,
6679 vector unsigned __int128);
6680 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6681 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6682 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6683 BU_ALTIVEC_OVERLOAD_X definitions.
6684 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6685 * doc/extend.texi: Update the built-in documentation file for the new
6688 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6691 * diagnostic-show-locus.c (layout::layout): Use start and finish
6692 spelling location for the start and finish of each range.
6693 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6694 Add unused aspect param.
6695 * input.c (expand_location_1): Add "aspect" param, and use it
6696 to access the correct part of the location.
6697 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6699 (expand_location_to_spelling_point): Likewise.
6700 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6701 param, and pass it to expand_location_1.
6703 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6705 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6706 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6707 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6708 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6709 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6710 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6711 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6712 _mm_maskz_getmant_ss): New intrinsics.
6713 (__builtin_ia32_getexpss128_mask): Changed to ...
6714 __builtin_ia32_getexpss128_round ... this.
6715 (__builtin_ia32_getexpsd128_mask): Changed to ...
6716 __builtin_ia32_getexpsd128_round ... this.
6717 * config/i386/i386-builtin-types.def
6718 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6719 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6720 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6721 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6722 __builtin_ia32_getmantss_mask_round): New builtins.
6723 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6724 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6725 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6726 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6727 * config/i386/sse.md
6728 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6729 avx512f_sgetexp<mode><mask_scalar_name>
6730 <round_saeonly_scalar_name> ... this.
6731 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6732 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6733 vgetexp<ssescalarmodesuffix>
6734 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6735 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6736 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6737 avx512f_vgetmant<mode><mask_scalar_name>
6738 <round_saeonly_scalar_name> ... this.
6739 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6740 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6741 vgetmant<ssescalarmodesuffix>
6742 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6743 %0<mask_scalar_operand4>, %1, %2
6744 <round_saeonly_scalar_mask_op4>, %3} ... this.
6745 * config/i386/subst.md (mask_scalar_operand4,
6746 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6747 round_saeonly_scalar_nimm_predicate): New subst attributes.
6749 2017-07-06 Julia Koval <julia.koval@intel.com>
6751 * config/i386/i386.c (ix86_erase_embedded_rounding):
6752 Remove code for old rounding pattern.
6754 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6756 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6758 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6760 * doc/sourcebuild.texi (Test Directives, Variants of
6761 dg-require-support): Add documentation for dg-require-stack-check.
6763 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6765 * config/i386/subst.md (mask_scalar, round_scalar,
6766 round_saeonly_scalar): New meta-templates.
6767 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6768 round_scalar_mask_operand3, round_scalar_mask_op3,
6769 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6770 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6771 round_saeonly_scalar_constraint,
6772 round_saeonly_scalar_prefix): New subst attribute.
6773 * config/i386/sse.md
6774 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6775 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6776 <round_scalar_name> ... this.
6777 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6778 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6779 <round_scalar_name> ... this.
6780 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6781 <sse>_vm<code><mode>3<mask_scalar_name>
6782 <round_saeonly_scalar_name> ... this.
6783 (v<plusminus_mnemonic><ssescalarmodesuffix>
6784 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6785 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6786 v<plusminus_mnemonic><ssescalarmodesuffix>
6787 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6788 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6789 (v<multdiv_mnemonic><ssescalarmodesuffix>
6790 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6791 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6792 v<multdiv_mnemonic><ssescalarmodesuffix>
6793 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6794 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6795 (v<maxmin_float><ssescalarmodesuffix>
6796 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6797 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6798 v<maxmin_float><ssescalarmodesuffix>
6799 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6800 %0<mask_scalar_operand3>, %1, %<iptr>2
6801 <round_saeonly_scalar_mask_op3>} ... this.
6803 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6805 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6806 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6808 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6809 Alan Hayward <alan.hayward@arm.com>
6810 David Sherwood <david.sherwood@arm.com>
6812 * combine.c (simplify_if_then_else): Remove "enum" before
6814 * compare-elim.c (can_eliminate_compare): Likewise.
6815 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6817 (aarch64_lookup_simd_builtin_type): Likewise.
6818 (aarch64_simd_builtin_type): Likewise.
6819 (aarch64_init_simd_builtin_types): Likewise.
6820 (aarch64_simd_expand_args): Likewise.
6821 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6823 (aarch64_reverse_mask): Likewise.
6824 (aarch64_simd_emit_reg_reg_move): Likewise.
6825 (aarch64_gen_adjusted_ldpstp): Likewise.
6826 (aarch64_ccmp_mode_to_code): Likewise.
6827 (aarch64_operands_ok_for_ldpstp): Likewise.
6828 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6829 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6831 (aarch64_min_divisions_for_recip_mul): Likewise.
6832 (aarch64_reassociation_width): Likewise.
6833 (aarch64_get_condition_code_1): Likewise.
6834 (aarch64_simd_emit_reg_reg_move): Likewise.
6835 (aarch64_simd_attr_length_rglist): Likewise.
6836 (aarch64_reverse_mask): Likewise.
6837 (aarch64_operands_ok_for_ldpstp): Likewise.
6838 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6839 (aarch64_gen_adjusted_ldpstp): Likewise.
6840 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6842 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6843 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6844 (arm_lookup_simd_builtin_type): Likewise.
6845 (arm_simd_builtin_type): Likewise.
6846 (arm_init_simd_builtin_types): Likewise.
6847 (arm_expand_builtin_args): Likewise.
6848 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6849 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6850 (ft32_setup_incoming_varargs): Likewise.
6851 (ft32_function_arg): Likewise.
6852 (ft32_function_arg_advance): Likewise.
6853 (ft32_pass_by_reference): Likewise.
6854 (ft32_arg_partial_bytes): Likewise.
6855 (ft32_valid_pointer_mode): Likewise.
6856 (ft32_addr_space_pointer_mode): Likewise.
6857 (ft32_addr_space_legitimate_address_p): Likewise.
6858 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6860 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6861 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6862 (iamcu_alignment): Likewise.
6863 (canonicalize_vector_int_perm): Likewise.
6864 (ix86_noce_conversion_profitable_p): Likewise.
6865 (ix86_mpx_bound_mode): Likewise.
6866 (ix86_operands_ok_for_move_multiple): Likewise.
6867 * config/microblaze/microblaze-protos.h
6868 (microblaze_expand_conditional_branch_reg): Likewise.
6869 * config/microblaze/microblaze.c
6870 (microblaze_expand_conditional_branch_reg): Likewise.
6871 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6873 (rs6000_reassociation_width): Likewise.
6874 (rs6000_invalid_binary_op): Likewise.
6875 (fusion_p9_p): Likewise.
6876 (emit_fusion_p9_load): Likewise.
6877 (emit_fusion_p9_store): Likewise.
6878 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6880 (riscv_hard_regno_mode_ok_p): Likewise.
6881 (riscv_address_insns): Likewise.
6882 (riscv_split_symbol): Likewise.
6883 (riscv_legitimize_move): Likewise.
6884 (riscv_function_value): Likewise.
6885 (riscv_hard_regno_nregs): Likewise.
6886 (riscv_expand_builtin): Likewise.
6887 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6888 (riscv_build_integer): Likewise.
6889 (riscv_split_integer): Likewise.
6890 (riscv_legitimate_constant_p): Likewise.
6891 (riscv_cannot_force_const_mem): Likewise.
6892 (riscv_regno_mode_ok_for_base_p): Likewise.
6893 (riscv_valid_base_register_p): Likewise.
6894 (riscv_valid_offset_p): Likewise.
6895 (riscv_valid_lo_sum_p): Likewise.
6896 (riscv_classify_address): Likewise.
6897 (riscv_legitimate_address_p): Likewise.
6898 (riscv_address_insns): Likewise.
6899 (riscv_load_store_insns): Likewise.
6900 (riscv_force_binary): Likewise.
6901 (riscv_split_symbol): Likewise.
6902 (riscv_force_address): Likewise.
6903 (riscv_legitimize_address): Likewise.
6904 (riscv_move_integer): Likewise.
6905 (riscv_legitimize_const_move): Likewise.
6906 (riscv_legitimize_move): Likewise.
6907 (riscv_address_cost): Likewise.
6908 (riscv_subword): Likewise.
6909 (riscv_output_move): Likewise.
6910 (riscv_canonicalize_int_order_test): Likewise.
6911 (riscv_emit_int_order_test): Likewise.
6912 (riscv_function_arg_boundary): Likewise.
6913 (riscv_pass_mode_in_fpr_p): Likewise.
6914 (riscv_pass_fpr_single): Likewise.
6915 (riscv_pass_fpr_pair): Likewise.
6916 (riscv_get_arg_info): Likewise.
6917 (riscv_function_arg): Likewise.
6918 (riscv_function_arg_advance): Likewise.
6919 (riscv_arg_partial_bytes): Likewise.
6920 (riscv_function_value): Likewise.
6921 (riscv_pass_by_reference): Likewise.
6922 (riscv_setup_incoming_varargs): Likewise.
6923 (riscv_print_operand): Likewise.
6924 (riscv_elf_select_rtx_section): Likewise.
6925 (riscv_save_restore_reg): Likewise.
6926 (riscv_for_each_saved_reg): Likewise.
6927 (riscv_register_move_cost): Likewise.
6928 (riscv_hard_regno_mode_ok_p): Likewise.
6929 (riscv_hard_regno_nregs): Likewise.
6930 (riscv_class_max_nregs): Likewise.
6931 (riscv_memory_move_cost): Likewise.
6932 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6933 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6934 (rl78_addr_space_address_mode): Likewise.
6935 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6937 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6938 (rs6000_reassociation_width): Likewise.
6939 (rs6000_invalid_binary_op): Likewise.
6940 (fusion_p9_p): Likewise.
6941 (emit_fusion_p9_load): Likewise.
6942 (emit_fusion_p9_store): Likewise.
6943 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6944 (ok_for_simple_move_operands): Likewise.
6945 (ok_for_simple_move_strict_operands): Likewise.
6946 (ok_for_simple_arith_logic_operands): Likewise.
6947 (visium_legitimize_reload_address): Likewise.
6948 (visium_select_cc_mode): Likewise.
6949 (output_cbranch): Likewise.
6950 (visium_split_double_move): Likewise.
6951 (visium_expand_copysign): Likewise.
6952 (visium_expand_int_cstore): Likewise.
6953 (visium_expand_fp_cstore): Likewise.
6954 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6955 (visium_function_arg): Likewise.
6956 (visium_function_arg_advance): Likewise.
6957 (visium_libcall_value): Likewise.
6958 (visium_setup_incoming_varargs): Likewise.
6959 (visium_legitimate_constant_p): Likewise.
6960 (visium_legitimate_address_p): Likewise.
6961 (visium_legitimize_address): Likewise.
6962 (visium_secondary_reload): Likewise.
6963 (visium_register_move_cost): Likewise.
6964 (visium_memory_move_cost): Likewise.
6965 (prepare_move_operands): Likewise.
6966 (ok_for_simple_move_operands): Likewise.
6967 (ok_for_simple_move_strict_operands): Likewise.
6968 (ok_for_simple_arith_logic_operands): Likewise.
6969 (visium_function_value_1): Likewise.
6970 (rtx_ok_for_offset_p): Likewise.
6971 (visium_legitimize_reload_address): Likewise.
6972 (visium_split_double_move): Likewise.
6973 (visium_expand_copysign): Likewise.
6974 (visium_expand_int_cstore): Likewise.
6975 (visium_expand_fp_cstore): Likewise.
6976 (visium_split_cstore): Likewise.
6977 (visium_select_cc_mode): Likewise.
6978 (visium_split_cbranch): Likewise.
6979 (output_cbranch): Likewise.
6980 (visium_print_operand_address): Likewise.
6981 * expmed.c (flip_storage_order): Likewise.
6982 * expmed.h (emit_cstore): Likewise.
6983 (flip_storage_order): Likewise.
6984 * genrecog.c (validate_pattern): Likewise.
6985 * hsa-gen.c (gen_hsa_addr): Likewise.
6986 * internal-fn.c (expand_arith_overflow): Likewise.
6987 * ira-color.c (allocno_copy_cost_saving): Likewise.
6988 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6989 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6990 (process_invariant_for_inheritance): Likewise.
6991 * lra-eliminations.c (move_plus_up): Likewise.
6992 * omp-low.c (lower_oacc_reductions): Likewise.
6993 * simplify-rtx.c (simplify_subreg): Likewise.
6994 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6995 (TARGET_CHKP_BOUND_MODE): Likewise..
6996 * targhooks.c (default_chkp_bound_mode): Likewise.
6997 (default_setup_incoming_vararg_bounds): Likewise.
6998 * targhooks.h (default_chkp_bound_mode): Likewise.
6999 (default_setup_incoming_vararg_bounds): Likewise.
7000 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7001 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7002 (have_whole_vector_shift): Likewise.
7003 * tree-vect-stmts.c (vectorizable_load): Likewise.
7004 * doc/tm.texi: Regenerate.
7006 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7008 Graceful degrade if Binutils PR21472 is not available.
7011 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7012 .rodata in flash test fails.
7013 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7014 * confgure: Regenerate.
7015 * config.in: Regenerate.
7016 * config/avr/avr.c (avr_asm_named_section)
7017 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7018 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7019 (avr_asm_init_sections): Same.
7021 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7023 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7024 (fma<VH:mode>4_intrinsic): Likewise.
7025 (*fmsub<VCVTF:mode>4): Likewise.
7026 (*fmsub<VH:mode>4_intrinsic): Likewise.
7028 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7031 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7032 Don't depend on "optimize > 0".
7033 (out_movhi_r_mr, out_movqi_mr_r): Same.
7034 (out_movhi_mr_r, out_movqi_r_mr): Same.
7035 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7036 io_address_operand on "optimize > 0".
7038 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7040 * tree-loop-distribution.c: Add general explanantion on the pass.
7041 (generate_loops_for_partition): Mark distributed loop.
7042 (pg_add_dependence_edges): New parameter. Handle alias data
7043 dependence specially and record it in the parameter if asked.
7044 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7045 (init_partition_graph_vertices, add_partition_graph_edge): New.
7046 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7047 (free_partition_graph_vdata, build_partition_graph): New.
7048 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7049 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7050 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7051 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7052 (version_for_distribution_p, finalize_partitions): New.
7053 (distribute_loop): Handle alias data dependence specially. Factor
7054 out loop fusion code as functions and call these functions.
7056 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7058 * tree-loop-distribution.c (classify_partition): New parameter and
7059 better handle reduction statement.
7060 (rdg_build_partitions): Revise comment.
7061 (distribute_loop): Compute statements in all partitions and pass it
7062 to classify_partition.
7064 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7066 * tree-loop-distribution.c (enum partition_type): New.
7067 (struct partition): New field type.
7068 (partition_merge_into): Add parameter. Update partition type.
7069 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7070 (build_rdg_partition_for_vertex): Compute partition type.
7071 (rdg_build_partitions): Dump partition type.
7072 (distribute_loop): Update calls to partition_merge_into.
7074 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7076 * tree-loop-distribution.c (struct ddr_hasher): New.
7077 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7079 (classify_partition): Call get_data_dependence.
7080 (pg_add_dependence_edges): Ditto.
7081 (distribute_loop): Release data dependence hash table.
7083 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7085 * tree-loop-distribution.c (ref_base_address): Delete.
7086 (similar_memory_accesses): Rename ...
7087 (share_memory_accesses): ... to this. Check if partitions access
7088 the same memory reference.
7089 (distribute_loop): Call share_memory_accesses.
7091 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7093 * tree-loop-distribution.c (struct partition): New field recording
7095 (partition_alloc, partition_free): Init and release data refs.
7096 (partition_merge_into): Merge data refs.
7097 (build_rdg_partition_for_vertex): Collect data refs for partition.
7098 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7100 (distribute_loop): Remve data refs from vertice data of partition
7103 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7105 * tree-loop-distribution.c (params.h): Include header file.
7106 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7107 (datarefs_vec): New global var.
7108 (create_rdg_vertices): Use datarefs_vec directly.
7109 (free_rdg): Don't free data references.
7110 (build_rdg): Update use. Don't free data references.
7111 (distribute_loop): Compute global variable for data references.
7112 Bail out if there are too many data references.
7114 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7116 * tree-loop-distribution.c (loop_nest): New global var.
7117 (build_rdg): Use loop directly, rather than loop nest.
7118 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7120 (distribute_loop): Compute global variable loop nest. Update use.
7122 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7124 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7125 (partition_merge_into): New parameter. Dump reason for fusion.
7126 (distribute_loop): Update use of partition_merge_into.
7128 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7130 * tree-loop-distribution.c (bb_top_order_index): New.
7131 (bb_top_order_index_size, bb_top_order_cmp): New.
7132 (stmts_from_loop): Use topological order.
7133 (pass_loop_distribution::execute): Compute and release topological
7134 order for basic blocks.
7136 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7138 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7141 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7143 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7144 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7145 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7146 * internal-fn.def (LOOP_DIST_ALIAS): New.
7147 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7148 (fold_loop_internal_call): ... this.
7149 (vect_loop_dist_alias_call): New function.
7150 (set_uid_loop_bbs): Call fold_loop_internal_call.
7151 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7154 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7157 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7158 Require dead FLAGS_REG at the beginning of a peephole.
7160 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7163 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7164 arguments in the call to __builtin_ia32_sbb_u32.
7165 (_subborrow_u64): Swap _X and _Y arguments in the call to
7166 __builtin_ia32_sbb_u64.
7168 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7171 * tree-vrp.c (compare_assert_loc): Turn into a function template
7172 with stable template parameter. Only test if a->e is NULL,
7173 !a->e == !b->e has been verified already. Use e == NULL or
7174 e != NULL instead of e or ! e tests. If stable is true, don't use
7175 iterative_hash_expr, on the other side allow a or b or both NULL
7176 and sort the NULLs last.
7177 (process_assert_insertions): Sort using compare_assert_loc<false>
7178 instead of compare_assert_loc, later sort using
7179 compare_assert_loc<true> before calling process_assert_insertions_for
7180 in a loop. Use break instead of continue once seen NULL pointer.
7182 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7184 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7185 Cortex-R7 and Cortex-R8 processors.
7187 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7189 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7190 uninitialized while src is not.
7192 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7194 * common/config/arm/arm-common.c: Adjust include path for
7196 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7197 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7198 (arm-cpu-data.h): Likewise.
7199 (arm-cpu-cdata.h): Likewise.
7200 * config/arm/arm-cpu.h: Delete.
7201 * config/arm/arm-cpu-cdata.h: Delete.
7202 * config/arm/arm-cpu-data.h: Delete.
7204 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7206 * config/arm/arm-cpus.in (cortex-a55): New.
7207 (cortex-a75): Likewise.
7208 (cortex-a75.cortex-a55): Likewise.
7209 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7211 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7212 * config/arm/arm-cpu-cdata.h: Regenerate.
7213 * config/arm/arm-cpu-data.h: Regenerate.
7214 * config/arm/arm-cpu.h: Regenerate.
7215 * config/arm/arm-tables.opt: Regenerate.
7216 * config/arm/arm-tune.md: Regenerate.
7218 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7220 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7222 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7224 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7227 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7229 PR tree-optimization/81292
7230 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7231 full_string_p, also call adjust_related_strinfos if the adjustment
7232 is simple, otherwise invalidate related strinfos.
7234 2017-07-04 Martin Liska <mliska@suse.cz>
7237 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7238 newly created variable as DECL_IGNORED_P.
7240 2017-07-04 Martin Liska <mliska@suse.cz>
7243 * ipa-inline.c (inline_small_functions):
7244 Use xstrdup_for_dump.
7246 2017-07-04 Tom de Vries <tom@codesourcery.com>
7248 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7250 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7253 * config/darwin.c (darwin_function_switched_text_sections):
7256 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7258 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7260 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7262 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7266 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7267 min_profitable_iters, and th as inclusive lower bounds.
7268 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7269 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7270 for min_profitable_iters and min_profitable_estimate.
7271 (vect_transform_loop): Treat th as an inclusive lower bound.
7272 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7274 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7277 * config/darwin.c (darwin_function_switched_text_sections):
7278 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7279 in two pieces, and suppress the use of buf.
7281 2017-07-03 Nathan Sidwell <nathan@acm.org>
7283 * hash-table.h (hash_table_mod1): Fix indentation.
7285 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7288 * predict.c (force_edge_cold): Be more careful about propagation
7290 * profile-count.h (profile_probability::guessed,
7291 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7293 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7295 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7297 * doc/invoke.texi (rcpc architecture extension): Document it.
7299 2017-07-03 Richard Biener <rguenther@suse.de>
7301 PR tree-optimization/60510
7302 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7303 the scalar reduction PHI and use it.
7304 (vectorizable_reduction): Properly guard the single_defuse_cycle
7305 path for non-SLP reduction chains where we cannot use it.
7306 Rework reduc_def/index and vector type deduction. Rework
7307 vector operand gathering during reduction op code-gen.
7308 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7309 chains dissolve the chain and leave it to non-SLP reduction
7312 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7314 * tree-data-ref.h (dr_alignment): Declare.
7315 * tree-data-ref.c (dr_alignment): New function.
7316 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7317 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7319 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7321 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7323 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7324 and base_misalignment fields.
7325 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7326 * tree-data-ref.c: Include builtins.h.
7327 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7328 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7329 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7330 * tree-vect-data-refs.c: Include tree-cfg.h.
7331 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7332 fields instead of calculating an alignment here.
7333 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7334 innermost_loop_behavior fields.
7336 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7338 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7340 (DR_STEP_ALIGNMENT): New macro.
7341 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7342 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7343 (create_data_ref): Print it.
7344 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7345 to tell whether the step preserves vector (mis)alignment.
7346 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7347 Move the check for an integer step and generalise to all INTEGER_CST.
7348 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7349 Print the outer step alignment.
7351 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7353 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7354 with offset_alignment.
7355 (DR_ALIGNED_TO): Delete.
7356 (DR_OFFSET_ALIGNMENT): New macro.
7357 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7358 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7359 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7360 (create_data_ref): Likewise.
7361 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7362 (vect_analyze_data_refs): Likewise.
7363 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7364 creating dummy innermost behavior.
7366 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7368 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7369 with a "innermost_loop_behavior *" and refeence tree.
7370 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7371 (create_data_ref): Update call accordingly.
7372 * tree-predcom.c (find_looparound_phi): Likewise.
7374 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7376 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7377 fields with dr_wrt_vec_loop.
7378 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7379 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7380 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7381 (vect_dr_behavior): New function.
7382 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7383 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7384 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7385 track whether the step preserves the misalignment.
7386 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7387 Use vect_dr_behavior.
7388 (vect_setup_realignment): Update call accordingly.
7389 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7390 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7391 call to vect_create_addr_base_for_vector_ref.
7392 (vect_create_cond_for_align_checks): Likewise.
7393 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7394 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7395 (vect_recog_mask_conversion_pattern): Likewise.
7396 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7397 (new_stmt_vec_info): Remove redundant zeroing.
7399 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7401 * common/config/arm/arm-common.c (arm_be8_option): New function.
7402 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7403 (ISA_ARMv6): Add isa_bit_be8.
7404 * config/arm/arm.h (arm_be8_option): Add prototype.
7405 (BE8_SPEC_FUNCTION): New define.
7406 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7407 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7408 (mlittle-endian): Similarly.
7409 (mbe8, mbe32): New options.
7410 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7411 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7413 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7415 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7417 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7419 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7420 (cleanup_tree_cfg_bb): Use it.
7421 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7423 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7425 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7428 * loop-doloop.c (add_test): Update profile.
7430 2017-07-03 Martin Liska <mliska@suse.cz>
7433 * sanopt.c (rewrite_usage_of_param): New function.
7434 (sanitize_rewrite_addressable_params): Likewise.
7435 (pass_sanopt::execute): Call rewrite_usage_of_param.
7437 2017-07-03 Richard Biener <rguenther@suse.de>
7439 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7440 back to using VIEW_CONVERT_EXPR.
7442 2017-07-03 Martin Liska <mliska@suse.cz>
7445 * doc/extend.texi: Document when a resolver function is
7446 generated for target_clones.
7448 2017-07-03 Martin Liska <mliska@suse.cz>
7450 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7451 and shadow memory of auto variables which are subject of
7452 use-after-scope sanitization.
7453 (asan_expand_mark_ifn): Add do set only when is_poison.
7455 2016-07-03 Richard Biener <rguenther@suse.de>
7457 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7459 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7460 (vectorizable_reduction): Perform reduction PHI creation when
7461 visiting a reduction PHI and adjust and simplify code generation
7462 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7463 (vect_transform_loop): Visit reduction PHIs.
7464 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7465 defs into the SLP tree.
7466 (vect_build_slp_tree): Reduction defs terminate the recursion.
7467 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7469 (vect_get_vec_defs_for_stmt_copy): Export.
7470 (vect_get_vec_defs): Likewise.
7471 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7473 (vect_get_vec_defs_for_stmt_copy): Declare.
7474 (vect_get_vec_defs): Likewise.
7476 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7478 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7479 parameter with a "loop" parameter and use it instead of the
7480 loop containing DR_STMT. Don't check simple_iv when doing
7481 BB analysis. Describe the two analysis modes in the comment.
7483 2017-07-03 Tom de Vries <tom@codesourcery.com>
7485 PR tree-optimization/69468
7486 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7487 (find_same_succ_bb): Handle ignore_edge_flags.
7489 2017-07-03 Tom de Vries <tom@codesourcery.com>
7491 PR tree-optimization/81192
7492 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7494 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7496 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7498 2017-07-03 Tom de Vries <tom@codesourcery.com>
7500 PR tree-optimization/81192
7501 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7502 BB_SAME_SUCC (bb) == NULL.
7504 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7506 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7509 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7511 * dumpfile.c: Include profile-count.h
7512 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7514 (insert_cond_bb): Update profile.
7515 * tree-cfg.h (insert_cond_bb): Update prototype.
7516 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7517 * tree-dump.c: Do not include tree-cfg.
7519 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7521 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7523 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7525 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7528 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7530 * tree-complex.c (expand_complex_div_wide): update profile.
7532 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7533 Alan Hayward <alan.hayward@arm.com>
7534 David Sherwood <david.sherwood@arm.com>
7536 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7537 (CORETYPES_H): New define.
7538 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7539 (insn-modes-inline.h, s-modes-inline-h): New rules.
7540 (generated_files): Add insn-modes-inline.h.
7541 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7542 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7543 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7544 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7545 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7546 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7547 (build/gencodes.o, build/genconditions.o): Likewise.
7548 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7549 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7550 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7551 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7552 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7553 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7554 * coretypes.h: Include everything up to real.h for generators.
7555 Include insn-modes.h first. Include wide-int-print.h after
7556 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7557 * machmode.h: Don't include insn-modes.h here.
7558 * function-tests.c: Remove includes of signop.h, machmode.h,
7559 double-int.h and wide-int.h.
7561 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7563 * optc-save-gen.awk: Likewise.
7564 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7565 * godump.c: Remove include of wide-int-print.h.
7566 * pretty-print.h: Likewise.
7567 * wide-int-print.cc: Likewise.
7568 * wide-int.cc: Likewise.
7569 * hash-map-tests.c: Remove include of signop.h.
7570 * hash-set-tests.c: Likewise.
7571 * rtl-tests.c: Likewise.
7572 * mkconfig.sh: Remove include of machmode.h.
7573 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7575 (emit_insn_modes_inline_h): ...this new function. Emit the code
7576 into an insn-modes-inline.h header file, adding appropriate
7577 include guards and end comments.
7578 (emit_insn_modes_c_header): Remove include of machmode.h.
7579 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7581 (main): Handle -i flag and call emit_insn_modes_inline_h when
7584 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7586 * tree-ssa-strlen.c (strinfo): Rename the length field to
7587 nonzero_chars. Add a full_string_p field.
7588 (compare_nonzero_chars, zero_length_string_p): New functions.
7589 (get_addr_stridx): Add an offset_out parameter.
7590 Use compare_nonzero_chars.
7591 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7592 (new_strinfo): Update after above changes to strinfo.
7593 (set_endptr_and_length): Set full_string_p.
7594 (get_string_length): Update after above changes to strinfo.
7595 (unshare_strinfo): Update call to new_strinfo.
7596 (maybe_invalidate): Likewise.
7597 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7598 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7599 as a uhwi instead of an shwi. Update after above changes to
7600 strinfo and new_strinfo.
7601 (zero_length_string): Assert that chainsi contains full strings.
7602 Use zero_length_string_p. Update call to new_strinfo.
7603 (adjust_related_strinfos): Update after above changes to strinfo.
7604 Copy full_string_p from origsi.
7605 (adjust_last_stmt): Use zero_length_string_p.
7606 (handle_builtin_strlen): Update after above changes to strinfo and
7607 new_strinfo. Install the lhs as the string length if the previous
7608 entry didn't describe a full string.
7609 (handle_builtin_strchr): Update after above changes to strinfo
7611 (handle_builtin_strcpy): Likewise.
7612 (handle_builtin_strcat): Likewise.
7613 (handle_builtin_malloc): Likewise.
7614 (handle_pointer_plus): Likewise.
7615 (handle_builtin_memcpy): Likewise. Track nonzero characters
7616 that aren't necessarily followed by a nul terminator.
7617 (handle_char_store): Likewise.
7619 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7621 PR tree-optimization/80769
7622 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7623 for malloc and calloc. Document the new invariant that all related
7624 strinfos have delayed lengths or none do.
7625 (verify_related_strinfos): Move earlier in file.
7626 (set_endptr_and_length): New function, split out from...
7627 (get_string_length): ...here. Also set the lengths of related
7629 (zero_length_string): Assert that chainsi has known (rather than
7631 (adjust_related_strinfos): Likewise.
7633 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7635 PR tree-optimization/81136
7636 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7637 assert that two references with the same misalignment have the same
7638 compile-time misalignment if those compile-time misalignments
7641 2017-07-01 Andi Kleen <ak@linux.intel.com>
7643 * print-tree.c (print_node): Print all attributes.
7645 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7647 * cfg.c (scale_bbs_frequencies): New function.
7648 * cfg.h (scale_bbs_frequencies): Declare it.
7649 * cfgloopanal.c (single_likely_exit): Cleanup.
7650 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7652 (scale_loop_profile): Likewise.
7653 (loop_version): Likewise.
7654 (create_empty_loop_on_edge): Update.
7655 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7656 scale_loop_frequencies, scale_loop_profile, loopify,
7657 loop_version): Update prototypes.
7658 * modulo-sched.c (sms_schedule): Update.
7659 * predict.c (unlikely_executed_edge_p): Also check probability.
7660 (probably_never_executed_edge_p): Fix typo.
7661 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7662 * tree-parloops.c (gen_parallel_loop): Update.
7663 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7664 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7665 * tree-ssa-loop-split.c (split_loop): Update.
7666 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7667 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7668 (vect_loop_versioning): Update.
7669 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7671 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7673 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7675 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7677 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7678 to keep profile consistent.
7680 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7682 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7683 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7684 * profile-count.h (max_safe_multiplier): Make unsigned.
7685 (profile_count::guessed_zero): New.
7687 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7689 * bb-reorder.c (fix_up_crossing_landing_pad,
7690 fix_crossing_conditional_branches): Use make_single_succ_edge
7691 to keep profile consistent.
7693 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7695 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7698 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7701 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7702 the right scopes, make sure cond_jump isn't preserved between multiple
7703 iterations. Search for fallthru edge whenever there are 3+ edges and
7704 use find_fallthru_edge for it.
7706 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7708 Patch by Alexander Monakov <amonakov@ispras.ru>
7709 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7710 probabilities consistently.
7712 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7714 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7715 emit_cmp_and_jump_insns.
7717 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7720 * tree-inline.c (expand_call_inline): Combine profile statuses.
7722 2017-06-30 Andrew Pinski <apinski@cavium.com>
7724 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7725 fold_stmt returned true.
7727 2017-06-30 Nathan Sidwell <nathan@acm.org>
7729 * ggc.h (empty_string): Delete.
7730 * cfgexpand.c (expand_asm_stmt): Use plain "".
7731 * optabs.c (expand_asm_memory_barrier): Likewise.
7732 * stringpool.c (empty_string): Delete.
7733 (digit_vector, digit_string): Delete.
7734 (ggc_alloc_string): Use plain "", don't optimize single digit
7735 strings. Use ggc_alloc_atomic.
7737 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7739 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7740 comparison set and one other set, use the cost of the non-comparison
7743 2017-06-30 Nathan Sidwell <nathan@acm.org>
7745 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7748 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7750 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7751 loops. Remove now unneeded calls to gimple_switch_set_label() that
7752 just set removed labels to NULL_TREE.
7754 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7756 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7757 (set_range_info): ...here. Only call set_range_info_raw if domain
7759 (set_nonzero_bits): Call set_range_info_raw.
7760 * tree-ssanames.h (set_range_info_raw): New.
7762 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7765 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7766 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7767 of nonimmediate_operand and <store_mask_constraint> instead of m
7768 for the input operand. For V8FI iterator, always split if input
7769 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7770 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7771 <store_mask_predicate> instead of register_operand and
7772 <store_mask_constraint> instead of v for the input operand. Make
7773 sure both operands aren't MEMs for if not <mask_applied>.
7775 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7777 * lto-wrapper.c (copy_file) Close both file descriptors before
7780 2017-06-30 Martin Liska <mliska@suse.cz>
7783 * multiple_target.c (create_dispatcher_calls): Make ifunc
7784 also for function that don't have calls or are not referenced.
7786 2017-06-30 Richard Biener <rguenther@suse.de>
7788 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7789 analyze the first scalar stmt. Move vector type computation
7790 for the BB case here from ...
7791 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7792 live operation processing in the SLP case properly.
7794 2017-06-30 Richard Biener <rguenther@suse.de>
7796 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7798 2017-06-30 Martin Liska <mliska@suse.cz>
7801 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7802 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7804 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7806 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7809 2017-06-30 Richard Biener <rguenther@suse.de>
7811 PR tree-optimization/81249
7812 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7813 condition reduction result to original scalar type.
7815 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7817 * profile-count.h (enum profile_quality): Fix typos and whitespace
7820 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7822 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7823 type for branch probabilities.
7825 2017-06-29 Julian Brown <julian@codesourcery.com>
7826 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7828 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7829 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7830 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7831 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7833 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7835 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7836 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7837 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7838 CC usage from generic code to here.
7839 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7840 CC usage into the target macros.
7842 2017-06-29 Maya Rashish <coypu@sdf.org>
7844 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7847 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7849 * arm/arm-builtins.c: Include profile-count.h
7850 * except.c (sjlj_emit_function_enter): Use
7851 profile_probability::unlikely.
7853 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7855 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7856 and tocrel_offset be pointer args rather than implicitly using
7858 (legitimate_constant_pool_address_p, rs6000_emit_move,
7859 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7860 tocrel_offset and use in toc_relative_expr_p call.
7861 (print_operand, print_operand_address): Use static tocrel_base_oac
7862 and tocrel_offset_oac.
7863 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7866 2017-06-29 Maya Rashish <coypu@sdf.org>
7868 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7870 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7872 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7873 objects, take into account only the alignment of 'op0' and 'mode1' if
7876 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7878 * ccmp.c (ccmp_tree_comparison_p): New function.
7879 (ccmp_candidate_p): Update to use above function.
7880 (get_compare_parts): New function.
7881 (expand_ccmp_next): Update to use new functions.
7882 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7884 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7885 take mode as argument.
7886 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7887 * expr.c (expand_expr_real_1): Pass mode as argument.
7889 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7891 * combine.c (combine_instructions): Print insns to dump_file, together
7894 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7896 * asan.c (asan_emit_stack_protection): Update.
7897 (create_cond_insert_point): Update.
7898 * auto-profile.c (afdo_propagate_circuit): Update.
7899 * basic-block.h (struct edge_def): Turn probability to
7900 profile_probability.
7901 (EDGE_FREQUENCY): Update.
7902 * bb-reorder.c (find_traces_1_round): Update.
7903 (better_edge_p): Update.
7904 (sanitize_hot_paths): Update.
7905 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7906 (make_single_succ_edge): Update.
7907 (check_bb_profile): Update.
7908 (dump_edge_info): Update.
7909 (update_bb_profile_for_threading): Update.
7910 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7911 probabilitycount to 0.
7912 * cfgbuild.c (compute_outgoing_frequencies): Update.
7913 * cfgcleanup.c (try_forward_edges): Update.
7914 (outgoing_edges_match): Update.
7915 (try_crossjump_to_edge): Update.
7916 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7917 (expand_gimple_tailcall): Update.
7918 (construct_init_block): Use make_single_succ_edge.
7919 (construct_exit_block): Use make_single_succ_edge.
7920 * cfghooks.c (verify_flow_info): Update.
7921 (redirect_edge_succ_nodup): Update.
7922 (split_edge): Update.
7923 (account_profile_record): Update.
7924 * cfgloopanal.c (single_likely_exit): Update.
7925 * cfgloopmanip.c (scale_loop_profile): Update.
7926 (set_zero_probability): Remove.
7927 (duplicate_loop_to_header_edge): Update.
7928 * cfgloopmanip.h (loop_version): Update prototype.
7929 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7930 (force_nonfallthru_and_redirect): Update.
7931 (update_br_prob_note): Update.
7932 (rtl_verify_edges): Update.
7933 (purge_dead_edges): Update.
7934 (rtl_lv_add_condition_to_bb): Update.
7935 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7936 * cgraphunit.c (init_lowered_empty_function): Update.
7937 (cgraph_node::expand_thunk): Update.
7938 * cilk-common.c: Include profile-count.h
7939 * dojump.c (inv): Remove.
7940 (jumpifnot): Update.
7941 (jumpifnot_1): Update.
7942 (do_jump_1): Update.
7944 (do_jump_by_parts_greater_rtx): Update.
7945 (do_compare_rtx_and_jump): Update.
7946 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7947 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7948 * dwarf2cfi.c: Include profile-count.h
7949 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7950 (sjlj_emit_dispatch_table): Likewise.
7951 * explow.c: Include profile-count.h
7952 * expmed.c (emit_store_flag_force): Update.
7953 (do_cmp_and_jump): Update.
7954 * expr.c (compare_by_pieces_d::generate): Update.
7955 (compare_by_pieces_d::finish_mode): Update.
7956 (emit_block_move_via_loop): Update.
7957 (store_expr_with_bounds): Update.
7958 (store_constructor): Update.
7959 (expand_expr_real_2): Update.
7960 (expand_expr_real_1): Update.
7961 * expr.h (try_casesi, try_tablejump): Update prototypes.
7962 * gimple-pretty-print.c (dump_probability): Update.
7963 (dump_profile): New.
7964 (dump_gimple_label): Update.
7965 (dump_gimple_bb_header): Update.
7966 * graph.c (draw_cfg_node_succ_edges): Update.
7967 * hsa-gen.c (convert_switch_statements): Update.
7968 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7969 (find_if_case_1): Update.
7970 (find_if_case_2): Update.
7971 * internal-fn.c (expand_arith_overflow_result_store): Update.
7972 (expand_addsub_overflow): Update.
7973 (expand_neg_overflow): Update.
7974 (expand_mul_overflow): Update.
7975 (expand_vector_ubsan_overflow): Update.
7976 * ipa-cp.c (good_cloning_opportunity_p): Update.
7977 * ipa-split.c (split_function): Use make_single_succ_edge.
7978 * ipa-utils.c (ipa_merge_profiles): Update.
7979 * loop-doloop.c (add_test): Update.
7980 (doloop_modify): Update.
7981 * loop-unroll.c (compare_and_jump_seq): Update.
7982 (unroll_loop_runtime_iterations): Update.
7983 * lra-constraints.c (lra_inheritance): Update.
7984 * lto-streamer-in.c (input_cfg): Update.
7985 * lto-streamer-out.c (output_cfg): Update.
7986 * mcf.c (adjust_cfg_counts): Update.
7987 * modulo-sched.c (sms_schedule): Update.
7988 * omp-expand.c (expand_omp_for_init_counts): Update.
7989 (extract_omp_for_update_vars): Update.
7990 (expand_omp_ordered_sink): Update.
7991 (expand_omp_for_ordered_loops): Update.
7992 (expand_omp_for_generic): Update.
7993 (expand_omp_for_static_nochunk): Update.
7994 (expand_omp_for_static_chunk): Update.
7995 (expand_cilk_for): Update.
7996 (expand_omp_simd): Update.
7997 (expand_omp_taskloop_for_outer): Update.
7998 (expand_omp_taskloop_for_inner): Update.
7999 * omp-simd-clone.c (simd_clone_adjust): Update.
8000 * optabs.c (expand_doubleword_shift): Update.
8001 (expand_abs): Update.
8002 (emit_cmp_and_jump_insn_1): Update.
8003 (expand_compare_and_swap_loop): Update.
8004 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8005 * predict.c (predictable_edge_p): Update.
8006 (edge_probability_reliable_p): Update.
8007 (set_even_probabilities): Update.
8008 (combine_predictions_for_insn): Update.
8009 (combine_predictions_for_bb): Update.
8010 (propagate_freq): Update.
8011 (estimate_bb_frequencies): Update.
8012 (force_edge_cold): Update.
8013 * profile-count.c (profile_count::dump): Add missing space into dump.
8014 (profile_count::debug): Add newline.
8015 (profile_count::differs_from_p): Explicitly convert to unsigned.
8016 (profile_count::stream_in): Update.
8017 (profile_probability::dump): New member function.
8018 (profile_probability::debug): New member function.
8019 (profile_probability::differs_from_p): New member function.
8020 (profile_probability::differs_lot_from_p): New member function.
8021 (profile_probability::stream_in): New member function.
8022 (profile_probability::stream_out): New member function.
8023 * profile-count.h (profile_count_quality): Rename to ...
8024 (profile_quality): ... this one.
8025 (profile_probability): New.
8026 (profile_count): Update.
8027 * profile.c (compute_branch_probabilities): Update.
8028 * recog.c (peep2_attempt): Update.
8029 * sched-ebb.c (schedule_ebbs): Update.
8030 * sched-rgn.c (find_single_block_region): Update.
8031 (compute_dom_prob_ps): Update.
8032 (schedule_region): Update.
8033 * sel-sched-ir.c (compute_succs_info): Update.
8034 * stmt.c (struct case_node): Update.
8035 (do_jump_if_equal): Update.
8036 (get_outgoing_edge_probs): Update.
8037 (conditional_probability): Update.
8038 (emit_case_dispatch_table): Update.
8039 (expand_case): Update.
8040 (expand_sjlj_dispatch_table): Update.
8041 (emit_case_nodes): Update.
8042 * targhooks.c: Update.
8043 * tracer.c (better_p): Update.
8044 (find_best_successor): Update.
8045 * trans-mem.c (expand_transaction): Update.
8046 * tree-call-cdce.c: Update.
8047 * tree-cfg.c (gimple_split_edge): Upate.
8048 (move_sese_region_to_fn): Upate.
8049 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8050 * tree-eh.c (lower_resx): Upate.
8051 (cleanup_empty_eh_move_lp): Upate.
8052 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8053 * tree-inline.c (copy_edges_for_bb): Update.
8054 (copy_cfg_body): Update.
8055 * tree-parloops.c (gen_parallel_loop): Update.
8056 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8057 (gimple_gen_time_profiler): Update.
8058 * tree-ssa-dce.c (remove_dead_stmt): Update.
8059 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8060 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8061 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8062 (unloop_loops): Update.
8063 (try_peel_loop): Update.
8064 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8065 * tree-ssa-loop-split.c (connect_loops): Update.
8066 (split_loop): Update.
8067 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8068 (hoist_guard): Update.
8069 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8070 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8071 (value_replacement): Update.
8072 * tree-ssa-reassoc.c (branch_fixup): Update.
8073 * tree-ssa-tail-merge.c (replace_block_by): Update.
8074 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8075 (create_edge_and_update_destination_phis): Update.
8076 (compute_path_counts): Update.
8077 (recompute_probabilities): Update.
8078 (update_joiner_offpath_counts): Update.
8079 (freqs_to_counts_path): Update.
8080 (duplicate_thread_path): Update.
8081 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8082 (struct switch_conv_info): Update.
8083 (gen_inbound_check): Update.
8084 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8085 (vect_do_peeling): Update.
8086 (vect_loop_versioning): Update.
8087 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8088 (optimize_mask_stores): Update.
8089 * ubsan.c (ubsan_expand_null_ifn): Update.
8090 * value-prof.c (gimple_divmod_fixed_value): Update.
8091 (gimple_divmod_fixed_value_transform): Update.
8092 (gimple_mod_pow2): Update.
8093 (gimple_mod_pow2_value_transform): Update.
8094 (gimple_mod_subtract): Update.
8095 (gimple_mod_subtract_transform): Update.
8096 (gimple_ic): Update.
8097 (gimple_stringop_fixed_value): Update.
8098 (gimple_stringops_transform): Update.
8099 * value-prof.h: Update.
8101 2017-06-29 Carl Love <cel@us.ibm.com>
8103 * config/rs6000/rs6000-c.c: Add support for built-in functions
8104 vector signed int vec_signed (vector float);
8105 vector signed long long vec_signed (vector double);
8106 vector signed int vec_signed2 (vector double, vector double);
8107 vector signed int vec_signede (vector double);
8108 vector signed int vec_signedo (vector double);
8109 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8110 instruction generator.
8111 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8112 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8113 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8115 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8116 vunsignede_v2df): Add define_expands.
8117 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8118 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8119 VEC_UNSIGNEDO): Add definitions.
8120 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8121 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8122 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8123 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8124 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8125 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8126 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8127 vec_unsignede, vec_unsignedo): Add builtin defines.
8128 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8130 * doc/extend.texi: Update the built-in documentation file for the
8131 new built-in functions.
8133 2017-06-29 Richard Biener <rguenther@suse.de>
8135 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8136 reduction chains to LOOP_VINFO_REDUCTIONS.
8137 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8138 SLP reductions after processing reduction chains.
8140 2017-06-29 Nathan Sidwell <nathan@acm.org>
8142 * builtins.c (fold_builtin_FUNCTION): Use
8143 lang_hooks.decl_printable_name.
8145 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8148 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8149 with only one label.
8150 * stmt.c (expand_case): Assert NCASES is greater than one.
8152 2017-06-29 Richard Biener <rguenther@suse.de>
8154 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8156 (group_case_labels): Likewise.
8157 (find_taken_edge): Push sanity checking on val to workers...
8158 (find_taken_edge_cond_expr): ... here
8159 (find_taken_edge_switch_expr): ... and here, handle cases
8160 with just a default label.
8161 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8162 (group_case_labels): Likewise.
8163 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8164 group_case_labels does anything cleanup the CFG again.
8166 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8168 PR tree-optimization/81196
8169 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8170 exit condition comparing two IVs.
8172 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8174 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8175 profile to the dummy entry at the end of the list of architectures.
8176 * config/arm/arm-cpu-cdata.h: Regenerated.
8178 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8179 Michael Collison <michael.collison@arm.com>
8182 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8184 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8185 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8186 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8187 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8188 of shift when the shift amount is masked with constant equal to
8189 the size of the mode.
8190 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8193 2017-06-29 Martin Liska <mliska@suse.cz>
8195 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8197 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8200 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8203 (edge_growth_cache_entry::edge_growth_cache_entry): New
8205 (reset_edge_growth_cache): Update to use constructor.
8207 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8209 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8210 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8211 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8213 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8215 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8216 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8218 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8220 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8221 (*-linux-uclibc*): Add t-uclibc tmake_file.
8222 * config/t-musl: New.
8223 * config/t-uclibc: New.
8225 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8227 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8229 (gen_comm_data): Emit architectural setting of arch_prof.
8230 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8232 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8233 (armv8-m.base, armv8-m.main): Likewise.
8234 * arm-protos.h (arm_build_target): Add profile field.
8235 (arch_option): Likewise.
8236 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8238 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8239 arm_active_target.profile.
8241 2017-06-28 Richard Biener <rguenther@suse.de>
8244 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8245 TYPE_OVERFLOW_WRAPS.
8246 * match.pd (negate_expr_p): Likewise.
8247 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8248 fold_build2, not fold_binary.
8250 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8252 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8253 Convert memory address to Pmode.
8254 (aarch64_print_operand): Assert MEM operands are always Pmode.
8256 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8259 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8260 Remove redundant if.
8261 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8262 * config/arm/aarch-common-protos.h
8263 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8264 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8266 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8269 * multiple_target.c (create_dispatcher_calls): Set the default
8270 clone to be static, not public.
8272 2017-06-28 Richard Biener <rguenther@suse.de>
8274 * tree-vect-loop.c (vectorizable_reduction): Move special
8275 cond reduction IV var creation ...
8276 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8277 parameter. Use STMT_VINFO_VECTYPE.
8278 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8281 2017-06-28 Martin Liska <mliska@suse.cz>
8284 * ipa-visibility.c (non_local_p): Handle visibility.
8286 2017-06-28 Martin Liska <mliska@suse.cz>
8289 * common.opt: Add IntegerRange to various options.
8290 * opt-functions.awk (integer_range_info): New function.
8291 * optc-gen.awk: Add integer_range_info to cl_options struct.
8292 * opts-common.c (decode_cmdline_option): Handle
8293 CL_ERR_INT_RANGE_ARG.
8294 (cmdline_handle_error): Likewise.
8295 * opts.c (print_filtered_help): Show valid interval in
8296 when --help is provided.
8297 * opts.h (struct cl_option): Add range_min and range_max fields.
8298 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8300 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8302 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8303 (x * C EQ/NE y * C): New transformation.
8305 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8307 * genmultilib (combination_space): Accept '+' in option names.
8309 2017-06-28 Martin Liska <mliska@suse.cz>
8312 * asan.c (instrument_derefs): Bail out inner references
8313 that are hard register variables.
8315 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8318 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8319 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8321 2017-06-28 Richard Biener <rguenther@suse.de>
8323 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8324 (vect_get_slp_defs): Adjust.
8325 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8327 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8329 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8330 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8331 (vectorizable_reduction): Adjust.
8332 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8334 (vect_get_slp_defs): Likewise.
8335 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8336 (vectorizable_bswap): Adjust.
8337 (vectorizable_call): Likewise.
8338 (vectorizable_conversion): Likewise.
8339 (vectorizable_assignment): Likewise.
8340 (vectorizable_shift): Likewise.
8341 (vectorizable_operation): Likewise.
8342 (vectorizable_store): Likewise.
8343 (vectorizable_condition): Likewise.
8344 (vectorizable_comparison): Likewise.
8346 2017-06-28 Michael Collison <michael.collison@arm.com>
8349 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8351 (arm_gen_movmemqi): Removed unused variable 'i'.
8352 Convert 'for' loop into 'while' loop.
8353 (arm_expand_prologue): Remove last unnecessary set of insn.
8354 (thumb_pop): Remove unused variable 'pushed_words'.
8355 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8357 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8359 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8360 * config/s390/s390-protos.h: Add prototype of
8361 s390_rel_address_ok_p.
8362 * config/s390/s390.c (s390_got_symbol): New function.
8363 (s390_rel_address_ok_p): New function.
8364 (legitimize_pic_address): Use s390_rel_address_ok_p.
8365 (s390_load_got): Use s390_got_symbol.
8366 (s390_option_override): Issue error if
8367 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8368 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8370 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8372 2017-06-27 Andrew Pinski <apinski@cavium.com>
8374 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8375 (X * copysign (1.0, X)): New pattern.
8376 (X * copysign (1.0, -X)): New pattern.
8377 (copysign (-1.0, CST)): New pattern.
8379 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8381 * genmultilib (combination_space): Remove variable.
8382 Validate reuse rules against regular expression for any sequence
8383 of multilib options in any order.
8385 2017-06-27 Michael Collison <michael.collison@arm.com>
8387 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8388 call aarch64_split_simd_combine.
8389 * (aarch64_combine_internal<mode>): Delete pattern.
8390 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8391 Allow register and subreg operands.
8393 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8395 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8396 specific need, just fallback on defaults.
8397 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8399 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8400 Olivier Hainque <hainque@adacore.com>
8402 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8404 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8405 targets. Pick a default if no particular attempt applied.
8406 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8409 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8411 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8412 (x86_64-wrs-vxworks7): Likewise.
8414 2017-06-27 Marek Polacek <polacek@redhat.com>
8417 * ubsan.c (instrument_null): Check get_base_address's result for null.
8419 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8421 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8423 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8425 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8426 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8427 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8428 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8430 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8431 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8432 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8433 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8434 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8435 BUILT_IN_FEUPDATEENV): New builtins.
8436 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8437 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8438 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8439 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8441 (builtin_structptr_types): Adjust size.
8442 * tree.c (builtin_structptr_types): Add four entries.
8444 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8445 Olivier Hainque <hainque@adacore.com>
8447 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8448 (TLS_SYM): New local macro, forcing reference to __tls__ on
8449 link command lines for VxWorks 7 RTPs, triggering initialization
8451 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8452 OS features TLS support, true for RTPs on VxWorks 7.
8453 * config/vxworks.c (vxworks_override_options): Setup emutls
8456 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8458 * predict.c (test_prediction_value_range): Use -1U instead of -1
8459 to avoid narrowing conversion warning.
8460 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8461 to avoid narrowing conversion warning.
8462 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8464 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8466 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8468 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8469 64bit configurations.
8470 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8471 (SIZE_TYPE): Likewise.
8472 * config/vxworks.c (vxworks_emutls_var_fields): Use
8473 long_unsigned_type_node instead of unsigned_type_node as the offset
8474 field type, which is "pointer" mode in emutls.c.
8476 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8479 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8482 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8483 gimple_vuse copying separately from gimple_vdef copying.
8485 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8487 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8490 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8491 Olivier Hainque <hainque@adacore.com>
8493 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8494 for all vxworks7 targets.
8495 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8496 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8497 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8498 variations for VX6/VX7 and 32/64bits later on in ...
8499 (VXWORKS_LIB_SPEC): Leverage new macros.
8500 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8501 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8503 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8505 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8506 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8508 2017-06-26 Carl Love <cel@us.ibm.com>
8510 * config/rs6000/rs6000-c.c: Add support for built-in functions
8511 vector bool char vec_reve (vector bool char);
8512 vector signed char vec_reve (vector signed char);
8513 vector unsigned char vec_reve (vector unsigned char);
8514 vector bool int vec_reve (vector bool int);
8515 vector signed int vec_reve (vector signed int);
8516 vector unsigned int vec_reve (vector unsigned int);
8517 vector bool long long vec_reve (vector bool long long);
8518 vector signed long long vec_reve (vector signed long long);
8519 vector unsigned long long vec_reve (vector unsigned long long);
8520 vector bool short vec_reve (vector bool short);
8521 vector signed short vec_reve (vector signed short);
8522 vector double vec_reve (vector double);
8523 vector float vec_reve (vector float);
8524 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8525 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8526 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8527 (altivec_vreve): New pattern.
8528 * config/rs6000/altivec.h (vec_reve): New define.
8529 * doc/extend.texi (vec_rev): Update the built-in documentation file
8530 for the new built-in functions.
8532 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8534 PR tree-optimization/71815
8535 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8537 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8539 (slsr_process_phi): Likewise.
8540 (replace_uncond_cands_and_profitable_phis): Don't replace a
8541 multiply candidate with a stride of 1 (copy or cast).
8542 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8544 (lowest_cost_path): Likewise.
8545 (total_savings): Likewise.
8547 2017-06-26 Richard Biener <rguenther@suse.de>
8550 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8551 Use def_builtin_pure for all gather builtins.
8553 2017-06-26 Richard Biener <rguenther@suse.de>
8555 PR tree-optimization/81203
8556 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8559 2017-06-26 Marek Polacek <polacek@redhat.com>
8562 * doc/invoke.texi: Document -Wmultistatement-macros.
8564 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8566 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8567 arm_neon_ok_no_float_abi effective target.
8569 2017-06-26 Richard Biener <rguenther@suse.de>
8571 PR tree-optimization/80928
8572 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8573 (copy_bbs): Set BB_DUPLICATED flag early.
8574 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8576 (execute_on_shrinking_pred): Likewise.
8577 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8578 BB_DUPLICATED blocks.
8579 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8580 iterate over all PHIs considering removal of *gsi.
8582 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8584 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8587 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8589 * config/rs6000/rs6000-string.c: (expand_block_clear,
8590 do_load_for_compare, select_block_compare_mode,
8591 compute_current_alignment, expand_block_compare,
8592 expand_strncmp_align_check, expand_strn_compare,
8593 expand_block_move, rs6000_output_load_multiple)
8594 Move functions related to string/block move/compare
8596 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8597 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8598 for this function which is now used in two files.
8599 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8600 * config.gcc: Add rs6000-string.o to extra_objs for
8601 targets powerpc*-*-* and rs6000*-*-*.
8603 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8606 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8607 32-bit, since indexed is not valid for DImode.
8608 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8609 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8610 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8611 (define_peephole2 for Altivec d-form store): Likewise.
8614 * multiple_target.c (create_dispatcher_calls): Only create the
8615 dispatcher call if the function is the default clone of a
8618 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8621 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8622 a call, force the call to not be a tail call.
