1 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4 rs6000_emit_move_from_cr and call renamed function.
5 (rs6000_emit_prologue): Call renamed functions.
6 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7 movesi_from_cr, remove volatile CRs.
9 2017-08-30 Jon Beniston <jon@beniston.com>
10 Richard Biener <rguenther@suse.de>
12 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13 of VECTOR_MODE_P check.
14 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19 * df.h (df_read_modify_subreg_p): Remove in favor of...
20 * rtl.h (read_modify_subreg_p): ...this new function. Take a
21 const_rtx instead of an rtx.
22 * cprop.c (local_cprop_find_used_regs): Update accordingly.
23 * df-problems.c (df_word_lr_mark_ref): Likewise.
24 * ira-lives.c (mark_pseudo_reg_live): Likewise.
25 (mark_pseudo_reg_dead): Likewise.
26 (mark_ref_dead): Likewise.
27 * reginfo.c (init_subregs_of_mode): Likewise.
28 * sched-deps.c (sched_analyze_1): Likewise.
29 * df-scan.c (df_def_record_1): Likewise.
30 (df_uses_record): Likewise.
31 (df_read_modify_subreg_p): Remove in favor of...
32 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
33 const_rtx instead of an rtx.
35 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
36 Alan Hayward <alan.hayward@arm.com>
37 David Sherwood <david.sherwood@arm.com>
39 * rtl.h (partial_subreg_p): New function.
40 * caller-save.c (save_call_clobbered_regs): Use it.
41 * calls.c (expand_call): Likewise.
42 * combine.c (combinable_i3pat): Likewise.
43 (simplify_set): Likewise.
44 (make_extraction): Likewise.
45 (make_compound_operation_int): Likewise.
46 (gen_lowpart_or_truncate): Likewise.
47 (force_to_mode): Likewise.
48 (make_field_assignment): Likewise.
49 (reg_truncated_to_mode): Likewise.
50 (record_truncated_value): Likewise.
51 (move_deaths): Likewise.
52 * cse.c (record_jump_cond): Likewise.
54 * cselib.c (cselib_lookup_1): Likewise.
55 * expmed.c (extract_bit_field_using_extv): Likewise.
56 * function.c (assign_parm_setup_reg): Likewise.
57 * ifcvt.c (noce_convert_multiple_sets): Likewise.
58 * ira-build.c (create_insn_allocnos): Likewise.
59 * lra-coalesce.c (merge_pseudos): Likewise.
60 * lra-constraints.c (match_reload): Likewise.
61 (simplify_operand_subreg): Likewise.
62 (curr_insn_transform): Likewise.
63 * lra-lives.c (process_bb_lives): Likewise.
64 * lra.c (new_insn_reg): Likewise.
65 (lra_substitute_pseudo): Likewise.
66 * regcprop.c (mode_change_ok): Likewise.
67 (maybe_mode_change): Likewise.
68 (copyprop_hardreg_forward_1): Likewise.
69 * reload.c (push_reload): Likewise.
70 (find_reloads): Likewise.
71 (find_reloads_subreg_address): Likewise.
72 * reload1.c (alter_reg): Likewise.
73 (eliminate_regs_1): Likewise.
74 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
76 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
78 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
79 back to if statements, including unpack.
81 2017-08-30 Martin Liska <mliska@suse.cz>
84 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
86 (func_checker::compare_asm_inputs_outputs): ... this function.
87 (func_checker::compare_gimple_asm): Use the function to compare
89 * ipa-icf-gimple.h: Rename the function.
91 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
92 Alan Hayward <alan.hayward@arm.com>
93 David Sherwood <david.sherwood@arm.com>
95 * coretypes.h (complex_mode): New type.
96 * gdbhooks.py (build_pretty_printer): Handle it.
97 * machmode.h (complex_mode): New class.
98 (complex_mode::includes_p): New function.
99 (is_complex_int_mode): Likewise.
100 (is_complex_float_mode): Likewise.
101 * genmodes.c (get_mode_class): Handle complex mode classes.
102 * function.c (expand_function_end): Use is_complex_int_mode.
104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * coretypes.h (scalar_mode_pod): New typedef.