8624 2017-06-23 Jeff Law <law@redhat.com>
8626 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8629 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8631 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8632 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8633 handling for early expansion of vector shifts (sl,sr,sra,rl).
8634 (builtin_function_type): Add vector shift right instructions
8635 to the unsigned argument list.
8637 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8639 rtl-optimizatoin/79286
8640 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8641 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8642 trap. PIC register plus a const unspec without offset can never trap.
8644 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8646 * tree.h (builtin_structptr_type): New type.
8647 (builtin_structptr_types): Declare new array.
8648 * tree.c (builtin_structptr_types): New array.
8649 (free_lang_data, build_common_tree_nodes): Use it.
8651 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8654 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8657 2017-06-22 Matt Turner <mattst88@gmail.com>
8659 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8660 Lake models to skylake case. Assume skylake for unknown
8661 models with clflushopt.
8663 2017-06-22 Jeff Law <law@redhat.com>
8665 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8666 frame sizes that do not satisfy aarch64_uimm12_shift.
8668 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8670 * profile-count.h (apply_probability,
8671 apply_scale, probability_in): Fix checks for zero.
8673 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8675 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8676 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8678 2016-06-22 Richard Biener <rguenther@suse.de>
8680 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8681 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8682 REDUC_MAX_EXPR support.
8683 (vectorizable_reduction): Likewise.
8684 (vect_create_epilog_for_reduction): Likewise.
8686 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8688 * match.pd (A / (1 << B) -> A >> B): New.
8689 * generic-match-head.c: Include optabs-tree.h.
8690 * gimple-match-head.c: Likewise.
8691 * optabs-tree.h (target_supports_op_p): New.
8692 * optabs-tree.c (target_supports_op_p): New.
8694 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8696 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8697 $gcc_cv_ld --help output.
8698 (gcc_cv_ld_demangle): Likewise.
8699 (gcc_cv_ld_eh_frame_hdr): Likewise.
8700 (gcc_cv_ld_pie): Likewise.
8701 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8702 (gcc_cv_ld_buildid): Likewise.
8703 (gcc_cv_ld_sysroot): Likewise.
8704 (ld_bndplt_support): Likewise.
8705 (ld_pushpopstate_support): Likewise.
8706 * configure: Regenerate.
8707 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8709 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8712 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8713 operands indexes to avoid gap between operands and match_dups.
8715 2017-06-21 Andrew Pinski <apinski@cavium.com>
8717 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8718 Increment Arith_shift and Arith_shift_reg by 1.
8719 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8721 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8722 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8723 (aarch64_strip_extend): Add new argument and test for it.
8724 (aarch64_cheap_mult_shift_p): New function.
8725 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8726 add a cost if it is true.
8727 Update calls to aarch64_strip_extend.
8728 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8730 2017-06-21 Andrew Pinski <apinski@cavium.com>
8732 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8734 (thunderxt88): Likewise.
8735 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8736 (thunderx_prefetch_tune): New variable.
8737 (thunderx2t99_prefetch_tune): Update for the correct values.
8738 (thunderxt88_tunings): New variable.
8739 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8740 generic_prefetch_tune.
8741 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8743 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8745 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8746 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8747 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8748 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8749 (aarch64_atomic_cas<mode>, GPI): Likewise.
8751 2017-06-21 Martin Liska <mliska@suse.cz>
8753 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8754 statements on cold and hot labels.
8755 * predict.c (tree_estimate_probability_bb): Remove the
8756 prediction from this place.
8758 2017-06-21 Martin Liska <mliska@suse.cz>
8760 PR tree-optimization/79489
8761 * gimplify.c (maybe_add_early_return_predict_stmt): New
8763 (gimplify_return_expr): Call the function.
8764 * predict.c (tree_estimate_probability_bb): Remove handling
8766 * predict.def: Update comment about early return predictor.
8767 * gimple-predict.h (is_gimple_predict): New function.
8768 * predict.def: Change default value of early return to 66.
8769 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8771 * passes.def: Put pass_strip_predict_hints to the beginning of
8774 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8776 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8777 FUNCTION_DECL declarations.
8778 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8780 (dwaf2out_decl): Likewise.
8781 * godump.c (go_early_global_decl): Skip call to the real debug hook
8782 for FUNCTION_DECL declarations.
8783 * passes.c (rest_of_decl_compilation): Skip call to the
8784 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8785 -fdump-go-spec is passed.
8787 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8789 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8790 Reorder for compactness.
8791 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8792 (def_builtin_pure, def_builtin_pure2): New functions.
8793 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8795 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8797 * match.pd (nop_convert): New predicate.
8798 ((A +- CST1) +- CST2): Allow some NOP conversions.
8800 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8803 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8804 with ctors/dtors if GOVD_SHARED is set.
8806 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8808 * config/aarch64/aarch64.md (movti_aarch64):
8809 Emit mov rather than orr.
8810 (movtf_aarch64): Likewise.
8811 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8812 Emit mov rather than orr.
8814 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8816 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8817 Swap alternatives, make integer dup more expensive.
8819 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8821 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8822 Return true for non-tls symbols.
8824 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8826 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8827 (cortex-a75): Likewise.
8828 (cortex-a75.cortex-a55): Likewise.
8829 * config/aarch64/aarch64-tune.md: Regenerate.
8830 * doc/invoke.texi (-mtune): Document new values for -mtune.
8832 2017-06-21 Tom de Vries <tom@codesourcery.com>
8834 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8836 (Effective-Target Keywords, Other attributes): Suggest using
8837 dg-add-options stack_size feature to get stack limit in stack_size
8838 effective target documentation.
8840 2017-06-21 Julian Brown <julian@codesourcery.com>
8841 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8843 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8844 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8845 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8847 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8848 attribute type list for neon_multiply.
8849 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8850 attribute type list for neon_multiply.
8851 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8852 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8853 attribute type list for neon_multiply.
8854 * config/arm/types.md (crypto_pmull): Add.
8855 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8856 attribute type list.
8858 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8860 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8863 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8865 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8866 to make sure not to dereference a NULL cost_classes_ptr pointer.
8868 2017-06-20 Carl Love <cel@us.ibm.com>
8870 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8871 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8872 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8873 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8874 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8875 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8876 VMULOSW): New enum "unspec" values.
8877 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8878 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8879 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8880 altivec_vmulosw): New patterns.
8881 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8882 VMULOSW): Add definitions.
8884 2017-06-20 Julia Koval <julia.koval@intel.com>
8886 * config/i386/i386.c: Fix rounding expand for new pattern.
8887 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8889 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8891 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8892 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8894 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8896 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8899 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8901 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8902 architecture, then by implementer ID.
8903 * config/aarch64/aarch64-tune.md: Regenerate.
8905 2017-06-20 Richard Biener <rguenther@suse.de>
8908 * fold-const.c (split_tree): Fold to type before negating.
8910 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8912 * diagnostic-show-locus.c
8913 (selftest::test_fixit_deletion_affecting_newline): New function.
8914 (selftest::diagnostic_show_locus_c_tests): Call it.
8916 2017-06-20 Andreas Schwab <schwab@suse.de>
8919 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8922 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8924 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8925 __ARM_FEATURE_COPROC according to support.
8927 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8929 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8930 Rewritten to avoid overflow for > 32-bit pointers.
8933 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8934 by removing enum keyword.
8935 (ubsan_type_descriptor): Likewise. Formatting fix.
8938 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8939 splitter): Require TARGET_SSE2 in the condition.
8941 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8944 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8945 for doing vector set of SFmode on ISA 3.0.
8946 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8947 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8949 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8950 SFmode value into a V4SF variable that was extracted from another
8951 V4SF variable without converting the element to double precision
8952 and back to single precision vector format.
8953 (vsx_insert_extract_v4sf_p9_2): Likewise.
8955 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8957 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8958 in UWHI to avoid undefined overflow.
8961 * ubsan.h (enum ubsan_encode_value_phase): New.
8962 (ubsan_encode_value): Change second argument to
8963 enum ubsan_encode_value_phase with default value of
8964 UBSAN_ENCODE_VALUE_GENERIC.
8965 * ubsan.c (ubsan_encode_value): Change second argument to
8966 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8967 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8968 create_tmp_var_raw instead of create_tmp_var and use a
8970 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8971 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8972 ubsan_encode_value callers.
8975 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8976 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8977 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8979 2017-06-19 Richard Biener <rguenther@suse.de>
8982 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8983 estimates if we changed anything.
8985 2017-06-19 Richard Biener <rguenther@suse.de>
8987 PR tree-optimization/80887
8988 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8989 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8990 simplified lookups, then reset mprts_hook.
8991 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8993 (try_to_simplify): Likewise.
8995 2017-06-19 Martin Liska <mliska@suse.cz>
8998 * gimplify.c (gimplify_switch_expr):
8999 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9001 2017-06-19 Martin Liska <mliska@suse.cz>
9003 * doc/install.texi: Document that PGO runs in 4 stages.
9005 2017-06-19 Martin Liska <mliska@suse.cz>
9008 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9009 to dispatcher function name.
9010 * multiple_target.c (replace_function_decl): New function.
9011 (create_dispatcher_calls): Redirect both edges and references.
9013 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9015 * profile-count.c (profile_count::dump): Dump quality.
9016 (profile_count::differs_from_p): Update for unsigned val.
9017 * profile-count.h (profile_count_quality): New enum.
9018 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9020 2017-06-19 Richard Biener <rguenther@suse.de>
9022 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9023 struct function as arg.
9024 (estimate_numbers_of_iterations): Export overload with loop arg.
9025 (free_numbers_of_iterations_estimates_loop): Use an overload of
9026 free_numbers_of_iterations_estimates instead.
9027 * tree-cfg.c (remove_bb): Adjust.
9028 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9029 * tree-parloops.c (gen_parallel_loop): Likewise.
9030 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9032 (tree_unroll_loops_completely): Likewise.
9033 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9034 Use an overload instead and export.
9035 (estimated_loop_iterations): Adjust.
9036 (max_loop_iterations): Likewise.
9037 (likely_max_loop_iterations): Likewise.
9038 (estimate_numbers_of_iterations): Take struct function as arg
9040 (loop_exits_before_overflow): Adjust.
9041 (free_numbers_of_iterations_estimates_loop): Use an overload.
9042 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9043 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9045 2017-06-19 Richard Biener <rguenther@suse.de>
9048 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9049 RANGE_EXPR conservatively.
9051 2017-06-16 Carl Love <cel@us.ibm.com>
9053 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9054 definitions for vec_float, vec_float2, vec_floato,
9055 vec_floate built-ins.
9056 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9057 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9059 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9060 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9061 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9062 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9063 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9064 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9065 vec_floato): Add builtin defines.
9066 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9067 Update the built-in documentation file for the new built-in
9070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9072 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9073 (mthumb): Mark as the negative of -marm.
9075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9077 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9079 (ARM Options, -mtune): Document that this accepts the same
9080 extension options as -mcpu.
9081 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9085 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9086 permitted extensions.
9088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9090 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9091 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9092 (armv8-m.main): Add option +nodsp.
9093 * config/arm/arm-cpu-cdata.h: Regenerated.
9095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9097 * config/arm/t-fuchsia: New file.
9098 * config.gcc (arm*-*-fuchsia*): Use it.
9100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9102 * config/arm/t-symbian: Rewrite for new option infrastructure.
9104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9106 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9107 (MULTILIB_REQUIRED): Likewise.
9109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9111 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9112 (MULTILIB_RESUE): Likewise.
9113 (MULTILIB_MATCHES): Likewise.
9114 (MULTLIB_REQUIRED): Likewise.
9116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9118 * config/arm/t-rtems: Rewrite for new option framework.
9120 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9122 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9123 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9124 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9125 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9126 * config/arm/t-multilib: ... here.
9127 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9128 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9129 armv7-a and armv8*-a when A-profile libraries have not been built.
9130 * config/arm/t-rmprofile: Rewrite.
9132 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9134 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9135 with a backslash. Remove the backslash after substituting unescaped
9137 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9141 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9142 the driver to pass through the default setting of -mfloat-abi.
9143 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9145 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9146 (all_feat_combs): New rule.
9147 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9149 * config/arm/t-aprofile: Rewrite.
9151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9153 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9154 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9155 fpu is not specified by the user/command-line.
9156 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9157 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9158 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9159 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9160 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9161 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9165 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9166 * config/arm/t-arm-elf: Rewritten.
9168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9170 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9171 have some floating-point instructions.
9172 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9173 (TARGET_MAYBE_HARD_FLOAT): New macro.
9174 * config/arm/arm-builtins.c (arm_init_builtins): Use
9175 TARGET_MAYBE_HARD_FLOAT.
9176 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9178 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9180 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9181 (configargs.h): Include it.
9182 (arm_print_hint_for_fpu_option): New function.
9183 (arm_parse_fpu_option): New function.
9184 (candidate_extension): New class.
9185 (arm_canon_for_multilib): New function.
9186 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9187 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9188 (ARCH_CANONICAL_SPECS): New macro.
9189 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9193 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9194 are set after handling multilib fragments. Set target_cpu_default2
9197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9199 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9201 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9202 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9204 (check_arch): Likewise.
9205 * config/arm/arm.c (arm_configure_build_target): Handle
9206 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9207 options in the default.
9209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9211 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9212 when an option is an alias of another.
9213 * config/arm/parsecpu.awk (optalias): New parser token.
9214 (gen_comm_data): Mark non-alias options as such. Emit entries
9215 for extension aliases.
9216 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9217 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9218 (armv6kz, armv6zk, armv6t2): Likewise.
9219 (armv7): Make vfpv3-d16 an alias.
9220 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9222 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9223 Sort in canonical order.
9224 (armv8-a): Sort in canonical order.
9225 (armv8.1-a, armv8.2-a): Likewise.
9226 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9228 (cortex-a9): Sort in canonical order.
9229 * config/arm/arm.c (selftests.h): Include it.
9230 (arm_test_cpu_arch_data): New function.
9231 (arm_run_self_tests): New function.
9232 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9233 (targetm): Move declaration to the end of the file.
9234 * arm-cpu-cdata.h: Regenerated.
9236 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9238 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9239 call to target_mode_check describing the type of option passed.
9240 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9241 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9242 arm_parse_cpu_option_name to match parameters against list of
9244 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9245 arm_arch_core_flags data structure.
9246 * config/arm/arm-cpu_cdata.h: Regenerated.
9248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9250 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9252 (arm_print_hint_for_cpu_option): Likewise.
9253 (arm_print_hint_for_arch_option): Likewise.
9254 (arm_parse_cpu_option_name): Likewise.
9255 (arm_parse_arch_option_name): Likewise.
9256 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9257 of entries in the all_fpus list.
9258 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9259 (arm_parse_cpu_option_name): Declare.
9260 (arm_parse_arch_option_name): Declare.
9261 (arm_parse_option_features): Declare.
9262 (arm_intialize_isa): Declare.
9263 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9265 (gen_comm_data): ... here. Make definitions non-static.
9266 * config/arm/arm-cpu-data.h: Regenerated.
9267 * config/arm/arm-cpu-cdata.h: Regenerated.
9269 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9271 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9272 (cpu_arch_extension): New structure.
9273 (cpu_arch_option, arch_option, cpu_option): New structures.
9274 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9276 (gen_data): Generate new format data tables.
9277 * config/arm/arm.c (cpu_tune): New structure.
9278 (cpu_option, processors): Delete.
9279 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9280 (arm_print_hint_for_cpu_option): ... this and ...
9281 (arm_print_hint_for_arch_option): ... this.
9282 (arm_parse_arch_cpu_name): Delete. Replace with ...
9283 (arm_parse_cpu_option_name): ... this and ...
9284 (arm_parse_arch_option_name): ... this.
9285 (arm_unrecognized_feature): Change type of target parameter to
9287 (arm_parse_arch_cpu_features): Delete. Replace with ...
9288 (arm_parse_option_features): ... this.
9289 (arm_configure_build_target): Rework to use new configuration data
9291 (arm_print_tune_info): Rework for new configuration data tables.
9292 * config/arm/arm-cpu-data.h: Regenerated.
9293 * config/arm/arm-cpu.h: Regenerated.
9295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9297 * Makefile.in (OBJS): Move sbitmap.o from here ...
9298 (OBJS-libcommon): ... to here.
9300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9302 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9303 (ISA_ALL_CRYPTO): New macro.
9304 (ISA_ALL_SIMD): New macro
9305 (ISA_ALL_FP): New macro.
9306 * config/arm/arm.c (fpu_bitlist): Update initializer.
9307 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9309 (arm9e): Add fpu. Add option for nofp
9310 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9311 (arm926ej-s, arm1026ej-s): Likewise.
9312 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9313 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9314 neon-fp16, neon-vfpv4, nofp and nosimd.
9315 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9316 (cortex-a8): Add fpu. Add option for nofp.
9317 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9318 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9319 (cortex-r4f): Add fpu.
9320 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9321 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9323 (cortex-r8): Likewise.
9324 (cortex-m4): Add fpu. Add option for nofp.
9325 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9326 (cortex-a17.cortex-a7): Likewise.
9327 (cortex-a32): Add fpu. Add options for crypto and nofp.
9328 (cortex-a35, cortex-a53): Likewise.
9329 (cortex-a57): Add fpu. Add option for crypto.
9330 (cortex-a72, cortex-a73): Likewise.
9331 (exynos-m1): Likewise.
9332 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9333 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9334 (cortex-m33): Add fpu. Add option for nofp.
9335 * config/arm/arm-cpu-cdata.h: Regenerated
9336 * config/arm/arm-cpu-data.h: Regenerated.
9338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9340 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9341 (armv5te, armv5tej): Likewise.
9342 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9343 (armv7): Add options fp and vfpv3-d16.
9344 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9345 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9347 (armv7ve): Likewise.
9348 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9349 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9350 (armv8-a): Add nocrypto option.
9351 (armv8.1-a, armv8.2-a): Likewise.
9352 (armv8-m.main): add options fp, fp.dp and nofp.
9354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9356 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9358 (armv8-a+crc): Delete.
9359 (armv8.1-a): Add options simd, crypto and nofp.
9360 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9361 (armv8.2-a+fp16): Delete.
9362 (armv8-m.main): Add option dsp.
9363 (armv8-m.main+dsp): Delete.
9364 (cortex-a8): Add fpu. Add nofp option.
9365 (cortex-a9): Add fpu. Add nofp and nosimd options.
9366 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9367 link to main cpu and architecture data structures.
9368 (gen_comm_data): Only put isa attributes from the main architecture
9370 (option): New statement for architecture and CPU entries.
9371 * arm.c (struct cpu_option): New structure.
9372 (struct processors): Add entry for options.
9373 (arm_unrecognized_feature): New function.
9374 (arm_parse_arch_cpu_name): Ignore any characters after the first
9376 (arm_parse_arch_cpu_feature): New function.
9377 (arm_configure_build_target): Separate out any CPU and architecture
9378 features and parse separately. Don't error out if -mfpu=auto is
9379 used with only an architecture string.
9380 (arm_print_asm_arch_directives): New function.
9381 (arm_file_start): Call it.
9382 * config/arm/arm-cpu-cdata.h: Regenerated.
9383 * config/arm/arm-cpu-data.h: Likewise.
9384 * config/arm/arm-tables.opt: Likewise.
9386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9388 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9389 assembler when it is not -mfpu=auto.
9391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9393 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9394 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9395 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9396 (ASM_CPU_SPEC): Rewrite.
9397 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9398 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9399 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9400 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9401 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9402 copied string is NUL-terminated. Also strip any characters prefixed
9404 (arm_rewrite_selected_arch): New function.
9405 (arm_rewrite_march): New function.
9407 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9409 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9410 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9411 (march, mcpu, mtune): Convert to string-based options.
9412 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9413 (arm_parse_arch_cpu_name): New function.
9414 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9415 identify selected architecture or CPU.
9416 (arm_option_save): New function.
9417 (TARGET_OPTION_SAVE): Redefine.
9418 (arm_option_restore): Restore string options.
9419 (arm_option_print): Print string options.
9421 2017-06-16 Martin Sebor <msebor@redhat.com>
9423 PR tree-optimization/80933
9424 PR tree-optimization/80934
9425 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9426 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9427 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9428 (gimple_fold_builtin): Call them.
9430 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9432 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9433 as unlikely; update profile.
9435 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9437 * predict.c (force_edge_cold): Handle declaring edges impossible
9440 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9442 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9444 (try_unroll_loop_completely): Fix reporting.
9446 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9448 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9450 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9453 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9454 if given a non-constant argument for an intrinsic which requires a
9457 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9459 * profile.c (compare_freqs): New function.
9460 (branch_prob): Sort edge list.
9461 (find_spanning_tree): Assume that the list is priority sorted.
9463 2017-06-16 Richard Biener <rguenther@suse.de>
9465 PR tree-optimization/81090
9466 * passes.def (pass_record_bounds): Remove.
9467 * tree-pass.h (make_pass_record_bounds): Likewise.
9468 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9469 make_pass_record_bounds): Likewise.
9470 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9471 not free niter estimates at the beginning but at the end.
9472 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9474 2017-06-16 Richard Biener <rguenther@suse.de>
9476 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9477 initializer to workaround ICE in host GCC 4.8.
9479 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9481 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9483 (clone_inlined_nodes): Update.
9485 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9487 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9488 prefetch settings, and enable prefetching by default at -O3.
9490 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9492 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9493 Set flag_prefetch_loop_arrays according to tuning data.
9495 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9497 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9499 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9500 [Unrelated to main purpose of the patch] Place the pointer field last
9501 to enable type checking errors when tune structure are wrongly merged.
9502 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9503 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9504 (thunderx2t99_prefetch_tune): New tune constants.
9505 (tune_params *_tunings): Update all tunings (no functional change).
9506 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9507 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9508 from tunings structures.
9510 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9513 * ubsan.c (instrument_null): Add T argument, use it instead
9514 of computing it based on IS_LHS.
9515 (instrument_object_size): Likewise.
9516 (pass_ubsan::execute): Adjust instrument_null and
9517 instrument_object_size callers to pass gimple_get_lhs or
9518 gimple_assign_rhs1 result to it. Use instrument_null instead of
9519 calling get_base_address and instrument_mem_ref. Handle
9520 aggregate call arguments for object-size sanitization.
9522 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9524 PR tree-optimization/81089
9525 * tree-vrp.c (is_masked_range_test): Validate operands of
9528 2017-06-15 Martin Sebor <msebor@redhat.com>
9531 * dumpfile.c (dump_register): Avoid calling memset to initialize
9532 a class with a default ctor.
9533 * gcc.c (struct compiler): Remove const qualification.
9534 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9535 * hash-table.h: Ditto.
9536 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9538 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9539 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9541 * params.h (struct param_info): Make struct members non-const.
9542 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9543 with default initialization.
9544 * vec.h (vec_copy_construct, vec_default_construct): New helper
9546 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9547 with vec_copy_construct.
9548 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9549 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9550 * doc/invoke.texi (-Wclass-memaccess): Document.
9552 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9554 * emit-rtl.h (is_leaf): Update comment about local
9557 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9560 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9561 for a variable to have a section before checking if the section has a
9563 Set section to.persistent if persistent attribute is set.
9564 Warn if .persistent attribute is used on an automatic variable.
9566 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9568 PR rtl-optimization/80474
9569 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9571 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9573 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9576 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9578 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9579 (casesi): Emit bounds checking as RTL.
9580 (casesi_internal_mips16_<mode>): Remove bounds checking.
9582 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9584 * config/xtensa/xtensa.c (xtensa_option_override): Append
9585 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9586 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9587 xtensa_doloop_hooks): Define unconditionally.
9588 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9590 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9591 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9592 for it in xtensa_option_override.
9593 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9594 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9596 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9598 * doc/cppopts.texi: Document '-' special value to -MF.
9600 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9602 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9603 (cortex_a53_fconst): Likewise.
9604 (cortex_a53_fpmul): Likewise.
9605 (cortex_a53_f_load_64): Likewise.
9606 (cortex_a53_f_load_many): Likewise.
9607 (cortex_a53_advsimd_alu): Likewise.
9608 (cortex_a53_advsimd_alu_q): Likewise.
9609 (cortex_a53_advsimd_mul): Likewise.
9610 (cortex_a53_advsimd_mul_q): Likewise.
9611 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9612 Add missing fmul, r2f_cvt and fconst cases.
9614 2017-06-14 Richard Biener <rguenther@suse.de>
9617 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9619 (fold_binary_loc): When associating do not treat pre-existing
9620 TREE_OVERFLOW on literal constants as a reason to allow
9621 TREE_OVERFLOW on associated literal constants.
9623 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9625 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9626 (MASK_FEATURES): New macro.
9627 * config/sparc/sparc.c (sparc_option_override): Remove the special
9628 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9630 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9632 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9633 a division of 0 if non-call exceptions are enabled.
9635 2017-06-14 Andrew Pinski <apinski@cavium.com>
9636 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9639 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9640 Improve vector initialization code gen for only variable case.
9642 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9644 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9646 2017-06-14 Richard Biener <rguenther@suse.de>
9648 PR tree-optimization/81083
9649 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9652 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9654 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9655 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9656 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9657 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9658 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9659 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9661 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9663 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9664 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9666 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9668 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9670 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9672 * config/rs6000/t-rtems: Don't handle SPE.
9674 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9676 * config/rs6000/t-linux: Don't handle SPE.
9678 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9680 * config/rs6000/eabispe.h: Delete file.
9682 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9684 * config/rs6000/t-spe: Delete file.
9686 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9688 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9689 (rs6000_legitimate_offset_address_p): Return false for anything in
9690 V2SImode or V2SFmode.
9692 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9694 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9695 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9696 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9698 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9699 (rs6000_legitimate_offset_address_p): Ditto.
9700 (rs6000_emit_move): Ditto.
9701 (rs6000_init_builtins): Remove V4HI_type_node.
9703 2017-06-13 Martin Liska <mliska@suse.cz>
9706 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9707 (gate_asan): Likewise.
9708 * asan.h (asan_no_sanitize_address_p): Remove the function.
9709 (sanitize_flags_p): New function.
9710 * builtins.def: Fix coding style.
9711 * common.opt: Use renamed enum value.
9712 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9713 * doc/extend.texi: Document no_sanitize attribute.
9714 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9715 to SANITIZE_UNDEFINED_NONDEFAULT.
9716 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9717 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9718 Use sanitize_flags_p.
9719 * gimplify.c (gimplify_function_tree): Likewise.
9720 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9721 * opts.c (parse_no_sanitize_attribute): New function.
9722 (common_handle_option): Use renamed enum value.
9723 * opts.h (parse_no_sanitize_attribute): Declare.
9724 * tree.c (sanitize_flags_p): New function.
9725 * tree.h: Declared here.
9726 * tsan.c: Use sanitize_flags_p.
9727 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9728 (instrument_mem_ref): Likewise.
9729 (instrument_bool_enum_load): Likewise.
9730 (do_ubsan_in_current_function): Remove the function.
9731 (pass_ubsan::execute): Use sanitize_flags_p.
9732 * ubsan.h: Remove do_ubsan_in_current_function
9733 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9734 (dump_function_to_file): Use it here.
9736 2017-06-13 Martin Jambor <mjambor@suse.cz>
9738 PR tree-optimization/80803
9739 PR tree-optimization/81063
9740 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9741 (propagate_subaccesses_across_link): Enqueue subtree whenever
9742 necessary instead of relying on the caller.
9744 2017-06-13 Martin Jambor <mjambor@suse.cz>
9746 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9747 that have a first_link.
9748 (sort_and_splice_var_accesses): Do not check first_link before
9750 (subtree_mark_written_and_enqueue): Likewise.
9751 (propagate_all_subaccesses): Likewise and do not stop at first
9752 parent with a first_link.
9754 2017-06-13 Martin Jambor <mjambor@suse.cz>
9756 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9759 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9761 * match.pd: New pattern.
9763 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9765 * tree-vrp.c (is_masked_range_test): New function.
9766 (register_edge_assert_for): Determine ranges for
9769 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9771 PR tree-optimization/67328
9772 * fold-const.c (maskable_range_p): New function.
9773 (build_range_check): Generate bittests if possible.
9775 2017-06-13 Martin Liska <mliska@suse.cz>
9777 * gimple-pretty-print.c (dump_probability): Add new argument.
9778 (dump_edge_probability): Dump both probability and count.
9779 (dump_gimple_label): Likewise.
9780 (dump_gimple_bb_header): Likewise.
9782 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9785 * config/avr/avr-devices.c: Fix indentation.
9786 * config/avr/gen-avr-mmcu-specs.c: Dito.
9788 2017-06-13 Richard Biener <rguenther@suse.de>
9790 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9791 instead get vector type from stmt_info.
9792 (vectorizable_reduction): Adjust. Remove dead code.
9794 2017-06-13 Richard Biener <rguenther@suse.de>
9797 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9798 case of C * (x * C2 + C3).
9799 (fold_addr_of_array_ref_difference): Properly fold index difference.
9801 2017-06-12 David S. Miller <davem@davemloft.net>
9804 * config/sparc/sparc.md (return expander): Emit frame blockage if
9805 function uses alloca.
9807 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9809 * combine.c (make_field_assignment): Check len rather than the mode
9810 precision when calling force_to_mode.
9812 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9814 Support multilibs and devices that see flash in RAM address range.
9817 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9818 (avr_mcu_t) <flash_pm_offset>: New field.
9819 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9820 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9821 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9822 (AVR_TINY_PM_OFFSET): Remove macro.
9823 * config/avr/avr.opt (-mshort-calls): New option.
9824 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9825 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9826 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9827 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9828 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9829 instead of avr_arch->have_jmp_call.
9830 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9831 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9832 avr_arch->flash_pm_offset to define.
9833 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9834 new field flash_pm_offset. Add entry for avrxmega3.
9835 (avr_texinfo): Add entry for avrxmega3.
9836 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9837 attiny212, attiny214,
9838 attiny412, attiny414, attiny416, attiny417,
9839 attiny814, attiny816, attiny817,
9840 attiny1614, attiny1616, attiny1617,
9841 attiny3214, attiny3216, attiny3217.
9842 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9843 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9844 (avr_print_operand_address) [AVR_TINY]: Same.
9845 (avr_asm_init_sections) <readonly_data_section>: Only patch
9846 callback if avr_arch->flash_pm_offset = 0.
9847 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9848 for rodata if avr_arch->flash_pm_offset != 0.
9849 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9850 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9851 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9852 (m_options): Append opt_rcall.
9853 (m_dirnames): Append dir_rcall.
9854 * config/avr/t-multilib: Regenerate.
9856 * configure.ac [target=avr]: Check whether avrxmega3 default
9857 linker description file works as needed.
9858 * configure: Regenerate.
9859 * doc/avr-mmcu.texi: Regenerate.
9860 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9861 <__AVR_ARCH__>: Document avrxmega3 and 103.
9862 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9863 <__AVR_SHORT_CALLS__>: Document it.
9864 <__AVR_PM_BASE_ADDRESS__>: Document it.
9865 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9866 (AVR Variable Attributes) <progmem>: Document this is
9867 not needed for avrxmega3.
9868 (AVR Named Address Spaces) <__flash>: Dito.
9870 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9872 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9874 2017-06-12 Doug Rupp <rupp@adacore.com>
9876 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9877 Append vxworks-stdint.h to the tm_file list.
9878 * config/vxworks-stdint.h: New file.
9880 2017-06-12 Martin Liska <mliska@suse.cz>
9882 PR tree-optimization/81041
9883 * tree-profile.c (gimple_gen_ic_func_profiler):
9884 Create an extra BB in profile-generate
9885 (gimple_gen_time_profiler): Likewise.
9887 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9889 PR tree-optimization/81003
9890 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9891 (update_range_test): Use it instead of force_gimple_operand_gsi.
9893 2017-06-12 Richard Biener <rguenther@suse.de>
9895 PR tree-optimization/81053
9896 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9897 with backedge value not defined in loop. Simplify def stmt
9900 2017-06-11 Tom de Vries <tom@codesourcery.com>
9903 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9905 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9906 nvptx_cannot_force_const_mem.
9908 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9910 * opts.c (finish_options): Move test for flag_split_stack after
9911 it has been initialized.
9913 2017-06-11 Jason Merrill <jason@redhat.com>
9915 * tree.h (id_equal): New.
9916 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9917 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9918 instead of strcmp of IDENTIFIER_POINTER.
9920 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9922 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9923 (mark_all_inlined_calls_cdtor): Fix formating.
9924 (inline_transform): Rescale profile before inlining.
9926 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9928 * cgraph.h (cgraph_edge::clone): Update prototype.
9929 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9930 (cgraph_node::create_clone): Update.
9931 (cgraph_node::create_version_clone): Update.
9932 * tree-inline.c (copy_bb): Update.
9933 (expand_call_inline): Update.
9935 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9937 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9938 factored out from ...
9939 (rs6000_emit_prologue): ... here.
9941 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9943 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9944 factored out from ...
9945 (rs6000_emit_prologue): ... here.
9947 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9949 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9951 (handle_missing_profiles): Fix computation of tp_first_run.
9952 (counts_to_freqs): Do not touch freqs when count is 0.
9954 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9956 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9959 2017-06-10 Tom de Vries <tom@codesourcery.com>
9961 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9962 attributes): Document signal effective target.
9964 2017-06-10 Tom de Vries <tom@codesourcery.com>
9966 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9967 Document effective target stack_size.
9969 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9971 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9972 to the edit_context if they can be auto-applied.
9974 2017-06-9 Ian Lance Taylor <iant@golang.org>
9976 * opts.c (finish_options): If -fsplit-stack, disable implicit
9977 -forder-blocks-and-partition.
9978 * doc/invoke.texi (Optimize Options): Document that when using
9979 -fsplit-stack -forder-blocks-and-partition is not implicitly
9982 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9984 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9985 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9986 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9987 * builtins.def (abort, trap, unreachable): Declare cold.
9988 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9989 * tree-core.h (ECF_COLD): New.
9990 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9991 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9993 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9995 * predict.c (unlikely_executed_stmt_p): Cleanup.
9997 2017-06-09 Richard Biener <rguenther@suse.de>
9999 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10000 model if the ref is always written to.
10002 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10004 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10006 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10008 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10011 2017-06-09 Tom de Vries <tom@codesourcery.com>
10014 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10015 "target cannot support label values" when encountering LABEL_REF.
10017 2017-06-09 Martin Liska <mliska@suse.cz>
10019 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10020 (gimple_gen_ic_func_profiler): Emit direct comparison
10021 of __gcov_indirect_call_callee with NULL.
10022 (gimple_gen_time_profiler): Change probability from
10023 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10025 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10027 * profile.c (edge_gcov_counts): Turn to pointer.
10028 (compute_branch_probabilities, compute_branch_probabilities): Update.
10029 (branch_prob): Do not clear edge_gcov_count.
10030 * profile.h (edge_gcov_counts): Turn to pointer.
10031 (edge_gcov_count): Update.
10033 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10035 * gimple.h (gimple_check_failed): Mark cold.
10037 2017-06-09 Richard Biener <rguenther@suse.de>
10039 PR tree-optimization/66623
10040 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10041 refactor check_reduction into two parts, properly computing
10042 whether we have to check reduction validity for outer loop
10045 2017-06-09 Richard Biener <rguenther@suse.de>
10047 PR tree-optimization/79483
10048 * graphite-scop-detection.c (order): New global.
10049 (get_order): Compute bb to order mapping that satisfies code
10050 generation constraints.
10051 (cmp_pbbs): New helper.
10052 (build_scops): Start domwalk at entry block, sort generated
10055 2017-06-09 Richard Biener <rguenther@suse.de>
10057 PR middle-end/81007
10058 * ipa-polymorphic-call.c
10059 (ipa_polymorphic_call_context::restrict_to_inner_class):
10060 Skip FIELD_DECLs with error_mark_node type.
10061 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10064 2017-06-09 Martin Liska <mliska@suse.cz>
10066 * predict.c (struct branch_predictor): New struct.
10067 (test_prediction_value_range): New test.
10068 (predict_c_tests): New function.
10069 * selftest-run-tests.c (selftest::run_tests): Run the function.
10070 * selftest.h: Declare new tests.
10072 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10075 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10076 gen_add3_insn did not fail.
10077 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10078 r0, construct that number in a temporary reg and add that reg to r0.
10079 If asked to put the result in r0 as well, fail.
10081 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10083 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10084 for early expansion of vec_eqv.
10086 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10088 PR middle-end/81005
10089 * ubsan.c (instrument_null): Avoid pointless code temporary.
10090 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10092 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10096 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10099 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10102 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10104 * predict.c (unlikely_executed_edge_p): Move ahead.
10105 (probably_never_executed_edge_p): Use it.
10107 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10109 PR middle-end/79988
10110 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10111 gimple_call_builtin_p call.
10113 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10115 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10116 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10117 rtl_check_failed_type2, rtl_check_failed_code1,
10118 rtl_check_failed_code2, rtl_check_failed_code_mode,
10119 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10120 rtvec_check_failed_bounds, rtl_check_failed_flag,
10121 _fatal_insn_not_found, _fatal_insn): Likewise.
10122 * tree.h (tree_contains_struct_check_failed,
10123 tree_check_failed, tree_not_check_failed,
10124 tree_class_check_failed, tree_range_check_failed,
10125 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10126 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10127 tree_operand_check_failed, omp_clause_check_failed,
10128 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10131 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10133 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10134 flag_branch_probabilities.
10135 * ipa-inline.c (edge_badness): Likewise.
10136 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10137 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10138 * predict.c (maybe_hot_frequency_p): Likewise.
10139 (probably_never_executed): Likewise.
10140 * sched-ebb.c (schedule_ebbs): Likewise.
10141 * sched-rgn.c (find_single_block_region): Likewise.
10142 * tracer.c (tail_duplicate): Likewise.
10144 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10146 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10147 longer requires x_flag_profile_use.
10149 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10151 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10152 instead of flag_reorder_blocks_and_partition.
10153 * dbxout.c (dbxout_function_end): Likewise.
10154 * dwarf2out.c (gen_subprogram_die): Likewise.
10155 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10156 * hw-doloop.c (reorg_loops): Likewise.
10157 * varasm.c (assemble_start_function,
10158 assemble_end_function): Likewise.
10159 (decide_function_section): Do not check for
10160 flag_reorder_blocks_and_partition.
10162 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10164 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10166 (chkp_get_hard_register_fake_addr_expr): Ditto.
10167 (chkp_build_addr_expr): Add check for hard reg case.
10168 (chkp_parse_array_and_component_ref): Ditto.
10169 (chkp_find_bounds_1): Ditto.
10170 (chkp_process_stmt): Don't generate bounds store for
10173 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10175 * predict.c (maybe_hot_bb_p): Do not check profile status.
10176 (maybe_hot_edge_p): Likewise.
10177 (probably_never_executed): Check for zero counts even if profile
10179 (unlikely_executed_edge_p): New function.
10180 (unlikely_executed_stmt_p): New function.
10181 (unlikely_executed_bb_p): New function.
10182 (set_even_probabilities): Use unlikely predicates.
10183 (combine_predictions_for_bb): Likewise.
10184 (predict_paths_for_bb): Likewise.
10185 (predict_paths_leading_to_edge): Likewise.
10186 (determine_unlikely_bbs): New function.
10187 (estimate_bb_frequencies): Use it.
10188 (compute_function_frequency): Use zero counts even if profile is
10190 * profile-count.h: Fix typo.
10192 2017-08-08 Julia Koval <julia.koval@intel.com>
10194 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10195 _mm512_mask_cvtsepi16_storeu_epi8,
10196 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10197 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10198 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10199 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10200 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10201 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10202 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10203 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10204 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10205 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10206 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10207 __builtin_ia32_pmovuswb256mem_mask,
10208 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10209 __builtin_ia32_pmovwb512mem_mask): New builtins.
10211 2017-08-08 Julia Koval <julia.koval@intel.com>
10213 PR target/73350,80862
10214 * config/i386/subst.md (round): Fix round pattern.
10215 * config/i386/i386.c (ix86_erase_embedded_rounding):
10216 Fix erasing rounding for the fixed pattern.
10218 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10220 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10222 2017-06-08 Martin Liska <mliska@suse.cz>
10224 PR gcov-profile/80911
10225 * gcov.c (block_info::block_info): New constructor.
10227 2017-06-07 Carl Love <cel@us.ibm.com>
10229 * config/rs6000/rs6000-c: The return type of the following
10230 built-in functions was implemented as int not long long. Fix sign
10231 of return value for the unsigned version of vec_mulo and vec_mule.
10232 vector unsigned long long vec_bperm (vector unsigned long long,
10233 vector unsigned char)
10234 vector signed long long vec_mule (vector signed int,
10236 vector unsigned long long vec_mule (vector unsigned int,
10237 vector unsigned int)
10238 vector signed long long vec_mulo (vector signed int,
10240 vector unsigned long long vec_mulo (vector unsigned int,
10241 vector unsigned int)
10242 * doc/extend.texi: Fix the documentation for the built-in
10245 2017-06-07 Carl Love <cel@us.ibm.com>
10248 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10251 2017-06-07 Carl Love <cel@us.ibm.com>
10253 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10254 support, Generate doublehv for signed int/float for BE case only.
10256 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10258 * doc/invoke.texi (mcx16): Rewrite.
10260 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10262 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10263 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10264 *mov<mode>_softfloat, and an anonymous splitter): Use
10265 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10267 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10269 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10271 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10272 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10273 (rs6000_debug_reg_global): Adjust.
10274 (rs6000_init_hard_regno_mode_ok): Adjust.
10275 (rs6000_dbx_register_number): Adjust.
10276 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10277 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10278 Remove SPE_ACC and SPEFSCR.
10279 (REG_ALLOC_ORDER): Ditto.
10280 (FRAME_POINTER_REGNUM): Change to 111.
10281 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10282 (REG_CLASS_NAMES): Ditto.
10283 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10284 (REGISTER_NAMES): Ditto.
10285 (ADDITIONAL_REG_NAMES): Ditto.
10286 (rs6000_reg_names): Ditto.
10287 * config/rs6000/rs6000.md: Renumber some register number
10290 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10292 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10294 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10295 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10297 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10298 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10299 Delete the SPE high registers.
10300 (REG_ALLOC_ORDER): Ditto.
10301 (enum reg_class): Remove SPE_HIGH_REGS.
10302 (REG_CLASS_NAMES): Ditto.
10303 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10304 (REGISTER_NAMES): Ditto.
10305 (rs6000_reg_names): Ditto.
10306 * doc/tm.texi.in: Remove SPE as example.
10307 * doc/tm.texi: Regenerate.
10309 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10311 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10312 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10313 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10314 * config/rs6000/rs6000.md (type): Remove "brinc".
10316 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10318 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10319 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10320 * config/rs6000/linuxspe.h: Delete file.
10321 * config/rs6000/rs6000.md: Don't include spe.md.
10322 * config/rs6000/spe.h: Delete file.
10323 * config/rs6000/spe.md: Delete file.
10324 * config/rs6000/t-rs6000: Remove spe.md.
10326 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10328 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10329 (reg_or_none500mem_operand): Delete.
10330 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10331 instead of reg_or_none500mem_operand.
10333 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10336 handling of SPE flags.
10337 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10339 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10341 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10343 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10344 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10345 paired_divv2sf3): Similar.
10346 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10347 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10348 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10350 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10351 Rename the paired_* instruction patterns.
10352 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10354 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10355 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10356 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10357 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10358 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10359 PAIRED_VECTOR_MODE.
10360 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10361 (spe_func_has_64bit_regs_p): Delete.
10362 (spe_expand_predicate_builtin): Delete.
10363 (spe_expand_evsel_builtin): Delete.
10364 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10365 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10366 (invalid_e500_subreg): Delete.
10367 (rs6000_legitimize_address): Always force_reg op2 as well, for
10368 paired single memory accesses.
10369 (rs6000_member_type_forces_blk): Delete.
10370 (rs6000_spe_function_arg): Delete.
10371 (rs6000_expand_unop_builtin): Delete SPE handling.
10372 (rs6000_expand_binop_builtin): Ditto.
10373 (spe_expand_stv_builtin): Delete.
10374 (bdesc_2arg_spe): Delete.
10375 (spe_expand_builtin): Delete.
10376 (spe_expand_predicate_builtin): Delete.
10377 (spe_expand_evsel_builtin): Delete.
10378 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10379 (spe_init_builtins): Delete.
10380 (spe_func_has_64bit_regs_p): Delete.
10381 (savres_routine_name): Delete "info" parameter. Adjust callers.
10382 (rs6000_emit_stack_reset): Ditto.
10383 (rs6000_dwarf_register_span): Delete.
10384 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10385 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10386 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10388 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10390 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10391 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10392 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10393 mulv2sf3, divv2sf3): Delete expanders.
10395 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10397 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10399 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10401 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10402 * config/rs6000/rs6000.c: Ditto.
10404 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10406 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10407 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10408 comparison_operator.
10410 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10412 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10413 * config/rs6000/rs6000.opt: Ditto.
10414 * config/rs6000/t-rtems: Ditto.
10416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10418 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10419 TARGET_E500_SINGLE by 0, simplify.
10420 * config/rs6000/rs6000.c: Ditto.
10421 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10422 (spe_build_register_parallel): Delete.
10423 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10424 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10425 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10426 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10427 (E500_CONVERT): Delete.
10428 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10430 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10432 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10433 * config/rs6000/dfp.md: Ditto.
10434 (negdd2, *negdd2_fpr): Merge.
10435 (absdd2, *absdd2_fpr): Merge.
10436 (negtd2, *negtd2_fpr): Merge.
10437 (abstd2, *abstd2_fpr): Merge.
10438 * config/rs6000/e500.h: Delete file.
10439 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10440 TARGET_FPRS by 1 and simplify.
10441 * config/rs6000/rs6000-c.c: Ditto.
10442 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10443 TARGET_DF_SPE by 0.
10444 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10446 * config/rs6000/rs6000.md: Ditto.
10447 (floatdidf2, *floatdidf2_fpr): Merge.
10448 (move_from_CR_gt_bit): Delete.
10449 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10450 (E500_CR_IOR_COMPARE): Delete.
10451 (All patterns that require !TARGET_FPRS): Delete.
10452 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10454 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10456 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10458 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10460 * graphds.c (add_edge): Intitialize edge's attached data.
10461 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10462 pointer parameter. Call pointed function on each edge during
10463 graph traversing. Skip traversing the edge when the function
10465 (graphds_dfs, graphds_scc): Ditto.
10466 (for_each_edge): New parameter. Pass the new parameter to callback
10468 * graphds.h (skip_edge_callback): New function pointer type.
10469 (graphds_dfs, graphds_scc): New function pointer parameter.
10470 (graphds_edge_callback, for_each_edge): New parameter.
10472 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10474 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10475 out code checking if runtime alias check is possible to below ...
10476 Call the new function.
10477 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10478 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10480 2017-06-07 Marek Polacek <polacek@redhat.com>
10483 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10484 TYPE_OVERFLOW_WRAPS check.
10486 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10488 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10489 if versioning is required.
10490 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10491 peeling with the check for versioning.
10493 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10495 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10496 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10497 Set true to new parameter if new ssa variable is defined.
10498 (vect_gen_vector_loop_niters): Refactor. Set range information
10499 for the new vector loop bound variable.
10500 (vect_do_peeling): Ditto.
10502 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10504 * tree-affine.c (ssa.h): Include header file.
10505 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10506 has wrapping overflow behavior.
10508 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10510 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10512 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10514 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10515 (tree_to_aff_combination): ... here.
10517 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10519 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10520 reg_pressure model function.
10521 (ivopts_global_cost_for_size): Delete.
10522 (determine_set_costs, iv_ca_recount_cost): Call new model function
10523 ivopts_estimate_reg_pressure.
10525 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10527 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10528 expensive than udiv. Remove floating point cases from mod.
10530 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10532 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10533 Increase idiv cost.
10535 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10537 * config/aarch64/aarch64.md
10538 (copysignsf3): Fix mask generation.
10540 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10542 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10544 (class dump_manager): Add register_dumps method.
10545 * dumpfile.c: Include langhooks.h.
10546 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10547 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10548 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10549 (dump_manager::dump_register): Start with 512 entries instead of 32.
10550 (dump_manager::register_dumps): New method.
10551 * toplev.c (general_init): Instead of invoking register_dumps
10552 langhook, invoke register_dumps method on the dump manager.
10553 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10556 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10558 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10559 Say that # requires an associated define_split to exist, and that
10560 the define_split must be suitable for use after register allocation.
10562 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10564 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10565 (compute_outgoing_frequencies): Also initialize zero counts.
10566 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10567 around loops; preserve more of profile when nothing changes.
10569 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10571 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10573 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10574 * config/arm/arm-cpu-cdata.h: Regenerate.
10575 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10576 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10577 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10578 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10579 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10581 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10583 * config/arm/t-rmprofile: Likewise.
10584 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10586 2017-06-06 David S. Miller <davem@davemloft.net>
10589 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10590 blockage if function uses alloca.
10592 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10594 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10595 New "uid" fields to hold pretty-print IDs of group and ref.
10596 Memory references are now identified as <group_id>:<ref_id>
10597 instead of using [random] addresses.
10598 (dump_mem_details): Simplify, no functional change.
10599 (dump_mem_ref): Simplify and make output more concise.
10600 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10601 (find_or_create_group): Initialize group uid.
10602 (record_ref): Initialize ref uid. Improve debug output.
10603 (prune_group_by_reuse, should_issue_prefetch_p,)
10604 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10605 (mark_nontemporal_store, determine_loop_nest_reuse):
10606 Improve debug output.
10608 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10610 * dbgcnt.def (prefetch): New debug counter.
10611 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10612 (schedule_prefetches): Stop issueing prefetches if debug counter
10615 2017-06-06 Tom de Vries <tom@codesourcery.com>
10617 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10618 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10620 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10622 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10623 Use aarch64_reg_or_zero predicate for operand 4.
10624 (aarch64_compare_and_swap<mode> define_insn_and_split):
10625 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10626 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10628 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10630 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10631 (arm_compute_save_core_reg_mask): This.
10632 (thumb1_compute_save_reg_mask): Rename into ...
10633 (thumb1_compute_save_core_reg_mask): This.
10634 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10635 (arm_compute_frame_layout): Likewise.
10637 2017-06-06 Richard Biener <rguenther@suse.de>
10639 PR tree-optimization/80974
10640 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10641 keep or clear leaders SSA info.
10643 2017-06-06 Tom de Vries <tom@codesourcery.com>
10645 * config/nvptx/nvptx.c (split_mode_p): New function.
10646 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10648 2017-06-06 Tom de Vries <tom@codesourcery.com>
10650 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10652 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10655 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10658 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10660 * shrink-wrap.c (handle_simple_exit): Update profile.
10661 (try_shrink_wrapping): Upate profile.
10663 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10665 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10666 (tree_guess_outgoing_edge_probabilities): New.
10667 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10668 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10670 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10672 * ipa-split.c (split_function): Initialize return bb profile.
10674 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10676 * profile.c (compute_branch_probabilities): Also initialize
10677 EXIT_BLOCK profile.
10679 2017-06-06 Richard Biener <rguenther@suse.de>
10681 PR tree-optimization/80928
10682 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10683 (vect_analyze_loop_operations): Properly guard analysis for
10685 (vect_transform_loop): Likewise.
10686 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10687 (vect_model_induction_cost): Do not cost for pure SLP.
10688 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10689 of induction in inner loop vectorization.
10690 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10691 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10692 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10694 (vect_analyze_slp_cost_1): Cost induction.
10695 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10696 (vect_get_slp_vect_defs): Likewise.
10697 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10698 (vect_transform_stmt): Handle SLP reductions.
10699 * tree-vectorizer.h (vectorizable_induction): Adjust.
10701 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10703 * config/rs6000/rs6000.c (make_resolver_func): Update
10704 init_lowered_empty_function call.
10706 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10708 * doc/invoke.texi: Document the -fprofile-abs-path option.
10709 * common.opt (fprofile-abs-path): New option.
10710 * gcov-io.h (gcov_write_filename): Declare.
10711 * gcov-io.c (gcov_write_filename): New function.
10712 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10713 * profile.c (output_location): Likewise.
10715 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10717 * shring-wrap.c: Revert accidental commit.
10719 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10721 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10723 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10725 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10727 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10728 profile in callgraph edge.
10729 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10730 (apply_scale): Likewise.
10731 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10732 Also scale profile when inlining function with zero profile.
10733 (initialize_cfun): Update exit block profile even when it is zero.
10734 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10735 when profile is read.
10737 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10739 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10740 (CLONE_*): New constants to define the processors we can generate
10741 code for with the target_clone attribute.
10742 (rs6000_clone_map): New array to identify which clone processors
10743 the current program is running on.
10744 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10745 target_clone attribute.
10746 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10747 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10748 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10749 (cpu_expand_builtin): Add support for target_clone attribute.
10750 (rs6000_valid_attribute_p): Allow "default" attribute.
10751 (get_decl_name): New debug function to simplify printing the
10752 current function name in debugging statements.
10753 (rs6000_clone_priority): New functions to support the target_clone
10754 attribute, and be able to generate code to switch between ISA 2.05
10755 through ISA 3.0 (power6 through power9).
10756 (rs6000_compare_version_priority): Likewise.
10757 (rs6000_get_function_versions_dispatcher): Likewise.
10758 (make_resolver_func): Likewise.
10759 (add_condition_to_bb): Likewise.
10760 (dispatch_function_versions): Likewise.
10761 (rs6000_generate_version_dispatcher_body): Likewise.
10762 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10763 (fusion_gpr_load_p): Fix a spacing issue.
10764 * doc/extend.texi (Common Function Attributes): Document that the
10765 PowerPC supports the target_clone attribute.
10767 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10769 * config/arm/arm.h: explain F symbol found in description of ARM
10770 register allocation in its legend.
10772 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10774 * config/mips/frame-header-opt.c: Include profile-count.h.
10775 * config/riscv/riscv.c: Include profile-count.h
10777 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10779 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10781 (sm_set_flag_if_changed): Add bbs field.
10782 (execute_sm_if_changed_flag_set): Pass BBS.
10783 (execute_sm): Update.
10785 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10787 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10790 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10792 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10793 (peephole2): New peephole2 to emit the above.
10794 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10796 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10798 * config/aarch64/aarch64.c (define_peephole2 above
10799 *sub_<shift>_<mode>): New peephole.
10801 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10803 * config/i386/i386.c (make_resolver_func): Update.
10804 * Makefile.in: Add profile-count.h and profile-count.o
10805 * auto-profile.c (afdo_indirect_call): Update to new API.
10806 (afdo_set_bb_count): Update.
10807 (afdo_propagate_edge): Update.
10808 (afdo_propagate_circuit): Update.
10809 (afdo_calculate_branch_prob): Update.
10810 (afdo_annotate_cfg): Update.
10811 * basic-block.h: Include profile-count.h
10812 (struct edge_def): Turn count to profile_count.
10813 (struct basic_block_def): Likewie.
10814 (REG_BR_PROB_BASE): Move to profile-count.h
10815 (RDIV): Move to profile-count.h
10816 * bb-reorder.c (max_entry_count): Turn to profile_count.
10817 (find_traces): Update.
10818 (rotate_loop):Update.
10819 (connect_traces):Update.
10820 (sanitize_hot_paths):Update.
10821 * bt-load.c (migrate_btr_defs): Update.
10822 * cfg.c (RDIV): Remove.
10823 (init_flow): Use alloc_block.
10824 (alloc_block): Uninitialize count.
10825 (unchecked_make_edge): Uninitialize count.
10826 (check_bb_profile): Update.
10827 (dump_edge_info): Update.
10828 (dump_bb_info): Update.
10829 (update_bb_profile_for_threading): Update.
10830 (scale_bbs_frequencies_int): Update.
10831 (scale_bbs_frequencies_gcov_type): Update.
10832 (scale_bbs_frequencies_profile_count): New.
10833 * cfg.h (update_bb_profile_for_threading): Update.
10834 (scale_bbs_frequencies_profile_count): Declare.
10835 * cfgbuild.c (compute_outgoing_frequencies): Update.
10836 (find_many_sub_basic_blocks): Update.
10837 * cfgcleanup.c (try_forward_edges): Update.
10838 (try_crossjump_to_edge): Update.
10839 * cfgexpand.c (expand_gimple_tailcall): Update.
10840 (construct_exit_block): Update.
10841 * cfghooks.c (verify_flow_info): Update.
10842 (dump_bb_for_graph): Update.
10843 (split_edge): Update.
10844 (make_forwarder_block): Update.
10845 (duplicate_block): Update.
10846 (account_profile_record): Update.
10847 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10848 (get_estimated_loop_iterations): Update.
10849 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10850 (single_likely_exit): Update.
10851 * cfgloopmanip.c (scale_loop_profile): Update.
10853 (set_zero_probability): Update.
10854 (lv_adjust_loop_entry_edge): Update.
10855 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10856 (purge_dead_edges): Update.
10857 (rtl_account_profile_record): Update.
10858 * cgraph.c (cgraph_node::create): Uninitialize count.
10859 (symbol_table::create_edge): Uninitialize count.
10860 (cgraph_update_edges_for_call_stmt_node): Update.
10861 (cgraph_edge::dump_edge_flags): Update.
10862 (cgraph_node::dump): Update.
10863 (cgraph_edge::maybe_hot_p): Update.
10864 * cgraph.h: Include profile-count.h
10865 (create_clone), create_edge, create_indirect_edge): Update.
10866 (cgraph_node): Turn count to profile_count.
10867 (cgraph_edge0: Likewise.
10868 (make_speculative, clone): Update.
10869 (create_edge): Update.
10870 (init_lowered_empty_function): Update.
10871 * cgraphclones.c (cgraph_edge::clone): Update.
10872 (duplicate_thunk_for_node): Update.
10873 (cgraph_node::create_clone): Update.
10874 * cgraphunit.c (cgraph_node::analyze): Update.
10875 (cgraph_node::expand_thunk): Update.
10876 * final.c (dump_basic_block_info): Update.
10877 * gimple-streamer-in.c (input_bb): Update.
10878 * gimple-streamer-out.c (output_bb): Update.
10879 * graphite.c (print_global_statistics): Update.
10880 (print_graphite_scop_statistics): Update.
10881 * hsa-brig.c: Include basic-block.h.
10882 * hsa-dump.c: Include basic-block.h.
10883 * hsa-gen.c (T sum_slice): Update.
10884 (convert_switch_statements):Update.
10885 * hsa-regalloc.c: Include basic-block.h.
10886 * ipa-chkp.c (chkp_produce_thunks): Update.
10887 * ipa-cp.c (struct caller_statistics): Update.
10888 (init_caller_stats): Update.
10889 (gather_caller_stats): Update.
10890 (ipcp_cloning_candidate_p): Update.
10891 (good_cloning_opportunity_p): Update.
10892 (get_info_about_necessary_edges): Update.
10893 (dump_profile_updates): Update.
10894 (update_profiling_info): Update.
10895 (update_specialized_profile): Update.
10896 (perhaps_add_new_callers): Update.
10897 (decide_about_value): Update.
10898 (ipa_cp_c_finalize): Update.
10899 * ipa-devirt.c (struct odr_type_warn_count): Update.
10900 (struct decl_warn_count): Update.
10901 (struct final_warning_record): Update.
10902 (possible_polymorphic_call_targets): Update.
10903 (ipa_devirt): Update.
10904 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10905 * ipa-icf.c (sem_function::merge): Update.
10906 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10907 * ipa-inline.c (compute_uninlined_call_time): Update.
10908 (compute_inlined_call_time): Update.
10909 (want_inline_small_function_p): Update.
10910 (want_inline_self_recursive_call_p): Update.
10911 (edge_badness): Update.
10912 (lookup_recursive_calls): Update.
10913 (recursive_inlining): Update.
10914 (inline_small_functions): Update.
10915 (dump_overall_stats): Update.
10916 (dump_inline_stats): Update.
10917 * ipa-profile.c (ipa_profile_generate_summary): Update.
10918 (ipa_propagate_frequency): Update.
10919 (ipa_profile): Update.
10920 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10921 * ipa-utils.c (ipa_merge_profiles): Update.
10922 * loop-doloop.c (doloop_modify): Update.
10923 * loop-unroll.c (report_unroll): Update.
10924 (unroll_loop_runtime_iterations): Update.
10925 * lto-cgraph.c (lto_output_edge): Update.
10926 (lto_output_node): Update.
10927 (input_node): Update.
10928 (input_edge): Update.
10929 (merge_profile_summaries): Update.
10930 * lto-streamer-in.c (input_cfg): Update.
10931 * lto-streamer-out.c (output_cfg): Update.
10932 * mcf.c (create_fixup_graph): Update.
10933 (adjust_cfg_counts): Update.
10934 (sum_edge_counts): Update.
10935 * modulo-sched.c (sms_schedule): Update.
10936 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10937 * predict.c (maybe_hot_count_p): Update.
10938 (probably_never_executed): Update.
10939 (dump_prediction): Update.
10940 (combine_predictions_for_bb): Update.
10941 (propagate_freq): Update.
10942 (handle_missing_profiles): Update.
10943 (counts_to_freqs): Update.
10944 (rebuild_frequencies): Update.
10945 (force_edge_cold): Update.
10946 * predict.h: Include profile-count.h
10947 (maybe_hot_count_p, counts_to_freqs): UPdate.
10948 * print-rtl-function.c: Do not include cfg.h
10949 * print-rtl.c: Include basic-block.h
10950 * profile-count.c: New file.
10951 * profile-count.h: New file.
10952 * profile.c (is_edge_inconsistent): Update.
10953 (correct_negative_edge_counts): Update.
10954 (is_inconsistent): Update.
10955 (set_bb_counts): Update.
10956 (read_profile_edge_counts): Update.
10957 (compute_frequency_overlap): Update.
10958 (compute_branch_probabilities): Update; Initialize and deinitialize
10960 (branch_prob): Update.
10961 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10962 (edge_gcov_count): New.
10963 (bb_gcov_count): New.
10964 * shrink-wrap.c (try_shrink_wrapping): Update.
10965 * tracer.c (better_p): Update.
10966 * trans-mem.c (expand_transaction): Update.
10967 (ipa_tm_insert_irr_call): Update.
10968 (ipa_tm_insert_gettmclone_call): Update.
10969 * tree-call-cdce.c: Update.
10970 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10971 (gimple_duplicate_sese_tail): Update.
10972 (gimple_account_profile_record): Update.
10973 (execute_fixup_cfg): Update.
10974 * tree-inline.c (copy_bb): Update.
10975 (copy_edges_for_bb): Update.
10976 (initialize_cfun): Update.
10977 (freqs_to_counts): Update.
10978 (copy_cfg_body): Update.
10979 (expand_call_inline): Update.
10980 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10981 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10982 (try_unroll_loop_completely): Update.
10983 (try_peel_loop): Update.
10984 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10985 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10986 * tree-ssa-loop-split.c (connect_loops): Update.
10987 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10988 * tree-ssa-reassoc.c (branch_fixup): Update.
10989 * tree-ssa-tail-merge.c (replace_block_by): Update.
10990 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10991 (compute_path_counts): Update.
10992 (update_profile): Update.
10993 (recompute_probabilities): Update.
10994 (update_joiner_offpath_counts): Update.
10995 (estimated_freqs_path): Update.
10996 (freqs_to_counts_path): Update.
10997 (clear_counts_path): Update.
10998 (ssa_fix_duplicate_block_edges): Update.
10999 (duplicate_thread_path): Update.
11000 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11001 (struct switch_conv_info): Update.
11002 * tree-tailcall.c (decrease_profile): Update.
11003 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11004 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11005 * value-prof.c (check_counter): Update.
11006 (gimple_divmod_fixed_value): Update.
11007 (gimple_mod_pow2): Update.
11008 (gimple_mod_subtract): Update.
11009 (gimple_ic_transform): Update.
11010 (gimple_stringop_fixed_value): Update.
11011 * value-prof.h (gimple_ic): Update.
11013 2017-06-02 Carl Love <cel@us.ibm.com>
11015 * config/rs6000/rs6000-c: Add support for built-in functions
11016 vector double vec_doublee (vector signed int);
11017 vector double vec_doublee (vector unsigned int);
11018 vector double vec_doublee (vector float);
11019 vector double vec_doubleh (vector signed int);
11020 vector double vec_doubleh (vector unsigned int);
11021 vector double vec_doubleh (vector float);
11022 vector double vec_doublel (vector signed int);
11023 vector double vec_doublel (vector unsigned int);
11024 vector double vec_doublel (vector float);
11025 vector double vec_doubleo (vector signed int);
11026 vector double vec_doubleo (vector unsigned int);
11027 vector double vec_doubleo (vector float);.
11028 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11029 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11031 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11032 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11033 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11035 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11036 vec_doublel, vec_doubleh.
11037 * doc/extend.texi: Update the built-in documentation file for the
11038 new built-in functions.
11040 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11043 * ipa-inline-analysis.c (free_growth_caches): Set
11044 edge_removal_hook_holder to NULL after removing it.
11046 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11048 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11049 comparision with zero.
11051 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11052 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11053 for early expansion of vec_min and vec_max builtins.
11054 (builtin_function_type): Add min/max unsigned variants to those
11055 identified as having unsigned arguments.
11057 2017-06-02 Olivier Hainque <hainque@adacore.com>
11059 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11061 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11063 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11064 Use VALL_F16 iterator rather than VALL.
11066 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11068 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11069 Emit CBNZ inside loop when doing a strong exchange and comparing
11070 against zero. Generate the CC flags after the loop.
11072 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11074 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11075 (dl_section_ref): New.
11076 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11077 On AIX, append an expression to subtract the size of the
11078 section length to dl_section_ref.
11080 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11082 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11083 for early expansion of vector absolute builtins.
11085 2017-06-02 Richard Biener <rguenther@suse.de>
11087 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11088 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11090 2017-06-02 Richard Biener <rguenther@suse.de>
11092 PR tree-optimization/80948
11093 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11094 stmt order as well.
11096 2017-06-02 Richard Biener <rguenther@suse.de>
11098 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11100 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11101 for inductions as relevant.
11103 2017-06-02 Richard Biener <rguenther@suse.de>
11105 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11106 (vectorizable_induction): ... this. Remove dead code.
11108 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11110 * builtins. (expand_builtin_alloca): Remove second parameter and
11111 infer its value from the first parameter instead.
11112 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11114 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11116 PR rtl-optimization/80903
11117 * loop-doloop.c (add_test): Unshare sequence.
11119 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11121 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11123 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11125 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11127 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11128 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11129 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11130 (xlogue_layout::get_stub_rtx): Make static.
11131 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11132 (xlogue_layout::compute_stub_managed_regs): Rename to...
11133 (xlogue_layout::count_stub_managed_regs): ...this.
11134 (xlogue_layout::is_stub_managed_reg): New function.
11135 (xlogue_layout::m_stub_names): Rename to...
11136 (xlogue_layout::s_stub_names): ...this, make static.
11137 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11138 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11139 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11140 xlogue_layout::s_stub_names): Instantiate statics.
11141 (stub_managed_regs): Remove.
11142 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11143 (disable_call_ms2sysv_xlogues): Rename to...
11144 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11145 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11147 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11148 change after reload_completed.
11149 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11151 (ix86_expand_prologue): Likewise.
11152 (ix86_expand_epilogue): Likewise.
11153 (ix86_expand_split_stack_prologue): Likewise.
11154 (ix86_compute_frame_layout): Remove frame parameter ...
11155 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11156 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11158 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11159 (ix86_frame): Move from here ...
11160 * config/i386/i386.h (ix86_frame): ... to here.
11161 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11162 complete ix86_frame data structure instead. Remove some_ld_name.
11164 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11166 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11167 symbols that hold a DECL_VALUE_EXPR.
11169 2017-06-01 Martin Jambor <mjambor@suse.cz>
11171 PR tree-optimization/80898
11172 * tree-sra.c (process_subtree_disqualification): Removed.
11173 (disqualify_candidate): Do not acll
11174 process_subtree_disqualification.
11175 (subtree_mark_written_and_enqueue): New function.
11176 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11177 RHS has been disqualified and re-queue LHS if necessary. Apart
11178 from that, ignore disqualified RHS.
11180 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11182 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11183 address fetch for z10 or later.
11185 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11187 * config/arc/arc.md (tst_movb): Add guard when splitting.
11189 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11191 * config/arc/arc.c (arc_can_eliminate): Test against
11192 arc_frame_pointer_needed.
11194 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11196 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11197 to prevent store reordering.
11198 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11199 (type): Add block type.
11200 (stack_tie): Define special instruction to be used in
11203 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11205 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11206 constraint. It is not valid for the pattern.
11207 (noncommutative_binary_comparison): Likewise.
11209 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11211 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11214 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11216 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11217 be used by the reg-alloc.
11219 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11221 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11222 reg-alloc when having mul64 or mul32x16 instructions.
11223 (mulsidi3): Likewise.
11224 (umulsidi3): Likewise.
11225 (mulsi32x16): New pattern.
11226 (mulsi64): Likewise.
11227 (mulsidi64): Likewise.
11228 (umulsidi64): Likewise.
11229 (MUL32x16_REG): Define.
11230 (mul64_600): Use MUL32x16_REG.
11231 (mac64_600): Likewise.
11232 (umul64_600): Likewise.
11233 (umac64_600): Likewise.
11235 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11237 * config/arc/arc.md (mulsi3_700): Make it commutative.
11239 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11241 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11243 (*sign_extendsidi2_insn): Likewise for movstosw.
11245 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11247 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11248 the type of the input discriminant value. Convert the
11249 discriminant value of signedness vary.
11251 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11253 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11254 Add to -Wall section.
11256 2017-06-01 Richard Biener <rguenther@suse.de>
11258 PR middle-end/66313
11259 * fold-const.c (fold_plusminus_mult_expr): If the factored
11260 factor may be zero use a wrapping type for the inner operation.
11261 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11262 and handle moved defs.
11263 (process_assignment): Properly guard the unary op case. Return a
11264 tri-state indicating that moving the stmt before the call may allow
11265 to continue. Pass through to_move.
11266 (find_tail_calls): Handle moving unrelated defs before
11269 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11272 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11273 splitter result in the canonical way.
11275 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11277 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11278 also for 32bit target. Update insn attributes.
11279 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11281 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11283 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11284 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11285 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11286 (_mm_maskz_min_ss): New intrinsics.
11288 2017-05-31 Martin Liska <mliska@suse.cz>
11290 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11291 Change comment style to one we normally use.
11292 (vectorizable_reduction): Likewise.
11293 (vectorizable_induction): Likewise.
11294 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11295 (vectorizable_call): Likewise.
11296 (vectorizable_simd_clone_call): Likewise.
11297 (vectorizable_conversion): Likewise.
11298 (vectorizable_assignment): Likewise.
11299 (vectorizable_shift): Likewise.
11300 (vectorizable_operation): Likewise.
11301 (vectorizable_store): Likewise.
11302 (vectorizable_load): Likewise.
11303 * tree-vectorizer.h: Likewise.
11305 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11307 * passes.c (emergency_dump_function): New.
11308 * tree-pass.h (emergency_dump_function): Declare.
11309 * plugin.c (plugins_internal_error_function): Remove.
11310 * plugin.h (plugins_internal_error_function): Remove declaration.
11311 * toplev.c (internal_error_function): New static function. Use it...
11312 (general_init): ...here.
11314 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11316 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11317 (arc_rtx_costs): Add costs for new patterns.
11318 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11319 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11321 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11323 * tree-ssa-strlen.c (get_next_strinfo): New function.
11324 (get_stridx_plus_constant): Use it.
11325 (zero_length_string): Likewise.
11326 (adjust_related_strinfos): Likewise.
11327 (adjust_last_stmt): Likewise.
11329 2017-05-31 Richard Biener <rguenther@suse.de>
11332 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11333 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11335 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11337 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11338 loop_vinfo argument and use of dependence distance vectors.
11339 Check instead whether the two references differ only in their
11340 initial value and assume that they have the same alignment if the
11341 difference is a multiple of the vector alignment.
11342 (vect_analyze_data_refs_alignment): Update call accordingly.
11344 2017-05-31 Martin Liska <mliska@suse.cz>
11347 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11349 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11351 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11352 (create_intersect_range_checks): Move from ...
11353 * tree-data-ref.c (create_intersect_range_checks_index)
11354 (create_intersect_range_checks): ... to here.
11355 (create_runtime_alias_checks): New function factored from ...
11356 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11357 here. Call above function.
11358 * tree-data-ref.h (create_runtime_alias_checks): New function.
11360 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11362 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11363 segment length for dr_b and compute it in wide_int.
11365 2017-05-31 Richard Biener <rguenther@suse.de>
11367 PR tree-optimization/80906
11368 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11369 and pass through iv_map.
11370 (copy_bb_and_scalar_dependences): Adjust.
11371 (translate_pending_phi_nodes): Likewise.
11372 (copy_loop_close_phi_args): Handle code-generating IVs instead
11375 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11377 * diagnostic-color.c (color_dict): Add "type-diff".
11378 (parse_gcc_colors): Update comment.
11379 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11380 -fdiagnostics-show-template-tree and -fno-elide-type.
11381 (GCC_COLORS): Add type-diff to example.
11383 (-fdiagnostics-show-template-tree): New.
11384 (-fno-elide-type): New.
11385 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11386 the pp_format_decoder callback. Call any m_format_postprocessor's
11388 (pretty_printer::pretty_printer): Initialize
11389 m_format_postprocessor.
11390 (pretty_printer::~pretty_printer): Delete any
11391 m_format_postprocessor.
11392 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11393 (class format_postprocessor): New class.
11394 (struct pretty_printer::format_decoder): Document the new parameters.
11395 (struct pretty_printer::m_format_postprocessor): New field.
11396 * tree-diagnostic.c (default_tree_printer): Update for new
11397 bool and const char ** params.
11398 * tree-diagnostic.h (default_tree_printer): Likewise.
11400 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11402 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11403 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11405 rs6000_gen_cell_microcode code.
11406 (rs6000_final_prescan_insn): Delete.
11407 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11408 "warn-cell-microcode" entries.
11409 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11410 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11411 throughout. Change cc_reg_not_micro_cr0_operand to
11412 cc_reg_not_cr0_operand throughout.
11413 (*extendhi<mode>2_noload): Delete.
11414 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11415 (mwarn-cell-microcode): Delete.
11416 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11417 -mgen-cell-microcode and -mwarn-cell-microcode.
11419 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11422 * config/i386/constraints.md (Yd): New constraint.
11424 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11425 and (?Yd, r) alternatives. Update insn attributes.
11426 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11427 and (?*Yd, r) alternatives. Update insn attributes.
11428 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11430 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11432 * gimplify.c (gimplify_modify_expr): Don't create a
11433 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11436 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11438 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11440 2017-05-30 Richard Biener <rguenther@suse.de>
11442 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11443 and reduc_def fields.
11444 (STMT_VINFO_REDUC_TYPE): New define.
11445 (STMT_VINFO_REDUC_DEF): Likewise.
11446 (vect_force_simple_reduction): Adjust prototype.
11447 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11448 (vect_is_simple_reduction): Remove check_reduction argument.
11449 (vect_force_simple_reduction): Adjust and set
11450 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11451 (vectorizable_reduction): Do not re-do reduction analysis
11452 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11453 * tree-parloops.c (gather_scalar_reductions): Adjust.
11455 2017-05-30 Richard Biener <rguenther@suse.de>
11457 PR middle-end/80901
11458 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11461 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11463 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11464 Introduce unknown_misalignment parameter and remove vf.
11465 (vect_peeling_hash_get_lowest_cost):
11466 Pass unknown_misalignment parameter.
11467 (vect_enhance_data_refs_alignment):
11468 Fix unsupportable data ref treatment.
11470 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11472 * tree-vect-data-refs.c (vect_get_data_access_cost):
11473 Workaround for SLP handling.
11474 (vect_enhance_data_refs_alignment):
11475 Compute costs for doing no peeling at all, compare to the best
11476 peeling costs so far and avoid peeling if cheaper.
11478 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11480 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11481 Return peeling info and set costs to zero for unlimited cost
11483 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11484 with unknown misalignment. Compute and costs for unknown
11485 misalignment, compare them to the costs for known misalignment
11486 and choose the cheapest for peeling.
11488 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11490 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11491 (vect_get_peeling_costs_all_drs): Create function.
11492 (vect_peeling_hash_get_lowest_cost):
11493 Use vect_get_peeling_costs_all_drs.
11494 (vect_peeling_supportable): Create function.
11495 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11497 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11499 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11500 DR_HAS_NEGATIVE_STEP.
11501 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11502 (vect_enhance_data_refs_alignment): Use.
11503 (vect_duplicate_ssa_name_ptr_info): Use.
11504 * tree-vectorizer.h (dr_misalignment): Use.
11505 (known_alignment_for_access_p): Use.
11507 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11510 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11512 (has_section_name): New function.
11514 2017-05-30 Martin Liska <mliska@suse.cz>
11517 * auto-profile.c (get_function_decl_from_block): Fix
11520 2017-05-30 Richard Biener <rguenther@suse.de>
11522 PR middle-end/80876
11523 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11525 2017-05-30 Martin Liska <mliska@suse.cz>
11527 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11528 * dumpfile.h (struct dump_file_info): Remove ctors.
11530 2017-05-30 Martin Liska <mliska@suse.cz>
11532 * predict.def: Fix GNU coding style.
11534 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11536 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11537 Mark 'to' argument with ATTRIBUTE_UNUSED.
11539 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11541 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11542 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11543 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11546 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11548 * doc/install.texi (Options specification): Restore entry of
11549 --enable-sjlj-exceptions.
11551 2017-05-27 Michael Eager <eager@eagercon.com>
11554 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11556 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11558 * config/microblaze/microblaze.h
11559 (FIXED_REGISTERS): Update in macro.
11560 (CALL_USED_REGISTERS): Update in macro.
11562 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11564 * doc/install.texi: Add links to macOS binary distributions.
11566 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11570 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11572 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11574 2017-05-26 Martin Liska <mliska@suse.cz>
11576 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11578 2017-05-26 Martin Liska <mliska@suse.cz>
11580 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11581 always leading ';; '.
11582 (dump_bb_info): Likewise.
11583 (brief_dump_cfg): Likewise.
11584 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11585 * dumpfile.c: Remove usage of TDF_VERBOSE.
11586 * dumpfile.h (enum dump_kind): Likewise.
11587 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11588 * print-tree.c (debug_verbose): Remove.
11589 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11590 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11591 * tree-diagnostic.c (default_tree_printer): Replace
11592 TDF_DIAGNOSTIC with TDF_SLIM.
11594 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11596 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11597 in parameter loop, rather than loop_vinfo.
11598 (create_intersect_range_checks): Ditto.
11599 (vect_create_cond_for_alias_checks): Update call to above functions.
11601 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11603 PR tree-optimization/80815
11604 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11605 for merging runtime alias checks. Handle negative DR_STEPs.
11607 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11609 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11611 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11612 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11613 out code pruning runtime alias checks.
11614 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11615 factored out from above.
11616 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11618 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11620 (prune_runtime_alias_test_list): New decalaration.
11622 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11624 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11625 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11626 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11627 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11628 (operator==, comp_dr_with_seg_len_pair): Ditto.
11629 (vect_prune_runtime_alias_test_list): Ditto.
11631 2017-05-26 Martin Liska <mliska@suse.cz>
11634 * params.def: Bound partial-inlining-entry-probability param.
11636 2017-05-26 Marek Polacek <polacek@redhat.com>
11639 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11642 2017-05-26 Richard Biener <rguenther@suse.de>
11644 PR tree-optimization/80842
11645 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11648 2017-05-26 Richard Biener <rguenther@suse.de>
11650 PR tree-optimization/80844
11651 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11653 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11655 * doc/md.texi (Machine Constraints): Update x86 family
11656 machine constraints section to match 'config/i386/constraints.md'.
11658 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11660 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11662 2017-05-25 Nathan Sidwell <nathan@acm.org>
11664 * doc/invoke.texi (--enable-languages): Update documentation.
11666 2017-05-25 Martin Liska <mliska@suse.cz>
11668 * dumpfile.c: Add TDF_FOLDING.
11669 * dumpfile.h (enum dump_kind): Likewise.
11670 * genmatch.c (dt_simplify::gen_1): Use it.
11672 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11674 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11676 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11678 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11679 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11681 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11683 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11684 * match.pd (X == C): Rewrite it here.
11685 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11686 with_certain_nonzero_bits2): New predicates.
11687 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11689 2017-05-24 Nathan Sidwell <nathan@acm.org>
11691 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11694 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11697 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11699 * config/powerpcspe: New port. Files are copied from the rs6000
11700 port, with "rs6000" in filenames replaced by "powerpcspe".
11702 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11704 PR rtl-optimization/80754
11705 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11707 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11709 * config/sparc/sparc.md (length): Return the correct value for -mflat
11710 sibcalls to match output_sibcall.
11712 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11716 * config/rs6000/rs6000.c (struct machine_function): Add new field
11718 (rs6000_get_separate_components): Init that field, use it.
11719 (rs6000_components_for_bb): Use the field.
11721 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11723 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11725 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11727 PR middle-end/80823
11728 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11730 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11733 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11734 address against address_operand predicate.
11735 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11737 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11739 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11740 parameters passed indirectly.
11742 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11744 * config/i386/i386.md (*movdi_internal): Remove SSE4
11745 alternative 18 (?r, *v). Update insn attributes.
11746 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11747 Update insn attributes.
11748 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11749 Update insn attributes.
11750 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11751 alternative 1 (r, v). Remove isa attribute.
11752 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11753 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11754 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11756 2017-05-23 Tom de Vries <tom@codesourcery.com>
11758 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11761 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11763 * cgraphunit.c (symbol_table::process_new_functions): Update.
11764 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11765 (inline_generate_summary): Rename to ...
11766 (ipa_fn_summary_generate): ... this one.
11767 (inline_read_summary): Rename to ...
11768 (ipa_fn_summary_read): ... this one.
11769 (inline_write_summary): Rename to ...
11770 (ipa_fn_summary_write): ... this one.
11771 (inline_free_summary): Rename to ...
11772 (ipa_free_fn_summary): ... this one.
11773 (pass_data_local_fn_summary, pass_local_fn_summary,
11774 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11775 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11776 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11777 make_pass_ipa_fn_summary): New.
11778 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11779 inline_write_summary, inline_free_summary): Remove.
11780 (ipa_free_fn_summary) : New.
11781 * ipa-inline.c (ipa_inline): Update.
11782 (pass_ipa_inline): Do not generate summaries.
11783 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11785 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11786 and add pass_ipa_fn_summary.
11787 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11789 (make_pass_inline_parameters): Remove.
11791 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11793 * omp-low.c (struct omp_context): Remove "default_kind" member.
11796 * omp-offload.c (execute_oacc_device_lower): Remove the
11797 parallelism dimensions function attributes for unparallelized
11798 OpenACC kernels constructs.
11800 2017-05-23 Martin Liska <mliska@suse.cz>
11802 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11804 (cgraph_edge::make_speculative): Likewise.
11805 (cgraph_edge::resolve_speculation): Likewise.
11806 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11807 (cgraph_node::dump): Likewise.
11808 * cgraph.h: Likewise.
11809 * cgraphunit.c (analyze_functions): Likewise.
11810 (symbol_table::compile): Likewise.
11811 * ipa-cp.c (print_all_lattices): Likewise.
11812 (determine_versionability): Likewise.
11813 (initialize_node_lattices): Likewise.
11814 (ipcp_verify_propagated_values): Likewise.
11815 (estimate_local_effects): Likewise.
11816 (update_profiling_info): Likewise.
11817 (create_specialized_node): Likewise.
11818 (perhaps_add_new_callers): Likewise.
11819 (decide_about_value): Likewise.
11820 (decide_whether_version_node): Likewise.
11821 (identify_dead_nodes): Likewise.
11822 (ipcp_store_bits_results): Likewise.
11823 * ipa-devirt.c (dump_targets): Likewise.
11824 (ipa_devirt): Likewise.
11825 * ipa-icf.c (sem_item::dump): Likewise.
11826 (sem_function::equals): Likewise.
11827 (sem_variable::equals): Likewise.
11828 (sem_item_optimizer::read_section): Likewise.
11829 (sem_item_optimizer::execute): Likewise.
11830 (congruence_class::dump): Likewise.
11831 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11832 (dump_inline_summary): Likewise.
11833 (estimate_node_size_and_time): Likewise.
11834 (inline_analyze_function): Likewise.
11835 * ipa-inline-transform.c (inline_call): Likewise.
11836 * ipa-inline.c (report_inline_failed_reason): Likewise.
11837 (want_early_inline_function_p): Likewise.
11838 (edge_badness): Likewise.
11839 (update_edge_key): Likewise.
11840 (inline_small_functions): Likewise.
11841 * ipa-profile.c (ipa_profile): Likewise.
11842 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11843 (ipa_make_edge_direct_to_target): Likewise.
11844 (remove_described_reference): Likewise.
11845 (ipa_impossible_devirt_target): Likewise.
11846 (propagate_controlled_uses): Likewise.
11847 (ipa_print_node_params): Likewise.
11848 (ipcp_transform_function): Likewise.
11849 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11850 (propagate_pure_const): Likewise.
11851 * ipa-reference.c (generate_summary): Likewise.
11852 (read_write_all_from_decl): Likewise.
11853 (propagate): Likewise.
11854 (ipa_reference_read_optimization_summary): Likewise.
11855 * ipa-utils.c (ipa_merge_profiles): Likewise.
11856 * ipa.c (walk_polymorphic_call_targets): Likewise.
11857 (symbol_table::remove_unreachable_nodes): Likewise.
11858 (ipa_single_use): Likewise.
11859 * passes.c (execute_todo): Likewise.
11860 * predict.c (drop_profile): Likewise.
11861 * symtab.c (symtab_node::get_dump_name): New function.
11862 (symtab_node::dump_name): Likewise.
11863 (symtab_node::dump_asm_name): Likewise.
11864 (symtab_node::dump_references): Likewise.
11865 (symtab_node::dump_referring): Likewise.
11866 (symtab_node::dump_base): Likewise.
11867 (symtab_node::debug_symtab): Likewise.
11868 * tree-sra.c (convert_callers_for_node): Likewise.
11869 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11870 * value-prof.c (init_node_map): Likewise.
11872 2017-05-23 Martin Liska <mliska@suse.cz>
11874 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11875 and symtab_node::debug_symtab to symbol_table::debug.
11876 * cgraphunit.c (analyze_functions): Use the renamed function.
11877 (symbol_table::compile): Likewise.
11878 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11879 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11880 * passes.c (execute_todo): Likewise.
11881 * symtab.c (symbol_table::dump): New function.
11882 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11884 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11886 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11887 that nonconst implies exec.
11889 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11891 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11892 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11893 (inline_edge_summary_vec): Turn into ...
11894 (ipa_call_summaries): ... this one.
11895 (redirect_to_unreachable, edge_set_predicate,
11896 evaluate_properties_for_edge, inline_summary_alloc,
11897 reset_ipa_call_summary, reset_inline_summary,
11898 inline_summary_t::duplicate): Update.
11899 (inline_edge_duplication_hook): Turn to ...
11900 (ipa_call_summary_t::duplicate): ... this one.
11901 (inline_edge_removal_hook): Turn to ...
11902 (ipa_call_summary_t::remove): ... this one.
11903 (dump_inline_edge_summary): Turn to ...
11904 (dump_ipa_call_summary): ... this one.
11905 (estimate_function_body_sizes): Update.
11906 (inline_update_callee_summaries): Update.
11907 (remap_edge_change_prob): Update.
11908 (remap_edge_summaries): Update.
11909 (inline_merge_summary): Update.
11910 (do_estimate_edge_time): Update.
11911 (inline_generate_summary): Update.
11912 (inline_read_section): Update.
11913 (inline_read_summary): Update.
11914 (inline_free_summary): Update.
11915 * ipa-inline.c (can_inline_edge_p): Update.
11916 (compute_inlined_call_time): Update.
11917 (want_inline_small_function_p): Update.
11918 (edge_badness): Update.
11919 (early_inliner): Update.
11920 * ipa-inline.h (inline_edge_summary): Turn to ...
11921 (ipa_call_summary): ... this one.
11922 (ipa_call_summary_t): New class.
11923 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11924 (ipa_call_summaries): New.
11925 (inline_edge_summary): Remove.
11926 (estimate_edge_growth): Update.
11927 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11928 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11929 * ipa-split.c (execute_split_functions): Update.
11930 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11932 2017-05-23 Tom de Vries <tom@codesourcery.com>
11934 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11935 attributes): Document rdrand effective target.
11937 2017-05-23 Tom de Vries <tom@codesourcery.com>
11939 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11940 attributes): Sort alphabetically.
11942 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11944 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11946 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11949 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11950 V2DF/V2DI splat into two separate patterns, one that handles
11951 registers, and the other that only handles memory. Drop support
11952 for splatting from a GPR on ISA 2.07 and then splitting the
11953 splat into direct move and splat.
11954 (vsx_splat_<mode>_reg): Likewise.
11955 (vsx_splat_<mode>_mem): Likewise.
11957 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11959 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11961 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11963 PR middle-end/80809
11964 * omp-low.c (finish_taskreg_remap): New function.
11965 (finish_taskreg_scan): If unit size of ctx->record_type
11966 is non-constant, unshare the size expression and replace
11967 decls in it with possible outer var refs.
11969 PR middle-end/80809
11970 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11971 GOVD_SHARED rather than GOVD_PRIVATE with it.
11972 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11973 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11975 PR middle-end/80853
11976 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11977 as last argument to build_outer_var_ref for pointer bases of array
11978 section reductions.
11980 2017-05-19 Martin Sebor <msebor@redhat.com>
11982 * print-tree.c (print_node): Print DECL_READ_P flag.
11984 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11986 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11987 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11988 * cgraph.c: Likewise.
11989 * cgraphunit.c: Likewise.
11990 * gengtype.c: Likewise.
11991 * ipa-cp.c: Likewise.
11992 * ipa-devirt.c: Likewise.
11993 * ipa-icf.c: Likewise.
11994 * ipa-predicate.c: Likewise.
11995 * ipa-profile.c: Likewise.
11996 * ipa-prop.c: Likewise.
11997 * ipa-split.c: Likewise.
11999 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12000 edge_predicate_pool, dump_inline_hints,
12001 inline_summary::account_size_time, redirect_to_unreachable,
12002 edge_set_predicate, set_hint_predicate,
12003 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12004 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12005 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12006 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12007 ipa_call_summary_t::remove, initialize_growth_caches,
12008 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12009 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12010 mark_modified, unmodified_parm_1, unmodified_parm,
12011 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12012 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12013 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12014 will_be_nonconstant_predicate, record_modified_bb_info,
12015 get_minimal_bb, record_modified, param_change_prob,
12016 phi_result_unknown_predicate, predicate_for_phi_result,
12017 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12018 estimate_function_body_sizes, compute_inline_parameters,
12019 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12020 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12021 inline_update_callee_summaries, remap_edge_change_prob,
12022 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12023 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12024 inline_analyze_function, inline_summary_t::insert,
12025 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12026 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12027 inline_free_summary): Move to ipa-fnsummary.h
12028 (predicate_t): Remove.
12029 * ipa-fnsummary.c: New file.
12030 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12031 (enum inline_hints_vals, inline_hints, agg_position_info,
12032 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12033 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12034 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12035 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12036 inline_read_summary, inline_write_summary, inline_free_summary,
12037 inline_analyze_function, initialize_inline_failed,
12038 inline_merge_summary, inline_update_overall_summary,
12039 compute_inline_parameters): Move to ipa-fnsummary.h
12040 * ipa-fnsummary.h: New file.
12041 * ipa-inline-transform.h: Include ipa-inline.h.
12042 * ipa-inline.c: LIkewise.
12044 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12046 * ipa-inline.c (edge_badness): Use inlined_time instead of
12047 inline_summaries->get.
12049 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12051 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12053 2017-05-22 Nathan Sidwell <nathan@acm.org>
12055 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12056 (fdump-lang): Document 'raw' option.
12057 * dumpfile.h (TDI_tu): Delete.
12058 * dumpfile.c (dump_files): Remove translation-unit.
12059 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12061 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12063 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12064 command option from $(AWK) call.
12065 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12066 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12067 [FORMAT]: Remove handling of variable.
12068 * config/avr/t-multilib: Regenerate.
12070 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12072 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12074 (dump_inline_summary): Do not print self_time.
12075 (estimate_function_body_sizes): Do not set self_time.
12076 (compute_inline_parameters): Likewise.
12077 (inline_read_section, inline_write_summary): Do not stream self_time.
12078 * ipa-inline.h (inline_summary): Drop self_time.
12080 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12082 * ipa-inline-analysis.c (account_size_time): Rename to ...
12083 (inline_summary::account_size_time): ... this one.
12084 (reset_ipa_call_summary): Turn to ...
12085 (ipa_call_summary::reset): ... this one.
12086 (reset_inline_summary): Turn to ...
12087 (inline_summary::reset): ... this one.
12088 (inline_summary_t::remove): Update.
12089 (inline_summary_t::duplicate): Update.
12090 (ipa_call_summary_t::remove): Update.
12091 (dump_inline_summary): Update.
12092 (estimate_function_body_sizes): Update.
12093 (compute_inline_parameters): Update.
12094 (estimate_node_size_and_time): Update.
12095 (inline_merge_summary): Update.
12096 (inline_update_overall_summary): Update.
12097 (inline_read_section): Update.
12098 (inline_write_summary): Update.
12099 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12100 add account_size_time and reset member functions.
12101 (ipa_call_summary): Add reset function.
12102 * ipa-predicate.h (predicate::operator &): Constify.
12104 2017-05-22 Richard Biener <rguenther@suse.de>
12106 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12108 2017-05-19 Jason Merrill <jason@redhat.com>
12110 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12112 2017-05-19 Marek Polacek <polacek@redhat.com>
12115 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12116 TYPE_OVERFLOW_WRAPS checks.
12118 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12120 * tree-core.h (enum omp_clause_default_kind): Add
12121 "OMP_CLAUSE_DEFAULT_PRESENT".
12122 * tree-pretty-print.c (dump_omp_clause): Handle it.
12123 * gimplify.c (enum gimplify_omp_var_data): Add
12124 "GOVD_MAP_FORCE_PRESENT".
12125 (gimplify_adjust_omp_clauses_1): Map it to
12126 "GOMP_MAP_FORCE_PRESENT".
12127 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12129 * gimplify.c (oacc_default_clause): Clarify.
12131 2017-05-19 Nathan Sidwell <nathan@acm.org>
12133 LANG_HOOK_REGISTER_DUMPS
12134 * toplev.c (general_init): Call register dump lang hook.
12135 * doc/invoke.texi: Document -fdump-lang option family.
12136 * dumpfile.c (dump_files): Remove class dump here.
12137 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12138 * dumpfile.h (tree_dump_index): Remove TDI_class.
12139 * langhooks-def.h (lhd_register_dumps): Declare.
12140 (LANG_HOOKS_REGISTER_DUMPS): Define.
12141 (LANG_HOOKS_INITIALIZER): Add it.
12142 * langhooks.c (lhd_register_dumps): Define.
12143 * langhooks.h (struct lang_hooks): Add register_dumps.
12145 2017-05-19 Nathan Sidwell <nathan@acm.org>
12147 * context.h (context::set_passes): New.
12148 * context.c (context::context): Do not create pass manager.
12149 * toplev.c (general_init): Create pass manager here.
12151 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12153 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12154 use this splitter if two add or or instructions would also work for
12155 the constant we want to generate.
12157 2017-05-19 Richard Biener <rguenther@suse.de>
12160 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12161 predicate evaluation.
12163 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12165 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12167 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12170 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12172 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12173 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12174 (inline_edge_summary_vec): Turn into ...
12175 (ipa_call_summaries): ... this one.
12176 (redirect_to_unreachable, edge_set_predicate,
12177 evaluate_properties_for_edge, inline_summary_alloc,
12178 reset_ipa_call_summary, reset_inline_summary,
12179 inline_summary_t::duplicate): Update.
12180 (inline_edge_duplication_hook): Turn to ...
12181 (ipa_call_summary_t::duplicate): ... this one.
12182 (inline_edge_removal_hook): Turn to ...
12183 (ipa_call_summary_t::remove): ... this one.
12184 (dump_inline_edge_summary): Turn to ...
12185 (dump_ipa_call_summary): ... this one.
12186 (estimate_function_body_sizes): Update.
12187 (inline_update_callee_summaries): Update.
12188 (remap_edge_change_prob): Update.
12189 (remap_edge_summaries): Update.
12190 (inline_merge_summary): Update.
12191 (do_estimate_edge_time): Update.
12192 (inline_generate_summary): Update.
12193 (inline_read_section): Update.
12194 (inline_read_summary): Update.
12195 (inline_free_summary): Update.
12196 * ipa-inline.c (can_inline_edge_p): Update.
12197 (compute_inlined_call_time): Update.
12198 (want_inline_small_function_p): Update.
12199 (edge_badness): Update.
12200 (early_inliner): Update.
12201 * ipa-inline.h (inline_edge_summary): Turn to ...
12202 (ipa_call_summary): ... this one.
12203 (ipa_call_summary_t): New class.
12204 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12205 (ipa_call_summaries): New.
12206 (inline_edge_summary): Remove.
12207 (estimate_edge_growth): Update.
12208 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12209 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12210 * ipa-split.c (execute_split_functions): Update.
12211 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12213 2017-05-19 Richard Biener <rguenther@suse.de>
12215 PR middle-end/80764
12216 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12218 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12220 * config/rs6000/rs6000.c (struct machine_function): Add field
12221 fpr_is_wrapped_separately.
12222 (rs6000_get_separate_components): Use 64 components. Handle the
12223 new FPR components.
12224 (rs6000_components_for_bb): Handle the FPR components.
12225 (rs6000_emit_prologue_components): Handle the FPR components.
12226 (rs6000_emit_epilogue_components): Handle the FPR components.
12227 (rs6000_set_handled_components): Handle the FPR components.
12228 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12229 that are already separately shrink-wrapped.
12230 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12231 that are already separately shrink-wrapped.
12233 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12236 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12239 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12240 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12241 cases where the register allocator uses a move and an offsettable
12242 memory operation to/from a FPR register on ISA 2.06/2.07.
12243 (define_peephole2 for Altivec d-form store): Likewise.
12245 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12248 * config/i386/mmx.md (*mov<mode>_internal): Enable
12249 alternatives 11, 12, 13 and 14 also for 32bit targets.
12250 Remove alternatives 15, 16, 17 and 18.
12251 * config/i386/sse.md (vec_concatv2di): Change
12252 alternative (!x, *y) to (x, ?!*Yn).
12254 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12256 * dumpfile.h (enum dump_kind): Remove stray comma.
12258 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12260 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12261 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12262 predicate::num_conditions
12263 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12264 (CHANGED): turn into predicate::changed.
12265 (agg_position_info): Move to ipa-predicate.h
12266 (add_condition, predicate::add_clause, predicate::operator &=,
12267 predicate::or_with, predicate::evaluate, predicate::probability,
12268 dump_condition, dump_clause, predicate::dump,
12269 predicate::remap_after_duplication, predicate::remap_after_inlining,
12270 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12271 (evaluate_conditions_for_known_args): Update.
12272 (set_cond_stmt_execution_predicate): Update.
12273 * ipa-inline.h: Include ipa-predicate.h
12274 (condition, inline_param_summary, conditions, agg_position_info,
12275 predicate): Move to ipa-predicate.h
12276 * ipa-predicate.c: New file.
12277 * ipa-predicate.h: New file.
12279 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12281 * final.c (leaf_function_p): Check we are not in a sequence.
12283 2017-05-18 Martin Liska <mliska@suse.cz>
12285 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12286 * dumpfile.c (dump_register): Use new enum dump_kind.
12287 (get_dump_file_name): Likewise.
12288 (dump_enable_all): Likewise.
12289 (dump_switch_p_1): Likewise.
12290 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12291 * dumpfile.h (enum dump_kind): New enum type.
12292 (struct dump_file_info): Create constructor and
12293 format fields and comments.
12294 * passes.c (pass_manager::register_one_dump_file):
12296 * statistics.c (statistics_early_init): Likewise.
12297 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12298 TDF_TREE with TDF_SLIM.
12299 (gather_memory_references_ref): Likewise.
12301 2017-05-18 Martin Liska <mliska@suse.cz>
12303 * vec.h (struct vnull): Use it.
12305 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12307 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12308 (true_predicate, false_predicate, true_predicate_p,
12309 false_predicate_p): Remove.
12310 (single_cond_predicate, not_inlined_predicate): Turn to member function
12312 (add_condition): Update.
12313 (add_clause): Turn to...
12314 (predicate::add_clause): ... this one; update; allow passing NULL
12316 (and_predicates): Turn to ...
12317 (predicate::operator &=): ... this one.
12318 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12319 (or_predicates): Turn to ...
12320 (predicate::or_with): ... this one.
12321 (evaluate_predicate): Turn to ...
12322 (predicate::evaluate): ... this one.
12323 (predicate_probability): Turn to ...
12324 (predicate::probability): ... this one.
12325 (dump_condition): Update.
12326 (dump_predicate): Turn to ...
12327 (predicate::dump): ... this one.
12328 (account_size_time): Update.
12329 (edge_set_predicate): Update.
12330 (set_hint_predicate): UPdate.
12331 (evaluate_conditions_for_known_args): Update.
12332 (evaluate_properties_for_edge): Update.
12333 (remap_predicate_after_duplication): Turn to...
12334 (predicate::remap_after_duplication): ... this one.
12335 (remap_hint_predicate_after_duplication): Update.
12336 (inline_summary_t::duplicate): UPdate.
12337 (dump_inline_edge_summary): Update.
12338 (dump_inline_summary): Update.
12339 (set_cond_stmt_execution_predicate): Update.
12340 (set_switch_stmt_execution_predicate): Update.
12341 (compute_bb_predicates): Update.
12342 (will_be_nonconstant_expr_predicate): Update.
12343 (will_be_nonconstant_predicate): Update.
12344 (phi_result_unknown_predicate): Update.
12345 (predicate_for_phi_result): Update.
12346 (array_index_predicate): Update.
12347 (estimate_function_body_sizes): Update.
12348 (estimate_node_size_and_time): Update.
12349 (estimate_ipcp_clone_size_and_time): Update.
12350 (remap_predicate): Rename to ...
12351 (predicate::remap_after_inlining): ... this one.
12352 (remap_hint_predicate): Update.
12353 (inline_merge_summary): Update.
12354 (inline_update_overall_summary): Update.
12355 (estimate_size_after_inlining): Update.
12356 (read_predicate): Rename to ...
12357 (predicate::stream_in): ... this one.
12358 (read_inline_edge_summary): Update.
12359 (write_predicate): Rename to ...
12360 (predicate::stream_out): ... this one.
12361 (write_inline_edge_summary): Update.
12362 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12363 (clause_t): Turn to uint32_t
12364 (predicate): Turn to class; implement constructor and operators
12366 (size_time_entry): Update.
12367 (inline_summary): Update.
12368 (inline_edge_summary): Update.
12370 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12372 * fold-const.c (fold_binary_loc): Move transformation...
12373 * match.pd (C - X CMP X): ... here.
12375 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12377 * config/sparc/sparc.c (sparc_option_override): Set function
12378 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12379 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12381 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12383 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12385 2017-05-18 Marek Polacek <polacek@redhat.com>
12388 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12389 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12390 gimple_assign_load_p.
12392 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12394 PR middle-end/80692
12395 * real.c (do_compare): Give decimal_do_compare preference over
12396 comparing just the signs.
12398 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12400 * doc/md.texi (Canonicalization of Instructions): Describe the
12401 canonical form of instructions that inherently set a condition
12404 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12406 PR middle-end/80775
12407 * tree-cfg.c: Move deletion of unreachable case statements to after
12408 the merging of consecutive case labels.
12410 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12412 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12413 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12414 restoring of callee-saved registers.
12416 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12418 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12419 * config/visium/visium.c (single_set_and_flags): Likewise.
12420 * config/visium/visium.md (Substitutions): Likewise.
12422 2017-05-17 Martin Liska <mliska@suse.cz>
12424 * cfg.c: Introduce dump_flags_t type and
12425 use it instead of int type.
12427 * cfghooks.c: Likewise.
12428 * cfghooks.h (struct cfg_hooks): Likewise.
12429 * cfgrtl.c: Likewise.
12430 * cfgrtl.h: Likewise.
12431 * cgraph.c (cgraph_node::get_body): Likewise.
12432 * coretypes.h: Likewise.
12433 * domwalk.c: Likewise.
12434 * domwalk.h: Likewise.
12435 * dumpfile.c (struct dump_option_value_info): Likewise.
12436 (dump_enable_all): Likewise.
12437 (dump_switch_p_1): Likewise.
12438 (opt_info_switch_p): Likewise.
12439 * dumpfile.h (enum tree_dump_index): Likewise.
12440 (struct dump_file_info): Likewise.
12441 * genemit.c: Likewise.
12442 * generic-match-head.c: Likewise.
12443 * gengtype.c (open_base_files): Likewise.
12444 * gimple-pretty-print.c: Likewise.
12445 * gimple-pretty-print.h: Likewise.
12446 * graph.c (print_graph_cfg): Likewise.
12447 * graphite-scop-detection.c (dot_all_sese): Likewise.
12448 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12449 * loop-unroll.c (report_unroll): Likewise.
12450 * passes.c (pass_manager::register_one_dump_file): Likewise.
12451 * print-tree.c: Likewise.
12452 * statistics.c: Likewise.
12453 * tree-cfg.c: Likewise.
12454 * tree-cfg.h: Likewise.
12455 * tree-dfa.c: Likewise.
12456 * tree-dfa.h: Likewise.
12457 * tree-dump.c (dump_function): Likewise.
12458 * tree-dump.h (struct dump_info): Likewise.
12459 * tree-pretty-print.c: Likewise.
12460 * tree-pretty-print.h: Likewise.
12461 * tree-ssa-live.c: Likewise.
12462 * tree-ssa-live.h: Likewise.
12463 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12464 * tree-vect-loop.c: Likewise.
12465 * tree-vect-slp.c: Likewise.
12467 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12468 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12470 PR tree-optimization/80457
12471 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12472 of all arguments to a statement as scalar_to_vec operations.
12473 (vectorizable_call): Adjust call to vect_model_simple_cost for
12475 (vectorizable_conversion): Likewise.
12476 (vectorizable_assignment): Likewise.
12477 (vectorizable_shift): Likewise.
12478 (vectorizable_operation): Likewise.
12479 (vectorizable_comparison): Likewise.
12480 (vect_is_simple_cond): Record the def types for operands.
12481 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12482 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12483 for statement argument count.
12485 2017-05-16 Carl Love <cel@us.ibm.com>
12487 * config/rs6000/rs6000-c: Add support for built-in functions
12488 vector unsigned long long vec_bperm (vector unsigned long long,
12489 vector unsigned char)
12490 vector signed long long vec_mule (vector signed int,
12492 vector unsigned long long vec_mule (vector unsigned int,
12493 vector unsigned int)
12494 vector signed long long vec_mulo (vector signed int,
12496 vector unsigned long long vec_mulo (vector unsigned int,
12497 vector unsigned int)
12498 vector signed char vec_sldw (vector signed char,
12499 vector signed char,
12501 vector unsigned char vec_sldw (vector unsigned char,
12502 vector unsigned char,
12504 vector signed short vec_sldw (vector signed short,
12505 vector signed short,
12507 vector unsigned short vec_sldw (vector unsigned short,
12508 vector unsigned short,
12510 vector signed int vec_sldw (vector signed int,
12513 vector unsigned int vec_sldw (vector unsigned int,
12514 vector unsigned int,
12516 vector signed long long vec_sldw (vector signed long long,
12517 vector signed long long,
12519 vector unsigned long long vec_sldw (vector unsigned long long,
12520 vector unsigned long long,
12522 * config/rs6000/rs6000-c: Add support for built-in functions
12523 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12524 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12525 * doc/extend.texi: Update the built-in documentation for the
12526 new built-in functions.
12528 2017-05-16 Marek Polacek <polacek@redhat.com>
12532 * tree.c (save_expr): Don't fold the expression.
12534 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12536 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12537 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12538 and (?*y,m). Update insn attributes.
12540 2017-05-16 Martin Liska <mliska@suse.cz>
12542 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12543 flags argument of print_gimple_stmt, print_gimple_expr,
12544 print_generic_stmt and print_generic_expr.
12545 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12546 * coretypes.h: Likewise.
12547 * except.c (dump_eh_tree): Likewise.
12548 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12549 * gimple-pretty-print.h: Likewise.
12550 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12551 (backprop::push_to_worklist): Likewise.
12552 (backprop::pop_from_worklist): Likewise.
12553 (backprop::process_use): Likewise.
12554 (backprop::intersect_uses): Likewise.
12555 (note_replacement): Likewise.
12556 * gimple-ssa-store-merging.c
12557 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12558 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12559 (pass_store_merging::execute): Likewise.
12560 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12561 (ssa_base_cand_dump_callback): Likewise.
12562 (dump_incr_vec): Likewise.
12563 (replace_refs): Likewise.
12564 (replace_mult_candidate): Likewise.
12565 (create_add_on_incoming_edge): Likewise.
12566 (create_phi_basis): Likewise.
12567 (insert_initializers): Likewise.
12568 (all_phi_incrs_profitable): Likewise.
12569 (introduce_cast_before_cand): Likewise.
12570 (replace_one_candidate): Likewise.
12571 * gimplify.c (gimplify_expr): Likewise.
12572 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12573 (set_rename): Likewise.
12574 (rename_uses): Likewise.
12575 (copy_loop_phi_nodes): Likewise.
12576 (add_close_phis_to_merge_points): Likewise.
12577 (copy_loop_close_phi_args): Likewise.
12578 (copy_cond_phi_args): Likewise.
12579 (graphite_copy_stmts_from_block): Likewise.
12580 (translate_pending_phi_nodes): Likewise.
12581 * graphite-poly.c (print_pdr): Likewise.
12582 (dump_gbb_cases): Likewise.
12583 (dump_gbb_conditions): Likewise.
12584 (print_scop_params): Likewise.
12585 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12586 (build_cross_bb_scalars_use): Likewise.
12587 (gather_bbs::before_dom_children): Likewise.
12588 * hsa-dump.c (dump_hsa_immed): Likewise.
12589 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12590 (get_replacement_map): Likewise.
12591 * ipa-inline-analysis.c (dump_condition): Likewise.
12592 (estimate_function_body_sizes): Likewise.
12593 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12594 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12595 * ipa-prop.c (ipa_dump_param): Likewise.
12596 (ipa_print_node_jump_functions_for_edge): Likewise.
12597 (ipa_modify_call_arguments): Likewise.
12598 (ipa_modify_expr): Likewise.
12599 (ipa_dump_param_adjustments): Likewise.
12600 (ipa_dump_agg_replacement_values): Likewise.
12601 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12602 * ipa-pure-const.c (check_stmt): Likewise.
12603 (pass_nothrow::execute): Likewise.
12604 * ipa-split.c (execute_split_functions): Likewise.
12605 * omp-offload.c (dump_oacc_loop_part): Likewise.
12606 (dump_oacc_loop): Likewise.
12607 * trans-mem.c (tm_log_emit): Likewise.
12608 (tm_memopt_accumulate_memops): Likewise.
12609 (dump_tm_memopt_set): Likewise.
12610 (dump_tm_memopt_transform): Likewise.
12611 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12612 (print_loop): Likewise.
12613 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12614 (chkp_gather_checks_info): Likewise.
12615 (chkp_get_check_result): Likewise.
12616 (chkp_remove_check_if_pass): Likewise.
12617 (chkp_use_outer_bounds_if_possible): Likewise.
12618 (chkp_reduce_bounds_lifetime): Likewise.
12619 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12620 (chkp_mark_completed_bounds): Likewise.
12621 (chkp_register_incomplete_bounds): Likewise.
12622 (chkp_mark_invalid_bounds): Likewise.
12623 (chkp_maybe_copy_and_register_bounds): Likewise.
12624 (chkp_build_returned_bound): Likewise.
12625 (chkp_get_bound_for_parm): Likewise.
12626 (chkp_build_bndldx): Likewise.
12627 (chkp_get_bounds_by_definition): Likewise.
12628 (chkp_generate_extern_var_bounds): Likewise.
12629 (chkp_get_bounds_for_decl_addr): Likewise.
12630 * tree-chrec.c (chrec_apply): Likewise.
12631 * tree-data-ref.c (dump_data_reference): Likewise.
12632 (dump_subscript): Likewise.
12633 (dump_data_dependence_relation): Likewise.
12634 (analyze_overlapping_iterations): Likewise.
12635 * tree-inline.c (expand_call_inline): Likewise.
12636 (tree_function_versioning): Likewise.
12637 * tree-into-ssa.c (dump_defs_stack): Likewise.
12638 (dump_currdefs): Likewise.
12639 (dump_names_replaced_by): Likewise.
12640 (dump_update_ssa): Likewise.
12641 (update_ssa): Likewise.
12642 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12643 * tree-parloops.c (build_new_reduction): Likewise.
12644 (try_create_reduction_list): Likewise.
12645 (ref_conflicts_with_region): Likewise.
12646 (oacc_entry_exit_ok_1): Likewise.
12647 (oacc_entry_exit_single_gang): Likewise.
12648 * tree-pretty-print.h: Likewise.
12649 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12650 (get_scalar_evolution): Likewise.
12651 (add_to_evolution): Likewise.
12652 (get_loop_exit_condition): Likewise.
12653 (analyze_evolution_in_loop): Likewise.