109 * gdbhooks.py (build_pretty_printer): Handle it.
110 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
111 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
112 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
113 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
117 Alan Hayward <alan.hayward@arm.com>
118 David Sherwood <david.sherwood@arm.com>
120 * machmode.h (mode_for_vector): Take a scalar_mode instead
122 * stor-layout.c (mode_for_vector): Likewise.
123 * explow.c (promote_mode): Use as_a <scalar_mode>.
124 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * target.def (preferred_simd_mode): Take a scalar_mode
131 instead of a machine_mode.
132 * targhooks.h (default_preferred_simd_mode): Likewise.
133 * targhooks.c (default_preferred_simd_mode): Likewise.
134 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
135 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
136 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
137 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
138 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
139 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
140 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
141 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
143 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
144 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
145 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
146 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
147 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
148 * doc/tm.texi: Regenerate.
149 * optabs-query.c (can_vec_mask_load_store_p): Return false for
152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
153 Alan Hayward <alan.hayward@arm.com>
154 David Sherwood <david.sherwood@arm.com>
156 * target.def (scalar_mode_supported_p): Take a scalar_mode
157 instead of a machine_mode.
158 * targhooks.h (default_scalar_mode_supported_p): Likewise.
159 * targhooks.c (default_scalar_mode_supported_p): Likewise.
160 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
161 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
162 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
163 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
164 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
165 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
166 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
167 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
168 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
169 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
170 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
171 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
173 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
174 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
175 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
176 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
177 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
179 * doc/tm.texi: Regenerate.
181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
182 Alan Hayward <alan.hayward@arm.com>
183 David Sherwood <david.sherwood@arm.com>
185 * coretypes.h (opt_scalar_mode): New typedef.
186 * gdbhooks.py (build_pretty_printers): Handle it.
187 * machmode.h (mode_iterator::get_2xwider): Add overload for
189 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
191 * expr.c (convert_mode_scalar): Likewise.
192 * omp-low.c (omp_clause_aligned_alignment): Likewise.
193 * optabs.c (expand_float): Likewise.
194 (expand_fix): Likewise.
195 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
198 Alan Hayward <alan.hayward@arm.com>
199 David Sherwood <david.sherwood@arm.com>
201 * optabs.c (expand_float): Explicitly check for scalars before
202 using a branching expansion.
203 (expand_fix): Likewise.
205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
206 Alan Hayward <alan.hayward@arm.com>
207 David Sherwood <david.sherwood@arm.com>
209 * expr.c (convert_mode): Split scalar handling out into...
210 (convert_mode_scalar): ...this new function. Treat the modes
213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
219 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
222 Alan Hayward <alan.hayward@arm.com>
223 David Sherwood <david.sherwood@arm.com>
225 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
226 rather than a machine_mode.
227 (fixed_from_string): Likewise.
228 (fixed_convert): Likewise.
229 (fixed_convert_from_int): Likewise.
230 (fixed_convert_from_real): Likewise.
231 (real_convert_from_fixed): Likewise.
232 * fixed-value.c (fixed_from_double_int): Likewise.
233 (fixed_from_string): Likewise.
234 (fixed_convert): Likewise.
235 (fixed_convert_from_int): Likewise.
236 (fixed_convert_from_real): Likewise.
237 (real_convert_from_fixed): Likewise.
238 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
241 Alan Hayward <alan.hayward@arm.com>
242 David Sherwood <david.sherwood@arm.com>
244 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
245 of separate mode class checks. Do not allow vector modes here.
246 (immed_wide_int_const): Use as_a <scalar_mode>.
247 * explow.c (trunc_int_for_mode): Likewise.
248 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
249 (wi::shwi): Likewise.
250 (wi::min_value): Likewise.
251 (wi::max_value): Likewise.
252 * dwarf2out.c (loc_descriptor): Likewise.
253 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
257 Alan Hayward <alan.hayward@arm.com>
258 David Sherwood <david.sherwood@arm.com>
260 * tree.h (SCALAR_TYPE_MODE): New macro.
261 * expr.c (expand_expr_addr_expr_1): Use it.
262 (expand_expr_real_2): Likewise.
263 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
264 (fold_convert_const_fixed_from_int): Likewise.