12654 (analyze_initial_condition): Likewise.
12655 (analyze_scalar_evolution): Likewise.
12656 (instantiate_scev): Likewise.
12657 (number_of_latch_executions): Likewise.
12658 (gather_chrec_stats): Likewise.
12659 (final_value_replacement_loop): Likewise.
12660 (scev_const_prop): Likewise.
12661 * tree-sra.c (dump_access): Likewise.
12662 (disqualify_candidate): Likewise.
12663 (create_access): Likewise.
12664 (reject): Likewise.
12665 (maybe_add_sra_candidate): Likewise.
12666 (create_access_replacement): Likewise.
12667 (analyze_access_subtree): Likewise.
12668 (analyze_all_variable_accesses): Likewise.
12669 (sra_modify_assign): Likewise.
12670 (initialize_constant_pool_replacements): Likewise.
12671 (find_param_candidates): Likewise.
12672 (decide_one_param_reduction): Likewise.
12673 (replace_removed_params_ssa_names): Likewise.
12674 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12675 * tree-ssa-copy.c (dump_copy_of): Likewise.
12676 (copy_prop_visit_cond_stmt): Likewise.
12677 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12678 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12679 (record_equivalences_from_stmt): Likewise.
12680 * tree-ssa-dse.c (compute_trims): Likewise.
12681 (delete_dead_call): Likewise.
12682 (delete_dead_assignment): Likewise.
12683 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12684 (forward_propagate_into_cond): Likewise.
12685 (pass_forwprop::execute): Likewise.
12686 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12687 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12689 (move_computations_worker): Likewise.
12690 (execute_sm): Likewise.
12691 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12692 (remove_exits_and_undefined_stmts): Likewise.
12693 (remove_redundant_iv_tests): Likewise.
12694 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12695 (adjust_iv_update_pos): Likewise.
12696 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12697 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12698 (value_replacement): Likewise.
12699 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12700 * tree-ssa-pre.c (print_pre_expr): Likewise.
12701 (get_representative_for): Likewise.
12702 (create_expression_by_pieces): Likewise.
12703 (insert_into_preds_of_block): Likewise.
12704 (eliminate_insert): Likewise.
12705 (eliminate_dom_walker::before_dom_children): Likewise.
12706 (eliminate): Likewise.
12707 (remove_dead_inserted_code): Likewise.
12708 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12709 * tree-ssa-reassoc.c (get_rank): Likewise.
12710 (eliminate_duplicate_pair): Likewise.
12711 (eliminate_plus_minus_pair): Likewise.
12712 (eliminate_not_pairs): Likewise.
12713 (undistribute_ops_list): Likewise.
12714 (eliminate_redundant_comparison): Likewise.
12715 (update_range_test): Likewise.
12716 (optimize_range_tests_var_bound): Likewise.
12717 (optimize_vec_cond_expr): Likewise.
12718 (rewrite_expr_tree): Likewise.
12719 (rewrite_expr_tree_parallel): Likewise.
12720 (linearize_expr): Likewise.
12721 (break_up_subtract): Likewise.
12722 (linearize_expr_tree): Likewise.
12723 (attempt_builtin_powi): Likewise.
12724 (attempt_builtin_copysign): Likewise.
12725 (transform_stmt_to_copy): Likewise.
12726 (transform_stmt_to_multiply): Likewise.
12727 (dump_ops_vector): Likewise.
12728 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12729 (print_scc): Likewise.
12730 (set_ssa_val_to): Likewise.
12731 (visit_reference_op_store): Likewise.
12732 (visit_use): Likewise.
12733 (sccvn_dom_walker::before_dom_children): Likewise.
12734 (run_scc_vn): Likewise.
12735 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12737 (expr_hash_elt::print): Likewise.
12738 (const_and_copies::pop_to_marker): Likewise.
12739 (const_and_copies::record_const_or_copy_raw): Likewise.
12740 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12741 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12742 (dump_predicates): Likewise.
12743 (find_uninit_use): Likewise.
12744 (warn_uninitialized_phi): Likewise.
12745 (pass_late_warn_uninitialized::execute): Likewise.
12746 * tree-ssa.c (verify_vssa): Likewise.
12747 (verify_ssa): Likewise.
12748 (maybe_optimize_var): Likewise.
12749 * tree-vrp.c (dump_value_range): Likewise.
12750 (dump_all_value_ranges): Likewise.
12751 (dump_asserts_for): Likewise.
12752 (register_edge_assert_for_2): Likewise.
12753 (vrp_visit_cond_stmt): Likewise.
12754 (vrp_visit_switch_stmt): Likewise.
12755 (vrp_visit_stmt): Likewise.
12756 (vrp_visit_phi_node): Likewise.
12757 (simplify_cond_using_ranges_1): Likewise.
12758 (fold_predicate_in): Likewise.
12759 (evrp_dom_walker::before_dom_children): Likewise.
12760 (evrp_dom_walker::push_value_range): Likewise.
12761 (evrp_dom_walker::pop_value_range): Likewise.
12762 (execute_early_vrp): Likewise.
12764 2017-05-16 Richard Biener <rguenther@suse.de>
12766 * dwarf2out.c (loc_list_from_tree_1): Do not create
12767 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12769 2017-05-16 Richard Biener <rguenther@suse.de>
12771 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12773 (note_variable_value_in_expr): If we resolved the decl ref
12774 do not push to the stack.
12776 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12778 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12779 operations in fast-math mode.
12780 (vaddq_f16): Likewise.
12781 (vmul_f16): Likewise.
12782 (vmulq_f16): Likewise.
12783 (vsub_f16): Likewise.
12784 (vsubq_f16): Likewise.
12785 * config/arm/neon.md (add<mode>3): New.
12787 (fma:<VH:mode>3): New. Also remove outdated comment.
12790 2017-05-16 Martin Liska <mliska@suse.cz>
12794 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12795 (odr_types_equivalent_p): Likewise.
12797 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12799 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12801 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12804 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12805 non-interunit SSE move alternatives with '?'.
12806 (zero-extendsidi peephole2): New peephole to skip intermediate
12807 general register in SSE zero-extend sequence.
12809 2017-05-15 Jeff Law <law@redhat.com>
12811 * reorg.c (relax_delay_slots): Create a new variable to hold
12812 the temporary target rather than clobbering TARGET_LABEL.
12814 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12815 missing argument to extract_bit_field call.
12816 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12818 2017-05-15 Martin Liska <mliska@suse.cz>
12821 * gcc.c (process_command): Do not allow empty argument of -o option.
12823 2017-05-15 Renlin Li <renlin.li@arm.com>
12825 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12826 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12827 * config/aarch64/constraints.md (Usf): Add long call check.
12828 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12829 (call_value): Likewise.
12830 (sibcall): Likewise.
12831 (sibcall_value): Likewise.
12833 (call_value_insn): New.
12834 (sibcall_insn): Update rtx pattern.
12835 (sibcall_value_insn): Likewise.
12836 (call_internal): Remove.
12837 (call_value_internal): Likewise.
12838 (sibcall_internal): Likewise.
12839 (sibcall_value_internal): Likewise.
12840 (call_reg): Likewise.
12841 (call_symbol): Likewise.
12842 (call_value_reg): Likewise.
12843 (call_value_symbol): Likewise.
12845 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12848 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12850 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12852 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12853 compatible with CCGOCmode and with CCZmode.
12855 2017-05-14 Martin Sebor <msebor@redhat.com>
12857 PR middle-end/77671
12858 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12859 (gimple_fold_builtin_snprintf): Same.
12860 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12861 (gimple_fold_builtin_snprintf): Same.
12862 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12863 of character types.
12864 (is_call_safe): New function.
12865 (try_substitute_return_value): Call it.
12866 (try_simplify_call): New function.
12867 (pass_sprintf_length::handle_gimple_call): Call it.
12869 2017-05-14 Martin Sebor <msebor@redhat.com>
12871 PR middle-end/80669
12872 * builtins.c (expand_builtin_stpncpy): Simplify.
12874 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12876 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12877 * config/i386/i386.h
12878 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12879 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12880 (struct machine_function): Add new members call_ms2sysv,
12881 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12882 (struct machine_frame_state): New fields sp_realigned and
12883 sp_realigned_offset.
12884 * config/i386/i386.c
12885 (enum xlogue_stub): New enum.
12886 (enum xlogue_stub_sets): New enum.
12887 (class xlogue_layout): New class.
12888 (struct ix86_frame): New fields stack_realign_allocate_offset,
12889 stack_realign_offset and outlined_save_offset. Modify comments to
12890 detail stack layout when using out-of-line stubs.
12891 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12892 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12893 -mcall-ms2sysv-xlogues.
12894 (stub_managed_regs): New static variable.
12895 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12896 registers managed by out-of-line stub.
12897 (disable_call_ms2sysv_xlogues): New function.
12898 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12899 m->call_ms2sysv when appropriate and compute frame layout for
12901 (sp_valid_at, fp_valid_at): New inline functions.
12902 (choose_basereg): New function.
12903 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12905 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12906 Use align parameter of choose_baseaddr to generated aligned SSE movs
12908 (pro_epilogue_adjust_stack): Modify to track
12909 machine_frame_state::sp_realigned.
12910 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12911 (ix86_nsaved_sseregs): Likewise.
12912 (ix86_emit_save_regs): Likewise.
12913 (ix86_emit_save_regs_using_mov): Likewise.
12914 (ix86_emit_save_sse_regs_using_mov): Likewise.
12915 (get_scratch_register_on_entry): Likewise.
12916 (gen_frame_set): New function.
12917 (gen_frame_load): Likewise.
12918 (gen_frame_store): Likewise.
12919 (emit_outlined_ms2sysv_save): Likewise.
12920 (emit_outlined_ms2sysv_restore): Likewise.
12921 (ix86_expand_prologue): Modify stack re-alignment code and call
12922 emit_outlined_ms2sysv_save when appropriate.
12923 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12924 parameter rtx_insn *insn, which allows the function to be used to only
12925 generate the notes.
12926 (ix86_expand_epilogue): Modify validity checks of frame and stack
12927 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12928 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12929 * config/i386/predicates.md
12930 (save_multiple): New predicate.
12931 (restore_multiple): Likewise.
12932 * config/i386/sse.md
12933 (save_multiple<mode>): New pattern.
12934 (save_multiple_realign<mode>): Likewise.
12935 (restore_multiple<mode>): Likewise.
12936 (restore_multiple_and_return<mode>): Likewise.
12937 (restore_multiple_leave_return<mode>): Likewise.
12938 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12940 2017-05-14 Julia Koval <julia.koval@intel.com>
12942 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12943 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12944 (__builtin_ia32_xsetbv): New builtins.
12945 * config/i386/i386.c (ix86_expand_special_args_builtin):
12947 (ix86_expand_builtin): Special expand for new intrinsics.
12948 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12949 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12950 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12954 * cfganal.c (inverted_post_order_compute): Change argument type
12956 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12957 * df-core.c (rest_of_handle_df_initialize): Adjust.
12958 (rest_of_handle_df_finish): Likewise.
12959 (df_analyze_1): Likewise.
12960 (df_analyze): Likewise.
12961 (loop_inverted_post_order_compute): Change argument to be a vec *.
12962 (df_analyze_loop): Adjust.
12963 (df_get_n_blocks): Likewise.
12964 (df_get_postorder): Likewise.
12965 * df.h (struct df_d): Change field to be a vec.
12966 * lcm.c (compute_laterin): Adjust.
12967 (compute_available): Likewise.
12968 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12969 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12970 * tree-ssa-pre.c (compute_antic): Likewise.
12972 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12974 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12975 (depth_first_search::depth_first_search): Change structure init
12976 function to this constructor.
12977 (depth_first_search::add_bb): Rename function to this member.
12978 (depth_first_search::execute): Likewise.
12979 (flow_dfs_compute_reverse_finish): Adjust.
12981 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12983 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12984 (longest_simple_path): Likewise.
12985 * shrink-wrap.c (spread_components): Likewise.
12986 (disqualify_problematic_components): Likewise.
12987 (emit_common_heads_for_components): Likewise.
12988 (emit_common_tails_for_components): Likewise.
12989 (insert_prologue_epilogue_for_components): Likewise.
12991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12993 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12996 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12998 * df-core.c (df_set_blocks): Start using auto_bitmap.
12999 (df_compact_blocks): Likewise.
13000 * df-problems.c (df_rd_confluence_n): Likewise.
13001 * df-scan.c (df_insn_rescan_all): Likewise.
13002 (df_process_deferred_rescans): Likewise.
13003 (df_update_entry_block_defs): Likewise.
13004 (df_update_exit_block_uses): Likewise.
13005 (df_entry_block_bitmap_verify): Likewise.
13006 (df_exit_block_bitmap_verify): Likewise.
13007 (df_scan_verify): Likewise.
13008 * lra-constraints.c (lra_constraints): Likewise.
13009 (undo_optional_reloads): Likewise.
13010 (lra_undo_inheritance): Likewise.
13011 * lra-remat.c (calculate_gen_cands): Likewise.
13012 (do_remat): Likewise.
13013 * lra-spills.c (assign_spill_hard_regs): Likewise.
13014 (spill_pseudos): Likewise.
13015 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13016 (bitmap_set_subtract_values): Likewise.
13018 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13020 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13021 management with auto_bitmap.
13022 (fix_inter_tick): Likewise.
13023 (fix_recovery_deps): Likewise.
13024 * ira.c (add_store_equivs): Likewise.
13025 (find_moveable_pseudos): Likewise.
13026 (split_live_ranges_for_shrink_wrap): Likewise.
13027 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13028 (rtx_reuse_manager::seen_def_p): Likewise.
13029 (rtx_reuse_manager::set_seen_def): Likewise.
13030 * print-rtl.h (class rtx_reuse_manager): Likewise.
13032 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13034 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13036 (migrate_btr_def): Likewise.
13037 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13038 * df-core.c (loop_post_order_compute): Likewise.
13039 (loop_inverted_post_order_compute): Likewise.
13040 * hsa-common.h: Likewise.
13041 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13042 * init-regs.c (initialize_uninitialized_regs): Likewise.
13043 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13044 (inline_small_functions): Likewise.
13045 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13046 * ira.c (combine_and_move_insns): Likewise.
13047 (build_insn_chain): Likewise.
13048 * loop-invariant.c (find_invariants): Likewise.
13049 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13050 * predict.c (tree_predict_by_opcode): Likewise.
13051 (predict_paths_leading_to): Likewise.
13052 (predict_paths_leading_to_edge): Likewise.
13053 (estimate_loops_at_level): Likewise.
13054 (estimate_loops): Likewise.
13055 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13056 (spread_components): Likewise.
13057 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13058 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13059 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13060 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13061 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13062 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13063 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13064 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13065 (mark_threaded_blocks): Likewise.
13066 (thread_through_all_blocks): Likewise.
13067 * tree-ssa.c (verify_ssa): Likewise.
13068 (execute_update_addresses_taken): Likewise.
13069 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13071 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13073 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13075 (post_order_compute): Likewise.
13076 (inverted_post_order_compute): Likewise.
13077 (pre_and_rev_post_order_compute_fn): Likewise.
13079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13081 * genrecog.c (int_set::int_set): Explicitly construct our
13082 auto_vec base class.
13083 * vec.h (auto_vec::auto_vec): New constructor.
13085 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13087 * bitmap.h (class auto_bitmap): New constructor taking
13088 bitmap_obstack * argument.
13090 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13092 * bitmap.h (class auto_bitmap): Change type of m_bits to
13093 bitmap_head, and adjust ctor / dtor and member operators.
13095 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13097 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13098 when returned register mode doesn't match original mode.
13100 2017-05-12 Jeff Law <law@redhat.com>
13101 Jakub Jelinek <jakub@redhat.com>
13103 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13104 we look for cc setter after the compare-elim changes.
13105 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13106 within the vector to match what compare-elim now expects.
13107 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13108 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13110 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13111 after the compare-elim changes.
13112 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13113 the vector to match what compare-elim now expects.
13114 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13115 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13116 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13117 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13118 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13120 * config/visium/visium.c (single_set_and_flags): Fix where
13121 we look for cc setter after the compare-elim changes.
13122 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13123 with the vector to match what compare-elim now expects.
13124 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13125 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13126 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13127 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13128 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13129 (neg<mode>2_insn_set_overflow): Likewise.
13131 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13133 PR middle-end/79794
13134 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13135 maybe_expand_insn call, set ops[0].target. If still set after call,
13136 set alt_rtl. Add extra arg to recursive calls.
13137 (extract_bit_field): Add alt_rtl argument. Pass to
13139 * expmed.h (extract_bit_field): Fix prototype.
13140 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13141 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13142 to extract_bit_field_calls.
13143 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13144 Pass alt_rtl to extract_bit_field calls.
13145 * calls.c (store_unaligned_arguments_into_psuedos)
13146 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13147 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13149 * optabs.h (struct expand_operand): Add target bitfield.
13151 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13153 * compare-elim.c (try_eliminate_compare): Canonicalize
13154 operation with embedded compare to
13155 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13156 (set (reg) (operation)].
13158 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13160 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13163 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13164 cost of adding a carry flag for ADC instruction.
13165 [case MINUS]: Ignore the cost of subtracting a carry flag
13166 for SBB instruction.
13168 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13170 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13172 * config/rs6000/bmiintrin.h: New file.
13173 * config/rs6000/bmi2intrin.h: New file.
13174 * config/rs6000/x86intrin.h: New file.
13176 2017-05-12 Jeff Law <law@redhat.com>
13178 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13181 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13183 PR middle-end/80707
13184 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13186 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13188 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13189 early expansion of vector divide builtins.
13190 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13191 builtins identified as having unsigned arguments.
13193 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13195 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13196 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13197 expansion of vector logical operations (and, andc, or, xor,
13200 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13202 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13203 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13205 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13207 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13208 early GIMPLE expansion of vector multiplies.
13210 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13212 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13213 TARGET_HAVE_MOVT conditional.
13214 (movt splitter): Likewise.
13216 2017-05-12 Richard Biener <rguenther@suse.de>
13218 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13219 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13220 Fold all stmts not inplace.
13222 2017-05-12 Richard Biener <rguenther@suse.de>
13224 PR tree-optimization/80713
13225 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13226 inserted_exprs bit for not removed stmts.
13228 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13230 PR middle-end/69921
13231 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13232 parallelized" attribute for parallelized OpenACC kernels.
13233 * omp-offload.c (execute_oacc_device_lower): Use it.
13235 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13236 Set "oacc kernels" attribute.
13237 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13238 parameter. Adjust all users.
13239 (oacc_fn_attrib_kernels_p): Remove function.
13240 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13241 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13242 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13243 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13244 assert "oacc kernels" attribute is set.
13246 2017-05-11 Carl Love <cel@us.ibm.com>
13248 * config/rs6000/rs6000-c: Add support for built-in functions
13249 vector unsigned char vec_popcnt (vector signed char)
13250 vector unsigned char vec_popcnt (vector unsigned char)
13251 vector unsigned short vec_popcnt (vector signed short)
13252 vector unsigned short vec_popcnt (vector unsigned short)
13253 vector unsigned int vec_popcnt (vector signed int)
13254 vector unsigned int vec_popcnt (vector unsigned int)
13255 vector unsigned long long vec_popcnt (vector signed long long)
13256 vector unsigned long long vec_popcnt (vector unsigned long long)
13257 vector signed long long vec_slo (vector signed long long,
13258 vector signed char)
13259 vector signed long long vec_slo (vector signed long long,
13260 vector unsigned char)
13261 vector unsigned long long vec_slo (vector unsigned long long,
13262 vector signed char)
13263 vector unsigned long long vec_slo (vector unsigned long long,
13264 vector unsigned char)
13265 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13266 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13267 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13268 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13269 * doc/extend.texi: Update the built-in documentation file for the
13270 new built-in functions.
13272 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13274 * attribs.h (sorted_attr_string): Move machine independent
13275 functions for target clone support from the i386 port to common
13276 code. Rename ix86_function_versions to common_function_versions.
13277 Rename make_name to make_unique_name.
13278 (common_function_versions): Likewise.
13279 (make_unique_name): Likewise.
13280 (make_dispatcher_decl): Likewise.
13281 (is_function_default_version): Likewise.
13282 * attribs.c (attr_strcmp): Likewise.
13283 (sorted_attr_string): Likewise.
13284 (common_function_versions): Likewise.
13285 (make_unique_name): Likewise.
13286 (make_dispatcher_decl): Likewise.
13287 (is_function_default_version): Likewise.
13288 * config/i386/i386.c (attr_strcmp): Likewise.
13289 (sorted_attr_string): Likewise.
13290 (ix86_function_versions): Likewise.
13291 (make_name): Likewise.
13292 (make_dispatcher_decl): Likewise.
13293 (is_function_default_version): Likewise.
13294 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13296 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13299 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13300 Account for direct move costs for vec_construct of integer
13303 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13306 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13307 (UNSPEC_STX_ATOMIC): Ditto.
13308 (loaddi_via_sse): New insn.
13309 (storedi_via_sse): Ditto.
13310 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13311 Update corresponding peephole2 patterns.
13312 (atomic_storedi_fpu): Ditto.
13314 2017-05-11 Julia Koval <julia.koval@intel.com>
13316 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13317 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13319 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13320 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13321 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13323 2017-05-11 Nathan Sidwell <nathan@acm.org>
13325 * graphite-poly.c: Include dumpfile.h.
13327 * dumpfle.h (dump_function): Declare here ...
13328 * tree-dump.h (dump_function): ... not here.
13329 * dumpfile.c: #include tree-cfg.h.
13330 (dump_function): Move here from ...
13331 * tree-dump.c (dump_function): ... here.
13332 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13333 * graphite-poly.c: Don't include tree-dump.h.
13334 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13335 * print-tree.c: Likewise.
13336 * stor-layout.c: Likewise.
13337 * tree-nested.c: Likewise.
13339 * dumpfile.c (dump_start): Use TDF_FLAGS.
13340 (dump_enable_all): Fix TDF_KIND check thinko.
13342 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13344 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13345 array entries to represent two legal parameterizations of the
13346 overloaded __builtin_cmpb function, as represented by the
13347 P6_OV_BUILTIN_CMPB constant.
13348 (altivec_resolve_overloaded_builtin): Add special case handling
13349 for the __builtin_cmpb function, as represented by the
13350 P6_OV_BUILTIN_CMPB constant.
13351 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13352 (BU_P6_64BIT_2): New macro.
13353 (BU_P6_OVERLOAD_2): New macro
13354 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13355 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13356 (CMPB): Add overload support to represent both 32-bit and 64-bit
13357 compare-bytes function.
13358 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13359 support for TARGET_CMPB.
13360 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13361 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13362 documentation of the __builtin_cmpb overloaded built-in function.
13364 2017-05-11 Richard Biener <rguenther@suse.de>
13366 PR tree-optimization/80705
13367 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13368 bases are not vectorizable.
13370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13372 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13373 when counting register pressure.
13375 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13377 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13378 (iv_ca_more_deps): Renamed to ...
13379 (iv_ca_compare_deps): ... this.
13380 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13382 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13384 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13386 (determine_group_iv_costs): ... here.
13387 (find_inv_vars_cb): Record inv var if it's not recorded before.
13389 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13391 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13392 (get_shiftadd_cost): Ditto.
13394 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13396 * tree-ssa-address.c: Include header file.
13397 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13399 (add_to_parts): Refactor.
13400 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13401 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13404 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13406 PR tree-optimization/53090
13407 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13409 (extract_cond_operands): Detect condition with IV on both sides
13410 and return COMP_IV_EXPR_2.
13411 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13412 (rewrite_use_compare): Simplify by removing call to function
13413 extract_cond_operands.
13415 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13417 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13418 (extract_cond_operands): Detect condition comparing against non-
13419 invariant bound and return appropriate enum value.
13420 (find_interesting_uses_cond): Update use of extract_cond_operands.
13421 Handle its return value accordingly.
13422 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13424 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13426 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13427 nonlinear iv_use computation in loop invariant sensitive way.
13429 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13431 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13432 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13436 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13437 (dump_cand): Support iv_cand.inv_exprs.
13438 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13440 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13443 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13445 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13447 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13448 as local function. Include necessary header files.
13449 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13453 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13455 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13457 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13458 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13459 RSHIFT_EXPR and BIT_NOT_EXPR.
13461 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13463 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13464 (adjust_setup_cost): New parameter supporting round up adjustment.
13465 (struct address_cost_data): Delete.
13466 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13467 (split_address_cost, ptr_difference_cost): Delete.
13468 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13469 (struct ainc_cost_data): New struct.
13470 (get_address_cost_ainc): New function.
13471 (get_address_cost, get_computation_cost): Reimplement.
13472 (determine_group_iv_cost_address): Record inv_expr for all uses of
13474 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13475 (iv_ca_has_deps): Reimplemented to ...
13476 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13478 (iv_ca_extend): Call iv_ca_more_deps.
13480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13482 * tree-ssa-address.c (struct mem_address): Move to header file.
13483 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13484 * tree-ssa-address.h (struct mem_address): Move from C file.
13485 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13487 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13489 * tree-affine.h (aff_combination_type): New interface.
13490 (aff_combination_zero_p): Remove static.
13491 (aff_combination_const_p): New interface.
13492 (aff_combination_singleton_var_p): New interfaces.
13494 2017-05-11 Richard Biener <rguenther@suse.de>
13496 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13497 Skip unreachable blocks and destinations.
13498 (eliminate): Move stmt removal and fixup ...
13499 (fini_eliminate): ... here. Skip inserted exprs.
13500 (pass_pre::execute): Move fini_pre after fini_eliminate.
13501 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13502 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13503 PRE to get rid of dead code that has invalid SSA form and
13504 split critical edges again.
13506 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13508 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13510 2017-05-11 Richard Biener <rguenther@suse.de>
13512 * passes.c (execute_function_todo): Verify loops if they are
13513 said to be up-to-date.
13514 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13515 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13517 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13520 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13521 handle calling assemble_external ourself.
13524 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13525 modes with zero size. Enhance comment.
13527 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13529 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13530 built-ins for vec_xl and vec_xst with short and char pointer
13533 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13535 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13536 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13537 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13538 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13539 (_mm_maskz_min_round_ss): New intrinsics.
13540 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13541 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13542 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13543 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13544 (__builtin_ia32_minss_mask_round): New builtins.
13545 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13546 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13547 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13549 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13550 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13552 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13555 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13557 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13558 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13559 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13560 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13561 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13562 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13563 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13564 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13565 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13566 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13567 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13568 (__builtin_ia32_mulss_mask_round): New builtins.
13569 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13570 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13571 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13573 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13574 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13576 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13579 2017-05-10 Julia Koval <julia.koval@intel.com>
13581 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13582 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13583 (_mm256_setr_m128i): New intrinsics.
13585 2017-05-10 Julia Koval <julia.koval@intel.com>
13587 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13588 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13589 (_mm_maskz_rcp14_ss): New intrinsics.
13590 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13591 (__builtin_ia32_rcp14ss_mask): New builtins.
13592 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13594 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13596 PR tree-optimization/51513
13597 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13598 (assert_unreachable_fallthru_edge_p): Use it.
13599 (group_case_labels_stmt): Likewise.
13600 * tree-cfg.h: Prototype it.
13601 * stmt.c: Include cfghooks.h and tree-cfg.h.
13602 (emit_case_dispatch_table) <gap_label>: New local variable.
13603 Use it to fill dispatch table gaps.
13604 Test for default_label before updating probabilities.
13605 (expand_case) <default_label>: Remove unneeded initialization.
13606 Test for unreachable default case statement and remove its edge.
13607 Set default_label accordingly.
13608 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13610 2017-05-10 Carl Love <cel@us.ibm.com>
13612 * config/rs6000/rs6000-c: Add support for built-in functions
13613 vector signed char vec_neg (vector signed char)
13614 vector signed short int vec_neg (vector short int)
13615 vector signed int vec_neg (vector signed int)
13616 vector signed long long vec_neg (vector signed long long)
13617 vector float vec_neg (vector float)
13618 vector double vec_neg (vector double)
13619 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13621 * config/rs6000/altivec.h: Add define for vec_neg
13622 * doc/extend.texi: Update the built-in documentation for the
13623 new built-in functions.
13625 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13627 PR tree-optimization/77644
13628 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13630 2017-05-10 Nathan Sidwell <nathan@acm.org>
13632 * dumpfile.h (TDI_lang_all): New.
13633 (TDF_KIND): New. Renumber others
13634 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13636 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13638 (get_dump_file_name): Adjust suffix generation.
13639 (dump_enable_all): Use TDF_KIND.
13640 * doc/invoke.texi (-fdump-lang-all): Document.
13642 * dumpfile.h: Tabify.
13644 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13647 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13648 Move member access before delete.
13650 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13652 * tree-inline.c (expand_call_inline): Split block at stmt
13655 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13658 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13659 are now unused after splitting mov{sf,sd}_hardfloat.
13660 (f32_lr2): Likewise.
13661 (f32_lm): Likewise.
13662 (f32_lm2): Likewise.
13663 (f32_li): Likewise.
13664 (f32_li2): Likewise.
13665 (f32_lv): Likewise.
13666 (f32_sr): Likewise.
13667 (f32_sr2): Likewise.
13668 (f32_sm): Likewise.
13669 (f32_sm2): Likewise.
13670 (f32_si): Likewise.
13671 (f32_si2): Likewise.
13672 (f32_sv): Likewise.
13673 (f32_dm): Likewise.
13674 (f32_vsx): Likewise.
13675 (f32_av): Likewise.
13676 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13677 For movsf, order stores so the VSX stores occur before the GPR
13678 store which encourages the register allocator to use a traditional
13679 FPR instead of a GPR. For movsd, order the stores so that the GPR
13680 store comes before the VSX stores to allow the power6 to work.
13681 This is due to the power6 not having a 32-bit integer store
13682 instruction from a FPR.
13683 (movsf_hardfloat): Likewise.
13684 (movsd_hardfloat): Likewise.
13686 2017-05-09 Martin Sebor <msebor@redhat.com>
13688 PR translation/80280
13689 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13692 PR translation/80280
13693 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13694 data member added in r247778.
13695 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13697 2017-05-09 Nathan Sidwell <nathan@acm.org>
13699 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13701 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13704 2017-05-09 Marek Polacek <polacek@redhat.com>
13706 * doc/invoke.texi: Fix typo.
13708 2017-05-09 Richard Biener <rguenther@suse.de>
13710 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13711 (vrp_val_is_min): Likewise.
13712 (set_value_range_to_value): Likewise.
13713 (set_value_range_to_nonnegative): Likewise.
13714 (gimple_assign_nonzero_p): Likewise.
13715 (gimple_stmt_nonzero_p): Likewise.
13716 (vrp_int_const_binop): Likewise. Remove unreachable case.
13717 (adjust_range_with_scev): Adjust comments.
13718 (compare_range_with_value): Likewise.
13719 (extract_range_from_phi_node): Likewise.
13720 (test_for_singularity): Likewise.
13722 2017-05-09 Richard Biener <rguenther@suse.de>
13724 * tree-vrp.c (get_single_symbol): Add assert that we don't
13725 get overflowed constants as invariant part.
13726 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13727 checks. Use wi::cmp instead of recursing for integer constants.
13728 (compare_values): Just ignore whether we assumed undefined
13729 overflow instead of failing the compare.
13730 (extract_range_for_var_from_comparison_expr): Add comment before the
13731 TREE_NO_WARNING sets.
13732 (test_for_singularity): Likewise.
13733 (extract_range_from_comparison): Do not disable optimization
13734 when we assumed undefined overflow.
13735 (extract_range_basic): Remove init of unused var.
13737 2017-05-09 Richard Biener <rguenther@suse.de>
13739 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13740 (extract_range_from_multiplicative_op_1): Adjust.
13741 (extract_range_from_binary_expr_1): Use int_const_binop.
13743 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13746 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13747 rs6000_store_data_bypass_p in seven define_bypass directives and
13748 in several comments.
13749 * config/rs6000/rs6000-protos.h: Add prototype for
13750 rs6000_store_data_bypass_p function.
13751 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13752 function implements slightly different (rs6000-specific) semantics
13753 than store_data_bypass_p, returning false rather than aborting
13754 with assertion error when arguments do not satisfy the
13755 requirements of store data bypass.
13756 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13757 rs6000_store_data_bypass_p.
13759 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13761 * config/xtensa/xtensa-protos.h
13762 (xtensa_initial_elimination_offset): New declaration.
13763 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13764 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13765 macro definition, add case for FRAME_POINTER_REGNUM when
13766 FRAME_GROWS_DOWNWARD.
13767 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13768 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13769 xtensa_initial_elimination_offset.
13771 2017-05-08 Nathan Sidwell <nathan@acm.org>
13773 * doc/invoke.texi: Alphabetize -fdump options.
13775 2017-05-08 Martin Sebor <msebor@redhat.com>
13777 PR translation/80280
13778 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13780 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13782 * target.def (compute_frame_layout): New optional target hook.
13783 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13784 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13785 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13787 * reload1.c (verify_initial_elim_offsets): Likewise.
13788 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13789 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13790 (arm_get_frame_offsets): Split up into this ...
13791 (arm_compute_frame_layout): ... and this function.
13793 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13795 * config/aarch64/constraints.md (Usa): New constraint.
13796 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13798 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13800 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13801 with_multilib_list after it has been checked.
13803 2017-05-08 Richard Biener <rguenther@suse.de>
13805 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13806 (bitmap_set_subtract_values): Likewise.
13808 2017-05-08 Richard Biener <rguenther@suse.de>
13810 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13811 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13813 (gimple_stmt_nonzero_warnv_p): Rename to ...
13814 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13816 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13817 (extract_range_basic): Adjust, do not disable propagation on
13818 strict overflow sensitive simplification.
13819 (vrp_visit_cond_stmt): Likewise.
13821 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13823 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13824 body size unconditionally.
13826 2017-05-07 Jeff Law <law@redhat.com>
13829 2017-05-06 Jeff Law <law@redhat.com>
13830 PR tree-optimization/78496
13831 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13834 PR tree-optimization/78496
13835 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13836 (simplify_stmt_using_ranges): Call it.
13837 (vrp_dom_walker::before_dom_children): Extract equivalences
13838 from an ASSERT_EXPR with an equality comparison against a
13841 2017-05-06 Jeff Law <law@redhat.com>
13843 PR tree-optimization/78496
13844 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13847 PR tree-optimization/78496
13848 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13849 (simplify_stmt_using_ranges): Call it.
13850 (vrp_dom_walker::before_dom_children): Extract equivalences
13851 from an ASSERT_EXPR with an equality comparison against a
13854 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13856 * lra-constraints.c (lra_copy_reg_equiv): New function.
13857 (split_reg): Use it to copy equivalence information from the
13858 original register to the spill register.
13860 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13862 PR rtl-optimization/75964
13863 * simplify-rtx.c (simplify_const_relational_operation): Remove
13864 invalid handling of comparisons of integer ABS.
13866 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13868 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13869 initialize to zero.
13870 (init_regs): Remove declaration.
13871 (function_arg_advance_32): Initialize error_p as boolean variable.
13873 2017-05-05 Nathan Sidwell <nathan@acm.org>
13875 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13876 lines. Use for (;;).
13878 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13880 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13881 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13882 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13883 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13884 VF=2 that require versioning.
13886 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13888 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13891 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13893 * diagnostic.h (diagnostic_override_option_index): Convert from
13894 macro to inline function.
13896 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13898 * diagnostic.c (last_module_changed_p): New function.
13899 (set_last_module): New function.
13900 (diagnostic_report_current_module): Convert macro usage to
13901 the above functions.
13902 * diagnostic.h (diagnostic_context::last_module): Strengthen
13903 from const line_map * to const line_map_ordinary *.
13904 (diagnostic_last_module_changed): Delete macro.
13905 (diagnostic_set_last_module): Delete macro.
13907 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13909 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13910 with diagnostic_report_diagnostic.
13911 (diagnostic_n_impl_richloc): Likewise.
13912 * diagnostic.h (report_diagnostic): Delete macro.
13913 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13914 with diagnostic_report_diagnostic.
13915 * substring-locations.c (format_warning_va): Likewise.
13917 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13919 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13920 save/restor of format_spec. Move option-printing code to...
13921 (print_option_information): ...this new function, and
13922 reimplement by simply printing to the pretty_printer,
13923 rather than appending to the format string.
13925 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13927 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13928 handling logic into...
13929 (update_effective_level_from_pragmas): ...this new function.
13931 2017-05-04 Andrew Waterman <andrew@sifive.com>
13933 * config/riscv/riscv.opt (mstrict-align): New option.
13934 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13935 (SLOW_UNALIGNED_ACCESS): Define.
13936 (riscv_slow_unaligned_access): Declare.
13937 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13939 (riscv_slow_unaligned_access): New variable.
13940 (rocket_tune_info): Set slow_unaligned_access to true.
13941 (optimize_size_tune_info): Set slow_unaligned_access to false.
13942 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13943 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13944 (riscv_option_override): Set riscv_slow_unaligned_access.
13945 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13947 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13949 * config/riscv/riscv.md: Unify indentation.
13951 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13956 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13958 (extendsi<mode>2): Add support for doing sign extension via
13959 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13960 don't have ISA 3.0 instructions.
13961 (extendsi<mode>2 splitter): Likewise.
13962 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13963 generate the normal insns since SImode can now go in vector
13964 registers. Disallow the special UNSPECs needed for previous
13965 machines to hide SImode being used. Add new insns
13966 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13967 (fix_trunc<mode>si2_stfiwx): Likewise.
13968 (fix_trunc<mode>si2_internal): Likewise.
13969 (fixuns_trunc<mode>si2): Likewise.
13970 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13971 (fctiw<u>z_<mode>_smallint): Likewise.
13972 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13973 of floating point to 32-bit integer from doing a direct move to
13974 the GPR registers to do a store.
13975 (fctiwz_<mode>): Break long line.
13977 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13979 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13980 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13981 (addr_list, addr_offset_valid_p): New.
13982 (split_address_groups): Check offset validity with above function.
13983 (gt-tree-ssa-loop-ivopts.h): Include header file.
13985 2017-05-05 Nathan Sidwell <nathan@acm.org>
13987 * config.gcc (arm*-*-*): Add missing 'fi'.
13989 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13991 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13992 included in -fopt-info does not matter.
13993 * doc/optinfo.texi (-fopt-info): Fix description of default
13994 behavour. Explicitly say order of options included in -fopt-info
13997 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13999 * config.gcc: Allow combinations of aprofile and rmprofile values for
14000 --with-multilib-list.
14001 * config/arm/t-multilib: New file.
14002 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14003 variables. Remove setting of ISA and floating-point ABI in
14004 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14005 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14006 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14007 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14009 * config/arm/t-rmprofile: Likewise except for the matches changes.
14010 * doc/install.texi (--with-multilib-list): Document the combination of
14011 aprofile and rmprofile values and warn about pitfalls in doing that.
14013 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14015 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14016 (movdi_aarch64): Likewise.
14018 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14020 PR tree-optimization/80632
14021 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14023 (build_arrays): Initialize it for virtual phis.
14024 (fix_phi_nodes): Use it for virtual phis.
14026 PR tree-optimization/80558
14027 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14028 [x, y] op z into [x op, y op z] for op & or | if conditions
14031 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14032 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14035 * config/arm/arm.md (use_literal_pool): Remove.
14036 (64-bit immediate split): No longer takes cost into consideration
14037 if arm_disable_literal_pool is enabled.
14038 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14039 used when arm_disable_literal_pool is enabled.
14040 (arm_max_const_double_inline_cost): Remove use of
14041 arm_disable_literal_pool.
14042 (push_minipool_fix): Add assert.
14043 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14044 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14045 (no_literal_pool_sf_immediate): New.
14047 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14049 PR tree-optimization/80613
14050 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14051 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14053 2017-05-05 Richard Biener <rguenther@suse.de>
14055 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14057 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14059 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14060 of this flag from insn conditions due to removal from r247495.
14062 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14064 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14066 (arm_early_store_addr_dep_ptr): Likewise.
14067 * config/arm/aarch-common-protos.h
14068 (arm_early_load_addr_dep_ptr): Add prototype.
14069 (arm_early_store_addr_dep_ptr): Likewise.
14070 * config/arm/cortex-a53.md: Add new bypasses.
14072 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14074 * tree.c (next_type_uid): Change type to unsigned.
14075 (type_hash_canon): Decrement back next_type_uid if
14076 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14077 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14080 2017-05-04 Martin Sebor <msebor@redhat.com>
14082 * builtins.c: Fix a trivial typo in a comment.
14084 PR middle-end/79234
14085 * builtins.c (check_sizes): Adjust to handle reading past the end.
14086 Avoid printing excessive upper bound of ranges. Use %E to print
14087 tree nodes instead of converting them to %wu.
14088 (expand_builtin_memchr): New function.
14089 (compute_dest_size): Rename...
14090 (compute_objsize): ...to this.
14091 (expand_builtin_memcpy): Adjust.
14092 (expand_builtin_mempcpy): Adjust.
14093 (expand_builtin_strcat): Adjust.
14094 (expand_builtin_strcpy): Adjust.
14095 (check_strncat_sizes): Adjust.
14096 (expand_builtin_strncat): Adjust.
14097 (expand_builtin_strncpy): Adjust and simplify.
14098 (expand_builtin_memset): Adjust.
14099 (expand_builtin_bzero): Adjust.
14100 (expand_builtin_memcmp): Adjust.
14101 (expand_builtin): Handle memcmp.
14102 (maybe_emit_chk_warning): Check strncat just once.
14104 2017-05-04 Martin Sebor <msebor@redhat.com>
14106 PR preprocessor/79214
14107 PR middle-end/79222
14108 PR middle-end/79223
14109 * builtins.c (check_sizes): Add inlining context and issue
14110 warnings even when -Wno-system-headers is set.
14111 (check_strncat_sizes): Same.
14112 (expand_builtin_strncat): Same.
14113 (expand_builtin_memmove): New function.
14114 (expand_builtin_stpncpy): Same.
14115 (expand_builtin): Handle memmove and stpncpy.
14117 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14119 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14120 which is not used any more.
14122 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14124 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14126 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14128 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14129 (cortexa53_tunings): Likewise.
14130 (cortexa57_tunings): Likewise.
14131 (cortexa72_tunings): Likewise.
14132 (cortexa73_tunings): Likewise.
14134 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14136 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14137 Set loop alignment to 8.
14139 2017-05-04 Martin Sebor <msebor@redhat.com>
14141 PR translation/80280
14142 * builtins.c (expand_builtin_object_size): Add missing quoting to
14143 %D and like directives.
14144 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14145 (hsa_type_for_tree_type): Same.
14146 (verify_function_arguments): Same.
14147 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14148 * varasm.c (get_section): Same.
14151 2017-05-04 Martin Sebor <msebor@redhat.com>
14153 PR translation/80280
14154 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14156 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14158 * config/aarch64/aarch64.c (generic_addrcost_table):
14159 Change HI/TI mode setting.
14161 2017-05-04 Martin Jambor <mjambor@suse.cz>
14163 PR tree-optimization/80622
14164 * tree-sra.c (comes_initialized_p): New function.
14165 (build_accesses_from_assign): Only set write lazily when
14166 comes_initialized_p is false.
14167 (analyze_access_subtree): Use comes_initialized_p.
14168 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14169 instead of testing for PARM_DECL.
14171 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14173 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14174 constraint on operand 0 to allow more general addressing modes.
14175 Adjust output template.
14176 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14178 * config/aarch64/aarch64-protos.h
14179 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14180 * config/aarch64/constraints.md (Dp): New address constraint.
14181 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14184 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14186 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14187 update use of estimate_ipcp_clone_size_and_time.
14188 (estimate_local_effects): Update use of
14189 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14190 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14191 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14192 Return nonspecialized time.
14194 2017-05-04 Richard Biener <rguenther@suse.de>
14196 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14197 for the last VUSE which def dominates the PHI. Directly call
14199 (get_continuation_for_phi_1): Remove.
14201 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14203 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14204 to explain the use of truncating division. Cap the number of
14205 iterations to the maximum given by nb_iterations_upper_bound,
14208 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14210 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14211 * configure: Regenerate.
14212 * config.in: Regenerate.
14213 * config/i386/driver-mingw32.c: new file.
14214 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14215 * config.host: Link driver-mingw32.o on MinGW host.
14216 * doc/install.texi: Document new --enable-mingw-wildcard configure
14219 2017-05-04 Marek Polacek <polacek@redhat.com>
14221 PR tree-optimization/80612
14222 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14224 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14225 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14227 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14228 (movt splitter): Likewise.
14229 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14230 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14231 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14232 block for Thumb-1 with MOVT.
14233 (thumb2_legitimate_address_p): Move code block ...
14234 (can_avoid_literal_pool_for_label_p): ... into this new function.
14235 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14237 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14238 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14239 "M-profile targets with the MOVT instruction".
14241 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14243 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14244 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14245 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14247 2017-05-04 Martin Liska <mliska@suse.cz>
14249 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14250 variable cond_code.
14252 2017-05-04 Richard Biener <rguenther@suse.de>
14254 * tree.c (array_at_struct_end_p): Handle arrays at struct
14255 end with flexarrays more conservatively. Refactor and treat
14256 arrays of arrays or aggregates more strict. Fix
14257 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14258 * tree.c (array_at_struct_end_p): Adjust prototype.
14259 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14260 * gimple-fold.c (get_range_strlen): Likewise.
14261 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14263 2017-05-04 Richard Biener <rguenther@suse.de>
14265 PR tree-optimization/31130
14266 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14268 (supports_overflow_infinity): Likewise.
14269 (is_negative_overflow_infinity): Likewise.
14270 (is_positive_overflow_infinity): Likewise.
14271 (is_overflow_infinity): Likewise.
14272 (stmt_overflow_infinity): Likewise.
14273 (overflow_infinity_range_p): Likewise.
14274 (usable_range_p): Remove as always returning true.
14275 (make_overflow_infinity): Remove.
14276 (negative_overflow_infinity): Likewise.
14277 (positive_overflow_infinity): Likewise.
14278 (avoid_overflow_infinity): Likewise.
14279 (set_value_range): Adjust accordingly.
14280 (set_value_range_to_nonnegative): Likewise, remove now unused
14281 overflow_infinity arg.
14282 (vrp_operand_equal_p): Adjust.
14283 (update_value_range): Likewise.
14284 (range_int_cst_singleton_p): Likewise.
14285 (operand_less_p): Likewise.
14286 (compare_values_warnv): Likewise.
14287 (extract_range_for_var_from_comparison_expr): Likewise.
14288 (vrp_int_const_binop): Likewise.
14289 (zero_nonzero_bits_from_vr): Likewise.
14290 (extract_range_from_multiplicative_op_1): Likewise.
14291 (extract_range_from_binary_expr_1): Likewise.
14292 (extract_range_from_unary_expr): Likewise.
14293 (extract_range_from_comparison): Likewise.
14294 (extract_range_basic): Likewise.
14295 (adjust_range_with_scev): Likewise.
14296 (compare_ranges): Likewise.
14297 (compare_range_with_value): Likewise.
14298 (dump_value_range): Likewise.
14299 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14301 (simplify_cond_using_ranges): Adjust.
14303 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14305 * brig-builtins.def: Added a builtin for class_f64.
14306 * builtin-types.def: Added a builtin type needed by class_f64.
14308 2017-05-03 Jason Merrill <jason@redhat.com>
14310 * timevar.def: Add TV_CONSTEXPR.
14312 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14314 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14316 2017-05-03 Martin Jambor <mjambor@suse.cz>
14318 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14319 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14320 * ipa-cp.c (ipcp_propagate_stage): Do not call
14321 ipa_update_after_lto_read.
14322 * ipa-inline.c (ipa_inline): Likewise.
14324 2017-05-03 Martin Jambor <mjambor@suse.cz>
14326 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14327 tag. Added a default constructor and a destructor.
14328 (ipa_edge_args_sum_t): New class;
14329 (ipa_edge_args_sum): Declare.
14330 (ipa_edge_args_vector): Remove declaration.
14331 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14332 (ipa_free_edge_args_substructures): Remove declaration.
14333 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14334 (ipa_edge_args_info_available_for_edge_p): Likewise.
14335 * ipa-prop.c (ipa_edge_args_vector): Removed.
14336 (edge_removal_hook_holder): Likewise.
14337 (edge_duplication_hook_holder): Likewise.
14338 (ipa_edge_args_sum): New variable.
14339 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14340 ipa_edge_args_vector.
14341 (ipa_free_edge_args_substructures): Likewise.
14342 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14343 ipa_edge_args_vector.
14344 (ipa_edge_removal_hook): Turned into method
14345 ipa_edge_args_sum_t::remove.
14346 (ipa_edge_duplication_hook): Turned into method
14347 ipa_edge_args_sum_t::duplicate.
14348 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14349 registering edge hooks.
14350 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14351 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14352 ipa_edge_args_sum instead of ipa_edge_args_vector.
14353 * ipa-profile.c (ipa_profile): Likewise.
14355 2017-05-03 Martin Jambor <mjambor@suse.cz>
14357 * symbol-summary.h (function_summary): New method exists.
14358 (function_summary::symtab_removal): Deallocate through release.
14359 (call_summary): New class.
14360 (gt_ggc_mx): New overload.
14361 (gt_pch_nx): Likewise.
14362 (gt_pch_nx): Likewise.
14364 2017-05-03 Jeff Law <law@redhat.com>
14366 PR tree-optimization/78496
14367 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14368 from simplify_cond_using_ranges. Split off code to walk
14369 backwards through casts into ...
14370 (simplify_cond_using_ranges_2): New function.
14371 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14372 (execute_vrp): After identifying jump threads, call
14373 simplify_cond_using_ranges_2.
14375 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14378 * ipa-inline.h (inline_summary): Add ctor.
14379 (create_ggc): Do not use ggc_cleared_alloc.
14381 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14382 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14384 * gcc.c (handle_braces): Support escaping in switch matching
14386 * doc/invoke.texi (Spec Files): Document it.
14387 Remove superfluous @code markup in items.
14389 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14391 * diagnostic-show-locus.c (struct column_range): New struct.
14392 (get_affected_columns): New function.
14393 (get_printed_columns): New function.
14394 (struct correction): New struct.
14395 (correction::ensure_capacity): New function.
14396 (correction::ensure_terminated): New function.
14397 (struct line_corrections): New struct.
14398 (line_corrections::~line_corrections): New dtor.
14399 (line_corrections::add_hint): New function.
14400 (layout::print_trailing_fixits): Reimplement in terms of the new
14402 (selftest::test_overlapped_fixit_printing): New function.
14403 (selftest::diagnostic_show_locus_c_tests): Call it.
14405 2017-05-03 Nathan Sidwell <nathan@acm.org>
14407 Canonicalize canonical type hashing
14408 * tree.h (type_hash_canon_hash): Declare.
14409 * tree.c (type_hash_list, attribute_hash_list): Move into
14410 type_hash_canon_hash.
14411 (build_type_attribute_qual_variant): Break out hash code calc into
14412 type_hash_canon_hash.
14413 (type_hash_canon_hash): New. Generic type hash computation.
14414 (build_range_type_1, build_array_type_1, build_function_type,
14415 build_method_type_directly, build_offset_type, build_complex_type,
14416 make_vector_type): Call it.
14418 2017-05-03 Richard Biener <rguenther@suse.de>
14420 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14421 When all DRs have unknown misaligned do not always peel
14422 when there is a store but apply the same costing model as if
14423 there were only loads.
14425 2017-05-03 Richard Biener <rguenther@suse.de>
14428 PR tree-optimization/80492
14429 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14430 compare_base_decls returning dont-know properly.
14432 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14434 * config/arm/iterators.md (CCSI): New mode iterator.
14435 (arch): New mode attribute.
14436 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14437 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14438 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14439 code iterator for success result mode.
14440 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14441 the corresponding new insn generators.
14443 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14446 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14447 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14449 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14451 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14452 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14453 (DDR_A): Wrap DDR argument in brackets.
14454 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14455 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14456 (DDR_REVERSED_P): Likewise.
14458 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14460 PR tree-optimization/79472
14461 * tree-switch-conversion.c (struct switch_conv_info): Add
14462 contiguous_range and default_case_nonstandard fields.
14463 (collect_switch_conv_info): Compute contiguous_range and
14464 default_case_nonstandard fields, don't clear final_bb if
14465 contiguous_range and only the default case doesn't have the required
14467 (check_all_empty_except_final): Set default_case_nonstandard instead
14468 of failing if contiguous_range and the default case doesn't have empty
14470 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14471 and only the default case doesn't have the required constants. Skip
14473 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14474 if default_case_nonstandard.
14475 (build_constructors): Build constant 1 just once. Assert that default
14476 values aren't inserted in between cases if contiguous_range. Skip
14478 (build_arrays): Skip virtual phis.
14479 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14480 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14481 Handle virtual phis.
14482 (gen_inbound_check): Handle default_case_nonstandard case.
14483 (process_switch): Adjust check_final_bb caller. Call
14484 gather_default_values with the first non-default case instead of
14485 default case if default_case_nonstandard.
14487 2017-05-02 Nathan Sidwell <nathan@acm.org>
14489 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14490 check. Fix formatting.
14492 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14494 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14495 errors when comparing specialized and unspecialized times.
14497 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14499 * diagnostic-show-locus.c
14500 (layout::should_print_annotation_line_p): Make private.
14501 (layout::print_annotation_line): Make private.
14502 (layout::annotation_line_showed_range_p): Make private.
14503 (layout::show_ruler): Make private.
14504 (layout::print_source_line): Make private. Pass in line and
14505 line_width, rather than calling location_get_source_line. Drop
14507 (layout::print_leading_fixits): New method.
14508 (layout::print_any_fixits): Rename to...
14509 (layout::print_trailing_fixits): ...this, and make private.
14510 Don't print newline fixits.
14511 (diagnostic_show_locus): Move logic for printing one row into...
14512 (layout::print_line): ...this new function. Move the
14513 location_get_source_line call and error-handling from
14514 print_source_line to here. Call print_leading_fixits, and rename
14515 print_any_fixits to print_trailing_fixits.
14516 (selftest::test_fixit_insert_containing_newline): Update now that
14517 newlines are partially supported.
14518 (selftest::test_fixit_insert_containing_newline_2): New test.
14519 (selftest::test_fixit_replace_containing_newline): Update comments.
14520 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14521 * edit-context.c (class added_line): New class.
14522 (class edited_line): Describe newline handling in comment.
14523 (edited_line::actually_edited_p): New method.
14524 (edited_line::print_content): Delete redundant decl.
14525 (edited_line::m_predecessors): New field.
14526 (edited_file::print_content): Call edited_line::print_content.
14527 (edited_file::print_diff): Update to support newlines.
14528 (edited_file::print_diff_hunk): Likewise.
14529 (edited_file::print_run_of_changed_lines): New function.
14530 (edited_file::print_diff_line): Convert to...
14531 (print_diff_line): ...this.
14532 (edited_file::get_effective_line_count): New function.
14533 (edited_line::edited_line): Initialize new field m_predecessors.
14534 (edited_line::~edited_line): Clean up m_predecessors.
14535 (edited_line::apply_fixit): Handle newlines.
14536 (edited_line::get_effective_line_count): New function.
14537 (edited_line::print_content): New function.
14538 (edited_line::print_diff_lines): New function.
14539 (selftest::test_applying_fixits_insert_containing_newline): New
14541 (selftest::test_applying_fixits_replace_containing_newline): New
14543 (selftest::insert_line): New function.
14544 (selftest::test_applying_fixits_multiple_lines): Add example of
14546 (selftest::edit_context_c_tests): Call the new tests.
14548 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14550 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14551 parameter cand. Update dump information.
14552 (get_computation_cost): Update uses.
14554 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14556 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14557 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14558 (get_computation_at, rewrite_use_address): Update use of
14559 get_computation_aff.
14561 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14563 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14564 (get_computation): Delete.
14565 (get_computation_cost): Implement like get_computation_cost_at.
14566 Use get_computation_at.
14567 (get_computation_cost_at): Delete.
14568 (rewrite_use_nonlinear_expr): Use get_computation_at.
14569 (rewrite_use_compare, remove_unused_ivs): Ditto.
14571 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14573 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14575 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14577 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14578 (ivopts_global_cost_for_size): Rename parameter and update uses.
14579 (iv_ca_recount_cost): Update uses.
14580 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14581 candidates seperately in n_invs and n_cands.
14582 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14584 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14586 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14587 (find_inv_vars_cb): New.
14588 (find_depends): Renamed to ...
14589 (find_inv_vars): ... this.
14590 (add_candidate_1, force_var_cost): Call find_inv_vars.
14591 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14593 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14595 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14596 inv_vars. Add inv_exprs.
14597 (struct iv_cand): Rename depends_on to inv_vars.
14598 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14599 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14600 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14601 (dump_cand): Dump inv_vars.
14602 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14603 (record_invariant, find_depends, add_candidate_1): Ditto.
14604 (set_group_iv_cost, force_var_cost): Ditto.