265 (fold_convert_const_fixed_from_real): Likewise.
266 (native_encode_fixed): Likewise
267 (native_encode_complex): Likewise
268 (native_encode_vector): Likewise.
269 (native_interpret_fixed): Likewise.
270 (native_interpret_real): Likewise.
271 (native_interpret_complex): Likewise.
272 (native_interpret_vector): Likewise.
273 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
274 (simd_clone_adjust_argument_types): Likewise.
275 (simd_clone_init_simd_arrays): Likewise.
276 (simd_clone_adjust): Likewise.
277 * stor-layout.c (layout_type): Likewise.
278 * tree.c (build_minus_one_cst): Likewise.
279 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
280 * tree-inline.c (estimate_move_cost): Likewise.
281 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
282 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
283 (vectorizable_reduction): Likewise.
284 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
285 (vect_recog_mixed_size_cond_pattern): Likewise.
286 (check_bool_pattern): Likewise.
287 (adjust_bool_pattern): Likewise.
288 (search_type_for_mask_1): Likewise.
289 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
290 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
291 (vectorizable_load): Likewise.
292 (vectorizable_store): Likewise.
293 * ubsan.c (ubsan_encode_value): Likewise.
294 * varasm.c (output_constant): Likewise.
296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
297 Alan Hayward <alan.hayward@arm.com>
298 David Sherwood <david.sherwood@arm.com>
300 * coretypes.h (scalar_mode): New class.
301 * machmode.h (scalar_mode): Likewise.
302 (scalar_mode::includes_p): New function.
303 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
304 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
305 * genmodes.c (get_mode_class): Handle remaining scalar modes.
306 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
307 * expmed.c (store_bit_field_1): Likewise.
308 (extract_bit_field_1): Likewise.
309 * expr.c (write_complex_part): Likewise.
310 (read_complex_part): Likewise.
311 (emit_move_complex_push): Likewise.
312 (expand_expr_real_2): Likewise.
313 * function.c (assign_parm_setup_reg): Likewise.
314 (assign_parms_unsplit_complex): Likewise.
315 * optabs.c (expand_binop): Likewise.
316 * rtlanal.c (subreg_get_info): Likewise.
317 * simplify-rtx.c (simplify_immed_subreg): Likewise.
318 * varasm.c (output_constant_pool_2): Likewise.
320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
321 Alan Hayward <alan.hayward@arm.com>
322 David Sherwood <david.sherwood@arm.com>
324 * expmed.c (extract_high_half): Use scalar_int_mode and remove
326 (expmed_mult_highpart_optab): Likewise.
327 (expmed_mult_highpart): Likewise.
329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
330 Alan Hayward <alan.hayward@arm.com>
331 David Sherwood <david.sherwood@arm.com>
333 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
334 instead of a machine_mode.
335 (builtin_memset_read_str): Likewise.
336 * builtins.c (c_readstr): Likewise.
337 (builtin_memcpy_read_str): Likewise.
338 (builtin_strncpy_read_str): Likewise.
339 (builtin_memset_read_str): Likewise.
340 (builtin_memset_gen_str): Likewise.
341 (expand_builtin_signbit): Use scalar_int_mode for local variables.
342 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
343 instead of a machine_mode.
344 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
346 (make_extraction): Likewise.
347 (try_widen_shift_mode): Take and return scalar_int_modes instead
349 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
350 a scalar_int_mode instead of a machine_mode.
351 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
352 (avr_addr_space_pointer_mode): Likewise.
353 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
354 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
355 (msp430_unwind_word_mode): Likewise.
356 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
357 (spu_addr_space_pointer_mode): Likewise.
358 (spu_addr_space_address_mode): Likewise.
359 (spu_libgcc_cmp_return_mode): Likewise.
360 (spu_libgcc_shift_count_mode): Likewise.
361 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
362 (rl78_addr_space_pointer_mode): Likewise.
363 (fl78_unwind_word_mode): Likewise.
364 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
366 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
367 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
368 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
369 (mips_valid_pointer_mode): Likewise.
370 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
371 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
372 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
374 (ft32_addr_space_address_mode): Likewise.
375 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
376 scalar_int_mode instead of a machine_mode.
377 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
379 (m32c_addr_space_address_mode): Likewise.
380 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
381 (rs6000_eh_return_filter_mode): Likewise.
382 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
383 (rs6000_eh_return_filter_mode): Likewise.
384 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
385 (s390_libgcc_shift_count_mode): Likewise.
386 (s390_unwind_word_mode): Likewise.
387 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
389 * target.def (mode_rep_extended): Likewise.
390 (valid_pointer_mode): Likewise.
391 (addr_space.valid_pointer_mode): Likewise.
392 (eh_return_filter_mode): Return a scalar_int_mode rather than
394 (libgcc_cmp_return_mode): Likewise.
395 (libgcc_shift_count_mode): Likewise.
396 (unwind_word_mode): Likewise.
397 (addr_space.pointer_mode): Likewise.
398 (addr_space.address_mode): Likewise.
399 * doc/tm.texi: Regenerate.
400 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
402 (do_jump): Use scalar_int_mode for local variables.
403 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
404 rather than a machine_mode.
405 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
406 (scompare_loc_descriptor_wide): Likewise.
407 (scompare_loc_descriptor_narrow): Likewise.
408 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
410 * except.c (sjlj_emit_dispatch_table): Likewise.
411 (expand_builtin_eh_copy_values): Likewise.
412 * explow.c (convert_memory_address_addr_space_1): Likewise.
413 Take a scalar_int_mode rather than a machine_mode.
414 (convert_memory_address_addr_space): Take a scalar_int_mode rather
416 (memory_address_addr_space): Use scalar_int_mode for local variables.
417 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
418 rather than a machine_mode.
419 * expmed.c (mask_rtx): Likewise.
420 (init_expmed_one_conv): Likewise.
421 (expand_mult_highpart_adjust): Likewise.
422 (extract_high_half): Likewise.
423 (expmed_mult_highpart_optab): Likewise.
424 (expmed_mult_highpart): Likewise.
425 (expand_smod_pow2): Likewise.
426 (expand_sdiv_pow2): Likewise.
427 (emit_store_flag_int): Likewise.
428 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
430 (extract_low_bits): Likewise.
431 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
433 * expr.c (pieces_addr::adjust): Likewise.
434 (can_store_by_pieces): Likewise.
435 (store_by_pieces): Likewise.
436 (clear_by_pieces_1): Likewise.
437 (expand_expr_addr_expr_1): Likewise.
438 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
439 (expand_expr_real_1): Likewise.
440 (try_casesi): Likewise.
441 * final.c (shorten_branches): Likewise.
442 * fold-const.c (fold_convert_const_int_from_fixed): Change the
443 type of "mode" to machine_mode.
444 * internal-fn.c (expand_arith_overflow_result_store): Take a
445 scalar_int_mode rather than a machine_mode.
446 (expand_mul_overflow): Use scalar_int_mode for local variables.
447 * loop-doloop.c (doloop_modify): Likewise.
448 (doloop_optimize): Likewise.
449 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
451 (expand_doubleword_shift_condmove): Likewise.
452 (expand_doubleword_shift): Likewise.
453 (expand_doubleword_clz): Likewise.
454 (expand_doubleword_popcount): Likewise.
455 (expand_doubleword_parity): Likewise.
456 (expand_absneg_bit): Use scalar_int_mode for local variables.
457 (prepare_float_lib_cmp): Likewise.
458 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
459 rather than a machine_mode.
460 (convert_memory_address_addr_space): Likewise.
461 (get_mode_bounds): Likewise.
462 (get_address_mode): Return a scalar_int_mode rather than a
464 * rtlanal.c (get_address_mode): Likewise.
465 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
467 * targhooks.c (default_mode_rep_extended): Likewise.
468 (default_valid_pointer_mode): Likewise.
469 (default_addr_space_valid_pointer_mode): Likewise.
470 (default_eh_return_filter_mode): Return a scalar_int_mode rather
472 (default_libgcc_cmp_return_mode): Likewise.
473 (default_libgcc_shift_count_mode): Likewise.
474 (default_unwind_word_mode): Likewise.
475 (default_addr_space_pointer_mode): Likewise.