14605 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14606 (get_computation_cost_at, get_computation_cost): Ditto.
14607 (determine_group_iv_cost_generic): Ditto.
14608 (determine_group_iv_cost_address): Ditto.
14609 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14610 (determine_group_iv_costs): Ditto.
14611 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14612 (iv_ca_set_remove_invariants): Renamed to ...
14613 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14614 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14615 (iv_ca_set_add_invariants): Renamed to ...
14616 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14617 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14618 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14619 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14620 (create_new_ivs): Remove useless dump.
14622 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14624 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14626 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14627 (iv_ca_set_no_cp, create_new_iv): Ditto.
14629 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14631 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14633 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14635 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14636 function tree_check2.
14638 2017-05-02 Martin Liska <mliska@suse.cz>
14640 * doc/gcov.texi: Add missing preposition.
14641 * gcov.c (function_info::function_info): Properly fill up
14642 all member variables.
14644 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14646 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14648 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14650 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14652 2017-05-02 Martin Liska <mliska@suse.cz>
14655 * lto-streamer-in.c (lto_read_tree_1): Remove
14656 LTO_STREAMER_DEBUG.
14657 * lto-streamer.c (struct tree_hash_entry): Likewise.
14658 (struct tree_entry_hasher): Likewise.
14659 (tree_entry_hasher::hash): Likewise.
14660 (tree_entry_hasher::equal): Likewise.
14661 (lto_streamer_init): Likewise.
14662 (lto_orig_address_map): Likewise.
14663 (lto_orig_address_get): Likewise.
14664 (lto_orig_address_remove): Likewise.
14665 * lto-streamer.h: Likewise.
14666 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14667 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14669 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14671 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14672 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14673 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14674 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14675 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14676 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14677 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14678 (mm_maskz_sub_ss): New intrinsics.
14679 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14680 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14681 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14682 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14683 (__builtin_ia32_subss_mask_round): New builtins.
14684 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14685 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14686 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14688 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14689 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14691 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14694 2017-05-02 Martin Jambor <mjambor@suse.cz>
14696 PR tree-optimization/78687
14697 * tree-sra.c (access): New field parent.
14698 (process_subtree_disqualification): New function.
14699 (disqualify_candidate): Call it.
14700 (build_accesses_from_assign): Reset write flag if creating an
14702 (build_access_subtree): Fill in parent field and also prpagate
14703 down grp_write flag.
14704 (create_artificial_child_access): New parameter set_grp_write, set
14705 grp_write to its value.
14706 (propagate_subaccesses_across_link): Also propagate grp_write flag
14708 (propagate_all_subaccesses): Push the closest parent back to work
14709 queue if add_access_to_work_queue returned true.
14711 2017-05-02 Richard Biener <rguenther@suse.de>
14713 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14714 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14715 -fstrict-overflow documentation.
14716 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14717 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14718 flag_strict_overflow.
14719 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14720 * lto-opts.c (lto_write_options): Do not stream it.
14721 * lto-wrapper.c (merge_and_complain): Do not handle it.
14722 * opts.c (default_options_table): Do not set -fstrict-overflow.
14723 (finish_options): Likewise do not clear it when sanitizing.
14724 * simplify-rtx.c (simplify_const_relational_operation): Do not
14725 test flag_strict_overflow.
14727 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14729 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14730 using enabled attribute.
14731 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14732 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14733 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14734 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14735 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14736 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14737 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14738 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14739 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14740 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14742 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14744 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14746 2017-05-02 Richard Biener <rguenther@suse.de>
14748 PR tree-optimization/80591
14750 2017-04-10 Richard Biener <rguenther@suse.de>
14752 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14755 2017-05-02 Richard Biener <rguenther@suse.de>
14757 PR tree-optimization/80549
14758 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14759 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14760 headers if they do not have a preheader.
14762 2017-05-02 Martin Liska <mliska@suse.cz>
14765 * common.opt: Fix typo.
14766 * doc/invoke.texi: Likewise.
14768 2017-05-01 Jan Beulich <jbeulich@suse.com>
14770 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14771 swapping, add (x,x,m,x,n) alternative.
14773 2017-05-01 Nathan Sidwell <nathan@acm.org>
14775 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14776 unnecessary unadjusted_alignment check.
14778 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14781 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14783 * gimplify.c (gimplify_cilk_detach): New function.
14784 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14785 * tree-core.h: Document EXPR_CILK_SPAWN.
14786 * tree.h (EXPR_CILK_SPAWN): Define.
14788 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14790 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14791 to use new fixit_hint representation, using the "replace" logic.
14792 (get_line_span_for_fixit_hint): Likewise.
14793 (layout::print_any_fixits): Likewise.
14794 (selftest::test_one_liner_many_fixits): Rename to...
14795 (selftest::test_one_liner_many_fixits_1): ...this, and update
14796 comment and expected output to reflect that the multiple fix-it
14797 hints are now consolidated into one insertion.
14798 (selftest::test_one_liner_many_fixits_2): New test.
14799 (selftest::test_diagnostic_show_locus_one_liner): Update for
14801 (selftest::test_fixit_consolidation): Update for fix-it API
14803 * diagnostic.c (print_parseable_fixits): Likewise.
14804 * edit-context.c (edited_line::m_line_events): Convert from
14805 auto_vec <line_event *> to auto_vec <line_event>.
14806 (class line_event): Convert from abstract base class to a concrete
14807 class, taking over the role of replace_event.
14808 (class insert_event): Delete.
14809 (class replace_event): Rename to class line_event. Convert to
14811 (edit_context::add_fixits): Reimplement.
14812 (edit_context::apply_insert): Delete.
14813 (edit_context::apply_replace): Rename to...
14814 (edit_context::apply_fixit): ...this. Convert to half-open range.
14815 (edited_file::apply_insert): Delete.
14816 (edited_file::apply_replace): Rename to...
14817 (edited_file::apply_fixit): ...this.
14818 (edited_line::~edited_line): Drop deletion of events.
14819 (edited_line::apply_insert): Delete.
14820 (edited_line::apply_replace): Rename to...
14821 (edited_line::apply_fixit): ...this. Convert to half-open range.
14822 Update for change to type of m_line_events.
14823 * edit-context.h (edit_context::apply_insert): Delete.
14824 (edit_context::apply_replace): Rename to...
14825 (edit_context::apply_fixit): ...this.
14827 2017-05-01 Martin Sebor <msebor@redhat.com>
14829 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14832 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14835 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14836 __get_cpuid_max returns 0.
14837 (__get_cpuid_count): Ditto.
14839 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14841 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14842 replacement expression is another instance of one of its arguments.
14844 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14847 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14848 check for stack push/pop autoinc.
14849 * config/i386/i386.c (ix86_agi_dependent): Return false
14850 if the only reason why modified_in_p returned true is that
14851 addr is SP based and set_insn is a push or pop.
14853 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14855 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14858 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14861 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14862 (account_size_time): Use two predicates - exec_pred and
14864 (evaluate_conditions_for_known_args): Compute both clause and
14866 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14867 (inline_summary_t::duplicate): Update.
14868 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14870 (compute_inline_parameters): Likewise.
14871 (estimate_edge_size_and_time): Update caluclation of time.
14872 (estimate_node_size_and_time): Compute both time and nonspecialized
14874 (estimate_ipcp_clone_size_and_time): Update.
14875 (inline_merge_summary): Update.
14876 (do_estimate_edge_time): Update.
14877 (do_estimate_edge_size): Update.
14878 (do_estimate_edge_hints): Update.
14879 (inline_read_section, inline_write_summary): Stream both new predicates.
14880 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14882 (compute_inlined_call_time): Cleanup.
14883 (big_speedup_p): Update.
14884 (edge_badness): Update.
14885 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14886 (size_time_entry): Replace predicate by exec_predicate and
14887 nonconst_predicate.
14888 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14889 (estimate_edge_time): Return also nonspec_time.
14890 (reset_edge_growth_cache): Update.
14892 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14894 PR rtl-optimization/80491
14895 * ifcvt.c (noce_process_if_block): When looking for x setter
14896 with missing else_bb, don't check only the insn right before
14897 cond_earliest, but look for the last insn that x is modified in
14898 within the same bb.
14900 PR rtl-optimization/80491
14901 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14903 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14905 PR tree-optimization/80487
14906 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14908 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14910 PR tree-optimization/79697
14911 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14912 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14913 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14915 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14916 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14918 2017-04-28 Martin Sebor <msebor@redhat.com>
14920 PR tree-optimization/80523
14921 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14922 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14924 (maybe_warn, format_directive, parse_directive): Use new functions.
14925 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14927 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14929 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14931 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14933 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14934 target_header_dir): Set correctly.
14935 * configure: Regenerated.
14936 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14937 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14938 instead of SYSTEM_HEADER_DIR.
14940 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14942 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14943 (estimate_local_effects): Likewise.
14944 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14945 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14946 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14947 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14948 do_estimate_edge_time, estimate_edge_time): Likewise.
14949 * ipa-inline-analysis.c (estimate_node_size_and_time,
14950 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14951 (estimate_time_after_inlining): Remove.
14953 2017-04-28 Martin Liska <mliska@suse.cz>
14955 * doc/gcov.texi: Enhance documentation of gcov.
14957 2017-04-28 Martin Liska <mliska@suse.cz>
14959 * doc/gcov.texi: Sort options in alphabetic order.
14960 * doc/gcov-dump.texi: Likewise.
14961 * doc/gcov-tool.texi: Likewise.
14962 * gcov.c (print_usage): Likewise.
14963 * gcov-dump.c (print_usage): Likewise.
14964 * gcov-tool.c (print_merge_usage_message): Likewise.
14965 (print_rewrite_usage_message): Likewise.
14966 (print_overlap_usage_message): Likewise.
14968 2017-04-28 Martin Liska <mliska@suse.cz>
14970 PR gcov-profile/53915
14971 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14973 2017-04-28 Martin Liska <mliska@suse.cz>
14975 PR gcov-profile/79891
14976 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14977 is marked by compiler as living on a line.
14978 (get_cycles_count): Remove usage of the union.
14979 (output_intermediate_file): Likewise.
14980 (find_source): Fix GNU coding style.
14981 (accumulate_line_counts): Remove old non-all block mode.
14982 (output_lines): Remove usage of the union.
14983 * profile.c (output_location): Include all BBs, even if
14984 belonging to a same line (and file) as a previous BB.
14986 2017-04-28 Martin Liska <mliska@suse.cz>
14988 * gcov.c (process_args): Handle new argument 'w'.
14989 (read_graph_file): Assign ID to BBs.
14990 (output_branch_count): Display BB # if verbose flag is set.
14991 (output_lines): Likewise for arcs.
14992 (print_usage): Add '--verbose' option help.
14993 * doc/gcov.texi: Document --verbose (-w) option.
14995 2017-04-28 Martin Liska <mliska@suse.cz>
14997 * gcov.c (struct block_location_info): New struct.
14998 (process_file): Fill up the new structure.
14999 (read_graph_file): Replace usage of encoding by the newly added
15001 (add_line_counts): Likewise.
15002 (accumulate_line_counts): Remove usage of the union.
15003 (function_info::function_info): New function.
15004 (function_info::~function_info): Likewise.
15005 (process_file): Call delete instead of release_function.
15006 (release_function): Release the function.
15007 (release_structures): Call delete instead of release_function.
15008 (solve_flow_graph): Replace usage of num_blocks.
15009 (find_exception_blocks): Likewise.
15010 (output_lines): Fix GNU coding style.
15012 2017-04-28 Martin Liska <mliska@suse.cz>
15015 * coverage.c (coverage_remove_note_file): New function.
15016 * coverage.h: Declare the function.
15017 * toplev.c (finalize): Clean if an error has been seen.
15019 2017-04-28 Martin Liska <mliska@suse.cz>
15021 PR gcov-profile/80031
15022 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15023 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15024 * gcov.c (read_graph_file): Read just number of blocks.
15025 * profile.c (branch_prob): Do not stream 0 flags per a basic
15028 2017-04-28 Martin Liska <mliska@suse.cz>
15030 * gcov-dump.c (tag_*): Add new argument to declarations.
15031 (dump_gcov_file): Likewise.
15032 (tag_blocks): Add and use new argument depth.
15033 (tag_arcs): Likewise.
15034 (tag_lines): Likewise.
15035 (tag_counters): Likewise.
15036 (tag_summary): Likewise.
15037 (dump_working_sets): Use depth to do a proper indentation.
15039 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15042 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15043 * symtab.c (symtab_node::debug_symtab): Move definition here.
15045 2017-04-28 Richard Biener <rguenther@suse.de>
15047 * lto-streamer.h (LTO_major_version): Bump to 7.
15049 2017-04-28 Richard Biener <rguenther@suse.de>
15051 * tree-vrp.c (assert_info): New struct.
15052 (add_assert_info): New helper.
15053 (register_edge_assert_for_2): Refactor to add asserts to a vector
15055 (register_edge_assert_for_1): Likewise.
15056 (register_edge_assert_for): Likewise.
15057 (finish_register_edge_assert_for): New helper actually registering
15058 asserts where live on edge.
15059 (find_conditional_asserts): Adjust.
15060 (find_switch_asserts): Likewise.
15061 (evrp_dom_walker::try_find_new_range): Generalize.
15062 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15064 2017-04-27 Marek Polacek <polacek@redhat.com>
15067 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15068 arg10 and arg11 to itype.
15070 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15072 * doc/extend.texi (Object Size Checking): Improve grammar.
15074 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15077 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15078 that the logic for permitting reciprocal estimates matches that
15081 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15084 * tree.c (type_cache_hasher::equal): Only compare
15085 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15086 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15087 non-aggregate element types.
15088 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15089 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15091 2017-04-27 Richard Biener <rguenther@suse.de>
15093 PR middle-end/80533
15094 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15095 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15096 keeping a reference to a trailing array.
15098 2017-04-27 Richard Biener <rguenther@suse.de>
15100 PR middle-end/80539
15101 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15102 being in loop-closed SSA form conservatively.
15103 (chrec_fold_multiply_poly_poly): Likewise.
15105 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15107 PR middle-end/79665
15108 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15109 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15111 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15114 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15115 (aarch64_function_arg_alignment): Return unsigned int again, but still
15116 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15117 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15118 Don't emit -Wpsabi note.
15119 (aarch64_function_arg_boundary): Likewise.
15120 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15123 2017-04-26 Nathan Sidwell <nathan@acm.org>
15125 * tree.h (crc32_unsigned_n): Declare.
15126 (crc32_unsigned, crc32_unsigned): Make inline.
15127 * tree.c (crc32_unsigned_bits): Replace with ...
15128 (crc32_unsigned_n): ... this.
15129 (crc32_unsigned, crc32_byte): Remove.
15130 (crc32_string): Remove unnecessary braces.
15132 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15134 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15135 * ipa-inline-analysis.c (MAX_TIME): Remove.
15136 (account_size_time): Use sreal for time.
15137 (dump_inline_summary): Update.
15138 (estimate_function_body_sizes): Update.
15139 (estimate_edge_size_and_time): Update.
15140 (estimate_calls_size_and_time): Update.
15141 (estimate_node_size_and_time): Update.
15142 (inline_merge_summary): Update.
15143 (inline_update_overall_summary): Update.
15144 (estimate_time_after_inlining): Update.
15145 (inline_read_section): Update.
15146 (inline_write_summary): Update.
15147 * ipa-inline.c (compute_uninlined_call_time): Update.
15148 (compute_inlined_call_time): Update.
15149 (recursive_inlining): Update.
15150 (inline_small_functions): Update.
15151 (dump_overall_stats): Update.
15152 * ipa-inline.h: Include sreal.h.
15153 (size_time_entry): Turn time to sreal.
15154 (inline_summary): Turn self_time nad time to sreal.
15156 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15158 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15160 (sreal::stream_out, sreal::stream_in): New.
15161 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15163 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15165 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15168 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15171 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15173 (dimode_scalar_chain::compute_convert_gain): Ditto.
15174 (dimode_scalar_chain::make_vector_copies): Ditto.
15175 (dimode_scalar_chain::convert_reg): Ditto.
15176 (dimode_scalar_chain::convert_insn): Ditto.
15177 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15178 (VI248_AVX512BW_1): New mode iterator.
15179 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15180 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15183 2017-04-25 Martin Sebor <msebor@redhat.com>
15185 PR tree-optimization/80497
15186 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15187 constants are representable in HOST_WIDE_INT.
15188 (parse_directive): Ditto.
15190 2017-04-25 Martin Sebor <msebor@redhat.com>
15193 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15194 (new_zero_array): Adjust signature.
15195 (dom_info::dom_init): Used unsigned rather that size_t.
15196 (dom_info::dom_info): Same.
15198 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15199 Jakub Jelinek <jakub@redhat.com>
15202 * config/arm/arm.c: Include gimple.h.
15203 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15204 returns negative, increment ncrn only if it returned positive.
15205 (arm_needs_doubleword_align): Return int instead of bool,
15206 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15207 members, but if there is any such non-FIELD_DECL
15208 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15209 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15210 returns negative, increment nregs only if it returned positive.
15211 (arm_setup_incoming_varargs): Likewise.
15212 (arm_function_arg_boundary): Emit -Wpsabi note if
15213 arm_needs_doubleword_align returns negative, return
15214 DOUBLEWORD_ALIGNMENT only if it returned positive.
15216 2017-04-25 Marek Polacek <polacek@redhat.com>
15219 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15220 first argument to type.
15222 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15225 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15226 type checks to test for compatibility instead of equality.
15228 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15229 Jakub Jelinek <jakub@redhat.com>
15232 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15234 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15235 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15236 the alignment computation, but return their maximum in warn_alignment.
15237 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15238 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15240 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15241 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15244 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15246 * config/arc/simdext.md (dmpyh): Fix typo.
15248 2017-04-25 Richard Biener <rguenther@suse.de>
15250 PR tree-optimization/80492
15251 * alias.c (compare_base_decls): Handle registers with asm
15252 specification conservatively.
15253 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15254 compare_base_decls returning dont-know properly.
15256 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15258 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15259 (legitimate_offset_address_p): New function.
15260 (arc_legitimate_address_p): Use above function.
15262 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15264 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15266 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15268 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15269 ACCH registers whenever they are available.
15271 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15273 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15274 double regs fix when not used.
15276 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15278 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15280 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15281 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15283 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15285 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15286 when emitting short ADD instructions.
15288 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15290 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15292 (cmpsi_cc_c_insn): Likewise.
15293 (cbranchsi4_scratch): Compute proper instruction length using
15294 compact_hreg_operand.
15295 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15297 2017-04-25 Richard Biener <rguenther@suse.de>
15299 PR middle-end/80509
15300 * passes.c (pass_manager::pass_manager): Initialize
15301 m_name_to_pass_map.
15303 2017-04-25 Richard Biener <rguenther@suse.de>
15305 PR tree-optimization/79201
15306 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15308 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15311 * config/s390/vector.md: Split MEM->GPR vector moves for
15312 non-s_operand addresses.
15314 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15317 * config/s390/predicates.md (reload_const_wide_int_operand): New
15319 * config/s390/s390.md ("movti"): Remove d/P alternative.
15320 ("movti_bigconst"): New pattern definition.
15322 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15325 * s390-protos.h (s390_expand_cs_hqi): Removed.
15326 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15327 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15328 modes as well as CCZ1mode and CCZmode.
15329 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15330 signature of s390_emit_compare_and_swap.
15331 (s390_expand_cs_hqi): Likewise, make static.
15332 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15333 compare-and-swap, in some cases.
15334 (s390_expand_cs): Wrapper function.
15335 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15337 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15338 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15339 patterns for small and large integers. Forbid symref memory operands.
15340 Move expander to s390.c. Require cc register.
15341 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15342 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15343 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15344 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15345 symref memory operands. Remove CC mode and call s390_match_ccmode
15347 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15349 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15351 * config/s390/s390.md (define_peephole2): New peephole to help
15352 combining the load-and-test pattern with volatile memory.
15354 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15356 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15357 with CCZmode for TARGET_Z196.
15359 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15361 PR rtl-optimization/80501
15362 * combine.c (make_compound_operation_int): Set subreg_code to SET
15363 even for AND with mask of the sign bit of mode.
15365 PR rtl-optimization/80500
15366 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15367 sum's initial value.
15369 2017-04-25 Julian Brown <julian@codesourcery.com>
15370 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15372 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15374 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15376 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15378 2017-04-25 Julian Brown <julian@codesourcery.com>
15379 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15381 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15382 (thunderx2t99_sha): New Reservation.
15384 2017-04-25 Julian Brown <julian@codesourcery.com>
15385 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15387 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15388 type for 1-element load.
15390 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15392 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15394 2017-04-24 Martin Jambor <mjambor@suse.cz>
15396 PR tree-optimization/80293
15397 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15398 char arrays not totally scalarizable if it is false.
15399 (analyze_all_variable_accesses): Pass correct value in the new
15400 parameter. Add a statistics counter.
15402 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15404 PR middle-end/79931
15405 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15407 2017-04-24 Richard Biener <rguenther@suse.de>
15409 PR tree-optimization/80494
15410 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15411 out for complex types.
15413 2017-04-24 Richard Biener <rguenther@suse.de>
15415 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15416 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15417 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15418 (DFS): Adjust and never fail.
15419 (sccvn_dom_walker::fail): Remove.
15420 (sccvn_dom_walker::before_dom_children): Adjust.
15421 (run_scc_vn): Likewise and never fail.
15422 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15423 (pass_fre::execute): Likewise.
15425 2017-04-24 Richard Biener <rguenther@suse.de>
15427 PR tree-optimization/79725
15428 * tree-ssa-sink.c (statement_sink_location): Return whether
15429 failure reason was zero uses. Move that check later.
15430 (sink_code_in_bb): Deal with zero uses by removing the stmt
15433 2017-04-24 Richard Biener <rguenther@suse.de>
15436 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15437 pointer-based references.
15439 2017-04-24 Richard Biener <rguenther@suse.de>
15442 * pass_manager.h (pass_manager::operator new): Remove.
15443 (pass_manager::operator delete): Likewise.
15444 * passes.c (pass_manager::operator new): Remove.
15445 (pass_manager::operator delete): Likewise.
15446 (pass_manager::pass_manager): Zero individual pass members.
15448 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15451 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15452 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15453 Check "XEXP (src, 1)" operand here.
15454 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15455 Check "XEXP (src, 1)" operand here.
15456 (dimode_scalar_chain::make_vector_copies): Detect count register
15457 of a shift instruction. Zero extend count register from QImode
15458 to DImode to satisfy vector shift pattern count operand predicate.
15459 Substitute vector shift count operand with a DImode copy.
15460 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15463 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15465 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15466 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15467 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15468 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15469 (UNSPEC_NOREX_MEM): Remove definition.
15471 2017-04-21 Richard Biener <rguenther@suse.de>
15473 PR tree-optimization/79547
15474 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15475 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15476 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15477 without any constraints.
15479 2017-04-21 Richard Biener <rguenther@suse.de>
15481 PR tree-optimization/78847
15482 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15484 2017-04-21 Richard Biener <rguenther@suse.de>
15486 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15487 (build_distinct_type_copy): Likewise.
15488 (build_variant_type_copy): Likewise.
15489 * tree.c (build_qualified_type): Pass down mem-stat info.
15490 (build_distinct_type_copy): Likewise.
15491 (build_variant_type_copy): Likewise.
15493 2017-04-21 Richard Biener <rguenther@suse.de>
15495 PR tree-optimization/80237
15496 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15498 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15499 for a simplified result.
15501 2016-04-21 Richard Biener <rguenther@suse.de>
15503 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15504 sth as strict as a simple_iv but a chrec without symbols and an
15505 operand defined in the loop we are peeling (and not some subloop).
15506 (propagate_constants_for_unrolling): Propagate all constants.
15508 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15511 * config/i386/i386.c (print_reg): Remove assert for disalowed
15512 regno values, call output_operand_lossage instead.
15514 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15517 * config/i386/constraints.md (Yc): New register constraint.
15518 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15519 Use Yc constraint for alternative 2 of operand 0. Remove
15520 preferred_for_speed attribute.
15522 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15524 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15525 lastprivate clauses in SIMT case.
15527 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15529 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15531 2017-04-20 Richard Biener <rguenther@suse.de>
15533 PR tree-optimization/57796
15534 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15535 as N scalar stores.
15536 (vect_model_load_cost): Cost gathers as N scalar loads.
15538 2017-04-20 Richard Biener <rguenther@suse.de>
15540 * ggc-page.c (ggc_allocated_p): Rename to ...
15541 (safe_lookup_page_table_entry): ... this and return the lookup
15543 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15545 2017-04-20 Richard Biener <rguenther@suse.de>
15547 PR tree-optimization/80453
15548 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15549 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15550 from the conditions.
15551 (vn_phi_eq): Pass them down.
15552 (vn_phi_lookup): Record them.
15553 (vn_phi_insert): Likewise.
15555 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15557 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15558 uninitialized variable warning to avoid buffer overrun.
15560 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15563 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15564 is suppressed for '{ 0 }' in C.
15566 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15568 * BASE-VER: Set to 8.0.0.
15570 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15572 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15573 priority .init_array and .fini_array section with SECTION_NOTYPE
15576 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15578 PR middle-end/80423
15579 * tree.h (build_array_type): Add typeless_storage default argument.
15580 * tree.c (type_cache_hasher::equal): Also compare
15581 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15582 (build_array_type): Add typeless_storage argument, set
15583 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15585 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15586 (build_array_type): Likewise. Add typeless_storage argument.
15588 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15589 Jakub Jelinek <jakub@redhat.com>
15591 PR tree-optimization/80426
15592 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15593 operation on symbolic operands, also compute the overflow for the
15594 invariant part when the operation degenerates into a negation.
15596 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15599 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15600 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15603 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15605 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15608 * config/avr/avr.c (tree.h): Include it.
15609 (cgraph.h): Include it.
15610 (avr_encode_section_info): Don't warn for uninitialized progmem
15611 variable if it's just an alias.
15613 2017-04-19 Richard Biener <rguenther@suse.de>
15616 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15617 when needed by AutoPGO.
15619 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15622 * doc/lto.texi: Remove an extra 'that'.
15624 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15626 PR rtl-optimization/80429
15627 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15628 are only used in debug insns.
15630 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15631 Vladimir Makarov <vmakarov@redhat.com>
15633 * config/sparc/predicates.md (input_operand): Add comment. Return
15634 true for any memory operand when LRA is in progress.
15635 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15637 2017-04-18 Jeff Law <law@redhat.com>
15640 * mips.md ({return,simple_return}_internal): Do not overwrite
15643 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15645 PR tree-optimization/80443
15646 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15647 instead of adding 1, subtract -1 and similarly instead of subtracting
15650 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15652 PR rtl-optimization/80357
15653 * haifa-sched.c (tmp_bitmap): New variable.
15654 (model_recompute): Handle duplicate use records.
15655 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15656 (free_global_sched_pressure_data): Free it.
15658 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15661 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15662 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15663 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15664 instead of SYSTEM_HEADER_DIR.
15666 2017-04-18 Jeff Law <law@redhat.com>
15668 PR middle-end/80422
15669 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15670 predecessors after walking up the insn chain.
15672 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15675 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15676 sizetype type into debug info.
15678 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15681 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15682 unneeded test for TARGET_UPPER_REGS_SF.
15683 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15685 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15688 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15689 instead of gsi_after_labels.
15691 2017-04-18 Jeff Law <law@redhat.com>
15693 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15697 2017-04-13 Jeff Law <law@redhat.com>
15698 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15699 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15701 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15704 * config/avr/avr.c (intl.h): Include it.
15705 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15707 2017-04-18 Martin Liska <mliska@suse.cz>
15709 PR gcov-profile/78783
15710 * gcov-tool.c (gcov_output_files): Validate that destination
15711 file is either removed by the tool or by a user.
15713 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15714 Guy Benyei <guybe@mellanox.com>
15716 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15717 block, and do not negate it, the stored id is already negative.
15719 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15721 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15723 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15726 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15727 masks of options that should be turned off if the VSX vector
15728 options are turned off.
15729 (OTHER_P8_VECTOR_MASKS): Likewise.
15730 (OTHER_VSX_VECTOR_MASKS): Likewise.
15731 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15732 rs6000_disable_incompatible_switches to validate no type switches
15734 (rs6000_incompatible_switch): New function to disallow turning on
15735 other vector options if -mno-vsx, -mno-power8-vector, or
15736 -mno-power9-vector are specified.
15738 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15740 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15742 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15744 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15745 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15746 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15747 (ARG_POINTER_CFA_OFFSET): Likewise.
15749 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15751 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15752 conditions to take advantage of various optimizations.
15754 2017-04-13 Jeff Law <law@redhat.com>
15756 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15757 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15758 (zero_extendsidi2_dext): Likewise.
15760 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15763 * fold-const.c (fold_ternary_loc): Revert
15764 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15767 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15769 PR rtl-optimization/80343
15770 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15771 new scratch pseudo.
15773 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15776 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15777 to ubsan_encode_value.
15779 2017-04-13 Jeff Law <law@redhat.com>
15781 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15782 appearing in DEBUG_INSNs.
15784 2017-04-13 Martin Liska <mliska@suse.cz>
15786 PR gcov-profile/80413
15787 * gcov-io.c (gcov_write_string): Copy to buffer just when
15788 allocated size is greater than zero.
15790 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15793 * dwarf2out.c (decls_for_scope): Ignore declarations of
15794 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15796 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15799 * ipa-visibility.c (non_local_p): Fix typos.
15800 (localize_node): When localizing symbol in same comdat group,
15801 dissolve the group only when we know external symbols are going
15803 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15805 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15807 PR tree-optimization/79390
15808 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15809 order does not result in usable sequence, retry with reversed operand
15815 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15816 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15817 op0 instead of fold_convert_loc (loc, type, arg0).
15819 2017-04-12 Jeff Law <law@redhat.com>
15821 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15822 has a delay slot in the generated code.
15824 * config/cris/cris.md (cris_preferred_reload_class): Return
15825 GENNONACR_REGS rather than GENERAL_REGS.
15827 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15830 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15831 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15832 signedness of the result type.
15834 2017-04-12 Richard Biener <rguenther@suse.de>
15835 Jeff Law <law@redhat.com>
15837 PR tree-optimization/80359
15838 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15839 trim stores to TARGET_MEM_REFs.
15841 2017-04-12 Richard Biener <rguenther@suse.de>
15843 PR tree-optimization/79390
15844 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15845 threading case even more.
15847 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15850 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15851 for quad_address_p for TImode, instead of just not indexed_address.
15853 2017-04-12 Richard Biener <rguenther@suse.de>
15854 Bernd Edlinger <bernd.edlinger@hotmail.de>
15856 PR middle-end/79671
15857 * alias.c (component_uses_parent_alias_set_from): Handle
15858 TYPE_TYPELESS_STORAGE.
15859 (get_alias_set): Likewise.
15860 * tree-core.h (tree_type_common): Add typeless_storage flag.
15861 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15862 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15863 for types containing members with TYPE_TYPELESS_STORAGE.
15864 (place_field): Likewise.
15865 (layout_type): Likewise for ARRAY_TYPE.
15866 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15867 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15868 TYPE_TYPELESS_STORAGE.
15869 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15871 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15874 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15875 first argument to type.
15877 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15881 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15882 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15883 (rs6000_expand_binop_builtin): Likewise.
15884 (rs6000_expand_ternop_builtin): Likewise; also add missing
15885 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15886 vshasigma built-ins.
15887 * doc/extend.texi: Document that vec_xxpermdi's third argument
15888 must be a constant.
15890 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15892 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15893 Use shift_const cost parameter when calculating gain of STV shifts.
15895 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15897 PR rtl-optimization/70478
15898 * lra-constraints.c (process_alt_operands): Check memory for
15899 disfavoring memory insn operand.
15901 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15903 PR middle-end/80100
15904 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15905 left shift in unsigned HOST_WIDE_INT type.
15907 PR rtl-optimization/80385
15908 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15909 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15912 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15913 if they have any depend clauses.
15915 2017-04-11 Martin Liska <mliska@suse.cz>
15918 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15919 * ipa-split.c (split_function): Create a local comdat symbol
15920 if caller is in a comdat group.
15922 2017-04-11 Martin Liska <mliska@suse.cz>
15925 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15928 2017-04-11 Martin Sebor <msebor@redhat.com>
15930 PR middle-end/80364
15931 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15932 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15934 (directive::set_width, directive::set_precision, format_character):
15936 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15939 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15942 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15943 conflict, set target->arch_name instead of target->cpu_name.
15945 2017-04-11 Richard Biener <rguenther@suse.de>
15947 PR tree-optimization/80374
15948 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15949 build_zero_cst, remove fold_convertible_p check again.
15951 2017-04-11 Martin Liska <mliska@suse.cz>
15954 * ubsan.c (instrument_object_size): Do not instrument register
15957 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15960 * config/i386/i386-builtin-types.def
15961 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15962 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15963 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15964 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15965 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15966 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15967 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15968 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15969 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15970 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15971 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15972 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15973 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15974 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15975 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15976 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15977 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15978 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15979 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15980 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15981 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15982 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15983 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15984 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15985 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15986 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15987 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15988 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15989 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15990 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15991 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15992 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15993 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15994 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15995 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15996 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15997 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15998 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15999 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16000 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16001 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16002 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16003 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16004 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16005 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16006 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16008 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16009 flag to second_arg_count, handle 4 argument function type _COUNT
16010 aliases, handle second_arg_count on second argument rather than last.
16012 2017-04-10 Jeff Law <law@redhat.com>
16014 PR tree-optimization/80374
16015 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16016 record anything if we can not convert integer_zero_node to the
16019 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16022 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16023 Enhance special handling given to the TARGET_P9_MINMAX option in
16024 relation to certain other options.
16026 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16028 PR tree-optimization/80153
16029 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16030 remove POINTER_PLUS_EXPR's base part directly, rather than through
16033 2017-04-10 Richard Biener <rguenther@suse.de>
16034 Bin Cheng <bin.cheng@arm.com>
16036 PR tree-optimization/80153
16037 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16038 the first element of pointer type aff_tree. Build result expr in
16040 (add_elt_to_tree): Convert to type unconditionally. Remove other
16041 fold_convert calls.
16042 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16043 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16045 2017-04-10 Richard Biener <rguenther@suse.de>
16047 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16050 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16052 PR rtl-optimization/70478
16053 * lra-constraints.c (curr_small_class_check): New.
16054 (update_and_check_small_class_inputs): New.
16055 (process_alt_operands): Update curr_small_class_check. Disfavor
16056 alternative insn memory operands. Check available regs for small
16059 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16062 * config/mips/mips.opt (-mvirt): Update description.
16063 * doc/invoke.texi (-mvirt): Likewise.
16065 2017-04-10 Richard Biener <rguenther@suse.de>
16067 PR middle-end/80362
16068 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16069 looking for NEGATE_EXPR in -A / -B to A / B folding.
16071 2017-04-10 Martin Liska <mliska@suse.cz>
16073 PR gcov-profile/80224
16074 * gcov.c (print_usage): Fix usage string.
16075 (get_gcov_intermediate_filename): Remove.
16076 (output_gcov_file): Use both for normal and intermediate format.
16077 (generate_results): Do not initialize special file for
16078 intermediate format.
16080 2017-04-10 Richard Biener <rguenther@suse.de>
16082 PR tree-optimization/80304
16083 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16086 2017-04-10 Nathan Sidwell <nathan@acm.org>
16089 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16090 (rs6000_init_builtins): Use it.
16092 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16094 * config/arm/arm.md (<mrc>): Add mode to SET source.
16095 (<mrrc>): Likewise.
16097 2017-04-10 Richard Biener <rguenther@suse.de>
16099 PR middle-end/80344
16100 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16102 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16105 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16106 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16107 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16108 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16109 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16110 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16111 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16112 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16113 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16114 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16115 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16116 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16117 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16118 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16119 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16120 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16121 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16122 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16123 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16124 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16125 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16126 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16127 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16129 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16131 PR rtl-optimization/70478
16132 * lra-constraints.c: Reverse the last patch.
16134 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16136 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16137 Add comment for WCHAR_T.
16139 2017-04-08 Martin Liska <mliska@suse.cz>
16142 2017-04-07 Martin Liska <mliska@suse.cz>
16145 * ipa-split.c (split_function): Add function part to a same comdat
16148 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16151 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16153 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16155 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16156 (vec_pairing): ...to this.
16157 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16158 (rs6000_sched_init): Adjust for name change.
16159 (struct rs6000_sched_context): Likewise.
16160 (rs6000_init_sched_context): Likewise.
16161 (rs6000_set_sched_context): Likewise.
16163 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16169 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16171 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16172 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16173 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16175 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16177 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16179 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16181 PR rtl-optimization/70703
16182 * ira-color.c (update_conflict_hard_regno_costs): Use
16183 int64_t instead of HOST_WIDE_INT.
16185 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16187 PR rtl-optimization/70478
16188 * lra-constraints.c (process_alt_operands): Disfavor alternative
16189 insn memory operands.
16191 2017-04-07 Jeff Law <law@redhat.com>
16193 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16194 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16196 2017-04-07 Martin Liska <mliska@suse.cz>
16199 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16200 Show error message instead of an ICE.
16202 2017-04-07 Martin Liska <mliska@suse.cz>
16205 * ipa-split.c (split_function): Add function part to a same comdat
16208 2017-04-07 Richard Biener <rguenther@suse.de>
16210 PR middle-end/80341
16211 * tree.c (get_unwidened): Also handle ! for_type case for
16213 * convert.c (do_narrow): Split out from ...
16214 (convert_to_integer_1): ... here. Do not pass final truncation
16215 type to get_unwidened for TRUNC_DIV_EXPR.
16217 2017-04-07 Richard Biener <rguenther@suse.de>
16219 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16221 (aff_combination_const): Adjust.
16222 (aff_combination_scale): Likewise.
16223 (aff_combination_add_elt): Likewise.
16224 (aff_combination_add_cst): Likewise.
16225 (aff_combination_convert): Likewise.
16226 (add_elt_to_tree): Likewise. Remove unused argument.
16227 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16229 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16231 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16233 * config/arm/arm.c (arm_default_short_enums): Use
16234 ARM_DEFAULT_SHORT_ENUMS.
16235 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16237 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16240 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16241 members with redundant out-of-class redeclaration.
16243 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16246 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16247 * config/i386/i386.md (*zero_extendsidi2):
16248 Add (?*x,*x) and (?*v,*v) alternatives.
16250 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16253 * config/i386/i386.c (ix86_expand_builtin)
16254 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16255 mode from insn data. Convert operands to insn operand mode.
16256 Copy operands that don't satisfy insn predicate to a register.
16258 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16260 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16263 2017-04-06 Richard Biener <rguenther@suse.de>
16265 PR tree-optimization/80334
16266 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16267 preserve alignment of accesses.
16269 2017-04-06 Richard Biener <rguenther@suse.de>
16271 PR tree-optimization/80262
16272 * tree-sra.c (build_ref_for_offset): Preserve address-space
16274 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16275 Drop useless address-space information on MEM_REF offsets.
16277 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16279 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16281 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16283 PR rtl-optimization/70703
16284 * ira-color.c (update_conflict_hard_regno_costs): Use
16285 HOST_WIDE_INT instead of long.
16287 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16290 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16291 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16293 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16294 for x86_64 target. Handle -m3dnowa option.
16296 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16298 PR rtl-optimization/70703
16299 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16300 (update_conflict_hard_regno_costs): Use long instead of unsigned
16301 arithmetic for cost calculation.
16303 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16304 Bernd Edlinger <bernd.edlinger@hotmail.de>
16307 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16310 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16313 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16314 ptr_mode with Pmode throughout.
16315 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16316 into probe_stack_range and use DImode.
16318 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16321 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16322 call_eh_return is true.
16324 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16326 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16327 Initialize last_match_fntype_index.
16329 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16332 * tree-nvr.c: Include internal-fn.h.
16333 (pass_return_slot::execute): Ignore internal calls without
16336 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16337 Richard Biener <rguenther@suse.de>
16340 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16341 captures used multiple times, except for the last use.
16342 * generic-match-head.c: Include gimplify.h.
16344 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16346 PR tree-optimization/79390
16347 * target.h (struct noce_if_info): Declare.
16348 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16349 * target.def (noce_conversion_profitable_p): New target hook.
16350 * ifcvt.h (struct noce_if_info): New type, moved from ...
16351 * ifcvt.c (struct noce_if_info): ... here.
16352 (noce_conversion_profitable_p): Renamed to ...
16353 (default_noce_conversion_profitable_p): ... this. No longer
16355 (noce_try_store_flag_constants, noce_try_addcc,
16356 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16357 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16358 instead of noce_conversion_profitable_p.
16359 * config/i386/i386.c: Include ifcvt.h.
16360 (ix86_option_override_internal): Don't override
16361 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16362 (ix86_noce_conversion_profitable_p): New function.
16363 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16364 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16365 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16366 * doc/tm.texi: Regenerated.
16368 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16370 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16373 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16376 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16377 instructions for small multiply cores.
16379 2017-04-04 Jeff Law <law@redhat.com>
16381 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16383 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16384 that are not set by the loop over the elements.
16386 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16389 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16390 int mode, convert_modes it to mode as unsigned, otherwise use
16391 lowpart_subreg to mode rather than SImode.
16392 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16393 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16394 Use DImode instead of SImode for the shift count operand.
16395 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16398 2017-04-04 Richard Biener <rguenther@suse.de>
16400 PR middle-end/80281
16401 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16402 arithmetic done for the negate or the plus. Simplify.
16403 (A - (-B) -> A + B): Likewise.
16404 * fold-const.c (split_tree): Make sure to not negate pointers.
16406 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16408 PR rtl-optimization/60818
16409 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16410 a compare of comparisons with the thing compared if this results
16411 in a different machine mode.
16413 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16415 * alias.c (base_alias_check): Fix typo in comment.
16416 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16417 * cgraphunit.c (symbol_table::compile): Likewise.
16418 * collect2.c (maybe_run_lto_and_relink): Likewise.
16419 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16420 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16421 * config/avr/avr.c (avr_map_op_t): Likewise.
16422 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16423 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16424 * config/epiphany/epiphany.md (movcc): Likewise.
16425 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16426 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16428 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16429 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16430 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16431 * config/sh/sh.c (sh_rtx_costs): Likewise.
16432 * fold-const.c (fold_truth_andor): Likewise.
16433 * genautomata.c (collapse_flag): Likewise.
16434 * gengtype.h (struct type::u::s): Likewise.
16435 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16436 * input.c (FORMAT_AMOUNT): Likewise.
16437 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16438 (known_aggs_to_agg_replacement_list): Likewise.
16439 * ipa-inline-analysis.c: Likewise.
16440 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16441 * ipa-polymorphic-call.c
16442 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16443 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16444 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16446 * modulo-sched.c (apply_reg_moves): Likewise.
16447 * omp-expand.c (build_omp_regions_1): Likewise.
16448 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16449 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16450 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16451 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16452 * value-prof.c: Likewise.
16453 * var-tracking.c (val_reset): Likewise.
16455 2017-04-03 Richard Biener <rguenther@suse.de>
16457 PR tree-optimization/80275
16458 * fold-const.c (split_address_to_core_and_offset): Handle
16461 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16463 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16464 descriptors is at least equal to that of functions.
16466 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16468 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16470 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16473 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16474 (mov<IMOD4:mode>): New expander.
16475 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16477 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16479 PR rtl-optimization/79405
16480 * fwprop.c (propagations_left): New variable.
16481 (forward_propagate_into): Decrement it.
16482 (fwprop_init): Initialize it.
16483 (fw_prop): If the variable has reached zero, stop propagating.
16484 (fwprop_addr): Ditto.
16486 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16489 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16490 a FUNCTION_DECL, pass it as decl instead of origin to
16493 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16495 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16498 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16501 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16502 TARGET_VSX_SMALL_INTEGER.
16504 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16506 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16507 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16509 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16511 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16512 extraction from odd-numbered MSA register.
16514 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16516 PR middle-end/80173
16517 * expmed.c (store_bit_field_1): Don't attempt to create
16518 a word subreg out of hard registers wider than word if they
16519 have HARD_REGNO_NREGS of 1 for their mode.
16521 PR middle-end/80163
16522 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16523 conversions to integer types wider than word and pointer.
16526 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16527 (rtx_equal_for_cselib_p): Pass 0 to it.
16528 * cselib.c (cselib_hasher::equal): Likewise.
16529 (rtx_equal_for_cselib_1): Add depth argument. If depth
16530 is 128, don't look up VALUE locs and punt. Increment
16531 depth in recursive calls when walking VALUE locs.
16533 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16535 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16536 (make_gcov_file_name): Use the canonical path name for generating
16538 (read_line): Fix handling of files with ascii null bytes.
16540 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16542 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16543 to initialise a vector register instead
16544 of using a const_int.
16546 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16548 PR translation/80189
16549 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16550 diagnostic messages.
16552 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16555 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16556 (dfp_diex_<mode>): Update mode of operand 1.
16557 * doc/extend.texi (dxex, dxexq): Document change to return type.
16558 (diex, diexq): Document change to argument type.
16560 2017-03-30 Martin Jambor <mjambor@suse.cz>
16563 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16564 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16565 it reflects the signature changes performed at the callee side.
16566 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16567 to cgraph_build_function_type_skip_args.
16568 (build_function_decl_skip_args): Adjust call to the above function.
16570 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16573 * config/i386/sse.md
16574 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16575 register as dest whenever it is a MEM not rtx_equal_p to the
16576 corresponding dup operand, and when forcing into reg move the
16577 reg into the memory afterwards.
16578 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16579 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16580 for the force_reg mode.
16581 (avx512vl_vextractf128<mode>): Use register as dest either
16582 always when a MEM, or when it is a MEM not rtx_equal_p to the
16583 corresponding dup operand, or even not when it is a CONST_VECTOR
16584 depending on the mode and lo vs. hi.
16585 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16587 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16588 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16589 Likewise. Require that operands[2] is even.
16590 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16591 Remove extraneous parens. Require that operands[2] is a multiple
16593 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16594 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16595 disallow memory then.
16597 2017-03-30 Richard Biener <rguenther@suse.de>
16599 PR tree-optimization/77498
16600 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16601 to non-constants over backedges.
16603 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16605 PR rtl-optimization/80233
16606 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16607 as last_combined_insn. Do not test for BARRIER_P separately.
16609 2017-03-29 Andreas Schwab <schwab@suse.de>
16612 * calls.c (prepare_call_address): Convert funexp to Pmode before
16613 copying to temp reg.
16615 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16617 PR tree-optimization/80158
16618 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16619 Handle possible future case of more than one alternate
16621 (replace_rhs_if_not_dup): Likewise.
16622 (replace_one_candidate): Likewise.
16624 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16626 PR rtl-optimization/80193
16627 * ira.c (ira): Do not check allocation for LRA.
16629 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16631 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16632 (nvptx_output_simt_exit): Declare.
16633 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16634 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16635 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16636 (nvptx_declare_function_name): ...here. Emit declaration of local
16637 memory space buffer for omp_simt_enter insn.
16638 (nvptx_output_unisimt_switch): New.
16639 (nvptx_output_softstack_switch): New.
16640 (nvptx_output_simt_enter): New.
16641 (nvptx_output_simt_exit): New.
16642 * config/nvptx/nvptx.h (struct machine_function): New fields
16643 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16644 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16645 (UNSPECV_SIMT_EXIT): Ditto.
16646 (omp_simt_enter_insn): New insn.
16647 (omp_simt_enter): New expansion.
16648 (omp_simt_exit): New insn.
16649 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16651 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16652 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16653 (expand_GOMP_SIMT_EXIT): New.
16654 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16655 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16656 (GOMP_SIMT_EXIT): Ditto.
16657 * target-insns.def (omp_simt_enter): New insn.
16658 (omp_simt_exit): Ditto.
16659 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16661 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16662 (lower_rec_input_clauses): Likewise.
16663 (lower_lastprivate_clauses): Handle SIMT privatization.
16665 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16666 (ompdevlow_adjust_simt_enter): New.
16667 (find_simtpriv_var_op): New.
16668 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16669 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16671 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16672 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16673 (copy_decl_for_dup_finish): Ditto.
16675 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16677 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16680 * config/i386/i386.c (ix86_option_override_internal): Always
16681 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16683 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16685 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16687 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16689 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16690 mark new edge's irreducible flag accordign to it.
16691 (vect_do_peeling): Check loop preheader edge's irreducible flag
16692 and pass it to function slpeel_add_loop_guard.
16694 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16696 PR tree-optimization/80218
16697 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16698 Update block frequencies and counts.
16700 2017-03-28 Richard Biener <rguenther@suse.de>
16702 PR tree-optimization/78644
16703 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16704 of a simplification result we may not use it at all.
16706 2017-03-28 Richard Biener <rguenther@suse.de>
16709 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16710 without arguments, generate default definition of a SSA name.
16712 2017-03-28 Richard Biener <rguenther@suse.de>
16714 PR middle-end/80222
16715 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16716 TYPE_REF_CAN_ALIAS_ALL references.
16717 * fold-const.c (fold_indirect_ref_1): Likewise.
16719 2017-03-28 Martin Liska <mliska@suse.cz>
16722 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16723 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16725 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16726 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16728 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16729 (EXTRA_SPECS): Define.
16730 (SUBTARGET_EXTRA_SPECS): Likewise.
16731 (SUBTARGET_CPP_SPEC): Likewise.
16732 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16733 SUBTARGET_EXTRA_SPECS.
16734 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16736 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16738 * config/arc/simdext.md (vst64_insn): Update pattern.
16739 (vld32wh_insn): Likewise.
16740 (vld32wl_insn): Likewise.
16741 (vld64_insn): Likewise.
16742 (vld32_insn): Likewise.
16744 2017-03-28 Marek Polacek <polacek@redhat.com>
16747 * fold-const.c (fold_comparison): Use protected_set_expr_location
16748 instead of SET_EXPR_LOCATION.
16750 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16752 * tree.c (add_expr): Avoid name lookup warning.
16754 2017-03-27 Jeff Law <law@redhat.com>
16756 PR tree-optimization/80216
16757 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16758 function name. Limit recursion depth.
16759 (record_temporary_equivalences): Corresponding changes.
16761 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16763 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16766 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16769 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16771 * cfgcleanup.c (reg_note_cfa_p): New array.
16772 (insns_have_identical_cfa_notes): New function.
16773 (old_insns_match_p): Don't cross-jump in between /f
16774 and non-/f instructions. If both i1 and i2 are frame related,
16775 verify all CFA notes, their order and content.
16777 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16780 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16781 HImode and SImode with zero extend to DImode to one insn.
16782 (bswap<mode>2_extenddi): Likewise.
16783 (bswapsi2_extenddi): Likewise.
16784 (bswaphi2_extendsi): Likewise.
16785 (bswaphi2): Combine bswap HImode and SImode into one insn.
16786 Separate memory insns from swapping register.
16787 (bswapsi2): Likewise.
16788 (bswap<mode>2): Likewise.
16789 (bswaphi2_internal): Delete, no longer used.
16790 (bswapsi2_internal): Likewise.
16791 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16792 store, and gpr<-gpr swap insns.
16793 (bswap<mode>2_store): Likewise.
16794 (bswaphi2_reg): Register only splitter, combine with the splitter.
16795 (bswaphi2 splitter): Likewise.
16796 (bswapsi2_reg): Likewise.
16797 (bswapsi2 splitter): Likewise.
16798 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16799 the insns into load, store, and register/register insns.
16800 (bswapdi2_ldbrx): Likewise.
16801 (bswapdi2_load): Likewise.
16802 (bswapdi2_store): Likewise.
16803 (bswapdi2_reg): Likewise.
16805 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16807 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16808 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16810 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16813 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16816 special handling for target option conflicts between dform
16817 options (-mpower9-dform, -mpower9-dform-vector,
16818 -mpower9-dform-scalar) and -mno-direct-move.
16820 2017-03-27 Richard Biener <rguenther@suse.de>
16822 PR tree-optimization/80181
16823 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16825 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16827 * config/arc/predicates.md (move_double_src_operand): Replace the
16828 call to move_double_src_operand with a call to address_operand.
16830 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16832 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16833 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16834 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16836 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16838 * config/arc/predicates.md (long_immediate_loadstore_operand):
16839 Consider scaled addresses cases.
16841 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16843 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16844 restored when in interrupt.
16845 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16846 doesn't have delay slot.
16848 2017-03-27 Richard Biener <rguenther@suse.de>
16851 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16852 inlined thunk clones.
16854 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16857 * asan.c (instrument_derefs): Copy over last operand from
16858 original COMPONENT_REF to the new COMPONENT_REF with
16859 DECL_BIT_FIELD_REPRESENTATIVE.
16860 * ubsan.c (instrument_object_size): Likewise.
16862 2017-03-27 Richard Biener <rguenther@suse.de>
16864 PR tree-optimization/80170
16865 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16866 sure DR/SCEV didnt fold in constants we do not see when looking
16867 at the reference base alignment.
16869 2017-03-27 Richard Biener <rguenther@suse.de>
16871 PR middle-end/80171
16872 * gimple-fold.c (fold_ctor_reference): Properly guard against
16873 NULL return value from canonicalize_constructor_val.
16875 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16878 * config/i386/i386.c (ix86_expand_builtin)
16879 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16880 flags reg setting and flags reg using instructions.
16881 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16882 clobbering instructions to zero extend op2.
16884 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16886 * doc/install.texi (Configuration) <--with-aix-soname>:
16887 Update link to AIX ld.
16889 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16891 PR rtl-optimization/80160
16892 PR rtl-optimization/80159
16893 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16894 reg_alternate_class into account.
16896 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16899 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16900 to consider in curr_insn_transform.
16902 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16904 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16905 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16906 and emit_mode_inner.
16908 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16910 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16911 argument to the overloaded builtin variants. Use the new flag to
16912 deprecate certain builtin variants.
16913 * config/s390/s390-builtin-types.def: Add new builtin types.
16914 * config/s390/s390-builtins.h: Support new flags field for
16915 overloaded builtins.
16916 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16917 (s390_macro_to_expand): Enable vector float data type.
16918 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16919 builtins by incrementing the __VEC__ version number.
16920 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16922 (s390_resolve_overloaded_builtin): Emit error messages depending
16923 on the builtin flags.
16924 * config/s390/s390.c (s390_expand_builtin): Support additional
16925 flags argument. Change error message to match the messages
16926 emitted in s390-c.c.
16927 * config/s390/s390.md: New UNSPEC_* constants.
16928 (op_type): Add new instruction types.
16929 * config/s390/vecintrin.h: Add new builtins and test data class
16931 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16932 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16933 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16934 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16935 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16936 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16938 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16939 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16940 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16941 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16943 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16944 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16945 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16946 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16948 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16949 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16950 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16951 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16952 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16953 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16954 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16956 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16957 ("vec_scatter_element<V_HW_4:mode>_DI")
16958 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16959 ("vec_fpint<mode>", "vflls")
16960 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16961 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16962 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16963 ("*vec_cmphe<mode>_cc"): ... these.
16965 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16966 mode constant instead of magic value.
16968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16970 * config/s390/s390.c (s390_expand_vec_compare): Support other
16971 vector floating point modes than just V2DF.
16972 (s390_expand_vcond): Likewise.
16973 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16974 (s390_cannot_change_mode_class): Prevent mode changes between TF
16975 and V1TF in vector registers.
16976 * config/s390/s390.md (DF, SF): New mode attributes.
16977 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16978 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16979 SFmode support for VRs.
16980 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16982 (VFT, VF_HW): New mode iterators.
16983 (vw, sdx): New mode attributes.
16984 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16985 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16986 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16987 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16988 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16989 also the new vector floating point modes. Renaming to ...
16991 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16992 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16993 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16994 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16995 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16996 ("vec_unordered<mode>"): ... these.
16998 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16999 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17000 ("*vec_extendv2df"): New insn definitions.
17002 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17004 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17005 ("mulditi3_2", "*muldi3_sign"): New patterns.
17006 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17007 rename the pattern definition.
17009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17011 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17013 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17015 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17017 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17018 instruction if possible.
17019 * config/s390/vector.md (vec_halfnumelts): New mode
17021 ("*vec_vllezlf<mode>"): New pattern.
17023 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17025 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17026 ("popcountv4si2", "popcountv2di2"): Rename to ...
17027 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17028 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17030 ("popcount<mode>2_vxe"): New pattern.
17032 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17034 * common/config/s390/s390-common.c (processor_flags_table): Add
17036 * config.gcc: Add arch12.
17037 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17038 Default to arch12 for unknown CPU model numbers.
17039 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17040 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17041 PROCESSOR_max sanity check.
17042 * config/s390/s390-opts.h (enum processor_type): Add
17044 * config/s390/s390.c (processor_table): Add arch12.
17045 (s390_expand_builtin): Add check for B_VXE flag.
17046 (s390_issue_rate): Add PROCESSOR_ARCH12.
17047 (s390_get_sched_attrmask): Likewise.