476 (default_addr_space_address_mode): Likewise.
477 * targhooks.h (default_eh_return_filter_mode): Likewise.
478 (default_libgcc_cmp_return_mode): Likewise.
479 (default_libgcc_shift_count_mode): Likewise.
480 (default_unwind_word_mode): Likewise.
481 (default_addr_space_pointer_mode): Likewise.
482 (default_addr_space_address_mode): Likewise.
483 (default_mode_rep_extended): Take a scalar_int_mode rather than
485 (default_valid_pointer_mode): Likewise.
486 (default_addr_space_valid_pointer_mode): Likewise.
487 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
489 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
490 rather than a machine_mode.
491 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
493 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
494 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
498 Alan Hayward <alan.hayward@arm.com>
499 David Sherwood <david.sherwood@arm.com>
501 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
502 the mode argument to scalar_int_mode.
503 (do_jump_by_parts_zero_rtx): Likewise.
504 (do_jump_by_parts_equality_rtx): Likewise.
505 (do_jump_by_parts_greater): Take a mode argument.
506 (do_jump_by_parts_equality): Likewise.
507 (do_jump_1): Update calls accordingly.
509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
510 Alan Hayward <alan.hayward@arm.com>
511 David Sherwood <david.sherwood@arm.com>
513 * is-a.h (safe_dyn_cast): New function.
514 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
515 (jump_table_for_label): Likewise.
516 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
517 instead of an rtx_insn *.
518 (shorten_branches): Use dyn_cast instead of LABEL_P and
519 JUMP_TABLE_DATA_P. Use jump_table_for_label and
520 rtx_jump_table_data::get_data_mode.
521 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
528 trying to combine a full-register integer set with a subreg
531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
532 Alan Hayward <alan.hayward@arm.com>
533 David Sherwood <david.sherwood@arm.com>
535 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
536 that is always either address_mode or pointer_mode.
538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
539 Alan Hayward <alan.hayward@arm.com>
540 David Sherwood <david.sherwood@arm.com>
542 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
543 when the two are known to be equal.
545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
546 Alan Hayward <alan.hayward@arm.com>
547 David Sherwood <david.sherwood@arm.com>
549 * simplify-rtx.c (simplify_const_unary_operation): Use
550 is_a <scalar_int_mode> instead of checking for a nonzero
551 precision. Forcibly convert op_mode to a scalar_int_mode
552 in that case. More clearly differentiate the operand and
553 result modes and use the former when deciding what the value
554 of a count-bits operation should be. Use is_int_mode instead
555 of checking for a MODE_INT. Remove redundant check for whether
556 this mode has a zero precision.
558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
559 Alan Hayward <alan.hayward@arm.com>
560 David Sherwood <david.sherwood@arm.com>
562 * optabs.c (widen_leading): Change the type of the mode argument
563 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
564 (widen_bswap): Likewise.
565 (expand_parity): Likewise.
566 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
567 (expand_ffs): Likewise.
568 (epand_unop): Check for scalar integer modes before calling the
571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
572 Alan Hayward <alan.hayward@arm.com>
573 David Sherwood <david.sherwood@arm.com>
575 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
577 (expand_expr_real_1): Update call accordingly.
579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
580 Alan Hayward <alan.hayward@arm.com>
581 David Sherwood <david.sherwood@arm.com>
583 * expmed.c (store_bit_field_using_insv): Add op0_mode and
584 value_mode arguments. Use scalar_int_mode internally.
585 (store_bit_field_1): Rename the new integer mode from imode
586 to op0_mode and use it instead of GET_MODE (op0). Update calls
587 to store_split_bit_field, store_bit_field_using_insv and
588 store_fixed_bit_field.
589 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
590 Use scalar_int_mode internally. Use a bit count rather than a mode
591 when calculating the largest bit size for get_best_mode.
592 Update calls to store_split_bit_field and store_fixed_bit_field_1.
593 (store_fixed_bit_field_1): Add mode and value_mode arguments.
594 Remove assertion that OP0 has a scalar integer mode.
595 (store_split_bit_field): Add op0_mode and value_mode arguments.
596 Update calls to extract_fixed_bit_field.
597 (extract_bit_field_using_extv): Add an op0_mode argument.