17048 (s390_get_unit_mask): Likewise.
17049 (s390_sched_score): Enable z13 scheduling for arch12.
17050 (s390_sched_reorder): Likewise.
17051 (s390_sched_variable_issue): Likewise.
17052 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17054 (s390_tune_attr): Use z13 scheduling also for arch12.
17055 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17056 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17057 (TARGET_VXE_P): New macros.
17058 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17059 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17060 * config/s390/s390.opt: Add arch12 as processor_type.
17062 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17064 * config/s390/s390.md
17065 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17066 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17067 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17069 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17070 Rename expanders to ...
17072 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17073 ("fixuns_truncdddi2_emu"): ... these.
17075 ("fixuns_trunc<mode>si2_emu"): New expander.
17077 ("*fixuns_truncdfdi2_z13"): Rename to ...
17078 ("*fixuns_truncdfdi2_vx"): ... this.
17080 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17082 * config/s390/2964.md: Remove the single element vector compare
17083 instructions which are no longer used.
17084 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17086 (s390_canonicalize_comparison): Remove handling of DFmode
17088 (s390_expand_vec_compare_scalar): Remove function.
17089 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17090 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17092 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17094 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17096 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17098 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17099 will anyway by matched by mov<mode>_64dfp.
17101 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17103 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17104 vlef/vstef. Add missing operand to vleif.
17106 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17108 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17109 pair for all vector types with 64 bit elements.
17110 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17111 * config/s390/vector.md (V_HW_64): ... here.
17112 (V_128_NOSINGLE): New mode iterator.
17113 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17114 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17115 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17116 ("*vec_load_pairv2di"): Change to ...
17117 ("*vec_load_pair<mode>"): ... this one.
17119 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17121 * config/s390/constraints.md: Add comments.
17122 (jKK): Reject element sizes > 8 bytes.
17123 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17125 * config/s390/s390.md: Add the s_operand checks formerly in
17126 s390_split_ok_p to various splitters where they are still
17128 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17129 for 128 bit vectors. Plus two splitters.
17131 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17133 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17136 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17139 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17140 error if the boundary argument is not constant.
17142 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17144 PR rtl-optimization/80112
17145 * loop-doloop.c (doloop_condition_get): Don't check condition
17146 if cmp isn't SET with IF_THEN_ELSE src.
17148 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17150 PR tree-optimization/80158
17151 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17152 replacing a candidate statement, also replace it for the
17153 candidate's alternate interpretation.
17154 (replace_rhs_if_not_dup): Likewise.
17155 (replace_one_candidate): Likewise.
17157 2017-03-24 Richard Biener <rguenther@suse.de>
17159 PR tree-optimization/80167
17160 * graphite-isl-ast-to-gimple.c
17161 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17163 (translate_isl_ast_to_gimple::get_rename): Likewise.
17165 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17167 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17168 handling of certain combinations of target options, including the
17169 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17170 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17172 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17175 * config/arm/arm.md (*load_multiple): Add reload_completed to
17176 matching condition.
17178 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17179 Richard Biener <rguenth@suse.de>
17181 PR tree-optimization/79908
17182 PR tree-optimization/80136
17183 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17184 been cast away, gimplify_and_add suffices.
17186 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17188 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17190 2017-03-23 Richard Biener <rguenther@suse.de>
17192 PR tree-optimization/80032
17193 * gimplify.c (gimple_push_cleanup): Forced unconditional
17194 cleanups still have to go to the conditional_cleanups
17197 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17199 PR tree-optimization/80072
17200 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17202 (next_operand_entry_id): Change type to unsigned int.
17203 (sort_by_operand_rank): Make sure to return the right return value
17204 even if unsigned fields are bigger than INT_MAX.
17205 (struct oecount): Change cnt and id type to unsigned int.
17206 (oecount_hasher::equal): Formatting fix.
17207 (oecount_cmp): Make sure to return the right return value
17208 even if unsigned fields are bigger than INT_MAX.
17209 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17212 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17213 TREE_READONLY on result if writing it more than once.
17216 * doc/invoke.texi (-fsanitize=thread): Document that with
17217 -fnon-call-exceptions atomics are not able to throw
17221 * tsan.c: Include tree-eh.h.
17222 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17223 maybe_clean_or_replace_eh_stmt where needed.
17224 (instrument_memory_accesses): Add cfg_changed argument.
17225 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17226 if it returned true.
17227 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17229 PR rtl-optimization/63191
17230 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17231 wrapper function, moved the whole old content into ...
17232 (ix86_delegitimize_address_1): ... this. New inline function.
17233 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17234 true as last argument instead of ix86_delegitimize_address.
17236 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17238 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17239 cortexa57_branch_cost.
17241 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17243 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17245 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17248 * doc/md.texi (Constraints): Document wA constraint.
17249 * config/rs6000/constraints.md (wA): New.
17250 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17251 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17252 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17253 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17255 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17258 * gimplify.c (is_oacc_declared): New function.
17259 (oacc_default_clause): Use it to set default flags for acc declared
17260 variables inside parallel regions.
17261 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17262 declared variables.
17263 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17264 declare attribute to any decl as necessary.
17266 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17269 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17270 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17271 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17272 (arm_arch_lpae): This.
17273 * config/arm/arm.c (arm_arch7ve): Rename into ...
17274 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17275 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17278 2017-03-22 Martin Liska <mliska@suse.cz>
17281 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17282 error message instead of an ICE.
17284 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17286 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17288 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17290 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17292 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17295 2017-03-21 Martin Sebor <msebor@redhat.com>
17297 * doc/extend.texi: Use "cannot" instead of "can't."
17298 * doc/hostconfig.texi: Same.
17299 * doc/install.texi: Same.
17300 * doc/invoke.texi: Same.
17301 * doc/loop.texi: Same.
17302 * doc/md.texi: Same.
17303 * doc/objc.texi: Same.
17304 * doc/rtl.texi: Same.
17305 * doc/tm.texi: Same.
17306 * doc/tm.texi.in: Same.
17307 * doc/trouble.texi: Same.
17309 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17312 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17313 (collect_checksum_attributes): Set it.
17314 (die_checksum_ordered): Use it.
17316 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17318 PR tree-optimization/79908
17319 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17320 change: For a VA_ARG whose LHS has been cast away, use
17321 force_gimple_operand to construct the side effects.
17323 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17325 PR translation/80001
17326 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17327 more amenable to translation.
17328 (oacc_loop_auto_partitions): Likewise.
17330 2017-03-21 Marek Polacek <polacek@redhat.com>
17331 Martin Sebor <msebor@redhat.com>
17333 PR tree-optimization/80109
17334 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17335 on INTEGRAL_TYPE_P.
17337 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17338 Segher Boessenkool <segher@kernel.crashing.org>
17341 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17342 check reg_used_between_p between insn and one of succ or succ2
17343 depending on if succ is artificial insn not inserted into insn
17346 2017-03-21 Martin Liska <mliska@suse.cz>
17348 PR gcov-profile/80081
17349 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17350 * doc/gcc.texi: Include gcov-dump stuff.
17351 * doc/gcov-dump.texi: New file.
17353 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17355 PR rtl-optimization/79150
17356 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17357 conditional jump, if the jump is the last insn of the loop.
17359 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17360 Richard Biener <rguenth@suse.de>
17362 PR tree-optimization/79908
17363 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17364 been cast away, use force_gimple_operand to construct the side
17367 2017-03-21 Martin Liska <mliska@suse.cz>
17369 PR libfortran/79956
17370 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17373 2017-03-21 Brad Spengler <spender@grsecurity.net>
17376 * plugin.c (htab_hash_plugin): New function.
17377 (add_new_plugin): Use it and adjust.
17378 (parse_plugin_arg_opt): Adjust.
17379 (init_one_plugin): Likewise.
17381 2017-03-21 Richard Biener <rguenther@suse.de>
17383 PR tree-optimization/80032
17384 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17385 if set force the cleanup to happen unconditionally.
17386 (gimplify_target_expr): Push inserted clobbers with force_uncond
17387 to avoid them being removed by control-dependent DCE.
17389 2017-03-21 Richard Biener <rguenther@suse.de>
17391 PR tree-optimization/80122
17392 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17393 va_arg_pack_len when the inlined call stmt requires pack
17395 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17397 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17400 * tsan.c (instrument_builtin_call): If the memory model argument
17401 is not a constant, assume it is valid.
17404 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17407 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17409 PR rtl-optimization/79910
17410 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17411 if its dest is used by an insn before I2 (other than the combined
17412 insns themselves, which are properly handled already).
17414 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17417 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17419 * combine.c (record_used_regs): New static function.
17420 (try_combine): Handle situations where there is an additional
17421 instruction between I2 and I3 which needs to have a LOG_LINK
17425 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17427 * combine.c (try_combine): Delete redundant i1 test. Call
17428 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17430 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17433 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17434 alternatives 13/14.
17436 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17438 PR tree-optimization/80054
17439 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17440 the optimization if a PHI or any of its arguments is not dominated
17441 by the candidate's basis. Use gphi* rather than gimple* as
17443 (replace_profitable_candidates): Clean up a gimple* variable that
17444 should be a gphi* variable.
17446 2017-03-20 Martin Sebor <msebor@redhat.com>
17449 * doc/extend.texi (attribute constructor): Document present limitation.
17451 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17454 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17455 __POWER9_VECTOR__ #ifdef control, change template definition to
17456 use Power9-specific built-in function.
17457 (vec_any_eq): Likewise.
17458 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17459 to control outcomes from this test.
17460 (vector_ae_<mode>p): For VEC_F modes, likewise.
17462 2017-03-20 Ian Lance Taylor <iant@google.com>
17464 * config/i386/i386.c (ix86_function_regparm): Save an extra
17465 register for -fsplit-stack with DECL_STATIC_CHAIN.
17467 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17470 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17471 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17473 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17475 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17477 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17480 2017-03-20 Marek Polacek <polacek@redhat.com>
17483 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17485 2017-03-20 Richard Biener <rguenther@suse.de>
17487 PR tree-optimization/80113
17488 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17489 allocate extra SSA name for PHI def.
17490 (add_close_phis_to_outer_loops): Likewise.
17491 (add_close_phis_to_merge_points): Likewise.
17492 (copy_loop_close_phi_args): Likewise.
17493 (copy_cond_phi_nodes): Likewise.
17495 2017-03-20 Martin Liska <mliska@suse.cz>
17497 PR middle-end/79753
17498 * tree-chkp.c (chkp_build_returned_bound): Do not build
17499 returned bounds for a LHS that's not a BOUNDED_P type.
17501 2017-03-20 Martin Liska <mliska@suse.cz>
17505 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17506 COMPLEX_CST and VECTOR_CST.
17508 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17511 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17512 target operand. A new splitter adds the clobber statement in case
17513 the target operand is dead anyway.
17515 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17517 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17518 to age-old versions of binutils and glibc.
17520 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17522 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17524 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17526 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17528 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17530 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17531 requirement for binutils 2.13.
17533 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17535 * combine.c (try_combine): Delete redundant i1 test. Call
17536 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17538 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17540 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17541 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17543 <riscv64-*-elf>: Re-arrange section
17544 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17545 <riscv32-*-linux>: Likewise.
17546 <riscv64-*-elf>: Likewise
17547 <riscv64-*-linux>: Likewise.
17549 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17552 * aarch64.opt(verbose-cost-dump): Fix typo.
17554 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17557 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17558 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17560 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17562 * reload.c (find_reloads): When reloading a nonoffsettable address,
17563 use RELOAD_OTHER for it and its address reloads.
17565 PR rtl-optimization/79910
17566 * combine.c (record_used_regs): New static function.
17567 (try_combine): Handle situations where there is an additional
17568 instruction between I2 and I3 which needs to have a LOG_LINK
17571 2017-03-17 Jeff Law <law@redhat.com>
17573 PR tree-optimization/71437
17574 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17575 conditional in the hash table first.
17576 (vrp_dom_walker::before_dom_children): Extract condition from
17577 ASSERT_EXPR. Record condition, its inverion and any implied
17578 conditions as well.
17580 2017-03-17 Marek Polacek <polacek@redhat.com>
17581 Markus Trippelsdorf <markus@trippelsdorf.de>
17583 PR tree-optimization/80079
17584 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17587 2017-03-17 Richard Biener <rguenther@suse.de>
17589 PR middle-end/80075
17590 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17591 Properly verify the LHS before the RHS possibly claims to be
17593 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17596 2017-03-17 Martin Jambor <mjambor@suse.cz>
17598 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17599 (List of -O2 options): Likewise.
17600 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17603 2017-03-17 Tom de Vries <tom@codesourcery.com>
17605 * gcov-dump.c (print_usage): Print bug_report_url.
17607 2017-03-17 Richard Biener <rguenther@suse.de>
17609 PR middle-end/80050
17610 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17611 (parser::peek): Likewise.
17613 2017-03-17 Richard Biener <rguenther@suse.de>
17615 PR tree-optimization/80048
17616 * sese.c (free_sese_info): Properly release rename_map and
17617 copied_bb_map elements.
17619 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17621 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17622 Add linked-list forward and backlinks. Insert on
17623 construction, remove on destruction.
17624 (class pass_store_merging): Add m_stores_head field.
17625 (pass_store_merging::terminate_and_process_all_chains):
17626 Iterate over m_stores_head list.
17627 (pass_store_merging::terminate_all_aliasing_chains):
17629 (pass_store_merging::execute): Check for debug stmts first.
17630 Push new chains onto the m_stores_head stack.
17632 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17635 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17636 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17637 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17639 2017-03-16 Jeff Law <law@redhat.com>
17641 PR tree-optimization/71437
17642 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17643 member function. Implementation moved into after_dom_children
17644 member function and into the threader's thread_outgoing_edges
17646 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17647 some code into new thread_outgoing_edges.
17648 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17649 definition. Simplify marker handling (do it here). Assume we always
17650 have the available expression and the const/copies tables.
17651 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17653 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17654 * tree-vrp.c (equiv_stack): No longer file scoped.
17655 (vrp_dom_walker): New class.
17656 (vrp_dom_walker::before_dom_children): New member function.
17657 (vrp_dom_walker::after_dom_children): Likewise.
17658 (identify_jump_threads): Setup domwalker. Use it rather than
17659 walking edges in a random order by hand. Simplify setup/finalization.
17660 (finalize_jump_threads): Remove.
17661 (vrp_finalize): Do not call identify_jump_threads here.
17662 (execute_vrp): Do it here instead and call thread_through_all_blocks
17665 PR tree-optimization/71437
17666 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17668 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17670 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17671 (dom_opt_dom_walker::thread_across_edge): Remove
17672 handle_dominating_asserts argument. All callers changed.
17673 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17674 changes. Remove calls to lhs_of_dominating_assert. Other
17675 uses of handle_dominating_asserts turn into unconditional code
17676 (simplify_control_stmt_condition_1): Likewise.
17677 (simplify_control_stmt_condition): Likewise.
17678 (thread_through_normal_block, thread_across_edge): Likewise.
17679 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17680 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17681 object if it is not an SSA_NAME.
17682 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17683 before calling into the VRP specific simplifiers.
17684 (identify_jump_threads): Remove handle_dominating_asserts
17687 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17690 * tree-diagnostic.c (default_tree_printer): No longer static.
17691 * tree-diagnostic.h (default_tree_printer): New prototype.
17693 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17695 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17696 Change ins into fmov.
17698 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17700 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17701 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17702 Use h_con constraint for operand 1.
17703 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17704 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17706 2017-03-15 Jeff Law <law@redhat.com>
17708 PR tree-optimization/71437
17709 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17710 (record_temporary_equivalences): Use it.
17712 PR tree-optimization/71437
17713 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17714 tree-ssa-scopedtables.
17715 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17716 (record_conditions, record_cond, vuse_eq): Likewise.
17717 (record_edge_info): Adjust to API tweak of record_conditions.
17718 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17719 (record_temporary_equivalences, optimize_stmt): Likewise.
17720 (eliminate_redundant_computations): Likewise.
17721 (record_equivalences_from_stmt): Likewise.
17722 * tree-ssa-scopedtables.c: Include options.h and params.h.
17723 (vuse_eq): New function, moved from tree-ssa-dom.c
17724 (build_and_record_new_cond): Likewise.
17725 (record_conditions): Likewise. Accept vector of conditions rather
17726 than edge_equivalence structure for first argument.
17727 for the first argument.
17728 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17729 from tree-ssa-dom.c.
17730 (avail_exprs_stack::record_cond): Likewise.
17731 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17732 from tree-ssa-dom.c.
17733 (avail_exprs_stack): Add new member functions lookup_avail_expr
17735 (record_conditions): Declare.
17737 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17740 * lra-constraints.c (process_alt_operands): Increase reject for
17741 reloading an input/output operand.
17743 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17746 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17747 insns to convert from signed/unsigned char/short to IEEE 128-bit
17749 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17751 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17754 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17755 subreg of inner mode for values already in registers.
17757 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17759 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17760 iteration reg is used after the loop.
17762 2017-03-14 Martin Sebor <msebor@redhat.com>
17764 PR tree-optimization/79800
17765 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17766 precision in negative-positive range.
17767 (format_floating): Call non-const overload with adjusted precision.
17769 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17772 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17775 2017-03-14 Martin Sebor <msebor@redhat.com>
17777 PR middle-end/80020
17778 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17779 * builtins.def (aligned_alloc): Use it.
17782 * Makefile.in (GTFILES): Add calls.c.
17783 * calls.c: Include "gt-calls.h".
17785 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17787 PR rtl-optimization/79728
17788 * regs.h (struct target_regs): New field
17789 x_contains_allocatable_regs_of_mode.
17790 (contains_allocatable_regs_of_mode): New macro.
17791 * reginfo.c (init_reg_sets_1): Initialize it, and change
17792 contains_reg_of_mode so it includes global regs as well.
17793 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17794 rather than contains_regs_of_mode.
17796 2017-03-14 Martin Liska <mliska@suse.cz>
17798 * doc/invoke.texi: Document options that can't be combined with
17799 -fcheck-pointer-bounds.
17801 2017-03-14 Martin Liska <mliska@suse.cz>
17803 PR middle-end/79831
17804 * doc/invoke.texi (-Wchkp): Document the option.
17806 2017-03-14 Martin Liska <mliska@suse.cz>
17808 * Makefile.in: Install gcov-dump.
17810 2017-03-14 Martin Liska <mliska@suse.cz>
17812 * multiple_target.c (expand_target_clones): Bail out for
17813 an invalid attribute.
17815 2017-03-14 Richard Biener <rguenther@suse.de>
17817 * alias.c (struct alias_set_entry): Pack properly.
17818 * cfgloop.h (struct loop): Likewise.
17819 * cse.c (struct set): Likewise.
17820 * ipa-utils.c (struct searchc_env): Likewise.
17821 * loop-invariant.c (struct invariant): Likewise.
17822 * lra-remat.c (struct cand): Likewise.
17823 * recog.c (struct change_t): Likewise.
17824 * rtl.h (struct address_info): Likewise.
17825 * symbol-summary.h (function_summary): Likewise.
17826 * tree-loop-distribution.c (struct partition): Likewise.
17827 * tree-object-size.c (struct object_size_info): Likewise.
17828 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17829 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17830 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17831 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17832 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17833 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17834 (struct _stmt_vec_info): Likewise.
17836 2017-03-14 Martin Liska <mliska@suse.cz>
17839 * multiple_target.c (create_dispatcher_calls): Check that
17840 a target can create a function dispatcher.
17842 2017-03-14 Martin Liska <mliska@suse.cz>
17845 * multiple_target.c (expand_target_clones): Drop local.local
17846 flag for default implementation.
17848 2017-03-14 Richard Biener <rguenther@suse.de>
17850 PR tree-optimization/80030
17851 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17853 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17855 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17856 gcc_fallthrough() instead of __attribute__((fallthrough));
17858 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17860 * doc/gcc.texi: Remove "up" link to (DIR).
17861 * doc/gccint.texi: Ditto.
17863 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17865 * doc/install.texi (Specific) <avr>: Remove reference to
17868 2017-03-13 Jeff Law <law@redhat.com>
17870 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17871 attribute rather than comments.
17873 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17874 match_scratch operand is highest.
17876 2017-03-13 Martin Liska <mliska@suse.cz>
17878 PR middle-end/78339
17879 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17880 is a CHKP clone, use original declaration.
17882 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17884 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17885 (arc_conditional_register_usage): Use a different allocation order
17886 when optimizing for size.
17887 * common/config/arc/arc-common.c (arc_option_optimization_table):
17888 Section anchors default on when optimizing for size.
17890 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17892 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17894 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17896 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17897 * config/arc/arc.md (cpu_facility): Add cd variant.
17898 (*movqi_insn): Add code density variant.
17899 (*movhi_insn): Likewise.
17900 (*movqi_insn): Likewise.
17901 (*addsi3_mixed): Likewise.
17902 (subsi3_insn): Likewise.
17904 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17906 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17908 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17910 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17911 expressions with MINUS and UNARY ops.
17913 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17916 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17918 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17919 between vec_select and vector argument.
17920 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17921 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17922 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17923 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17924 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17925 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17927 2017-03-13 Richard Biener <rguenther@suse.de>
17930 * params.def (vect-max-peeling-for-alignment): Fix typo.
17932 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17934 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17935 issue that only occurred with binutils below 2.18.
17937 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17939 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17940 refer to binutils 2.11/2.12 minimum.
17942 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17944 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17945 ftp.kernel.org and simplify binutils requirement.
17947 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17949 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17951 (Optimize Options): Ditto. Also remove redundancy.
17953 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17955 PR translation/79848
17956 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17958 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17959 to G_ to avoid double translation.
17961 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17963 PR translation/79923
17964 * auto-profile.c (get_combined_location): Convert leading
17965 character of diagnostics to lower case and remove trailing period.
17966 (read_profile): Likewise for various diagnostics.
17967 * config/arm/arm.c (arm_option_override): Remove trailing period
17968 from various diagnostics.
17969 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17970 (msp430_expand_delay_cycles): Likewise.
17972 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17975 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17976 full command-line argument, rather than just "str".
17977 (aarch64_validate_march): Likewise.
17978 (aarch64_validate_mtune): Likewise.
17980 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17982 PR rtl-optimization/78911
17983 * lra-assigns.c (must_not_spill_p): New function.
17984 (spill_for): Use it.
17986 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17988 PR tree-optimization/79981
17989 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17990 ATOMIC_COMPARE_EXCHANGE ifn result.
17991 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17992 IFN_ATOMIC_COMPARE_EXCHANGE.
17994 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17997 * opts.c (parse_sanitizer_options): Add missing question mark to
17998 "did you mean" message.
18000 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18002 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18004 (VMULEUH_UNS): Likewise.
18005 (VMULOUB_UNS): Likewise.
18006 (VMULOUH_UNS): Likewise.
18007 * config/rs6000/rs6000.c (builtin_function_type): Remove
18008 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18010 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18013 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18014 x with result of extra_parsing_for_operand_code_0.
18015 (function_reader::extra_parsing_for_operand_code_0): Convert
18016 return type from void to rtx, returning x. When reading
18017 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18018 larger size containing struct block_symbol.
18020 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18022 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18023 -mfloat128-hardware without -m64.
18025 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18028 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18029 entries to the case statement that marks unsigned arguments to
18030 overloaded functions.
18032 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18035 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18037 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18040 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18041 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18043 2017-03-10 Martin Liska <mliska@suse.cz>
18047 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18048 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18051 2017-03-10 Olivier Hainque <hainque@adacore.com>
18053 * tree-switch-conversion (array_value_type): Start by resetting
18054 candidate type to it's main variant.
18056 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18058 PR rtl-optimization/79909
18059 * combine.c (try_combine): Use simplify_replace_rtx on individual
18060 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18061 of the whole CALL_INSN_FUNCTION_USAGE.
18063 PR tree-optimization/79972
18064 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18065 get_range_info on SSA_NAMEs. Formatting fixes.
18067 2017-03-10 Richard Biener <rguenther@suse.de>
18068 Jakub Jelinek <jakub@redhat.com>
18070 PR tree-optimization/77975
18071 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18072 edge to be constant.
18073 (get_val_for): For constant x return it. Formatting fix.
18074 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18075 would use the same bases as the current one.
18077 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18079 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18080 instead of vec_select for V1TImode.
18081 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18083 (VSX_LE_128): Add V1TI to this mode iterator.
18084 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18085 (*vsx_le_perm_store_<mode>): Likewise.
18086 (pre-reload splitter for VSX stores): Likewise.
18087 (post-reload splitter for VSX stores): Likewise.
18088 (*vsx_xxpermdi2_le_<mode>): Likewise.
18089 (*vsx_lxvd2x2_le_<mode>): Likewise.
18090 (*vsx_stxvd2x2_le_<mode>): Likewise.
18092 2017-03-09 Michael Eager <eager@eagercon.com>
18094 Correct failures with --enable-checking=yes,rtl.
18096 * config/microblaze/microblaze.c (microblaze_expand_shift):
18097 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18098 test for const0_rtx.
18099 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18100 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18102 2017-03-09 Richard Biener <rguenther@suse.de>
18104 PR tree-optimization/79977
18105 * graphite-scop-detection.c (scop_detection::merge_sese):
18106 Handle the case of extra exits to blocks dominating the entry.
18108 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18110 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18113 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18115 PR rtl-optimization/79949
18116 * lra-constraints.c (process_alt_operands): Check memory when
18117 trying to predict a cycle. Print about the overall increase.
18119 2017-03-09 Richard Biener <rguenther@suse.de>
18121 PR middle-end/79971
18122 * gimple-expr.c (useless_type_conversion_p): Preserve
18123 TYPE_SATURATING for fixed-point types.
18125 2017-03-09 Richard Biener <rguenther@suse.de>
18128 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18129 alignment of BLKmode params.
18131 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18134 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18135 (VALL_NO_V2Q): Likewise.
18137 * config/aarch64/aarch64-simd.md
18138 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18140 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18141 VALL_NO_V2Q mode iterator.
18142 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18144 2017-03-09 Martin Liska <mliska@suse.cz>
18146 PR tree-optimization/79631
18147 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18148 tree_int_cst_sign_bit just for INTEGER constants.
18150 2017-03-09 Martin Liska <mliska@suse.cz>
18154 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18157 2017-03-09 Marek Polacek <polacek@redhat.com>
18160 * tree.c (inchash::add_expr): Handle TREE_VEC.
18162 2017-03-09 Martin Liska <mliska@suse.cz>
18165 (chkp_narrow_size_and_offset): New function.
18166 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18167 (void chkp_parse_bit_field_ref): New function.
18168 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18169 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18171 2017-03-09 Martin Liska <mliska@suse.cz>
18174 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18175 (chkp_find_bounds_1): Remove gcc_unreachable.
18177 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18180 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18181 BUILT_IN_SYNC*, determine the access type from the size suffix and
18182 always build a MEM_REF with that type. Handle forgotten
18183 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18186 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18187 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18188 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18189 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18190 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18191 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18192 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18193 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18194 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18195 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18196 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18197 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18198 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18199 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18200 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18201 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18202 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18203 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18204 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18205 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18206 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18207 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18208 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18209 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18210 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18211 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18212 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18213 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18214 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18215 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18216 definitions outside of __OPTIMIZE__ guarded section.
18219 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18220 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18221 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18222 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18225 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18227 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18228 ("vfenez<mode>"): Add missing constraints.
18230 2017-03-08 Martin Sebor <msebor@redhat.com>
18233 * config/nds32/nds32.c (nds32_option_override):
18234 Fix misspelled diagnostic.
18236 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18239 * gimplify.c (gimplify_omp_for): Replace index var in outer
18240 taskloop statement with an artificial variable and add
18241 OMP_CLAUSE_PRIVATE clause for it.
18243 2017-03-08 Richard Biener <rguenther@suse.de>
18245 PR tree-optimization/79955
18246 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18247 for accesses that are completely outside of the variable.
18249 2017-03-08 Andrew Haley <aph@redhat.com>
18251 PR tree-optimization/79943
18252 * tree-ssa-loop-split.c (compute_new_first_bound): When
18253 calculating the new upper bound, (END-BEG) should be added, not
18256 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18258 * config/avr/avr.md (setmemhi): Make sure match_dup
18259 operand number comes before match_scratch.
18261 2017-03-08 Richard Biener <rguenther@suse.de>
18263 PR tree-optimization/79920
18264 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18265 with ncopies == 1 to ...
18266 (vect_transform_slp_perm_load): ... here. Properly compute
18267 all element loads by iterating VF times over the group. Do
18268 not handle ncopies (computed in a broken way) in
18269 vect_create_mask_and_perm.
18271 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18274 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18275 is a uniform vector, use uniform_vector_p return value instead of
18276 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18278 2017-03-07 Marek Polacek <polacek@redhat.com>
18280 PR middle-end/79809
18281 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18282 (alloca_call_type): Likewise.
18284 2017-03-07 Martin Liska <mliska@suse.cz>
18286 * gcov.c (process_args): Put comment to correct location.
18288 2017-03-07 Martin Liska <mliska@suse.cz>
18290 PR middle-end/68270
18291 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18292 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18293 (chkp_narrow_bounds_for_field): Likewise.
18294 (chkp_parse_array_and_component_ref): Pass one more argument to
18297 2017-03-07 Richard Biener <rguenther@suse.de>
18299 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18302 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18304 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18305 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18307 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18310 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18311 to end of description.
18312 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18314 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18316 PR rtl-optimization/79901
18317 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18319 (*avx512f_<code><mode>3<mask_name>): ... this.
18320 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18321 iterator instead of VI8_AVX2_AVX512BW.
18323 PR rtl-optimization/79901
18324 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18325 min/max expander, expand it using expand_vec_cond_expr.
18328 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18331 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18334 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18335 to void * for PCH reasons.
18336 * dwarf2out.c (output_loc_operands, output_die): Cast
18337 v.val_vec.array to unsigned char *.
18339 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18342 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18345 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18347 PR rtl-optimization/79571
18348 * lra-constraints.c (process_alt_operands): Calculate static
18349 reject and subtract it from overall when only addresses will be
18352 2017-03-06 Julia Koval <julia.koval@intel.com>
18355 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18356 incoming stack boundary to 128 for 64-bit targets.
18358 2017-03-06 Richard Biener <rguenther@suse.de>
18360 PR tree-optimization/79894
18361 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18362 to NULL after folding it.
18364 2017-03-06 Richard Biener <rguenther@suse.de>
18366 PR tree-optimization/79824
18367 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18368 check disabling peeling for gaps.
18370 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18372 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18373 attributes): Document gettimeofday.
18375 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18377 * config/s390/s390.c (s390_option_override_internal): Set
18378 PARAM_MIN_VECT_LOOP_BOUND
18380 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18382 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18383 * config/s390/s390.md: Likewise.
18385 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18388 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18389 (<avx2_avx512>_perm<mode>): Rename to ...
18390 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18392 (<avx512>_perm<mode>_mask): Rename to ...
18393 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18395 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18396 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18397 instead of VI8F_256_512.
18398 (avx512f_perm<mode>): New define_expand.
18399 (avx512f_perm<mode>_mask): Likewise.
18400 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18401 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18403 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18405 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18406 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18408 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18410 2017-03-06 Martin Liska <mliska@suse.cz>
18413 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18414 when having a SSA NAME w/o VAR_DECL assigned to it.
18416 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18418 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18419 msa_dpsub_<su>_d): Fix MODE for vec_select.
18421 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18423 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18425 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18427 2017-03-06 Richard Biener <rguenther@suse.de>
18429 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18430 * plugin.c (register_plugin_info): Likewise.
18431 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18433 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18435 * config/i386/sse.md (sse_storehps, sse_storelps,
18436 avx_<castmode><avxsizesuffix>_<castmode>,
18437 avx512f_<castmode><avxsizesuffix>_<castmode>,
18438 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18439 in condition that at least one operand is not a MEM.
18441 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18443 PR middle-end/79805
18444 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18445 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18447 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18448 gimple_call_nothrow_p flag based on whether original builtin can throw.
18449 If it can, emit following stmts on the fallthrough edge.
18450 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18451 don't create new bb if inserting just debug stmts on the edge, try to
18452 insert them on the fallthru bb or just reset debug stmts.
18454 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18457 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18458 restore recog_data (including the operand rtxes inside it) around
18459 the call to get_insn_template.
18461 2017-03-03 Martin Sebor <msebor@redhat.com>
18463 PR tree-optimization/79699
18464 * context.c (context::~context): Free MPFR caches to avoid
18465 a memory leak on program exit.
18467 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18469 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18470 Use wide_int::ulow () instead of .elt (0).
18472 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18474 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18475 (*pushxf): Limit oF constraint to 32bit targets and add oC
18476 constraint for 64bit targets.
18477 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18478 (*pushdf): Change rmF constraint to rmC.
18480 2017-03-03 Martin Liska <mliska@suse.cz>
18482 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18483 Remove unused variable.
18485 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18488 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18489 is a memory operand, increase num_memory.
18490 (ix86_expand_args_builtin): Likewise.
18492 2017-03-03 Jan Hubicka <jh@suse.cz>
18495 * ipa-devirt.c (maybe_record_node): Properly handle
18496 __cxa_pure_virtual visibility.
18498 2017-03-03 Martin Liska <mliska@suse.cz>
18500 PR tree-optimization/79803
18501 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18503 (pass_loop_prefetch::execute): Disabled optimization if an
18504 assumption about L1 cache size is not met.
18506 2017-03-03 Martin Liska <mliska@suse.cz>
18508 PR rtl-optimization/79574
18509 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18510 (hash_scan_set): Likewise.
18511 (dump_hash_table): Likewise.
18512 (hoist_code): Likewise.
18514 2017-03-03 Richard Biener <rguenther@suse.de>
18516 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18517 in place of uhigh ().
18518 (fixed_convert_from_real): Likewise.
18520 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18523 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18525 2017-03-03 Richard Biener <rguenther@suse.de>
18527 PR middle-end/79818
18528 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18529 TYPE_OVERFLOW_UNDEFINED check.
18531 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18533 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18535 (vector_ae_<mode>_p): Likewise.
18536 (vector_nez_<mode>_p): Likewise.
18537 (vector_ne_v2di_p): Likewise.
18538 (vector_ae_v2di_p): Likewise.
18539 (vector_ne_<mode>_p): Likewise.
18540 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18542 (vsx_tsqrt<mode>2_fe): Likewise.
18544 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18547 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18549 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18551 PR rtl-optimization/79780
18552 * cprop.c (one_cprop_pass): When second and further conditional trap
18553 in a single basic block is turned into an unconditional trap, turn it
18554 into a deleted note to avoid RTL verification failures.
18556 2017-03-02 Richard Biener <rguenther@suse.de>
18558 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18560 2017-03-02 Richard Biener <rguenther@suse.de>
18562 PR tree-optimization/79345
18564 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18565 param and abort the walk, returning -1 if it is hit.
18566 (walk_aliased_vdefs): Take a limit param and pass it on.
18567 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18568 defaulting to 0 and return a signed int.
18569 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18570 (check_defs): New helper.
18571 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18572 about uninitialized memory.
18573 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18574 bogus uninitialized warning.
18575 (fixed_convert_from_real): Likewise.
18577 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18579 PR tree-optimization/66768
18580 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18581 iv_use if base object can't be determined.
18583 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18585 PR tree-optimization/79345
18586 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18587 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18588 (get_pattern_stats): Initialize it.
18589 * genemit.c (gen_expand): Verify match_scratch numbers come after
18590 match_operand/match_dup numbers.
18591 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18592 match_scratch numbers.
18593 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18595 * config/s390/s390.md (trunctdsd2): Likewise.
18597 2017-03-02 Richard Biener <rguenther@suse.de>
18599 * wide-int.h (wide_int_storage::operator=): Implement in terms
18602 2017-03-02 Richard Biener <rguenther@suse.de>
18604 PR tree-optimization/79777
18605 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18606 the to insert expression to sth existing.
18608 2017-03-01 Martin Sebor <msebor@redhat.com>
18610 PR middle-end/79692
18611 * gimple-ssa-sprintf.c
18612 (directive::known_width_and_precision): New function.
18613 (format_integer): Use it.
18614 (get_mpfr_format_length): Consider the full range of precision
18615 when computing %g output with the # flag. Set the likely byte
18616 count to 3 rather than 1 when precision is indeterminate.
18617 (format_floating): Correct the lower bound of precision.
18619 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18621 * doc/invoke.texi: Document default code model for 64-bit Linux.
18623 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18626 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18627 udiv rather than div since input pattern is unsigned.
18629 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18631 * config/i386/i386.c (print_reg): Warn for values of
18632 unsupported size in integer register.
18634 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18637 * config/rs6000/predicates.md (current_file_function_operand): Do
18638 not allow self calls to be local if the function is replaceable.
18640 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18643 * config/rs6000/altivec.h (vec_ctz and others): Change the
18644 preprocessor macro that controls conditional compilation from
18645 _ARCH_PWR9 to __POWER9_VECTOR__.
18646 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18647 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18648 control (instead of _ARCH_PWR9 control) so that template
18649 definition uses power9-specific function.
18650 (vec_any_eq): Likewise.
18651 (vec_all_ne): Change macro definition to use a power9-specific
18652 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18653 _ARCH_PWR9 control).
18654 (vec_any_eq) Likewise.
18655 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18656 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18657 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18658 support for xvcmpnedp instruction.
18659 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18660 macro expansion so that Power9 implementation of vec_all_ne does
18661 not use the AltiVec predicate framework.
18662 (VCMPNEH_P): Likewise.
18663 (VCMPNEW_P): Likewise.
18664 (VCMPNED_P): Likewise.
18665 (VCMPNEFP_P): Likewise.
18666 (VCMPNEDP_P): Likewise.
18667 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18668 implementation of vec_any_eq to not use AltiVec predicate
18670 (VCMPAEH_P): Likewise.
18671 (VCMPAEW_P): Likewise.
18672 (VCMPAED_P): Likewise.
18673 (VCMPAEFP_P): Likewise.
18674 (VCMPAEDP_P): Likewise.
18675 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18676 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18677 not use the AltiVec predicate framework.
18678 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18679 of vec_any_eq to not use AltiVec predicate framework.
18680 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18681 support for predefined __POWER9_VECTOR__ macro to indicate that
18682 Power9 instruction selection is enabled.
18683 (altivec_overloaded_builtins): Remove extraneous
18684 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18685 function argument types RS6000_BTI_bool_V16QI and
18686 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18687 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18688 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18689 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18690 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18691 Power9 for implementations of vec_cmpne. Change the signature for
18692 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18693 (representing vec_all_ne) to remove the previously described first
18694 argument of type RS6000_BTI_INTSI, as this was an artifact of
18695 reliance on the AltiVec predicate framework, which is no longer
18696 used in the implementation of these functions. Add
18697 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18698 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18699 since, unlike the AltiVec predicate framework implementation, we
18700 do not share function descriptors between vec_alle and vec_anyeq.
18701 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18702 set of modes that receive special treatment even when
18703 TARGET_P9_VECTOR is true. The special treatment emits code that
18704 does not depend on Power9 instructions.
18705 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18706 define_expand to not rely on AltiVec predicate framework.
18707 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18709 (vector_ne_v2di_p): Change this define_expand to not rely on
18710 AltiVec predicate framework.
18711 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18713 (vector_ne_<mode>_p): Change this define_expand to not rely on
18714 AltiVec predicate framework.
18715 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18717 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18718 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18719 define_insn pattern.
18720 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18721 define_insn pattern because the xvcmpne<VSs>. instruction is not
18723 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18724 instruction is not supported.
18726 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18728 * config/nvptx/nvptx.c: Include intl.h.
18730 2017-03-01 Martin Jambor <mjambor@suse.cz>
18733 * ipa-prop.h (ipa_bits): Removed field known.
18734 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18735 to pointers. Adjusted their comments to warn about their sharing.
18736 (ipcp_transformation_summary): Change bits to a vector of pointers.
18737 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18738 (ipa_get_ipa_bits_for_value): Declare.
18739 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18740 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18741 (ipa_bits_hash_table): Likewise.
18742 (ipa_vr_ggc_hash_traits): Likewise.
18743 (ipa_vr_hash_table): Likewise.
18744 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18745 being pointers and vr_known being removed.
18746 (ipa_set_jf_unknown): Likewise.
18747 (ipa_get_ipa_bits_for_value): New function.
18748 (ipa_set_jfunc_bits): Likewise.
18749 (ipa_get_value_range): New overloaded functions.
18750 (ipa_set_jfunc_vr): Likewise.
18751 (ipa_compute_jump_functions_for_edge): Use the above functions to
18752 construct bits and vr parts of jump functions.
18753 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18754 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18756 (ipcp_grow_transformations_if_necessary): Also allocate
18757 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18759 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18760 them. Fix too long lines.
18761 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18762 vr_known being removed.
18763 (ipa_read_jump_function): Use new setter functions to construct bits
18764 and vr parts of jump functions or set them to NULL.
18765 (write_ipcp_transformation_info): Adjust for bits being pointers.
18766 (read_ipcp_transformation_info): Likewise.
18767 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18769 Include gt-ipa-prop.h.
18770 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18772 (ipcp_store_bits_results): Likewise.
18773 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18774 Do not write to existing jump functions but use a temporary instead.
18776 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18779 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18780 attempt to use its first operand as BIT_FIELD_REF base.
18782 2017-03-01 Richard Biener <rguenther@suse.de>
18784 PR middle-end/79721
18785 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18786 interpolating formula in wrapping arithmetic.
18787 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18789 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18791 PR tree-optimization/79734
18792 * tree-vect-generic.c (expand_vector_condition): Optimize
18793 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18794 Handle VEC_COND_EXPR where comparison has different inner width from
18795 type's inner width.
18797 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18799 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18800 markup, and similar issues. Remove @opindex entries for things
18801 that aren't options. Add missing -mmpy-option entries.
18803 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18805 PR tree-optimization/79737
18806 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18807 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18808 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18809 instead of byte_size. Formatting fix.
18810 (shift_bytes_in_array_right): Formatting fix.
18812 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18815 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18816 condition on optimize for the leaf function test.
18818 2017-02-28 Martin Liska <mliska@suse.cz>
18821 * read-rtl-function.c (function_reader::handle_unknown_directive):
18822 Bail out when one uses -flto.
18824 2017-02-28 Martin Liska <mliska@suse.cz>
18826 * common.opt: Replace space with tabular for options of <number>
18828 * config/i386/i386.opt: Show <number> value for
18829 -mlarge-data-threshold.
18830 * opts.c (print_filtered_help): Do not display number in hexadecimal
18833 2017-02-28 Martin Liska <mliska@suse.cz>
18835 * common.opt: Fix --help=option -Q for options which are of
18838 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18840 * config/i386/i386.c (print_reg): Error out for values
18841 of 8-bit size in invalid integer register.
18843 2017-02-28 Martin Sebor <msebor@redhat.com>
18845 PR tree-optimization/79691
18846 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18848 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18851 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18852 gcc_unreachable with output_operand_lossage.
18854 2017-02-28 Richard Biener <rguenther@suse.de>
18856 PR tree-optimization/79740
18857 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18859 (visit_nary_op): Insert the nary into the hashtable if we
18860 pattern-matched sth.
18861 * tree-ssa-pre.c (eliminate_insert): Robustify.
18863 2017-02-28 Richard Biener <rguenther@suse.de>
18865 PR middle-end/79731
18866 * fold-const.c (decode_field_reference): Reject out-of-bound
18869 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18871 * config/i386/i386.c: Include intl.h.
18872 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18873 instead of just cond ? "..." : "...".
18874 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18875 * coverage.c (read_counts_file): Likewise.
18876 * omp-offload.c: Include intl.h.
18877 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18878 of just cond ? "..." : "...".
18879 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18880 of just cond ? "..." : "...".
18882 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18885 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18887 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18888 'tune for' CPU name.
18889 * config/arm/arm-cpu-data.h: Regenerated.
18891 2017-02-28 Richard Biener <rguenther@suse.de>
18893 PR tree-optimization/79732
18894 * tree-inline.c (expand_call_inline): Do not shadow var.
18896 2017-02-28 Richard Biener <rguenther@suse.de>
18898 PR tree-optimization/79723
18899 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18900 address-space properly.
18902 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18904 * doc/optinfo.texi (Optimization groups): Fix option used for
18906 * doc/invoke.texi (-fopt-info): Document "omp".
18907 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18908 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18909 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18910 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18911 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18913 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18915 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18916 OPTGROUP_OMP with "omp".
18918 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18921 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18922 for arithmetic shift of unsigned V2DI.
18924 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18926 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18927 arc/linux.h headers.
18928 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18929 (LINK_SPEC): Likewise.
18930 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18931 (EXTRA_SPECS): Likewise.
18932 (STARTFILE_SPEC): Likewise.
18933 (ENDFILE_SPEC): Likewise.
18934 (LIB_SPEC): Likewise.
18935 (TARGET_SDATA_DEFAULT): Likewise.
18936 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18937 (MULTILIB_DEFAULTS): Likewise.
18938 (DWARF2_UNWIND_INFO): Likewise.
18939 * config/arc/big.h: New file.
18940 * config/arc/elf.h: Likewise.
18941 * config/arc/linux.h: Likewise.
18942 * config/arc/t-uClibc: Remove.
18944 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18946 PR tree-optimization/77536
18947 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18948 (tree_transform_and_unroll_loop): Use above function to compute the
18949 estimated niter of unrolled loop and use it when scaling profile.
18950 Also use count info rather than frequency if it's non-zero.
18951 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18952 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18953 (vect_transform_loop): Call above function.
18955 2017-02-27 Richard Biener <rguenther@suse.de>
18957 PR tree-optimization/45397
18958 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18959 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18960 (visit_nary_op): Add pattern matching for CSEing sign-changed
18961 or truncated operations with wider ones.
18963 2017-02-27 Richard Biener <rguenther@suse.de>
18965 PR tree-optimization/79690
18966 * tree-vect-stmts.c (vectorizable_store): Use vector type
18967 built from the DR with address-space.
18969 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18971 * doc/invoke.texi (Optimize Options): Refine the description
18972 of asan-use-after-return.
18974 2017-02-25 Alan Modra <amodra@gmail.com>
18976 PR rtl-optimization/79584
18977 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18978 base, not ad->base_term, the reg within base. Remove assertion
18979 that ad->base == ad->base_term. Replace gen_int_mode using
18980 bogus mode with const0_rtx.
18982 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18984 PR middle-end/79396
18985 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18986 FMA_EXPR like tcc_binary or tcc_unary.
18988 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18991 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18993 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18994 (output_loc_operands): Handle DW_OP_call_ref and
18995 DW_OP_GNU_variable_value.
18996 (struct variable_value_struct): New type.
18997 (struct variable_value_hasher): Likewise.
18998 (variable_value_hash): New variable.
18999 (string_types): Remove.
19000 (copy_loc_descr): New function.
19001 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19002 (prepend_loc_descr_to_each): New function.
19003 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19004 instead of add_loc_descr_to_each if the first argument is single
19005 location list and the second has multiple.
19006 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19007 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19008 when looking for variable value which doesn't have other location info.
19009 (loc_list_from_tree): Formatting fix.
19010 (gen_array_type_die): Simplify DW_AT_string_length handling.
19011 (adjust_string_types): Remove.
19012 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19013 string_types. Call resolve_variable_values.
19014 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19015 (resolve_addr_in_expr): Likewise. Add A argument.
19016 (copy_deref_exprloc): Remove deref argument. Adjust for the
19017 original expression being DW_OP_GNU_variable_value with optionally
19018 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19019 optionally after it.
19020 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19021 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19022 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19023 (variable_value_hasher::hash, variable_value_hasher::equal): New
19025 (resolve_variable_value_in_expr, resolve_variable_value,
19026 resolve_variable_values, note_variable_value_in_expr,
19027 note_variable_value): New functions.
19028 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19031 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19034 * opts.h (handle_generated_option): Add GENERATED_P argument.
19035 * opts-common.c (handle_option): Adjust function comment.
19036 (handle_generated_option): Add GENERATED_P argument, pass it to
19038 (control_warning_option): Pass false to handle_generated_option
19040 * opts.c (maybe_default_option): Pass true to handle_generated_option
19042 * optc-gen.awk: Likewise.
19044 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19046 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19047 a REG, look at the REG it is a SUBREG of.
19048 (splitter for cmpeqsi_t): Ditto.
19050 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19052 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19053 the special USEs with the pattern of the insn, not the insn itself.
19055 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19058 * doc/invoke.texi: Document -mload-store-pairs.
19060 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19061 Sandra Loosemore <sandra@codesourcery.com>
19063 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19064 argument isn't a CONST_INT.
19065 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19066 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19067 (nios2_validate_compare): Bypass alternate compare logic if *op2
19068 is not a CONST_INT.
19069 (ldstwm_operation_p): Return false if first_base is not a REG or
19070 if first_offset is not a CONST_INT.
19072 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19074 * config/cris/cris.md: Use correct operand in a define_peephole2.
19076 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19078 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19080 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19082 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19083 this_insn if it is an INSN or JUMP_INSN.
19084 (force_offsettable): Look at base, not at addr.
19085 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19086 on things that aren't necessarily CONST_INTs.
19088 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19090 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19091 -mfpmath=sse is the default also for x86-32 targets with SSE2
19092 instruction set when @option{-ffast-math} is enabled
19094 2017-02-24 Jeff Law <law@redhat.com>
19096 PR rtl-optimizatoin/79286
19097 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19100 2017-02-24 Richard Biener <rguenther@suse.de>
19102 PR tree-optimization/79389
19103 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19106 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19108 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19109 function comment to reflect reality.
19110 (loop_exits_before_overflow): Fix typo in function description.
19112 2017-02-24 Richard Biener <rguenther@suse.de>
19114 PR tree-optimization/79389
19115 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19116 properly that a threading opportunity exists. Detect conditional
19117 copy/constant propagation opportunities.
19119 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19121 * config/visium/visium.md (type): Add trap.
19122 (b): New mode attribute.
19123 (*btst): Rename into...
19124 (*btst<mode>): ...this and adjust.
19125 (*cbranchsi4_btst_insn): Rename into...
19126 (*cbranch<mode>4_btst_insn): ...this and adjust.
19127 (trap): New define_insn.
19129 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19131 PR tree-optimization/79389
19132 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19133 (noce_reversed_cond_code): New function.
19134 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19135 reversed_comparison_code. Formatting fix.
19136 (noce_try_store_flag): Test rev_cond != NULL in addition to
19137 reversed_comparison_code.
19138 (noce_try_store_flag_constants): Likewise.
19139 (noce_try_store_flag_mask): Likewise.
19140 (noce_try_addcc): Use rev_cond if non-NULL instead of
19141 reversed_comparison_code.
19142 (noce_try_cmove_arith): Likewise. Formatting fixes.
19143 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19144 (noce_find_if_block): Initialize rev_cond.
19145 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19146 instead of false as last argument never attempt to reverse it
19149 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19151 PR tree-optimization/79663
19152 * tree-predcom.c (combine_chains): Process refs in reverse order
19153 only for ZERO length chains, and add explaining comment.
19155 2017-02-23 Jeff Law <law@redhat.com>
19157 PR tree-optimization/79578
19158 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19159 in call to operand_equal_p.
19161 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19164 * config/i386/cpuid.h: Fix another undefined behavior.
19166 2017-02-23 Richard Biener <rguenther@suse.de>
19168 PR tree-optimization/79683
19169 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19170 vector types for data-refs.
19172 2017-02-23 Martin Liska <mliska@suse.cz>
19174 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19176 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19178 PR middle-end/79665
19179 * internal-fn.c (get_range_pos_neg): Moved to ...
19180 * tree.c (get_range_pos_neg): ... here. No longer static.
19181 * tree.h (get_range_pos_neg): New prototype.
19182 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19183 are known to be in between 0 and signed maximum inclusive, try to
19184 expand both unsigned and signed divmod and use the cheaper one from
19187 2017-02-22 Jeff Law <law@redhat.com>
19189 PR tree-optimization/79578
19190 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19191 to compare base operands.
19193 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19196 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19197 gpc_reg_operand instead of fpr_reg_operand.
19199 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19201 * config/mips/mips.c (mips_return_in_memory): Force FP
19202 vector types to be returned in memory for o32 ABI.
19204 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19206 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19207 instead of DW_TAG_member for static data member declarations and don't
19208 set no_linkage_name for static inline data members.
19209 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19212 2017-02-22 Martin Liska <mliska@suse.cz>
19214 * doc/invoke.texi: Replace inequality signs with square brackets
19217 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19219 PR tree-optimization/68644
19220 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19222 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19225 * lra-constraints.c (simplify_operand_subreg): Handle
19226 WORD_REGISTER_OPERATIONS targets.
19228 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19231 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19232 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19233 elimination by swapping fld*.
19235 2017-02-22 Richard Biener <rguenther@suse.de>
19237 PR tree-optimization/79673
19238 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19239 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19240 irrelevant address-space qualifiers and avoiding a
19241 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19243 2017-02-22 Richard Biener <rguenther@suse.de>
19245 PR tree-optimization/79666
19246 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19247 to not symbolically negate if that may introduce undefined
19250 2017-02-22 Martin Liska <mliska@suse.cz>
19253 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19254 * data-streamer-out.c (streamer_write_gcov_count_stream):
19256 * value-prof.c (stream_out_histogram_value): Make assert more
19257 precise based on type of counter.
19259 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19262 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19263 Use nonimmediate_operand instead of memory_operand for operand 1.
19264 (float-extend standard_x87sse_constant_load splitter): Ditto.
19266 2017-02-21 Jeff Law <law@redhat.com>
19268 PR tree-optimization/79621
19269 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19270 blocks with edges to themselves.
19272 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19275 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19276 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19277 Use gimple_call_builtin_p.
19280 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19281 on temporarily removed DEBUG_INSNs.
19283 PR tree-optimization/79649
19284 * tree-loop-distribution.c (classify_partition): Give up on
19285 non-generic address space loads/stores.
19287 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19289 * doc/loop.texi (Loop manipulation): Remove nonexistent
19290 tree_ssa_loop_version from the documentation.
19291 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19293 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19296 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19297 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19298 * config/rs6000/rs6000.c: Include except.h.
19299 (rs6000_expand_split_stack_prologue): Call
19300 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19302 2017-02-21 Martin Jambor <mjambor@suse.cz>
19305 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19306 have been analyzed.
19308 2017-02-21 Martin Jambor <mjambor@suse.cz>
19310 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19311 for backward compatibility only.
19312 * doc/invoke.texi (Option Summary): Remove all references to
19313 -fipa-cp-alignment.
19315 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19319 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19321 * lra-constraints.c (curr_insn_transform): Handle
19322 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19324 2017-02-21 Martin Liska <mliska@suse.cz>
19326 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19328 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19330 PR translation/79638
19331 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19333 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19336 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19337 (arm_function_ok_for_sibcall): Return false for an indirect call by
19338 descriptor if all the argument registers are used.
19339 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19340 alignment of the function.
19342 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19344 PR tree-optimization/61441
19345 * simplify-rtx.c (simplify_const_unary_operation): For
19346 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19347 the sNaN unmodified.
19349 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19351 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19352 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19353 instead of SYSTEM_HEADER_DIR.
19355 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19356 Martin Liška <mliska@suse.cz>
19358 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19359 Fix typos and grammar, use active voice, and clarify.
19361 2017-02-20 Marek Polacek <polacek@redhat.com>
19363 PR middle-end/79537
19364 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19367 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19369 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19372 * config/i386/i386.c (ix86_expand_builtin): Handle
19373 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19374 ix86_builtins_isa[fcode].isa as a requirement of those
19375 flags and any other flag in the bitmask.
19376 (ix86_init_mmx_sse_builtins): Use 0 instead of
19377 ~OPTION_MASK_ISA_64BIT as mask.
19378 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19379 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19380 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19381 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19383 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19386 * lra-constraints.c (split_reg): Check requested split mode
19387 is supported by the register.
19389 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19391 * lra-constraints.c (simplify_operand_subreg): Remove early
19394 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19397 * lra-constraints.c (curr_insn_transform): Tighten condition
19398 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19400 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19403 * lra-constraints.c (curr_insn_transform): Handle
19404 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19406 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19409 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19411 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19413 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19416 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19419 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19421 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19424 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19426 * final.c (last_columnnum, override_columnnum): New variables.
19427 (final_start_function): Set last_columnnum, pass it to begin_prologue
19428 hook and pass 0 to dwarf2out_begin_prologue.
19429 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19430 to source_line debug hook.
19431 (notice_source_line): Compute last_columnnum and for debug_column_info
19432 return true on column changes.
19433 * debug.h (struct gcc_debug_hooks): Add column argument to
19434 source_line and begin_prologue hooks.
19435 (debug_nothing_int_charstar_int_bool): Remove prototype.
19436 (debug_nothing_int_int_charstar,
19437 debug_nothing_int_int_charstar_int_bool): New prototypes.
19438 (dwarf2out_begin_prologue): Add column argument.
19439 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19440 begin_prologue hooks.
19441 (debug_nothing_int_charstar_int_bool): Remove.
19442 (debug_nothing_int_int_charstar,
19443 debug_nothing_int_int_charstar_int_bool): New functions.
19444 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19445 through to dwarf2out_source_line.
19446 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19447 (dwarf2out_source_line): Add column argument, emit it if requested.
19448 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19450 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19451 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19452 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19453 through to dwarf2out_begin_prologue.
19454 (vmsdbgout_source_line): Add column argument, pass it through to
19455 dwarf2out_source_line.
19456 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19457 dbxout_source_line caller.
19458 (dbxout_source_line): Add column argument.
19460 * common.opt (gno-column-info, gcolumn-info): New options.
19461 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19462 (check_die): Also test for multiple DW_AT_decl_column attributes.
19463 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19464 DW_AT_decl_column if requested.
19465 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19467 (gen_variable_die): Likewise.
19468 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19469 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19472 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19473 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19474 (ix86_handle_option): Handle OPT_m3dnowa.
19475 * doc/invoke.texi (-m3dnowa): Document.
19476 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19477 -m3dnowa instead of -m3dnow -march=athlon.
19480 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19481 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19483 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19486 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19487 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19488 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19489 generator for vsx_xxpermdi_<mode>_be.
19490 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19491 force big-endian semantics.
19492 (vsx_xxpermdi_<mode>_be): New define_expand with same
19493 implementation as previous version of vsx_xxpermdi_<mode>.
19495 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19497 PR tree-optimization/79327
19498 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19499 variable, its initialization and use.
19501 2017-02-17 Julia Koval <julia.koval@intel.com>
19503 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19504 (OPTION_MASK_ISA_PKU_UNSET): New.
19505 (ix86_handle_option): Handle -mrdpid.
19506 * config/i386/cpuid.h (bit_RDPID): New.
19507 * config/i386/driver-i386.c (host_detect_local_cpu):
19508 Detect RDPID feature.
19509 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19510 * config/i386/i386-c.c (ix86_target_macros_internal):
19512 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19513 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19514 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19515 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19516 * config/i386/i386.md (define_insn "rdpid"): New.
19517 * config/i386/i386.opt Add -mrdpid.
19518 * config/i386/immintrin.h (_rdpid_u32): New.
19520 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19522 PR rtl-optimization/79541
19523 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19524 instead of transforming it into USE.
19526 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19528 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19529 If HONOR_SNANS (SFmode) force the input to a register.
19530 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19531 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19532 an frsp or similar insn.
19534 2017-02-17 Martin Liska <mliska@suse.cz>
19536 PR rtl-optimization/79577
19537 * params.def (selsched-max-sched-times): Increase minimum to 1.
19539 2017-02-17 Martin Liska <mliska@suse.cz>
19541 PR rtl-optimization/79574
19542 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19544 2017-02-17 Martin Liska <mliska@suse.cz>
19546 PR tree-optimization/79529
19547 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19548 ssa_defined_default_def_p to handle cases which are implicitly
19550 * tree-ssa.c (ssa_defined_default_def_p): New function.
19551 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19552 which are implicitly defined.
19553 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19555 2017-02-17 Richard Biener <rguenther@suse.de>
19557 PR middle-end/79576
19558 * params.def (max-ssa-name-query-depth): Limit to 10.
19560 2017-02-17 Richard Biener <rguenther@suse.de>
19562 PR tree-optimization/79552
19563 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19566 2017-02-17 Richard Biener <rguenther@suse.de>
19569 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19571 2017-02-17 Marek Polacek <polacek@redhat.com>
19573 PR middle-end/79536
19574 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19575 (fold_negate_expr): New wrapper.
19577 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19579 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19580 Correct terminology and de-emphasize pre-standard behavior.
19582 2017-02-16 Alan Modra <amodra@gmail.com>
19584 PR rtl-optimization/79286
19585 * ira.c (def_dominates_uses): New function.
19586 (update_equiv_regs): Don't create an equivalence for insns that
19587 may trap where the register def does not dominate the use.
19589 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19591 PR rtl-optimization/78127
19592 * lra.c (lra): Call lra_eliminate before finish the loop after
19595 2017-02-16 Richard Biener <rguenther@suse.de>
19597 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19599 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19600 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19601 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19602 (isl_val_int_from_wi): New function.
19603 (extract_affine_gmp): Rename to ...
19604 (extract_affine_wi): ... this, take a widest_int.
19605 (extract_affine_int): Just wrap extract_affine_wi.
19606 (add_param_constraints): Use isl_val_int_from_wi.
19607 (add_loop_constraints): Likewise, and extract_affine_wi.
19609 2017-02-15 Jeff Law <law@redhat.com>
19611 PR middle-end/79521
19612 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19613 ira_init_register_move_cost_if_necessary.
19615 2017-02-15 Martin Sebor <msebor@redhat.com>
19617 PR middle-end/32003
19618 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19619 removed in a prior commit.
19621 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19623 PR tree-optimization/79347
19624 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19625 counters during peeling.
19627 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19629 * Makefile.in (site.exp): Remove "set ISLVER".
19631 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19634 * real.c (real_from_integer): Call real_convert even for decimal.
19636 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19639 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19641 2017-02-14 Andrew Pinski <apinski@cavium.com>
19643 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19644 cores and change the partno/implementer to be correct.
19645 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19646 the 'B" as the implementer.
19647 * config/aarch64/aarch64-tune.md: Regenerate.
19649 2017-02-14 Carl Love <cel@us.ibm.com>
19651 * config/rs6000/rs6000.c: Add case statement entry to make the
19652 xvcvuxdsp built-in argument unsigned.
19653 * config/rs6000/vsx.md: Fix the source and return operand types so they
19654 match the instruction definitions from the ISA document. Fix typo
19655 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19658 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19661 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19662 member early_clobber_alts.
19663 * lra-lives.c (reg_early_clobber_p): New.
19664 (process_bb_lives): Use it.
19665 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19666 (debug_operand_data): Initialize early_clobber_alts.
19667 (setup_operand_alternative): Set up early_clobber_alts.
19668 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19669 alternatives to new_insn_reg.
19670 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19672 (lra_update_insn_regno_info): Pass the new arg.
19674 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19676 PR middle-end/79505
19677 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19678 (new_oacc_loop_raw): Don't clear already cleared fields.
19681 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19682 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19683 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19685 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19688 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19690 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19693 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19694 the extra instruction to the right place to store 128-bit constant
19697 2017-02-14 Martin Sebor <msebor@redhat.com>
19699 PR middle-end/79448
19700 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19701 warning for strings of unknown length.
19703 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19705 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19707 2017-02-14 Jeff Law <law@redhat.com>
19710 * ira-costs.c (scan_one_insn): Initialize register move costs
19711 for pseudos seen in USE/CLOBBER insns.
19713 PR tree-optimization/79095
19714 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19715 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19716 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19717 if the operands are known to be not equal, then the resulting range
19719 (intersect_ranges): If the new range is ~[0,0] and the old range is
19720 wide, then prefer ~[0,0].
19721 * tree-vrp.c (overflow_comparison_p_1): New function.
19722 (overflow_comparison_p): New function.
19723 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19724 if NAME is used in an overflow test.
19725 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19726 overflow check that can be expressed as an equality test, then adjust
19727 ops to be that equality test.
19729 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19731 * config/s390/s390-builtin-types.def: Remove flags argument.
19732 * config/s390/s390.c (s390_init_builtins): Likewise.
19734 2017-02-14 Martin Liska <mliska@suse.cz>
19736 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19737 vector. Fix trailing white spaces.
19739 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19741 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19744 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19746 PR rtl-optimization/68664
19747 * config/arm/arm.c (arm_sched_can_speculate_insn):
19748 New function. Declare prototype.
19749 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19751 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19753 PR rtl-optimization/68664
19754 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19756 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19758 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19760 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19761 max skip bytes for function, loop and jump.
19763 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19765 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19766 ABS_EXPR for gimple dump.
19768 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19771 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19773 PR tree-optimization/79408
19774 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19775 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19776 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19777 also if rhs1 is INTEGER_CST.
19779 2017-02-14 Richard Biener <rguenther@suse.de>
19781 PR middle-end/79432
19782 * tree-into-ssa.c (insert_phi_nodes): When the function can
19783 have abnormal edges rewrite SSA names with broken use-def
19784 dominance out of SSA and register them for PHI insertion.
19786 2017-02-13 Martin Sebor <msebor@redhat.com>
19788 PR middle-end/79496
19789 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19790 clearing info.nowrite flag when snprintf size argument is a range.
19792 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19794 * cprop.c (cprop_jump): Add missing space in string literal.
19795 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19796 (get_constraint_for_component_ref): Likewise.
19797 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19798 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19799 * lra-constraints.c (process_alt_operands): Likewise.
19800 * ipa-inline.c (inline_small_functions): Likewise.
19801 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19802 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19803 * trans-mem.c (diagnose_tm_1_op): Likewise.
19804 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19805 (grid_parallel_clauses_gridifiable): Likewise.
19807 * config/nvptx/mkoffload.c (process): Add space in between
19810 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19811 "MOD4_SSE_REGS" and "ALL_REGS".
19813 * spellcheck.c (test_data): Add , in between "foo" and "food".
19815 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19818 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19819 boundary crossing check and subsequent code generation agree.
19821 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19823 * config/aarch64/aarch64.c (has_memory_op): Delete.
19824 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19827 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19829 PR rtl-optimization/79388
19830 PR rtl-optimization/79450
19831 * combine.c (distribute_notes): When removing TEM_INSN for which
19832 corresponding dest has last value recorded, invalidate that last
19835 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19837 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19838 of explicit '@'. Add missing assembly comment marker on branch costs
19841 2017-02-13 Nathan Sidwell <nathan@acm.org>
19843 * gengtype-lex.l (<in_struct>): Add '/'.
19845 2017-02-13 Martin Liska <mliska@suse.cz>
19848 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19850 2017-02-13 Richard Biener <rguenther@suse.de>
19852 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19854 * configure: Re-generate.
19855 * config.in: Likewise.
19856 * graphite-dependences.c: Simplify as if
19857 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19858 * graphite-isl-ast-to-gimple.c: Likewise.
19859 * graphite-optimize-isl.c: Likewise.
19860 * graphite-poly.c: Likewise.
19861 * graphite-sese-to-poly.c: Likewise.
19862 * graphite.h: Likewise.
19863 * toplev.c: Include isl/version.h and use isl_version () for
19864 printing the ISL version.
19865 * doc/install.texi: Update ISL requirement.
19867 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19869 * doc/standards.texi (Standards): Update reference to
19872 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19874 * doc/extend.texi (Named Address Spaces): sourceware.org now
19876 * doc/install.texi (Binaries): Ditto.
19879 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19881 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19882 standard terminology "stringize"/"stringizing" throughout.
19883 * doc/cppinternals.texi: Likewise.
19885 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19887 * doc/extend.texi: Fix some spelling mistakes and typos.
19888 * doc/invoke.texi: Likewise.
19890 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19893 * params.def (inline-min-speedup) Change from 10 to 8.
19895 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19897 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19900 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19903 * ipa-inline-analysis.c (get_minimal_bb): New function.
19904 (record_modified): Use it.
19905 (remap_edge_change_prob): Handle also ancestor functions.
19907 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19909 * doc/contrib.texi (Contributors): Remove broken link into
19910 the Mauve CVS repository.
19912 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19914 PR middle-end/79454
19915 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19916 result computation whenever lhs doesn't have vector mode, not
19917 just when it has BLKmode.
19919 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19921 * doc/makefile.texi (profiledbootstrap): Refer to the
19922 installation instructions only in textual form.
19924 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19927 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19929 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19931 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19932 (Specific): Update mingw-w64 reference.
19934 (Specific): Remove broken link to Renesas RX processor.
19936 2017-02-10 Richard Biener <rguenther@suse.de>
19938 * toplev.c (process_options): Do not mention obsolete graphite
19939 options when printing sorry message about missing graphite support.
19940 Mention -floop-nest-optimize.
19942 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19944 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19945 (vtst_p16): Likewise.
19946 (vtstq_p8): Likewise.
19947 (vtstq_p16): Likewise.
19949 (vtstq_p64): Likewise.
19950 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19951 (vset_lane_p64): New.
19952 (vsetq_lane_p64): New.
19954 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19956 PR tree-optimization/79411
19957 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19958 stmt operands are SSA_NAMEs used in abnormal phis.
19959 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19962 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19965 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19968 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19970 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19972 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19974 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19978 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19979 not arbitrary TREE_CONSTANT.
19982 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19983 "omp declare target link" attribute unless is_global_var.
19984 * omp-offload.c (find_link_var_op): Likewise.
19986 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19987 Chung-Lin Tang <cltang@codesourcery.com>
19989 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19991 (gimplify_adjust_omp_clauses): Don't delete TILE.
19992 (gimplify_omp_for): Deal with TILE.
19993 * internal-fn.c (expand_GOACC_TILE): New function.
19994 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19996 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19997 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19999 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20000 avoid DIV for outermost collapse var.
20001 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20002 Remove out of date comments, fix whitespace.
20003 * omp-general.c (omp_extract_for_data): Deal with tiling.
20004 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20005 adjust OLF_DIM_BASE value.
20006 (struct omp_for_data): Add tiling field.
20007 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20008 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20009 for auto loops. Remove default auto determining, moved to
20010 oacc_loop_fixed_partitions.
20011 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20012 stmts, add e_mask field.
20013 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20014 (oacc_thread_numbers): Use oacc_dim_call.
20015 (oacc_xform_tile): New.
20016 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20017 (finish_oacc_loop): Adjust for ifns vector.
20018 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20019 add case for GOACC_TILE fns.
20020 (oacc_loop_xform_loop): Delete.
20021 (oacc_loop_process): Iterate over call list directly, and add
20022 handling for GOACC_TILE fns.
20023 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20025 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20026 vector partitioning to outer loops. Assign 2 partitions to loops
20027 when available. Add TILE handling.
20028 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20029 (execite_oacc_device_lower): Process GOACC_TILE fns,
20030 ignore unknown specs.
20031 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20032 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20033 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20035 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20037 * configure.ac (ACX_BUGURL): Update.
20038 * configure: Regenerate.
20040 2017-02-09 Richard Biener <rguenther@suse.de>
20042 PR tree-optimization/69823
20043 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20044 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20046 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20048 * config/arc/arc-c.def: Add __NPS400__ definition.
20049 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20050 (TARGET_NPS400): Define.
20052 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20054 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20056 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20057 pointer, arch_info.
20058 (arc_cpu_types): Fill the arch_info field with a pointer into the
20059 arc_arch_types table.
20060 (arc_selected_cpu): Declare.
20061 * config/arc/arc.c (arc_selected_cpu): Make global.
20062 (arc_selected_arch): Delete.
20063 (arc_base_cpu): Delete.
20064 (arc_override_options): Remove references to deleted variables,
20065 update access to arch information.
20066 (ARC_OPT): Update access to arch information.
20067 (ARC_OPTX): Likewise.
20068 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20069 (TARGET_ARC600): Update access to arch information.
20070 (TARGET_ARC601): Likewise.
20071 (TARGET_ARC700): Likewise.
20072 (TARGET_EM): Likewise.
20073 (TARGET_HS): Likewise.
20074 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20077 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20080 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20081 condition/operands for integer GE/LE/GEU/LEU operations.
20083 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20085 PR translation/79397
20086 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20089 2017-02-08 Martin Jambor <mjambor@suse.cz>
20092 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20093 whether allocation happened.
20094 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20095 nothing was allocated.
20097 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20099 PR tree-optimization/79408
20100 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20101 constant, but SSA_NAME with a known integer range, use the minimum
20102 of that range instead of op1 to determine if modulo can be replaced
20103 with its first operand.
20105 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20107 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20109 2017-02-08 Richard Biener <rguenther@suse.de>
20111 PR tree-optimization/71824
20112 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20113 Check all loops contained in the merged region.
20115 2017-02-07 Andrew Pinski <apinski@cavium.com>
20117 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20119 2017-02-07 Andrew Pinski <apinski@cavium.com>
20121 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20122 (thunderxt88): Likewise.
20123 (thunderxt81): Disable LSE and change v8.1 to v8.
20124 (thunderxt83): Likewise.
20126 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20127 Richard Biener <rguenther@suse.de>
20129 PR middle-end/79399
20130 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20131 type from int to size_t.
20132 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20134 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20136 PR rtl-optimization/79386
20137 * cprop.c (bypass_conditional_jumps): Initialize
20138 bypass_last_basic_block already before splitting bbs after
20139 unconditional traps...
20140 (bypass_conditional_jumps): ... rather than here.
20143 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20144 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20145 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20146 fix -masm=intel patterns.
20148 2017-02-07 Richard Biener <rguenther@suse.de>
20150 PR tree-optimization/79256
20151 PR middle-end/79278
20152 * builtins.c (get_object_alignment_2): Use min_align_of_type
20153 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20154 and ADJUST_FIELD_ALIGN.
20156 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20158 * doc/tm.texi: Regenerate.
20159 * stor-layout.c (layout_decl): Adjust.
20160 (update_alignment_for_field): Likewise.
20161 (place_field): Likewise.
20162 (min_align_of_type): Likewise.
20163 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20164 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20165 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20166 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20167 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20168 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20169 * config/i386/i386.c (x86_field_alignment): Likewise.
20170 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20171 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20172 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20173 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20174 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20175 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20179 2017-01-30 Richard Biener <rguenther@suse.de>
20181 PR tree-optimization/79256
20182 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20183 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20186 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20188 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20189 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20190 builtins to SImode and emit a zero-extend, if necessary.
20192 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20194 * docs/invoke.texi (RISC-V Options): Alphabetize.
20196 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20198 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20201 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20203 * config/riscv/riscv.c: New file.
20204 * common/config/riscv/riscv-common.c: Likewise.
20205 * config.gcc: Likewise.
20206 * config/riscv/constraints.md: Likewise.
20207 * config/riscv/elf.h: Likewise.
20208 * config/riscv/generic.md: Likewise.
20209 * config/riscv/linux.h: Likewise.
20210 * config/riscv/multilib-generator: Likewise.
20211 * config/riscv/peephole.md: Likewise.
20212 * config/riscv/pic.md: Likewise.
20213 * config/riscv/predicates.md: Likewise.
20214 * config/riscv/riscv-builtins.c: Likewise.
20215 * config/riscv/riscv-c.c: Likewise.
20216 * config/riscv/riscv-ftypes.def: Likewise.
20217 * config/riscv/riscv-modes.def: Likewise.
20218 * config/riscv/riscv-opts.h: Likewise.
20219 * config/riscv/riscv-protos.h: Likewise.
20220 * config/riscv/riscv.h: Likewise.
20221 * config/riscv/riscv.md: Likewise.
20222 * config/riscv/riscv.opt: Likewise.
20223 * config/riscv/sync.md: Likewise.
20224 * config/riscv/t-elf-multilib: Likewise.
20225 * config/riscv/t-linux: Likewise.
20226 * config/riscv/t-linux-multilib: Likewise.
20227 * config/riscv/t-riscv: Likewise.
20228 * configure.ac: Likewise.
20229 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20230 Waterman as RISC-V maintainers.
20231 * doc/install.texi: Add RISC-V entries.
20232 * doc/invoke.texi: Add RISC-V options section.
20233 * doc/md.texi: Add RISC-V constraints section.
20234 * configure: Regenerated.
20236 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20239 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20240 false values to be constant vectors with all 0 or all 1 bits set.
20241 (vcondu<mode><mode>): Likewise.
20242 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20244 (fpmask_comparison_operator): Update comment.
20245 (vecint_comparison_operator): New predicate.
20246 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20247 vector conditionals when the true and false values are constant
20248 vectors with all 0 bits or all 1 bits set.
20250 2017-02-06 Martin Sebor <msebor@redhat.com>
20252 PR tree-optimization/79376
20253 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20255 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20257 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20258 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20259 to simplify split condition.
20261 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20263 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20264 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20267 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20269 PR rtl-optimization/68664
20270 * target.def (can_speculate_insn): New hook.
20271 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20272 * doc/tm.texi: Regenerate.
20273 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20274 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20275 (rs6000_sched_can_speculate_insn): New function.
20277 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20279 PR tree-optimization/79284
20280 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20281 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20282 vectorizable_mask_load_store, vectorizable_operation,
20283 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20284 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20285 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20286 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20287 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20288 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20289 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20290 is_gimple_assign (stmt). Replace another such test with
20291 is_gimple_assign (stmt).
20293 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20296 * config/avr/avr.c (rtl-iter.h): Include it.
20297 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20298 (avr_legitimate_combined_insn): ...and implementation.
20300 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20302 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20303 * config/s390/s390.c (s390_const_operand_ok)
20304 (s390_canonicalize_comparison, s390_extract_part)
20305 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20306 (s390_contiguous_bitmask_p, s390_rtx_costs)
20307 (legitimize_pic_address): Likewise.
20308 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20309 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20310 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20311 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20312 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20314 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20316 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20317 REGNO($0) == REGNO($1).
20319 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20321 * config/s390/linux.h(SIZE_TYPE): Add comment.
20323 2017-02-06 Julian Brown <julian@codesourcery.com>
20324 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20325 Virendra Pathak <virendra.pathak@broadcom.com>
20327 * config/aarch64/aarch64-cores.def: Change the scheduler
20329 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20330 * config/aarch64/thunderx2t99.md: New file.
20332 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20334 * doc/standards.texi (Go Language): Update link to language
20337 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20339 * tree-eh.c (lower_resx): Sanitize profile.
20340 (cleanup_empty_eh_move_lp): Likewise.
20342 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20345 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20347 * cfgloopmanip.h (loop_version): Update prototype.
20348 * modulo-sched.c (sms_schedule): Update call of loop_version.
20349 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20350 * tree-parloops.c (gen_parallel_loop): Likewise.
20351 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20352 * tree-ssa-loop-split.c (split_loop): Likewise.
20353 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20354 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20356 2017-02-05 Martin Liska <mliska@suse.cz>
20359 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20361 (print_operand_address): Initialize a struct to zero.
20363 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20365 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20366 garbage collector only in textual form.
20368 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20370 * doc/extend.texi (x86 specific memory model extensions for
20371 transactional memory): Simplify a phrase.
20373 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20376 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20377 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20378 (atomic_storedi_1): Likewise.
20380 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20382 PR tree-optimization/79338
20383 * tree-parloops.c (gather_scalar_reductions): Don't call
20384 vect_analyze_loop_form for loop->inner before destroying loop's
20387 2017-02-03 Martin Sebor <msebor@redhat.com>
20389 PR tree-optimization/79327
20390 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20391 when precision has resulted in leading zeros.
20392 (format_integer): Adjust the likely counter to assume an unknown
20393 argument that may be zero is non-zero.
20395 2017-02-03 Jason Merrill <jason@redhat.com>
20398 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20399 avoid copying non-taken branch.
20401 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20403 PR tree-optimization/79340
20404 * tree-vect-loop.c (vectorizable_reduction): Release
20405 vec_defs elements after safe_splicing them into other vectors.
20408 PR tree-optimization/79327
20409 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20410 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20412 (format_integer): Use wide_int_to_tree instead of build_int_cst
20413 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20414 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20415 of shortest and longest sequence.
20417 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20419 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20420 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20422 2017-02-03 Walter Lee <walt@tilera.com>
20425 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20426 after initial stackframe link reg save.
20427 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20429 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20432 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20433 wu for stxssp alternative.
20435 2017-02-03 Martin Sebor <msebor@redhat.com>
20437 PR tree-optimization/79352
20438 * gimple-fold.c (get_range_strlen): Add argument.
20439 (get_range_strlen): Change return type to bool.
20440 (get_maxval_strlen): Pass in a dummy argument.
20441 * gimple-fold.h (get_range_strlen): Change return type to bool.
20442 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20443 * tree.h (array_at_struct_end_p): Add argument.
20444 * tree.c (array_at_struct_end_p): Handle it.
20446 2017-02-03 Martin Liska <mliska@suse.cz>
20449 * multiple_target.c (create_dispatcher_calls): Redirect edge
20450 from a caller of a dispatcher.
20451 (expand_target_clones): Make the clones local.
20452 (ipa_target_clone): Do both target clones and resolvers.
20453 (ipa_dispatcher_calls): Remove the pass.
20454 (pass_dispatcher_calls::gate): Likewise.
20455 (make_pass_dispatcher_calls): Likewise.
20456 * passes.def (pass_target_clone): Put as very first IPA early
20459 2017-02-03 Martin Liska <mliska@suse.cz>
20461 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20462 in case of a function with ifunc attribute.
20464 2017-02-03 Martin Liska <mliska@suse.cz>
20466 * cgraph.c (cgraph_node::dump): Dump function version info.
20467 * symtab.c (symtab_node::dump_base): Add missing new line.
20469 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20471 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20472 (ifcombine_ifandif): Use it.
20474 2017-02-03 Martin Liska <mliska@suse.cz>
20476 * doc/invoke.texi: Document default value for
20477 use-after-scope-direct-emission-threshold.
20479 2017-02-03 Martin Liska <mliska@suse.cz>
20481 PR tree-optimization/79339
20482 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20483 (format_floating): Likewise.
20485 2017-02-03 Martin Liska <mliska@suse.cz>
20488 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20490 (ipa_node_params_t::remove): Likewise.
20491 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20492 initialization from removed ipa_node_params_t::insert.
20493 (ipa_node_params::~ipa_node_params): Move from removed
20494 ipa_node_params_t::release.
20495 * symbol-summary.h (symbol_summary::m_released): New member.
20496 Do not release a summary twice. Do not allow to call finalizer
20497 for types of a summary that live in GGC memory.
20499 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20501 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20504 2017-02-02 Martin Sebor <msebor@redhat.com>
20506 PR middle-end/79275
20507 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20508 (format_string): Tighten up the range of output for non-constant
20509 strings and correct the expected range for wide non-constant strings.
20511 2017-02-02 Martin Sebor <msebor@redhat.com>
20513 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20515 PR middle-end/32003
20516 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20518 (-fdump-tree-@var): Add to index and document how to come up
20519 with pass-specific option and dump file names.
20520 (-fdump-passes): Clarify where to look for output.
20522 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20524 PR middle-end/77445
20525 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20526 statistics of the analyzed path; allow threading for speed when
20527 any of BBs along the path are optimized for speed.
20529 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20531 PR middle-end/78468
20532 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20533 settings of the virtual registers.
20536 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20538 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20539 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20542 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20544 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20545 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20547 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20549 * config/s390/s390.md: Add missing comments with the expanded
20551 * config/s390/vector.md: Likewise.
20552 * config/s390/vx-builtins.md: Likewise.
20554 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20557 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20558 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20559 conditions on a single line.
20561 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20563 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20564 __S390_VX__ to __VX__.
20566 2017-02-01 Andrew Pinski <apinski@cavium.com>
20568 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20569 stmt_info to record_stmt_cost.
20570 (vect_get_known_peeling_cost): Pass stmt_info if known to
20572 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20573 cpu_vector_cost field into
20574 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20575 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20576 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20577 splitting of scalar_stmt_cost and vec_stmt_cost.
20578 (thunderx_vector_cost): Likewise.
20579 (cortexa57_vector_cost): LIkewise.
20580 (exynosm1_vector_cost): Likewise.
20581 (xgene1_vector_cost): Likewise.
20582 (thunderx2t99_vector_cost): Improve after the splitting of the two
20584 (aarch64_builtin_vectorization_cost): Update for the splitting of
20585 scalar_stmt_cost and vec_stmt_cost.
20587 2017-02-01 Torvald Riegel <triegel@redhat.com>
20588 Richard Henderson <rth@redhat.com>
20590 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20591 conditional on existance of a fast atomic load.
20592 * optabs-query.c (can_atomic_load_p): New function.
20593 * optabs-query.h (can_atomic_load_p): Declare it.
20594 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20595 no fast atomic load is available for the particular size of access.
20596 (expand_atomic_compare_and_swap): Likewise.
20597 (expand_atomic_load): Likewise.
20598 (expand_atomic_store): Likewise.
20599 (expand_atomic_fetch_op): Likewise.
20600 * testsuite/lib/target-supports.exp
20601 (check_effective_target_sync_int_128): Remove x86 because it provides
20602 no fast atomic load.
20603 (check_effective_target_sync_int_128_runtime): Likewise.
20605 2017-02-01 Richard Biener <rguenther@suse.de>
20607 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20608 (graphite_transform_loops): Provide opt-info for optimized nests.
20609 * tree-parloop.c (parallelize_loops): Provide opt-info for
20610 parallelized loops.
20612 2017-02-01 Richard Biener <rguenther@suse.de>
20614 PR middle-end/79315
20615 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20616 was not set before.
20618 2017-02-01 Richard Biener <rguenther@suse.de>
20620 PR tree-optimization/71824
20621 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20622 Verify the loops are valid in the merged SESE region.
20623 (scop_detection::can_represent_loop_1): Check analyzing the
20624 evolution of the number of iterations in the region succeeds.
20626 2017-01-31 Ian Lance Taylor <iant@golang.org>
20628 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20629 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20631 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20633 PR preprocessor/79210
20634 * input.c (get_substring_ranges_for_loc): Replace line_width
20635 assertion with error-handling.
20637 2017-01-31 Richard Biener <rguenther@suse.de>
20639 PR tree-optimization/77318
20640 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20641 (create_pw_aff_from_tree): Take loop parameter.
20642 (add_condition_to_pbb): Pass loop of the condition to
20643 create_pw_aff_from_tree.
20645 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20647 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20648 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20650 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20654 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20656 (convert_int_to_float128): Likewise.
20657 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20658 (convert_int_to_float128): Likewise.
20659 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20660 (UNSPEC_IEEE128_CONVERT): Likewise.
20661 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20662 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20663 Use local variables for IBM extended format.
20664 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20665 (fix_trunc<mode>si2_fprs): Likewise.
20666 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20667 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20668 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20669 to know that we can now have integers of all sizes in vector
20671 (fix<uns>_<mode>di2_hw): Likewise.
20672 (float<uns>_<mode>si2_hw): Likewise.
20673 (fix_<mode>si2_hw): Likewise.
20674 (fixuns_<mode>si2_hw): Likewise.
20675 (float<uns>_<mode>di2_hw): Likewise.
20676 (float_<mode>di2_hw): Likewise.
20677 (float_<mode>si2_hw): Likewise.
20678 (floatuns_<mode>di2_hw): Likewise.
20679 (floatuns_<mode>si2_hw): Likewise.
20680 (xscvqp<su>wz_<mode>): Delete, no longer used.
20681 (xscvqp<su>dz_<mode>): Likewise.
20682 (xscv<su>dqp_<mode>): Likewise.
20683 (ieee128_mfvsrd_64bit): Likewise.
20684 (ieee128_mfvsrd_32bit): Likewise.
20685 (ieee128_mfvsrwz): Likewise.
20686 (ieee128_mtvsrw): Likewise.
20687 (ieee128_mtvsrd_64bit): Likewise.
20688 (ieee128_mtvsrd_32bit): Likewise.
20690 2017-01-31 Martin Liska <mliska@suse.cz>
20693 * ipa-prop.c (ipa_free_all_node_params): Call release method
20694 instead of ~sumbol_summary to not to trigger double times
20697 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20699 PR tree-optimization/71691
20700 * bitmap.h (class auto_bitmap): New.
20701 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20702 is_maybe_undefined instead of ssa_undefined_value_p.
20704 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20706 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20707 __S390_ARCH_LEVEL__ to __ARCH__.
20709 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20711 PR tree-optimization/79267
20712 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20713 if should_remove_lhs_p is true.
20715 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20718 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20719 (add_alignment_attribute): New.
20720 (base_type_die): Add alignment attribute.
20721 (subrange_type_die): Likewise.
20722 (modified_type_die): Likewise.
20723 (gen_array_type_die): Likewise.
20724 (gen_descr_array_type_die: Likewise.
20725 (gen_enumeration_type_die): Likewise.
20726 (gen_subprogram_die): Likewise.
20727 (gen_variable_die): Likewise.
20728 (gen_field_die): Likewise.
20729 (gen_ptr_to_mbr_type_die): Likewise.
20730 (gen_struct_or_union_type_die): Likewise.
20731 (gen_subroutine_type_die): Likewise.
20732 (gen_typedef_die): Likewise.
20733 (base_type_cmp): Compare alignment attribute.
20735 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20738 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20739 (setb_unsigned) New pattern for setb with CCUNS.
20740 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20741 subfc./subfe sequence to avoid overflow problems. Generate a
20742 shorter sequence with cmpld/setb for power9.
20743 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20744 for generating subfc. instruction.
20745 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20746 now uses this instruction.
20748 2017-01-30 Ian Lance Taylor <iant@google.com>
20751 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20752 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20754 2017-01-30 Martin Sebor <msebor@redhat.com>
20756 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20757 Move constant to the right of a relational operator.
20758 (get_mpfr_format_length, format_character, format_string): Ditto.
20759 (should_warn_p, maybe_warn): Same.
20761 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20763 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20766 * asan.c (get_translation_unit_decl): Remove function.
20767 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20769 2017-01-30 Martin Liska <mliska@suse.cz>
20771 PR gcov-profile/79259
20772 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20773 -fprofile-generate.
20775 2017-01-30 Martin Liska <mliska@suse.cz>
20778 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20779 Initialize variables with NULL value.
20781 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20784 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20786 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20788 2017-01-30 Richard Biener <rguenther@suse.de>
20790 PR tree-optimization/79276
20791 * tree-vrp.c (process_assert_insertions): Properly adjust common
20792 when removing a duplicate.
20794 * gcc.dg/torture/pr79276.c: New testcase.
20796 2017-01-30 Richard Biener <rguenther@suse.de>
20798 PR tree-optimization/79256
20799 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20800 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20802 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20804 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20807 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20808 ("*r<noxa>sbg_<mode>_sll_bitmask")
20809 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20810 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20811 Use contiguous_bitmask_nowrap_operand.
20813 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20816 * config/rs6000/altivec.h (vec_xl): Revise #define.
20817 (vec_xst): Likewise.
20819 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20821 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20823 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20825 PR rtl-optimization/79194
20826 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20827 traps before call to bypass_conditional_jumps.
20829 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20831 PR tree-optimization/71374
20832 * lra-constraints.c (check_conflict_input_operands): New.
20833 (match_reload): Use it.
20835 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20838 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20839 account to calculate conflict_set.
20841 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20843 PR rtl-optimization/78559
20844 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20845 other_insn in combine.
20847 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20849 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20850 uint16_type_node for BT_UINT16.
20852 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20854 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20855 "RTL Tests" to menu.
20856 (GIMPLE Tests): New node.
20857 (RTL Tests): New node.
20859 2017-01-27 Richard Biener <rguenther@suse.de>
20861 PR tree-optimization/79245
20862 * tree-loop-distribution.c (distribute_loop): Apply cost
20863 modeling also to detected patterns.
20865 2017-01-27 Richard Biener <rguenther@suse.de>
20867 PR tree-optimization/71433
20868 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20869 (compare_assert_loc): New function.
20870 (process_assert_insertions): Sort and optimize assert locations
20871 to remove duplicates and push down identical assertions on
20872 edges to their destination block.
20874 2017-01-27 Richard Biener <rguenther@suse.de>
20876 PR tree-optimization/79244
20877 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20878 out SSA names even if abnormal.
20880 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20882 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20883 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20884 instead of MPFR_RNDN.
20886 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20889 * arm.c (arm_option_override): Don't call build_target_option_node
20890 until after doing all option overrides.
20891 (arm_valid_target_attribute_tree): Likewise.
20893 2017-01-27 Martin Liska <mliska@suse.cz>
20895 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20896 for {cd}tors and C++ {cd}tors.
20898 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20900 * config/s390/s390.md ("*setmem_long_and")
20901 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20903 2017-01-26 Martin Sebor <msebor@redhat.com>
20905 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20908 2017-01-26 Martin Sebor <msebor@redhat.com>
20910 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20911 HAVE_DFmode before using XFmode or DFmode.
20912 (parse_directive): Avoid using the z length modifier to avoid
20913 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20915 PR middle-end/78703
20916 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20917 to accept adjustment as an array.
20918 (get_int_range): New function.
20919 (struct directive): Make width and prec arrays.
20920 (directive::set_width, directive::set_precision): Call get_int_range.
20921 (format_integer, format_floating): Handle width and precision ranges.
20922 (format_string, parse_directive): Same.
20924 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20927 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20928 comdat_type_p set, just clone them, but keep the children in the
20932 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20933 which have direct callers with -fvar-tracking-assignments enabled
20935 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20936 inside of type units.
20938 2017-01-26 Martin Sebor <msebor@redhat.com>
20940 PR middle-end/78703
20941 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20943 (struct format_result): Replace number_chars, number_chars_min,
20944 and number_chars_max with a single member of struct result_range.
20946 (format_result::operator+=): Adjust.
20947 (struct fmtresult): Remove bounded. Handle likely and unlikely
20949 (fmtresult::adjust_for_width_or_precision): New function.
20950 (fmtresult:type_max_digits): New function.
20951 (bytes_remaining): Handle likely and unlikely counters.
20952 (min_bytes_remaining): Remove.
20953 (format_percent): Simplify.
20954 (format_integer, format_floating): Set likely and unlikely counters.
20955 (get_string_length, format_character, format_string): Same.
20956 (format_plain, should_warn_p): New function.
20957 (maybe_warn): Call should_warn_p. Update diagnostic messages
20958 and handle those for all directives, including plain strings.
20959 (format_directive): Handle likely and unlikely counters.
20960 Remove unnecessary quoting from diagnostics. Add an informational
20962 (add_bytes): Remove.
20963 (pass_sprintf_length::compute_format_length): Simplify.
20964 (try_substitute_return_value): Handle likely and unlikely counters.
20966 2017-01-26 Carl Love <cel@us.ibm.com>
20968 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20969 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20971 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20974 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20975 endianess for subregs into account.
20976 * lra-constraints.c (lra_constraints): Do risky transformations
20977 always on the first iteration.
20978 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20979 last_call_used_reg_set.
20980 (process_bb_lives): Define and use last_call_used_reg_set.
20981 * lra.c (lra): Always continue after lra_constraints on the first
20984 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20986 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20988 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20990 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20992 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20993 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20994 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20995 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20996 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20997 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20998 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20999 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21000 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21002 2017-01-26 Marek Polacek <polacek@redhat.com>
21005 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21006 for the third operand.
21008 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21010 PR middle-end/79236
21011 * omp-low.c (struct omp_context): Add simt_stmt field.
21012 (scan_omp_for): Return omp_context *.
21013 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21014 context to the _simt_ SIMD stmt.
21015 (lower_omp_for): For combined SIMD with sibling _simt_
21016 SIMD, make sure to use the same decls in _looptemp_
21017 clauses as in the sibling.
21019 2017-01-26 David Sherwood <david.sherwood@arm.com>
21021 PR middle-end/79212
21022 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21025 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21028 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21029 emit fld b; fld a; if possible.
21031 * brig-builtins.def: Update copyright years.
21032 * config/arm/arm_acle_builtins.def: Update copyright years.
21034 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21037 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21038 constraint instead of o for the stxsd instruction.
21040 2017-01-25 Carl Love <cel@us.ibm.com>
21042 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21043 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21045 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21047 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21049 2017-01-25 Richard Biener <rguenther@suse.de>
21051 PR tree-optimization/69264
21052 * target.def (vector_alignment_reachable): Improve documentation.
21053 * doc/tm.texi: Regenerate.
21054 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21056 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21057 earlier changes with respect to TYPE_USER_ALIGN.
21058 (vector_alignment_reachable_p): Likewise. Improve dumping.
21060 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21063 * config/arm/arm.md (xordi3): Force constant operand into a register
21066 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21068 * doc/invoke.texi (-fstore-merging): Correct default optimization
21069 levels at which it is enabled.
21070 (-O): Move -fstore-merging from list to...
21073 2017-01-25 Richard Biener <rguenther@suse.de>
21076 * omp-expand.c: Include debug.h.
21077 (expand_omp_taskreg): Make sure to generate early debug before
21078 outlining anything from a function.
21079 (expand_omp_target): Likewise.
21080 (grid_expand_target_grid_body): Likewise.
21082 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21085 * asan.c (get_translation_unit_decl): New function.
21086 (asan_add_global): Extract modules file name from globals
21087 TRANSLATION_UNIT_DECL name.
21089 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21092 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21093 for long calls with APCS frame and VFP.
21095 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21097 * cfg.c (original_copy_tables_initialized_p): New function.
21098 * cfg.h (original_copy_tables_initialized_p): New decl.
21099 * cfgrtl.c (relink_block_chain): Guard the call to
21100 free_original_copy_tables with a call to
21101 original_copy_tables_initialized_p.
21102 * cgraph.h (symtab_node::native_rtl_p): New decl.
21103 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21104 (symtab_node::needed_p): Don't assert for early assembly output
21105 for __RTL functions.
21106 (cgraph_node::finalize_function): Set "force_output" for __RTL
21108 (cgraph_node::analyze): Bail out early for __RTL functions.
21109 (analyze_functions): Update assertion to support __RTL functions.
21110 (cgraph_node::expand): Bail out early for __RTL functions.
21111 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21113 * function.h (struct function): Update comment for field
21115 * gimple-expr.c: Include "tree-pass.h".
21116 (gimple_has_body_p): Return false for __RTL functions.
21117 * Makefile.in (OBJS): Add run-rtl-passes.o.
21118 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21120 (gcc::pass_manager::get_clean_slate): New accessor.
21121 * passes.c: Include "insn-addr.h".
21122 (should_skip_pass_p): Add logging. Update logic for running
21123 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21124 property-provider override so it is only done for gimple passes.
21126 (skip_pass): New function.
21127 (execute_one_pass): Call skip_pass when skipping passes.
21128 * read-md.c (md_reader::read_char): Support filtering
21129 the input to a subset of line numbers.
21130 (md_reader::md_reader): Initialize fields
21131 m_first_line and m_last_line.
21132 (md_reader::read_file_fragment): New function.
21133 * read-md.h (md_reader::read_file_fragment): New decl.
21134 (md_reader::m_first_line): New field.
21135 (md_reader::m_last_line): New field.
21136 * read-rtl-function.c (function_reader::create_function): Only
21137 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21138 curr_properties. Set DECL_INITIAL to a dummy block.
21139 (read_rtl_function_body_from_file_range): New function.
21140 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21142 * run-rtl-passes.c: New file.
21143 * run-rtl-passes.h: New file.
21145 2017-01-24 Jeff Law <law@redhat.com>
21147 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21150 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21152 PR tree-optimization/79159
21153 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21154 (record_nonwrapping_iv): Improve boundary using above function if no
21155 value range information.
21157 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21158 Martin Jambor <mjambor@suse.cz>
21160 * brig-builtins.def: New file.
21161 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21162 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21163 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21164 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21165 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21166 * builtin-types.def (BT_INT8): New.
21167 (BT_INT16): Likewise.
21168 (BT_UINT8): Likewise.
21169 (BT_UINT16): Likewise.
21170 (BT_FN_ULONG): Likewise.
21171 (BT_FN_UINT_INT): Likewise.
21172 (BT_FN_UINT_ULONG): Likewise.
21173 (BT_FN_UINT_LONG): Likewise.
21174 (BT_FN_UINT_PTR): Likewise.
21175 (BT_FN_ULONG_PTR): Likewise.
21176 (BT_FN_INT8_FLOAT): Likewise.
21177 (BT_FN_INT16_FLOAT): Likewise.
21178 (BT_FN_UINT32_FLOAT): Likewise.
21179 (BT_FN_UINT16_FLOAT): Likewise.
21180 (BT_FN_UINT8_FLOAT): Likewise.
21181 (BT_FN_UINT64_FLOAT): Likewise.
21182 (BT_FN_UINT16_UINT32): Likewise.
21183 (BT_FN_UINT32_UINT16): Likewise.
21184 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21185 (BT_FN_INT_PTR_INT): Likewise.
21186 (BT_FN_UINT_PTR_UINT): Likewise.
21187 (BT_FN_LONG_PTR_LONG): Likewise.
21188 (BT_FN_ULONG_PTR_ULONG): Likewise.
21189 (BT_FN_VOID_UINT64_UINT64): Likewise.
21190 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21191 (BT_FN_INT8_INT8_INT8): Likewise.
21192 (BT_FN_INT16_INT16_INT16): Likewise.
21193 (BT_FN_INT_INT_INT): Likewise.
21194 (BT_FN_UINT_FLOAT_UINT): Likewise.
21195 (BT_FN_FLOAT_UINT_UINT): Likewise.
21196 (BT_FN_ULONG_UINT_UINT): Likewise.
21197 (BT_FN_ULONG_UINT_PTR): Likewise.
21198 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21199 (BT_FN_UINT_UINT_UINT): Likewise.
21200 (BT_FN_VOID_UINT_PTR): Likewise.
21201 (BT_FN_UINT_UINT_PTR: Likewise.
21202 (BT_FN_UINT32_UINT64_PTR): Likewise.
21203 (BT_FN_INT_INT_UINT_UINT): Likewise.
21204 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21205 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21206 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21207 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21208 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21209 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21210 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21211 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21212 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21213 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21214 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21215 * doc/frontends.texi: List BRIG FE.
21216 * doc/install.texi (Testing): Add BRIG tesring requirements.
21217 * doc/invoke.texi (Overall Options): Mention BRIG.
21218 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21220 2017-01-24 Richard Biener <rguenther@suse.de>
21222 PR translation/79208
21223 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21225 2017-01-24 Martin Jambor <mjambor@suse.cz>
21228 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21229 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21230 and known_contexts.
21232 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21234 PR middle-end/79123
21235 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21236 casts from signed to unsigned really don't have a range.
21238 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21240 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21241 GMP_RNDx for compatiblity.
21243 2017-01-24 Martin Liska <mliska@suse.cz>
21246 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21247 that would prevent us to call alloca with -1 as argument.
21249 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21251 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21252 Avoid -Wformat-security warning.
21254 2017-01-23 Andrew Pinski <apinski@cavium.com>
21256 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21259 2017-01-23 Martin Sebor <msebor@redhat.com>
21261 PR middle-end/78703
21262 * gimple-ssa-sprintf.c (warn_level): New global.
21263 (format_integer): Use it here and throughout the rest of the file.
21264 Use the same switch to compute sign as base.
21265 (maybe_warn): New function.
21266 (format_directive): Factor out warnings into maybe_warn.
21267 Add debugging output. Use warn_level.
21268 (add_bytes): Use warn_level.
21269 (pass_sprintf_length::compute_format_length): Add debugging output.
21270 (try_substitute_return_value): Same.
21271 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21273 PR middle-end/78703
21274 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21275 (struct fmtresult, format_integer, format_floating): Adjust.
21276 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21277 (get_string_length, format_string,format_directive): Same.
21278 (pass_sprintf_length::compute_format_length): Same.
21279 (try_substitute_return_value): Simplify slightly.
21281 PR middle-end/78703
21282 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21283 (fmtresult::operator+=): Outlined.
21284 (struct fmtresult): Add ctors.
21285 (struct conversion_spec): Rename...
21286 (struct directive): ...to this. Add and remove data members.
21287 (directive::set_width, directive::set_precision): New functions.
21288 (format_percent): Use fmtresult ctor.
21289 (get_width_and_precision): Remove.
21290 (format_integer): Make naming changes. Avoid computing width and
21292 (format_floating): Same. Adjust indentation.
21293 (format_character, format_none): New functions.
21294 (format_string): Moved character handling to format_character.
21295 (format_directive): Remove arguments, change return type.
21296 (parse_directive): New function.
21297 (pass_sprintf_length::compute_format_length): Move directive
21298 parsing to parse_directive.
21300 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21302 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21303 (assign_assembler_name_if_needed): ... this.
21304 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21305 (assign_assembler_name_if_needed): ... this.
21306 (free_lang_data_in_cgraph): Adjust callers.
21307 * cgraphunit.c (cgraph_node::analyze): Likewise.
21308 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21311 2017-01-23 Richard Biener <rguenther@suse.de>
21313 PR tree-optimization/79088
21314 PR tree-optimization/79188
21315 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21316 resetting loop bounds after last path deletion. Reset loop
21317 bounds of the target loop, make code match the comments.
21318 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21319 Make sure loops need no fixups.
21321 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21323 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21324 exponent support with double type for first argument.
21325 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21326 type returned by __builtin_vec_extract_sig,
21327 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21328 functions from "vector int" to "vector unsigned int" or from
21329 "vector long long int" to "vector unsigned long long int".
21330 Changed type returned by __builtin_vec_extract_exp,
21331 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21332 functions from "vector int" to "vector unsigned int" or from
21333 "vector long long int" to "vector unsigned long long int".
21334 Changed return type of __builtin_vec_test_data_class,
21335 __builtin_vec_test_data_class_sp, and
21336 __builtin_vec_test_data_class_dp from "vector int" to
21337 "vector bool int" or from "vector long long int" to "vector bool
21338 long long int" and changed second argument type from "unsigned
21339 int" to "int". Added new overloaded function forms "vector float
21340 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21341 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21342 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21343 double, vector unsigned long long int)" and "vector double
21344 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21345 long int)". Changed return type of
21346 __builtin_scalar_test_data_class and
21347 __builtin_scalar_test_data_class_sp and
21348 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21349 int" and changed second argument from "unsigned int" to "int".
21350 Changed type returned by __builtin_scalar_test_neg,
21351 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21352 from "int" to "bool int". Added new overloaded function form
21353 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21354 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21355 exponent double-precision with floating point first argument.
21356 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21357 documentation of scalar_test_data_class, scalar_test_neg,
21358 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21359 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21360 vec_test_data_class built-in functions to reflect refinements in
21361 their type signatures.
21363 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21365 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21367 (aarch64_elf_asm_destructor): Likewise.
21369 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21371 PR rtl-optimization/78634
21372 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21373 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21374 * ifcvt.c (noce_try_cmove): Add missing cost check.
21376 PR rtl-optimization/71724
21377 * combine.c (if_then_else_cond): Look for situations where it is
21378 beneficial to undo the work of one of the recursive calls.
21380 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21382 PR tree-optimization/70754
21383 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21384 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21385 combined stmt before it if not NULL.
21386 (combine_chains): Process refs reversely and compute dominance point
21389 2017-01-23 Martin Liska <mliska@suse.cz>
21391 PR tree-optimization/79196
21392 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21393 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21395 (strlen_optimize_stmt): Call the renamed function.
21397 2017-01-23 Michael Matz <matz@suse.de>
21399 PR tree-optimization/78384
21400 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21402 2017-01-23 Richard Biener <rguenther@suse.de>
21404 PR tree-optimization/79186
21405 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21406 both incoming edges before moving an assert.
21408 2017-01-23 Martin Jambor <mjambor@suse.cz>
21410 * ipa-prop.c (load_from_param_1): Removed.
21411 (load_from_unmodified_param): Bits from load_from_param_1 put back
21413 (load_from_param): Removed.
21414 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21415 with stmt. Reverted back to use of load_from_unmodified_param.
21417 2017-01-23 Martin Jambor <mjambor@suse.cz>
21420 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21421 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21422 field a pointer to garbage collected vector, mark lattices and
21423 ipcp_orig_node with GTY((skip)).
21424 (ipa_get_param_count): Adjust to descriptors being a pointer.
21425 (ipa_get_param): Likewise.
21426 (ipa_get_type): Likewise.
21427 (ipa_get_param_move_cost): Likewise.
21428 (ipa_set_param_used): Likewise.
21429 (ipa_get_controlled_uses): Likewise.
21430 (ipa_set_controlled_uses): Likewise.
21431 (ipa_is_param_used): Likewise.
21432 (ipa_node_params_t): Move into garbage collector. New methods insert
21434 (ipa_node_params_sum): Annotate wth GTY(()).
21435 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21437 (ipa_load_from_parm_agg): Adjust declaration.
21438 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21439 * ipa-profile.c (ipa_profile): Likewise.
21440 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21441 (ipa_populate_param_decls): Make descriptors parameter garbage
21443 (ipa_dump_param): Adjust to descriptors being a pointer.
21444 (ipa_alloc_node_params): Likewise.
21445 (ipa_initialize_node_params): Likewise.
21446 (load_from_param_1): Make descriptors parameter garbage collected.
21447 (load_from_unmodified_param): Likewise.
21448 (load_from_param): Likewise.
21449 (ipa_load_from_parm_agg): Likewise.
21450 (ipa_node_params::~ipa_node_params): Removed.
21451 (ipa_free_all_node_params): Remove call to delete operator.
21452 (ipa_node_params_t::insert): New.
21453 (ipa_node_params_t::remove): Likewise.
21454 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21455 copy known_csts and known_contexts vectors.
21456 (ipa_read_node_info): Adjust to descriptors being a pointer.
21457 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21459 (ipcp_transform_function): Make descriptors variable garbage
21462 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21464 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21465 * config/i386/avx512dqintrin.h: Ditto.
21466 * config/i386/avx512fintrin.h: Ditto.
21467 * config/i386/i386.c: Handle new builtins.
21468 * config/i386/i386-builtin.def: Add new builtins.
21469 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21470 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21472 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21473 Martin Liska <mliska@suse.cz>
21475 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21476 * asan.c (asan_expand_poison_ifn): Support stores and use
21477 appropriate ASAN report function.
21478 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21479 * internal-fn.def (ASAN_POISON_USE): Declare.
21480 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21481 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21482 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21483 ASAN_POISON calls w/o LHS.
21484 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21485 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21486 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21487 * gimplify.c (asan_poison_variables): Add attribute
21488 use_after_scope_memory to variables that really needs to live
21490 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21491 having the attribute.
21493 2017-01-23 Martin Liska <mliska@suse.cz>
21495 * asan.c (create_asan_shadow_var): New function.
21496 (asan_expand_poison_ifn): Likewise.
21497 * asan.h (asan_expand_poison_ifn): New declaration.
21498 * internal-fn.c (expand_ASAN_POISON): Likewise.
21499 * internal-fn.def (ASAN_POISON): New builtin.
21500 * sanopt.c (pass_sanopt::execute): Expand
21501 asan_expand_poison_ifn.
21502 * tree-inline.c (copy_decl_for_dup_finish): Make function
21504 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21505 * tree-ssa.c (is_asan_mark_p): New function.
21506 (execute_update_addresses_taken): Rewrite local variables
21507 (identified just by use-after-scope as addressable) into SSA.
21509 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21511 * doc/install.texi (Specific): opensource.apple.com uses https
21512 now. Remove trailing slash.
21514 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21516 * README.Portability: Remove note on an Irix compatibility issue.
21518 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21520 * gcov.c (INCLUDE_ALGORITHM): Define.
21521 (INCLUDE_VECTOR): Define.
21522 No longer include <vector> and <algorithm> directly.
21524 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21526 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21528 * doc/invoke.texi (Code Gen Options): Ditto.
21530 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21533 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21535 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21537 rtl-optimization/79125
21538 * cprop.c (local_cprop_pass): Handle cases where we make an
21539 unconditional trap.
21541 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21545 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21546 read from, for big endian.
21548 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21550 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21551 register pauth builtins for LP64 only.
21553 2017-01-20 Marek Polacek <polacek@redhat.com>
21556 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21559 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21561 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21563 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21564 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21565 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21567 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21570 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21571 in big-endian mode when they are not a single duplicated value.
21573 2017-01-20 Richard Biener <rguenther@suse.de>
21575 * BASE-VER: Bump to 7.0.1.
21577 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21579 * omp-low.c (omplow_simd_context): New struct. Use it...
21580 (lower_rec_simd_input_clauses): ...here and...
21581 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21582 references to idx, lane, max_vf, is_simt.
21584 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21586 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21589 2017-01-20 Martin Jambor <mjambor@suse.cz>
21591 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21592 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21594 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21595 (GTFILES): Rename hsa.c to hsa-common.c.
21596 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21597 * hsa-dump.c: Likewise.
21598 * hsa-gen.c: Likewise.
21599 * hsa-regalloc.c: Likewise.
21600 * ipa-hsa.c: Likewise.
21601 * omp-expand.c: Likewise.
21602 * omp-low.c: Likewise.
21603 * toplev.c: Likewise.
21605 2017-01-20 Marek Polacek <polacek@redhat.com>
21608 * doc/invoke.texi: Document -Wduplicated-branches.
21609 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21610 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21611 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21612 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21613 return 0 only when not OEP_LEXICOGRAPHIC.
21614 (fold_build_cleanup_point_expr): Use the expression
21615 location when building CLEANUP_POINT_EXPR.
21616 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21617 * tree.c (add_expr): Handle error_mark_node.
21619 2017-01-20 Martin Liska <mliska@suse.cz>
21622 * tree-profile.c (init_ic_make_global_vars): Do not call
21624 (gimple_init_gcov_profiler): Likewise.
21626 2017-01-20 Martin Liska <mliska@suse.cz>
21629 * cgraph.h (maybe_create_reference): Remove argument and
21631 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21633 * ipa-cp.c (create_specialized_node): Likewise.
21634 * symtab.c (symtab_node::maybe_create_reference): Handle
21635 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21637 2017-01-20 Martin Liska <mliska@suse.cz>
21639 * read-rtl-function.c (function_reader::create_function): Use
21640 build_decl instread of build_decl_stat.
21642 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21644 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21645 * config/i386/avx512dqintrin.h: Ditto.
21646 * config/i386/avx512fintrin.h: Ditto.
21647 * config/i386/i386-builtin-types.def: Add new types.
21648 * config/i386/i386.c: Handle new types.
21649 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21650 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21651 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21652 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21653 (__builtin_ia32_kshiftridi): New.
21654 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21656 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21660 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21661 define to rs6000_init_stack_protect_guard.
21662 (rs6000_init_stack_protect_guard): New function.
21664 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21665 Yunqiang Su <yunqiang.su@imgtec.com>
21667 * config.gcc (supported_defaults): Add madd4.
21668 (with_madd4): Add validation.
21669 (all_defaults): Add madd4.
21670 * config/mips/mips.opt (mmadd4): New option.
21671 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21673 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21675 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21676 (ISA_HAS_FUSED_MADD4): Likewise.
21677 * doc/invoke.texi (-mmadd4): Document the new option.
21678 * doc/install.texi (--with-madd4): Document the new option.
21680 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21682 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21683 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21684 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21685 (aarch64_init_pauth_hint_builtins): New.
21686 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21687 (aarch64_expand_builtin): Expand new builtins.
21689 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21691 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21692 * combine-stack-adj.c (no_unhandled_cfa): Handle
21693 REG_CFA_TOGGLE_RA_MANGLE.
21694 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21695 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21696 info for return address signing.
21697 (aarch64_expand_epilogue): Likewise.
21699 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21701 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21702 * config/aarch64/aarch64-protos.h
21703 (aarch64_return_address_signing_enabled): New declaration.
21704 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21706 (aarch64_expand_prologue): Sign return address before it's pushed onto
21708 (aarch64_expand_epilogue): Authenticate return address fetched from
21710 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21711 (aarch64_attributes): New function attributes for "sign-return-address".
21712 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21713 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21714 ("*do_return"): Generate combined instructions according to key index.
21715 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21716 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21718 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21719 * config/aarch64/aarch64.opt (msign-return-address=): New.
21720 * doc/extend.texi (AArch64 Function Attributes): Documents
21721 "sign-return-address=".
21722 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21724 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21726 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21727 overall option summary.
21729 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21731 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21732 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21733 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21734 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21736 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21738 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21739 -mpower9-minmax by default for -mcpu=power9.
21740 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21741 128-bit floating point.
21743 2017-01-20 Alan Modra <amodra@gmail.com>
21745 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21746 optimizing for size.
21748 2017-01-20 Alan Modra <amodra@gmail.com>
21751 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21752 for strcmp and strncmp from corresponding builtin decl.
21754 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21756 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21757 instead of i386/rtems-64.h.
21758 * config/i386/rtems-64.h: Remove.
21760 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21764 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21766 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21768 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21770 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21771 Change int to HOST_WIDE_INT.
21772 * config/aarch64/aarch64-protos.h
21773 (aarch64_simd_gen_const_vector_dup): Likewise.
21774 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21776 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21778 * langhooks-def.h (lhd_type_for_size): New decl.
21779 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21780 * langhooks.c (lhd_type_for_size): New function, taken from
21783 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21785 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21786 define_bypass for CR latency.
21787 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21788 (power9-alu2): Add define_bypass for CR latency.
21790 (power9-mul): Update insn latency.
21791 (power9-mul-compare): Update insn latency, bypass latency and remove
21794 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21796 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21798 * config/aarch64/aarch64.md
21799 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21800 aarch64_nopcrelative_literal_loads.
21801 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21803 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21805 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21806 TARGET_LOONGSON_3A.
21807 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21809 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21812 * config.gcc (supported_defaults): Add lxc1-sxc1.
21813 (with_lxc1_sxc1): Add validation.
21814 (all_defaults): Add lxc1-sxc1.
21815 * config/mips/mips.opt (mlxc1-sxc1): New option.
21816 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21818 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21819 __mips_no_lxc1_sxc1.
21820 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21821 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21822 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21824 2017-01-19 Richard Biener <rguenther@suse.de>
21826 PR tree-optimization/72488
21827 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21828 sure to restore SSA info.
21829 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21831 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21833 PR rtl-optimization/79121
21834 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21835 of the inner type when shifting an extended value.
21837 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21840 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21841 interposable aliases.
21843 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21846 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21847 Use the evmergelohi instruction.
21848 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21849 (mov_sitf_e500_subreg8_2_be): Likewise.
21850 (mov_sitf_e500_subreg12_2_le): Likewise.
21851 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21852 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21853 (mov_sitf_e500_subreg8_2_le): Likewise.
21854 (mov_sitf_e500_subreg12_2_be): Likewise.
21856 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21858 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21859 attribute from vecsimple to vecperm.
21860 (altivec_vbpermq2): Likewise.
21862 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21865 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21867 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21868 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21869 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21870 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21871 case where N arg is SIZE_MAX.
21872 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21873 (cmpstrsi): Add pattern.
21875 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21877 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21878 __builtin_vec_revb builtins.
21879 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21880 built-in functions to support generation of the ISA 3.0 XXBR<x>
21881 vector byte reverse instructions.
21882 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21883 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21884 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21885 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21886 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21887 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21888 (P9V_BUILTIN_VEC_REVB): Likewise.
21889 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21890 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21891 (p9_xxbrq_v16qi): Likewise.
21892 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21893 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21894 (p9_xxbrh_v8hi): Likewise.
21895 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21896 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21897 vec_revb built-in functions.
21899 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21901 PR rtl-optimization/78952
21902 * config/i386/i386.md (any_extract): New code iterator.
21903 (*insvqi_2): Use any_extract for source operand.
21904 (*insvqi_3): Use any_shiftrt for source operand.
21906 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21908 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21910 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21912 2017-01-18 Matthias Klose <doko@ubuntu.com>
21914 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21916 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21918 * config/rs6000/altivec.h (vec_bperm): Change #define.
21919 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21920 (altivec_vbpermq2): New define_insn.
21921 (altivec_vbpermd): Likewise.
21922 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21923 function interface.
21924 (VBPERMD): Likewise.
21925 (VBPERM): New polymorphic function interface.
21926 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21927 Add entries for P9V_BUILTIN_VEC_VBPERM.
21928 * doc/extend.texi: Add interfaces for vec_bperm.
21930 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21932 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21933 first letter of error messages.
21934 (s390_resolve_overloaded_builtin): Likewise.
21935 * config/s390/s390.c (s390_expand_builtin): Likewise.
21936 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21937 (s390_valid_target_attribute_inner_p): Likewise.
21938 * config/s390/s390.md ("tabort"): Likewise.
21940 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21942 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21943 (ISA_AVOID_DIV_HILO): New macro.
21944 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21945 (ISA_HAS_DDIV): Likewise.
21947 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21949 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21951 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21953 * doc/invoke.texi (fabi-version): Spelling fix.
21955 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21958 * doc/invoke.texi (fabi-version): Mention mangling fix for
21961 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21964 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21966 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21969 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21970 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21972 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21974 (stack_protect_test): Ditto.
21975 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21976 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21978 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21979 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21980 -mstack-protector-guard-offset=.
21981 (RS/6000 and PowerPC Options): Ditto.
21983 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21985 * config/i386/i386.h (MASK_CLASS_P): New define.
21986 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21987 there are no registers from different register sets also when
21988 mask registers are used. Update function comment.
21989 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21990 to (*k/*r) and (*k/*km) alternatives.
21992 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21994 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21995 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21996 (EH_RETURN_HANDLER_RTX): New define.
21997 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21998 Force frame pointer in EH return functions.
21999 (aarch64_expand_epilogue): Add barrier for eh_return.
22000 (aarch64_final_eh_return_addr): Remove.
22001 (aarch64_eh_return_handler_rtx): New function.
22002 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22004 (aarch64_eh_return_handler_rtx): New prototype.
22006 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22008 * config/rs6000/altivec.h (vec_rlmi): New #define.
22009 (vec_vrlnm): Likewise.
22010 (vec_rlnm): Likewise.
22011 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22012 (UNSPEC_VRLNM): Likewise.
22013 (VIlong): New mode iterator.
22014 (altivec_vrl<VI_char>mi): New define_insn.
22015 (altivec_vrl<VI_char>nm): Likewise.
22016 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22018 (VRLDNM): Likewise.
22019 (RLNM): New polymorphic function entry.
22020 (VRLWMI): New monomorphic function entry.
22021 (VRLDMI): Likewise.
22022 (RLMI): New polymorphic function entry.
22023 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22024 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22025 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22028 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22031 * dwarf2out.c (field_byte_offset): Restore the
22032 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22033 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22034 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22037 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22040 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22042 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22045 * dwarf2out.c (add_data_member_location_attribute): For constant
22046 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22047 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22048 DW_AT_byte_size attributes.
22050 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22052 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22053 after forcing to constant memory when the code model is medium.
22055 2017-01-17 Julia Koval <julia.koval@intel.com>
22058 * config/i386/avx512fintrin.h
22059 (_mm512_i32gather_ps): Change __addr type to void const*.
22060 (_mm512_mask_i32gather_ps): Ditto.
22061 (_mm512_i32gather_pd): Ditto.
22062 (_mm512_mask_i32gather_pd): Ditto.
22063 (_mm512_i64gather_ps): Ditto.
22064 (_mm512_mask_i64gather_ps): Ditto.
22065 (_mm512_i64gather_pd): Ditto.
22066 (_mm512_mask_i64gather_pd): Ditto.
22067 (_mm512_i32gather_epi32): Ditto.
22068 (_mm512_mask_i32gather_epi32): Ditto.
22069 (_mm512_i32gather_epi64): Ditto.
22070 (_mm512_mask_i32gather_epi64): Ditto.
22071 (_mm512_i64gather_epi32): Ditto.
22072 (_mm512_mask_i64gather_epi32): Ditto.
22073 (_mm512_i64gather_epi64): Ditto.
22074 (_mm512_mask_i64gather_epi64): Ditto.
22075 (_mm512_i32scatter_ps): Change __addr type to void*.
22076 (_mm512_mask_i32scatter_ps): Ditto.
22077 (_mm512_i32scatter_pd): Ditto.
22078 (_mm512_mask_i32scatter_pd): Ditto.
22079 (_mm512_i64scatter_ps): Ditto.
22080 (_mm512_mask_i64scatter_ps): Ditto.
22081 (_mm512_i64scatter_pd): Ditto.
22082 (_mm512_mask_i64scatter_pd): Ditto.
22083 (_mm512_i32scatter_epi32): Ditto.
22084 (_mm512_mask_i32scatter_epi32): Ditto.
22085 (_mm512_i32scatter_epi64): Ditto.
22086 (_mm512_mask_i32scatter_epi64): Ditto.
22087 (_mm512_i64scatter_epi32): Ditto.
22088 (_mm512_mask_i64scatter_epi32): Ditto.
22089 (_mm512_i64scatter_epi64): Ditto.
22090 (_mm512_mask_i64scatter_epi64): Ditto.
22091 * config/i386/avx512pfintrin.h
22092 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22093 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22094 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22095 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22096 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22097 (_mm512_prefetch_i32scatter_ps): Ditto.
22098 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22099 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22100 (_mm512_prefetch_i64scatter_pd): Ditto.
22101 (_mm512_prefetch_i64scatter_ps): Ditto.
22102 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22103 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22104 * config/i386/avx512vlintrin.h
22105 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22106 (_mm_mmask_i32gather_ps): Ditto.
22107 (_mm256_mmask_i32gather_pd): Ditto.
22108 (_mm_mmask_i32gather_pd): Ditto.
22109 (_mm256_mmask_i64gather_ps): Ditto.
22110 (_mm_mmask_i64gather_ps): Ditto.
22111 (_mm256_mmask_i64gather_pd): Ditto.
22112 (_mm_mmask_i64gather_pd): Ditto.
22113 (_mm256_mmask_i32gather_epi32): Ditto.
22114 (_mm_mmask_i32gather_epi32): Ditto.
22115 (_mm256_mmask_i32gather_epi64): Ditto.
22116 (_mm_mmask_i32gather_epi64): Ditto.
22117 (_mm256_mmask_i64gather_epi32): Ditto.
22118 (_mm_mmask_i64gather_epi32): Ditto.
22119 (_mm256_mmask_i64gather_epi64): Ditto.
22120 (_mm_mmask_i64gather_epi64): Ditto.
22121 (_mm256_i32scatter_ps): Change __addr type to void*.
22122 (_mm256_mask_i32scatter_ps): Ditto.
22123 (_mm_i32scatter_ps): Ditto.
22124 (_mm_mask_i32scatter_ps): Ditto.
22125 (_mm256_i32scatter_pd): Ditto.
22126 (_mm256_mask_i32scatter_pd): Ditto.
22127 (_mm_i32scatter_pd): Ditto.
22128 (_mm_mask_i32scatter_pd): Ditto.
22129 (_mm256_i64scatter_ps): Ditto.
22130 (_mm256_mask_i64scatter_ps): Ditto.
22131 (_mm_i64scatter_ps): Ditto.
22132 (_mm_mask_i64scatter_ps): Ditto.
22133 (_mm256_i64scatter_pd): Ditto.
22134 (_mm256_mask_i64scatter_pd): Ditto.
22135 (_mm_i64scatter_pd): Ditto.
22136 (_mm_mask_i64scatter_pd): Ditto.
22137 (_mm256_i32scatter_epi32): Ditto.
22138 (_mm256_mask_i32scatter_epi32): Ditto.
22139 (_mm_i32scatter_epi32): Ditto.
22140 (_mm_mask_i32scatter_epi32): Ditto.
22141 (_mm256_i32scatter_epi64): Ditto.
22142 (_mm256_mask_i32scatter_epi64): Ditto.
22143 (_mm_i32scatter_epi64): Ditto.
22144 (_mm_mask_i32scatter_epi64): Ditto.
22145 (_mm256_i64scatter_epi32): Ditto.
22146 (_mm256_mask_i64scatter_epi32): Ditto.
22147 (_mm_i64scatter_epi32): Ditto.
22148 (_mm_mask_i64scatter_epi32): Ditto.
22149 (_mm256_i64scatter_epi64): Ditto.
22150 (_mm256_mask_i64scatter_epi64): Ditto.
22151 (_mm_i64scatter_epi64): Ditto.
22152 (_mm_mask_i64scatter_epi64): Ditto.
22153 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22154 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22155 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22156 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22157 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22158 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22159 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22160 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22161 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22162 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22163 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22164 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22165 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22166 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22167 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22168 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22169 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22170 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22171 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22172 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22173 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22174 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22175 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22176 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22177 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22178 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22179 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22180 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22181 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22182 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22183 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22184 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22185 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22186 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22187 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22188 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22189 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22190 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22191 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22192 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22193 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22194 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22195 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22196 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22197 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22198 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22199 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22200 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22201 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22202 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22203 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22204 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22205 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22206 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22207 definitions accordingly.
22209 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22210 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22213 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22216 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22219 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22220 hard regs for inner regno.
22222 2017-01-17 Martin Liska <mliska@suse.cz>
22225 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22226 assumption and add comment.
22228 2017-01-17 Nathan Sidwell <nathan@acm.org>
22230 * ipa-visibility.c (localize_node): New function, broken out of ...
22231 (function_and_variable_visibility): ... here. Call it.
22233 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22235 PR middle-end/77445
22236 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22237 correctly set frequency of oudgoing edge.
22238 (duplicate_thread_path): Fix profile updating.
22240 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22243 * configure.ac: Add GCC_BASE_VER.
22244 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22245 version from BASE-VER file.
22246 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22247 (gcc.o): Depend on $(BASEVER).
22248 * common.opt (dumpfullversion): New option.
22249 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22250 * doc/invoke.texi: Document -dumpfullversion.
22251 * doc/install.texi: Document --with-gcc-major-version-only.
22252 * configure: Regenerated.
22254 2017-01-17 Richard Biener <rguenther@suse.de>
22256 PR tree-optimization/71433
22257 * tree-vrp.c (register_new_assert_for): Merge same asserts
22258 on all incoming edges.
22259 (process_assert_insertions_for): Handle insertions at the
22262 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22264 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22265 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22267 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22270 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22273 2017-01-17 Alan Modra <amodra@gmail.com>
22276 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22277 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22278 symbolic stack limit when pic.
22280 2017-01-16 Martin Sebor <msebor@redhat.com>
22282 PR tree-optimization/78608
22283 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22285 2017-01-16 Jeff Law <law@redhat.com>
22288 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22289 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22290 for several include directories that may be relative to sysroot.
22291 * config/i386/x-mingw32 (gplus_includedir): Define.
22292 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22293 (native_system_includedir): Likewise.
22294 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22295 override if TARGET_SYSTEM_ROOT is defined.
22296 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22298 PR tree-optimization/79090
22299 PR tree-optimization/33562
22300 PR tree-optimization/61912
22301 PR tree-optimization/77485
22302 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22303 and computed trims into the dump file.
22305 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22307 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22309 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22312 * gimplify.c (gimplify_init_constructor): If want_value and
22313 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22317 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22318 sequence. Formatting fixes.
22319 (doloop_optimize): Formatting fixes.
22322 * gcc.c (debug_level_greater_than_spec_func): New function.
22323 (static_spec_functions): Add debug-level-gt spec function.
22324 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22326 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22327 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22328 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22329 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22330 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22331 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22333 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22335 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22336 QImode fixups to general and mask registers only.
22338 2017-01-16 Carl Love <cel@us.ibm.com>
22340 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22341 for built-in functions
22342 vector signed char vec_nabs (vector signed char)
22343 vector signed short vec_nabs (vector signed short)
22344 vector signed int vec_nabs (vector signed int)
22345 vector signed long long vec_nabs (vector signed long long)
22346 vector float vec_nabs (vector float)
22347 vector double vec_nabs (vector double)
22348 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22350 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22351 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22352 * doc/extend.texi: Update the documentation file for the new built-in
22355 2017-01-16 Martin Sebor <msebor@redhat.com>
22357 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22360 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22362 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22363 UNSPEC_VSX__XXSPLTD to require special splat handling.
22365 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22368 * system.h: Poison strndup.
22370 2017-01-16 Alan Modra <amodra@gmail.com>
22373 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22376 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22378 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22380 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22382 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22383 call recog here. Assert that INSN_CODE (insn) is non-negative.
22385 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22388 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22390 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22391 in the currently scheduled RTL fragment.
22393 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22395 PR rtl-optimization/78751
22396 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22399 2017-01-14 Jeff Law <law@redhat.com>
22401 PR tree-optimization/79090
22402 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22403 variable length stores.
22404 (compute_trims): Delete dead assignment to *trim_tail.
22405 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22408 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22410 PR rtl-optimization/78626
22411 PR rtl-optimization/78727
22412 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22413 of a block, and split such blocks after everything else is finished.
22415 2017-01-14 Alan Modra <amodra@gmail.com>
22418 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22419 target legitimate_combined_insn.
22420 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22421 (rs6000_legitimate_combined_insn): New function.
22422 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22424 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22425 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22426 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22428 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22430 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22432 2017-01-13 Jeff Law <law@redhat.com>
22434 PR tree-optimization/33562
22435 PR tree-optimization/61912
22436 PR tree-optimization/77485
22437 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22439 (delete_dead_assignment): Likewise.
22440 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22441 statement to delete_dead_call and delete_dead_assignment.
22443 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22446 * substring-locations.c (format_warning_va): Strengthen case 1 so
22447 that both endpoints of the substring must be within the format
22448 range for just the substring to be printed.
22450 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22452 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22453 * config/i386/i386.c (ix86_target_string): Add missing options
22454 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22455 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22456 flags_other and ix86_target_other to flags2_other. Display unknown
22458 (ix86_valid_target_attribute_inner_p): Add missing options and
22459 reorder options by implied ISAs, as in ix86_target_string.
22461 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22463 * hash-table.h (hash_table::too_empty_p): New function.
22464 (hash_table::expand): Use it.
22465 (hash_table::traverse): Likewise.
22466 (hash_table::empty_slot): Use sizeof (value_type) instead of
22467 sizeof (PTR) to convert bytes to elements. Shrink the table
22468 if the current size is excessive for the current number of
22471 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22473 * ira-costs.c (record_reg_classes): Break from the inner loop
22474 early once alt_fail is known to be true. Update outer loop
22475 handling accordingly.
22477 2017-01-13 Jeff Law <law@redhat.com>
22479 * tree-ssa-dse.c (decrement_count): New function.
22480 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22481 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22482 when we know the partially dead statement is a mem* function.
22484 PR tree-optimization/61912
22485 PR tree-optimization/77485
22486 * tree-ssa-dse.c: Include expr.h.
22487 (maybe_trim_constructor_store): New function.
22488 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22490 PR tree-optimization/33562
22491 PR tree-optimization/61912
22492 PR tree-optimization/77485
22493 * doc/invoke.texi: Document new dse-max-object-size param.
22494 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22495 * tree-ssa-dse.c: Include params.h.
22496 (dse_store_status): New enum.
22497 (initialize_ao_ref_for_dse): New, partially extracted from
22499 (valid_ao_ref_for_dse, normalize_ref): New.
22500 (setup_live_bytes_from_ref, compute_trims): Likewise.
22501 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22502 (maybe_trim_partially_dead_store): Likewise.
22503 (maybe_trim_complex_store): Likewise.
22504 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22505 Track what bytes live from the original store. Return tri-state
22506 for dead, partially dead or live.
22507 (dse_dom_walker): Add constructor, destructor and new private members.
22508 (delete_dead_call, delete_dead_assignment): New extracted from
22510 (dse_optimize_stmt): Make a member of dse_dom_walker.
22511 Use initialize_ao_ref_for_dse.
22513 PR tree-optimization/33562
22514 PR tree-optimization/61912
22515 PR tree-optimization/77485
22516 * sbitmap.h (bitmap_count_bits): Prototype.
22517 (bitmap_clear_range, bitmap_set_range): Likewise.
22518 * sbitmap.c (bitmap_clear_range): New function.
22519 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22521 2017-01-13 Martin Liska <mliska@suse.cz>
22524 * function.c (set_cfun): Add new argument force.
22525 * function.h (set_cfun): Likewise.
22526 * ipa-inline-transform.c (inline_call): Use the function when
22527 strict alising from is dropped for function we inline to.
22529 2017-01-13 Richard Biener <rguenther@suse.de>
22531 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22532 for dumping GIMPLE INTEGER_CSTs.
22534 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22536 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22537 to 201112L since C++17.
22539 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22542 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22543 if -fsanitize=kernel-address is present.
22545 2017-01-13 Richard Biener <rguenther@suse.de>
22547 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22548 as _Literal ( type ) number in case usual suffixes do not
22549 preserve all information.
22551 2017-01-13 Richard Biener <rguenther@suse.de>
22553 PR tree-optimization/77283
22554 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22555 and ssa-iterators.h.
22556 (is_feasible_trace): Implement a cost model based on joiner
22559 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22562 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22563 char or short to __float128/_Float128 directly.
22565 2017-01-12 Martin Sebor <msebor@redhat.com>
22567 to -Wformat-overflow.
22568 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22569 (min_bytes_remaining): Same.
22570 (get_string_length): Same.
22571 (format_string): Same.
22572 (format_directive): Same.
22574 (pass_sprintf_length::handle_gimple_call): Same.
22576 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22578 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22579 info.nowrite calls with no lhs that can't throw. Return bool
22580 whether gsi_remove has been called or not.
22581 (pass_sprintf_length::handle_gimple_call): Return bool whether
22582 try_substitute_return_value called gsi_remove. Formatting fix.
22583 (pass_sprintf_length::execute): Don't use gsi_remove if
22584 handle_gimple_call returned true.
22587 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22588 be removed due to side-effects, don't remove following barrier nor
22589 turn the successor edge into fallthru edge.
22591 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22594 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22595 element-reversing loads and stores as not swappable.
22597 2017-01-12 Nathan Sidwell <nathan@acm.org>
22598 Nicolai Stange <nicstange@gmail.com>
22600 * combine.c (try_combine): Don't ignore result of overlap checking
22601 loop. Combine overlap & asm check into single loop.
22603 2017-01-12 Richard Biener <rguenther@suse.de>
22605 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22606 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22608 2017-01-12 Richard Biener <rguenther@suse.de>
22610 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22611 and TS_TARGET_OPTION directly derive from TS_BASE.
22612 * tree-core.h (tree_optimization_option): Derive from tree_base.
22613 (tree_target_option): Likewise.
22615 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22617 * config/i386/i386.c (memory_address_length): Increase len
22618 only when rip_relative_addr_p returns false.
22620 2017-01-11 Julia Koval <julia.koval@intel.com>
22622 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22623 (OPTION_MASK_ISA_SGX_SET): New.
22624 (ix86_handle_option): Handle OPT_msgx.
22625 * config.gcc: Added sgxintrin.h.
22626 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22627 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22628 * config/i386/i386.c (ix86_target_string): Add -msgx.
22630 (ix86_option_override_internal): Handle new options.
22631 (ix86_valid_target_attribute_inner_p): Add sgx.
22632 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22633 * config/i386/i386.opt: Add msgx.
22634 * config/i386/sgxintrin.h: New file.
22635 * config/i386/x86intrin.h: Add sgxintrin.h.
22637 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22640 * fold-const.c (maybe_nonzero_address): Return 1 for function
22642 (tree_single_nonzero_warnv_p): Don't handle function local objects
22646 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22649 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22652 * opts.c: Include "spellcheck.h"
22653 (struct string_fragment): New struct.
22654 (struct edit_distance_traits<const string_fragment &>): New
22656 (get_closest_sanitizer_option): New function.
22657 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22659 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22661 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22663 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22664 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22665 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22666 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22667 for initial die_offset if dwarf_split_debug_info.
22668 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22669 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22671 (output_skeleton_debug_sections): Formatting fix. Use
22672 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22673 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22675 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22677 * config/arm/cortex-a53.md: Add bypasses for
22678 cortex_a53_r2f_cvt.
22679 (cortex_a53_r2f): Only use for transfers.
22680 (cortex_a53_f2r): Likewise.
22681 (cortex_a53_r2f_cvt): Add reservation for conversions.
22682 (cortex_a53_f2r_cvt): Likewise.
22684 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22686 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22687 to all inlined functions, change static to extern.
22689 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22692 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22694 (arm_assemble_integer): Likewise.
22696 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22698 * config.gcc: Use new awk script to check CPU, FPU and architecture
22699 parameters for --with-... options.
22700 * config/arm/parsecpu.awk: New file
22701 * config/arm/arm-cpus.in: New file.
22702 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22704 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22706 * config/arm/t-arm: Update dependency rules.
22707 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22708 of processing .def files.
22709 * config/arm/genopt.sh: Deleted.
22710 * config/arm/gentune.sh: Deleted.
22711 * config/arm/arm-cores.def: Deleted.
22712 * config/arm/arm-arches.def: Deleted.
22713 * config/arm/arm-fpus.def: Deleted.
22714 * config/arm/arm-tune.md: Regenerated.
22715 * config/arm/arm-tables.opt: Regenerated.
22716 * config/arm/arm-cpu.h: New generated file.
22717 * config/arm/arm-cpu-data.h: New generated file.
22718 * config/arm/arm-cpu-cdata.h: New generated file.
22720 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22723 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22725 (input_varpool_node): Unpack dynamically_initialized bit.
22727 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22729 PR rtl-optimization/79032
22730 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22731 the alignment of the adjusted memory reference against that of MODE,
22732 instead of the alignment of the original memory reference.
22734 2017-01-11 Martin Jambor <mjambor@suse.cz>
22736 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22738 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22739 decorated functions.
22741 2017-01-11 Richard Biener <rguenther@suse.de>
22743 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22744 set range/nonnull info for PHI results. Do not set it on
22745 stmts marked for removal.
22747 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22749 * expr.c (store_field): In the bitfield case, fetch the return value
22750 from the registers before applying a single big-endian adjustment.
22751 Always do a final load for a BLKmode value not larger than a word.
22753 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22756 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22757 that we correctly handle column numbers greater than
22758 LINE_MAP_MAX_COLUMN_NUMBER.
22760 2017-01-10 Martin Sebor <msebor@redhat.com>
22762 PR middle-end/78245
22763 * gimple-ssa-sprintf.c (get_destination_size): Call
22764 {init,fini}object_sizes.
22765 * tree-object-size.c (addr_object_size): Adjust.
22766 (pass_through_call): Adjust.
22767 (pass_object_sizes::execute): Adjust.
22768 * tree-object-size.h (fini_object_sizes): Declare.
22770 2017-01-10 Martin Sebor <msebor@redhat.com>
22772 PR tree-optimization/78775
22773 * builtins.c (get_size_range): Move...
22774 * calls.c: ...to here.
22775 (alloc_max_size): Accept zero argument.
22776 (operand_signed_p): Remove.
22777 (maybe_warn_alloc_args_overflow): Call get_size_range.
22778 * calls.h (get_size_range): Declare.
22780 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22782 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22783 from TI's devices.csv file as of September 2016.
22784 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22786 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22788 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22789 * doc/invoke.texi: Likewise.
22790 * doc/md.texi: Likewise.
22791 * doc/objc.texi: Likewise.
22793 2017-01-10 Joshua Conner <joshconner@google.com>
22795 * config/arm/fuchsia-elf.h: New file.
22796 * config/fuchsia.h: New file.
22797 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22798 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22800 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22802 2016-01-10 Richard Biener <rguenther@suse.de>
22804 PR tree-optimization/79034
22805 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22806 Propagate out degenerate PHIs in the joiner.
22808 2017-01-10 Martin Liska <mliska@suse.cz>
22810 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22811 (sort_congruence_classes_by_decl_uid): Likewise.
22812 (sort_congruence_class_groups_by_decl_uid): Likewise.
22813 (sem_item_optimizer::merge_classes): Sort class, groups in these
22814 classes and members in the groups by DECL_UID of declarations.
22815 This would make merge operations stable.
22817 2017-01-10 Martin Liska <mliska@suse.cz>
22819 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22820 usage of m_classes_vec.
22821 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22822 (sem_item_optimizer::get_group_by_hash): Likewise.
22823 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22824 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22825 (sem_item_optimizer::verify_classes): Likewise.
22826 (sem_item_optimizer::process_cong_reduction): Likewise.
22827 (sem_item_optimizer::dump_cong_classes): Likewise.
22828 (sem_item_optimizer::merge_classes): Likewise.
22829 * ipa-icf.h (congruence_class_hash): Rename from
22830 congruence_class_group_hash. Remove declaration of m_classes_vec.
22832 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22834 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22835 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22836 * config.gcc: Add avx512vpopcntdqintrin.h.
22837 * config/i386/avx512vpopcntdqintrin.h: New.
22838 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22839 * config/i386/i386-builtin-types.def: Add new types.
22840 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22841 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22842 __builtin_ia32_vpopcountq_v8di_mask): New.
22843 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22844 __AVX512VPOPCNTDQ__.
22845 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22846 (PTA_AVX512VPOPCNTDQ): Define.
22847 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22848 TARGET_AVX512VPOPCNTDQ_P): Define.
22849 * config/i386/i386.opt: Add mavx512vpopcntdq.
22850 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22851 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22853 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22855 PR middle-end/77484
22856 * predict.def (PRED_CALL): Set to 67.
22858 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22860 * expr.c (store_field): In the bitfield case, if the value comes from
22861 a function call and is of an aggregate type returned in registers, do
22862 not modify the field mode; extract the value in all cases if the mode
22863 is BLKmode and the size is not larger than a word.
22865 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22868 * config/i386/cpuid.h: Fix undefined behavior.
22870 2017-01-04 Jeff Law <law@redhat.com>
22872 PR tree-optimization/79007
22873 PR tree-optimization/67955
22874 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22875 conservative for pt.null when flag_non_call_exceptions is on.
22877 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22879 PR translation/79019
22880 PR translation/79020
22881 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22882 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22883 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22885 * config/avr/avr.opt (maccumulate-args): Likewise.
22886 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22887 * common.opt (freport-bug): Likewise.
22888 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22889 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22890 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22891 translatable string.
22892 * config/i386/i386.c (function_value_32): Likewise.
22893 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22894 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22896 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22897 * common/config/msp430/msp430-common.c (msp430_handle_option):
22899 * symtab.c (symtab_node::verify_base): Likewise.
22900 * opts.c (set_debug_level): Likewise.
22901 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22902 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22903 missing whitespace to translatable strings.
22904 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22905 * config/sh/superh.h: Likewise.
22906 * config/i386/xopintrin.h: Likewise.
22907 * config/i386/znver1.md: Likewise.
22908 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22909 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22910 * double-int.h (struct double_int): Likewise.
22911 * double-int.c (div_and_round_double): Likewise.
22912 * wide-int.cc: Likewise.
22913 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22914 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22915 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22916 (crossjumps_occurred): ... this.
22917 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22920 PR tree-optimization/78899
22921 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22922 returning bool return struct loop *, NULL for failure and the new
22924 (versionable_outer_loop_p): Don't version outer loop if it has
22925 dont_vectorized bit set.
22926 (tree_if_conversion): When versioning outer loop, ensure
22927 tree_if_conversion is performed also on the inner loop of the
22928 non-vectorizable outer loop copy.
22929 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22930 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22931 prevent vectorization of it.
22932 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22933 the outer loop vectorization of the non-scalar version is attempted
22934 before vectorization of the inner loop in scalar version. If
22935 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22936 vectorization of its inner loop.
22937 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22938 has 2 inner loops, rename also on edges from bb whose single pred
22939 is outer_loop->header. Fix typo in function comment.
22941 2017-01-09 Martin Sebor <msebor@redhat.com>
22944 * asan.c (asan_emit_stack_protection): Increase local buffer size
22945 to avoid snprintf truncation warning.
22947 2017-01-09 Andrew Pinski <apinski@cavium.com>
22949 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22950 to reference thunderx2t99 for the tuning structure
22951 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22953 (thunderx2t99_extra_costs): This.
22954 * config/aarch64/aarch64-tune.md: Regenerate.
22955 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22956 (vulcan_addrcost_table): This.
22957 (vulcan_regmove_cost): Rename to ...
22958 (thunderx2t99_regmove_cost): This.
22959 (vulcan_vector_cost): Rename to ...
22960 (thunderx2t99_vector_cost): this.
22961 (vulcan_branch_cost): Rename to ...
22962 (thunderx2t99_branch_cost): This.
22963 (vulcan_tunings): Rename to ...
22964 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22965 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22967 2017-01-09 Martin Jambor <mjambor@suse.cz>
22971 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22972 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22973 (propagate_vr_accross_jump_function): Use the above function for all
22974 value range computations for pass-through jump functions and type
22975 converasion from explicit value range values.
22976 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22977 parameters from TYPE_ARG_TYPES.
22978 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22979 (ipa_write_node_info): Stream type of the actual argument.
22980 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22982 2017-01-09 Martin Liska <mliska@suse.cz>
22985 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22986 (lookup_compiler): Do not show error message with have_E.
22988 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22990 PR tree-optimization/78938
22991 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22992 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22993 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22994 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22997 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22999 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23002 2017-01-09 Richard Biener <rguenther@suse.de>
23004 PR tree-optimization/78997
23005 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23006 name condition properly.
23008 2017-01-09 Richard Biener <rguenther@suse.de>
23011 * dwarf2out.c (is_cxx): New overload with context.
23012 (is_naming_typedef_decl): Use it.
23014 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23016 * invoke.texi (Option Summary): Correct spacing in option lists
23017 and add line breaks to fix over-long lines.
23019 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23021 PR middle-end/17660
23023 * extend.texi (Common Variable Attributes): Add xref to GCC
23024 Internals manual to explain mode attribute keywords.
23026 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23029 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23030 and Preprocessor Options.
23031 (Options for Linking): Document -pthread here....
23032 (RS/6000 and PowerPC Options): ...not here.
23033 (Solaris 2 Options): ...or here.
23034 * doc/cppopts.texi: Document -pthread.
23036 2017-01-08 Martin Sebor <msebor@redhat.com>
23038 PR middle-end/77708
23039 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23040 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23041 New member functions.
23042 (format_directive): Used them.
23044 (pass_sprintf_length::handle_gimple_call): Same.
23045 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23046 to avoid truncation for any argument.
23047 (extract_affine_mul): Same.
23048 * tree.c (get_file_function_name): Same.
23050 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23052 PR middle-end/77484
23053 * predict.def (PRED_INDIR_CALL): Set to 86.
23055 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23057 PR preprocessor/54124
23058 * doc/cppopts.texi: Reformat -d subtable to list the full name
23059 of the options. Add cross-reference to the docs for the general
23060 compiler -d options.
23061 * doc/invoke.texi (Developer Options): Add cross-reference to the
23062 preprocessor-specific -d option documentation.
23064 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23066 PR preprocessor/13498
23067 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23068 redudant material, and reflect new command-line options.
23069 (System Headers): Likewise.
23071 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23073 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23074 -isystem, and -idirafter. Copy-edit.
23075 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23076 default for -ftrack-macro-expansion. Delete obsolete and
23077 badly-formatted implementation details about -fdebug-cpp output.
23078 * doc/cppwarnopts.texi: Copy-edit.
23080 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23083 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23084 that the transition from a max line width >= 1<<10 to narrower
23085 lines works correctly.
23087 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23089 * doc/options.texi (PerFunction): New.
23090 * opt-functions.awk (switch_flags): Map both Optimization and
23091 PerFunction to CL_OPTIMIZATION.
23092 * opth-gen.awk: Test for PerFunction flag along with
23094 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23095 it only when the latter is present. Skip those that don't in
23096 the hash function generator.
23097 * common.opt (fvar-tracking): Mark as PerFunction instead of
23099 (fvar-tracking-assignments): Likewise.
23100 (fvar-tracking-assignments-toggle): Likewise.
23101 (fvar-tracking-uninit): Likewise.
23103 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23105 PR translation/79018
23106 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23109 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23112 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23113 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23114 TARGET_LONG_JUMP_TABLE_OFFSETS.
23115 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23116 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23117 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23118 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23119 * config/m68k/m68k.md (tablejump expander): Likewise.
23120 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23121 TARGET_LONG_JUMP_TABLE_OFFSETS.
23122 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23123 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23125 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23126 David Holsgrove <david.holsgrove@xilinx.com>
23128 * common/config/microblaze/microblaze-common.c
23129 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23130 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23132 * config/microblaze/microblaze.c (microblaze_must_save_register)
23133 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23135 (microblaze_eh_return): New function.
23136 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23137 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23138 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23139 * config/microblaze/microblaze.md (eh_return): New pattern.
23141 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23143 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23144 GCC_DIAGNOSTIC_STRINGIFY): Define.
23146 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23148 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23150 * config/arm/arm.md (<mcrr>): New.
23152 * config/arm/arm.c (arm_arch5te): New.
23153 (arm_option_override): Set arm_arch5te.
23154 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23156 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23157 (arm_mcrr_qualifiers): ... this. New.
23158 (MRRC_QUALIFIERS): Define to...
23159 (arm_mrrc_qualifiers): ... this. New.
23160 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23162 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23163 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23164 (MRRCI, mrrc, MRRC): New.
23165 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23166 VUNSPEC_MRRC2): New.
23168 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23170 * config/arm/arm.md (<mcr>): New.
23172 * config/arm/arm.c (arm_coproc_builtin_available): Add
23173 support for mcr, mrc, mcr2 and mrc2.
23174 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23175 (arm_mcr_qualifiers): ... this. New.
23176 (MRC_QUALIFIERS): Define to ...
23177 (arm_mrc_qualifiers): ... this. New.
23178 (MCR_QUALIFIERS): Define to ...
23179 (arm_mcr_qualifiers): ... this. New.
23180 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23182 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23183 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23184 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23185 VUNSPEC_MRC2): New.
23187 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23189 * config/arm/arm.md (*ldc): New.
23193 * config/arm/arm.c (arm_coproc_builtin_available): Add
23194 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23195 (arm_coproc_ldc_stc_legitimate_address): New.
23196 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23197 'qualifier_const_pointer'.
23198 (LDC_QUALIFIERS): Define to...
23199 (arm_ldc_qualifiers): ... this. New.
23200 (STC_QUALIFIERS): Define to...
23201 (arm_stc_qualifiers): ... this. New.
23202 * config/arm/arm-protos.h
23203 (arm_coproc_ldc_stc_legitimate_address): New.
23204 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23205 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23206 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23207 stc2, stcl, stc2l): New.
23208 * config/arm/constraints.md (Uz): New.
23209 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23210 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23211 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23212 VUNSPEC_STC2L): New.
23214 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23216 * config/arm/arm.md (<cdp>): New.
23217 * config/arm/arm.c (neon_const_bounds): Rename this ...
23218 (arm_const_bounds): ... this.
23219 (arm_coproc_builtin_available): New.
23220 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23221 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23222 (CDP_QUALIFIERS): Define to...
23223 (arm_cdp_qualifiers): ... this. New.
23225 (arm_expand_builtin_args): Add case for 6 arguments.
23226 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23227 (arm_const_bounds): ... this.
23228 (arm_coproc_builtin_available): New.
23229 * config/arm/arm_acle.h (__arm_cdp): New.
23231 * config/arm/arm_acle_builtins.def (cdp): New.
23233 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23234 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23235 'arm_const_bounds'.
23236 * config/arm/types.md (coproc): New.
23237 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23238 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23239 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23240 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23242 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23244 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23245 (UBINOP_QUALIFIERS): New.
23247 (acle_builtin_data): New. Change comment.
23248 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23249 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23250 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23251 arm_acle_builtins.def.
23252 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23253 (arm_init_acle_builtins): New.
23254 (CRC32_BUILTIN): Remove.
23255 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23256 crc32cb, crc32ch and crc32cw.
23257 (arm_init_crc32_builtins): Remove.
23258 (arm_init_builtins): Use arm_init_acle_builtins rather
23259 than arm_init_crc32_builtins.
23260 (arm_expand_acle_builtin): New.
23261 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23262 * config/arm/arm_acle_builtins.def: New.
23264 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23266 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23267 (arm_builtin_datum): ... this.
23268 (arm_init_neon_builtin): Rename to ...
23269 (arm_init_builtin): ... this. Add a new parameters PREFIX
23270 and USE_SIG_IN_NAME.
23271 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23272 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23273 'arm_builtin_datum'.
23274 (arm_init_vfp_builtins): Likewise.
23275 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23276 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23277 (arm_expand_neon_args): Rename to ...
23278 (arm_expand_builtin_args): ... this. Rename builtin_arg
23279 enum values and differentiate between ARG_BUILTIN_MEMORY
23280 and ARG_BUILTIN_NEON_MEMORY.
23281 (arm_expand_neon_builtin_1): Rename to ...
23282 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23283 values, arm_expand_builtin_args and add bool parameter NEON.
23284 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23285 (arm_expand_vfp_builtin): Likewise.
23286 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23288 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23290 PR middle-end/77484
23291 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23292 * predict.c (tree_estimate_probability_bb): Reverse direction of
23293 polymorphic call predictor.
23295 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23297 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23298 (determine_pass_name_match): ...this new function and...
23299 (should_skip_pass_p): ...this new function.
23301 2017-01-06 Nathan Sidwell <nathan@acm.org>
23303 * ipa-visibility.c (function_and_variable_visibility): Reformat
23304 comments and long lines. Remove extrneous if.
23305 * symtab.c (symtab_node::make_decl_local): Fix code format.
23306 (symtab_node::set_section_for_node): Fix comment typo.
23308 2017-01-06 Martin Liska <mliska@suse.cz>
23311 * lra-constraints.c: Rename invariant to lra_invariant.
23312 * predict.c (set_even_probabilities): Initialize e to NULL.
23314 2017-01-05 Martin Sebor <msebor@redhat.com>
23316 PR tree-optimization/78910
23317 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23318 (format_integer): Correct off-by-one error in the handling
23319 of precision with negative numbers in signed conversions..
23321 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23323 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23325 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23327 PR tree-optimization/71016
23328 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23329 factor_out_conditional_conversion. Formatting fix.
23330 (factor_out_conditional_conversion): Add cond_stmt argument.
23331 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23332 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23335 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23337 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23338 read-rtl-function.o, and selftest-rtl.o.
23339 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23340 (selftest::aarch64_test_loading_full_dump): New function.
23341 (selftest::aarch64_run_selftests): New function.
23342 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23343 selftest::aarch64_run_selftests.
23344 * config/i386/i386.c
23345 (selftest::ix86_test_loading_dump_fragment_1): New function.
23346 (selftest::ix86_test_loading_call_insn): New function.
23347 (selftest::ix86_test_loading_full_dump): New function.
23348 (selftest::ix86_test_loading_unspec): New function.
23349 (selftest::ix86_run_selftests): Call the new functions.
23350 * emit-rtl.c (maybe_set_max_label_num): New function.
23351 * emit-rtl.h (maybe_set_max_label_num): New decl.
23352 * function.c (instantiate_decls): Guard call to
23353 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23354 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23356 * gensupport.c (gen_reader::gen_reader): Pass "false"
23357 for new "compact" param of rtx_reader.
23358 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23359 rather than an empty string for NULL strings.
23360 * read-md.c: Potentially include config.h rather than bconfig.h.
23361 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23362 (have_error): New global, copied from errors.c.
23363 (md_reader::read_name): Rename to...
23364 (md_reader::read_name_1): ...this, adding "out_loc" param,
23365 and converting "missing name or number" to returning false, rather
23367 (md_reader::read_name): Reimplement in terms of read_name_1.
23368 (md_reader::read_name_or_nil): New function.
23369 (md_reader::read_string): Handle "(nil)" by returning NULL.
23370 (md_reader::md_reader): Add new param "compact".
23371 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23372 (md_reader::read_file): New method.
23373 * read-md.h (md_reader::md_reader): Add new param "compact".
23374 (md_reader::read_file): New method.
23375 (md_reader::is_compact): New accessor.
23376 (md_reader::read_name): Convert return type from void to file_location.
23377 (md_reader::read_name_or_nil): New decl.
23378 (md_reader::read_name_1): New decl.
23379 (md_reader::m_compact): New field.
23380 (noop_reader::noop_reader): Pass "false" for new "compact" param
23382 (rtx_reader::rtx_reader): Add new "compact" param.
23383 (rtx_reader::read_rtx_operand): Make virtual and convert return
23384 type from void to rtx.
23385 (rtx_reader::read_until): New decl.
23386 (rtx_reader::handle_any_trailing_information): New virtual function.
23387 (rtx_reader::postprocess): New virtual function.
23388 (rtx_reader::finalize_string): New virtual function.
23389 (rtx_reader::m_in_call_function_usage): New field.
23390 (rtx_reader::m_reuse_rtx_by_id): New field.
23391 * read-rtl-function.c: New file.
23392 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23393 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23394 (selftest::verify_three_block_rtl_cfg): New decl.
23395 * read-rtl-function.h: New file.
23396 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23397 For host, include function.h, memmodel.h, and emit-rtl.h.
23398 (one_time_initialization): New function.
23399 (struct compact_insn_name): New struct.
23400 (compact_insn_names): New array.
23401 (find_code): Handle insn codes in compact dumps.
23402 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23403 (bind_subst_iter_and_attr): Likewise.
23404 (add_condition_to_string): Likewise.
23405 (add_condition_to_rtx): Likewise.
23406 (apply_attribute_uses): Likewise.
23407 (add_current_iterators): Likewise.
23408 (apply_iterators): Likewise.
23409 (initialize_iterators): Guard usage of apply_subst_iterator with
23410 #ifdef GENERATOR_FILE.
23411 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23412 (md_reader::read_mapping): Likewise.
23413 (add_define_attr_for_define_subst): Likewise.
23414 (add_define_subst_attr): Likewise.
23415 (read_subst_mapping): Likewise.
23416 (check_code_iterator): Likewise.
23417 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23419 (one_time_initialization): New function.
23420 (rtx_reader::read_until): New method.
23421 (read_flags): New function.
23422 (parse_reg_note_name): New function.
23423 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23424 Handle reuse_rtx ids.
23425 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23426 Add parsing support for RTL dumps, mirroring the special-cases in
23427 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23428 values, and calling handle_any_trailing_information.
23429 (rtx_reader::read_rtx_operand): Convert return type from void
23430 to rtx, returning return_rtx. Handle case 'e'. Call
23431 finalize_string on XSTR and XTMPL fields.
23432 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23433 "(nil)" values were omitted. Call the postprocess vfunc on the
23435 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23436 class ctor. Initialize m_in_call_function_usage. Call
23437 one_time_initialization.
23438 * rtl-tests.c (selftest::test_uncond_jump): Call
23439 set_new_first_and_last_insn.
23440 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23441 * selftest-rtl.c: New file.
23442 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23443 (selftest::get_insn_by_uid): New decl.
23444 * selftest-run-tests.c (selftest::run_tests): Call
23445 read_rtl_function_c_tests.
23446 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23447 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23450 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23452 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23453 operands in a special way. Assert that pos+len <= mode precision.
23455 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23457 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23458 3 argument Alias with unlimited for the negative form.
23459 (fno-vect-cost-model): Removed.
23461 2017-01-05 Martin Liska <mliska@suse.cz>
23463 * hsa-gen.c (gen_hsa_divmod): New function.
23464 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23466 2017-01-05 Martin Liska <mliska@suse.cz>
23469 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23472 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23474 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23475 small constant length operands.
23477 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23479 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23480 between loop iterations.
23482 2017-01-05 Martin Liska <mliska@suse.cz>
23485 * gimplify.c (gimplify_decl_expr): Compare to
23486 asan_poisoned_variables instread of checking flags.
23487 (gimplify_target_expr): Likewise.
23488 (gimplify_expr): Likewise.
23489 (gimplify_function_tree): Conditionally initialize
23490 asan_poisoned_variables.
23492 2017-01-04 Jeff Law <law@redhat.com>
23494 PR tree-optimizatin/78812
23495 * rtl.h (contains_mem_rtx_p): Prototype.
23496 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23497 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23498 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23499 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23500 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23502 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23504 * input.c (assert_char_at_range): Default-initialize actual_range.
23506 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23508 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23509 to match the caller.
23511 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23513 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23514 insns after final jump in test to emit dummy move.
23516 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23518 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23519 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23521 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23523 * multiple_target.c (create_dispatcher_calls): Init e_next.
23524 * tree-ssa-loop-split.c (split_loop): Init border.
23525 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23528 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23533 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23534 (altivec_register_operand): Do not return true if the operand
23535 contains a SUBREG mixing SImode and SFmode.
23536 (vsx_register_operand): Likewise.
23537 (vsx_reg_sfsubreg_ok): New predicate.
23538 (vfloat_operand): Do not return true if the operand contains a
23539 SUBREG mixing SImode and SFmode.
23540 (vint_operand): Likewise.
23541 (vlogical_operand): Likewise.
23542 (gpc_reg_operand): Likewise.
23543 (int_reg_operand): Likewise.
23544 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23545 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23546 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23548 (rs6000_emit_move_si_sf_subreg): New helper function.
23549 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23550 fixup SUBREGs involving SImode and SFmode.
23551 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23552 numbers for the new peephole2 optimization.
23553 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23554 the GLIBC math library that do AND/IOR/XOR operations on single
23555 precision floating point.
23556 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23557 target macros to say whether we need to avoid SUBREGs mixing
23559 (TARGET_ALLOW_SF_SUBREG): Likewise.
23560 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23561 (UNSPEC_SI_FROM_SF): Likewise.
23562 (iorxor): Change spacing.
23563 (and_ior_xor): New iterator for AND, IOR, and XOR.
23564 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23565 (movdi_from_sf_zero_ext): Likewise.
23566 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23567 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23568 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23569 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23570 (fms<mode>4): Likewise.
23571 (fnma<mode>4): Likewise.
23572 (fnms<mode>4): Likewise.
23573 (nfma<mode>4): Likewise.
23574 (nfms<mode>4): Likewise.
23576 2017-01-04 Marek Polacek <polacek@redhat.com>
23579 * doc/invoke.texi: Document -Wpointer-compare.
23581 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23583 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23586 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23587 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23590 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23593 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23594 documentation of the powerpc_popcntb_ok attribute.
23595 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23596 code to issue warning messages if a requested CPU configuration is
23597 not supported by the binary (assembler and loader) toolchain.
23598 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23599 made to define a built-in function that has been disabled.
23600 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23601 made to define a built-in function that has been disabled.
23602 (altivec_init_builtins): Add comment explaining why definition
23603 of the DST built-in functions is not preceded by an assertion
23604 check. Add assertions to prevent ICE if attempts are made to
23605 define an altivec predicate or an abs* built-in function that has
23607 (htm_init_builtins): Add comment explaining why definition of the
23608 htm built-in functions is not preceded by an assertion check.
23610 2017-01-04 Jeff Law <law@redhat.com>
23612 PR tree-optimizatin/67955
23613 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23614 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23615 the points-to solution does not include pt_null. Use DECL_PT_UID
23618 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23620 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23621 Use gen_int_mode instead of gen_lopwart for const_int operands.
23623 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23625 PR tree-optimization/71563
23626 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23627 out of range value - has low bits known to be zero.
23629 2017-01-04 Alan Modra <amodra@gmail.com>
23631 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23632 * configure: Regenerate.
23633 * config.in: Regenerate.
23635 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23638 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23639 a substring of the message, but strcmp with the whole message. Ifdef
23640 ENABLE_NLS, translate the message first using dgettext.
23642 2017-01-03 Jeff Law <law@redhat.com>
23644 PR tree-optimizatin/78856
23645 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23646 (mark_threaded_blocks): Remove code to truncate thread paths that
23647 cross multiple loop headers. Instead invalidate the cached loop
23648 iteration information and handle case of a thread path walking
23649 into an irreducible region.
23651 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23654 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23655 assertions. Add support for doing the signbit if the IEEE 128-bit
23656 floating point value is in a GPR.
23657 * config/rs6000/rs6000.md (Fsignbit): Delete.
23658 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23659 Update the length attribute if the value is in a GPR.
23660 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23661 the sign or zero extension instruction, since the value is always 0/1.
23662 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23665 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23666 extracting SImode to a GPR register so that we can generate a
23667 store, limit the vector to be in a traditional Altivec register
23668 for the vextuwrx instruction.
23670 2017-01-03 Ian Lance Taylor <iant@google.com>
23672 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23674 2017-01-03 Martin Sebor <msebor@redhat.com>
23676 PR tree-optimization/78696
23677 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23678 precision. Use MPFR for %f for greater fidelity. Correct handling
23680 (pass_sprintf_length::compute_format_length): Set width and precision
23681 specified by asrerisk to void_node for vararg functions.
23682 (try_substitute_return_value): Adjust dump output.
23684 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23686 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23688 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23690 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23691 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23692 -mlra/-mno-lra was passed to the compiler.
23694 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23696 PR rtl-optimization/65618
23697 * emit-rtl.c (try_split): Move initialization of "before" and
23698 "after" to just before the call to emit_insn_after_setloc.
23700 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23702 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23704 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23706 * dwarf2out.c (gen_enumeration_type_die): When
23707 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23709 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23711 PR tree-optimization/78965
23712 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23713 Change first argument from const call_info & to call_info &. For %n
23714 set info.nowrite to false.
23716 PR middle-end/78901
23717 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23718 possibly throwing calls.
23720 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23721 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23722 and fns handling, rather than in a separate case SSA_NAME.
23724 2017-01-02 Jeff Law <law@redhat.com>
23726 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23727 fixes for first_period and second_period variables.
23729 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23732 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23733 (*insvqi_1): New insn pattern.
23734 (*insvqi_1_mem_rex64): Ditto.
23735 (*insvqi_2): Ditto.
23736 (*insvqi_3): Rename from *insvqi.
23738 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23740 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23742 * doc/cfg.texi (Edges): Remove reference to Java.
23743 (Maintaining the CFG): Ditto.
23745 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23747 PR middle-end/77674
23748 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23749 transparent aliases.
23751 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23753 PR middle-end/77484
23754 * predict.def (PRED_CALL): Update hitrate.
23755 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23756 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23757 into direct/indirect/polymorphic variants.
23759 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23761 Update copyright years.
23763 * gcc.c (process_command): Update copyright notice dates.
23764 * gcov-dump.c (print_version): Ditto.
23765 * gcov.c (print_version): Ditto.
23766 * gcov-tool.c (print_version): Ditto.
23767 * gengtype.c (create_file): Ditto.
23768 * doc/cpp.texi: Bump @copying's copyright year.
23769 * doc/cppinternals.texi: Ditto.
23770 * doc/gcc.texi: Ditto.
23771 * doc/gccint.texi: Ditto.
23772 * doc/gcov.texi: Ditto.
23773 * doc/install.texi: Ditto.
23774 * doc/invoke.texi: Ditto.
23776 Copyright (C) 2017 Free Software Foundation, Inc.
23778 Copying and distribution of this file, with or without modification,
23779 are permitted in any medium without royalty provided the copyright
23780 notice and this notice are preserved.