598 Use scalar_int_mode internally.
599 (extract_bit_field_1): Rename the new integer mode from imode to
600 op0_mode and use it instead of GET_MODE (op0). Update calls to
601 extract_split_bit_field, extract_bit_field_using_extv and
602 extract_fixed_bit_field.
603 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
604 to extract_split_bit_field and extract_fixed_bit_field_1.
605 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
606 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
608 (extract_split_bit_field): Add an op0_mode argument. Update call
609 to extract_fixed_bit_field.
611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
612 Alan Hayward <alan.hayward@arm.com>
613 David Sherwood <david.sherwood@arm.com>
615 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
616 * explow.c (hard_function_value): Likewise.
617 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
618 convert_to_mode call outside the loop.
619 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
620 for the mode iterator. Require the mode specified by max_pieces
622 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
624 (copy_blkmode_to_reg): Likewise.
625 (set_storage_via_setmem): Likewise.
626 * optabs.c (prepare_cmp_insn): Likewise.
627 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
628 * stor-layout.c (finish_bitfield_representative): Likewise.
630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
631 Alan Hayward <alan.hayward@arm.com>
632 David Sherwood <david.sherwood@arm.com>
634 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
635 * expr.c (convert_move): Use them.
636 (convert_modes): Likewise.
637 (store_expr_with_bounds): Likewise.
639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
640 Alan Hayward <alan.hayward@arm.com>
641 David Sherwood <david.sherwood@arm.com>
643 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
644 parameter for the mode of "x". Remove the "known_x", "known_mode"
645 and "known_ret" arguments. Change the type of the mode argument
647 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
648 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
649 (reg_num_sign_bit_copies_for_combine): Likewise.
650 * rtlanal.c (nonzero_bits1): Likewise.
651 (num_sign_bit_copies1): Likewise.
652 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
653 (reg_num_sign_bit_copies_general): Likewise.
654 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
655 (reg_nonzero_bits_general): Likewise.
657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
658 Alan Hayward <alan.hayward@arm.com>
659 David Sherwood <david.sherwood@arm.com>
661 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
662 than in subroutines. Return 1 for non-integer modes.
663 (cached_num_sign_bit_copies): Change the type of the mode parameter
665 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
666 classes. Handle CONST_INT_P first and then check whether X also
667 has a scalar integer mode. Check the same thing for inner registers
668 of a SUBREG and for values that are being extended or truncated.
670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
671 Alan Hayward <alan.hayward@arm.com>
672 David Sherwood <david.sherwood@arm.com>
674 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
675 in subroutines. Return the mode mask for non-integer modes.
676 (cached_nonzero_bits): Change the type of the mode parameter
678 (nonzero_bits1): Likewise. Remove early exit for other mode
679 classes. Handle CONST_INT_P first and then check whether X
680 also has a scalar integer mode.
682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
683 Alan Hayward <alan.hayward@arm.com>
684 David Sherwood <david.sherwood@arm.com>
686 * expr.c (widest_int_mode_for_size): Make the comment match the code.
687 Return a scalar_int_mode and assert that the size is greater than
689 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
690 (op_by_pieces_d::op_by_pieces_d): Likewise.
691 (op_by_pieces_d::run): Likewise.
692 (can_store_by_pieces): Likewise.
694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
695 Alan Hayward <alan.hayward@arm.com>
696 David Sherwood <david.sherwood@arm.com>
698 * combine.c (extract_left_shift): Add a mode argument and update
700 (make_compound_operation_int): Change the type of the mode parameter
701 to scalar_int_mode and update the call to extract_left_shift.
703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
704 Alan Hayward <alan.hayward@arm.com>
705 David Sherwood <david.sherwood@arm.com>
707 * combine.c (simplify_and_const_int): Change the type of the mode
708 parameter to scalar_int_mode.
709 (simplify_and_const_int_1): Likewise. Update recursive call.
711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
712 Alan Hayward <alan.hayward@arm.com>
713 David Sherwood <david.sherwood@arm.com>
715 * combine.c (simplify_compare_const): Check that the mode is a
716 scalar_int_mode (rather than VOIDmode) before testing its
718 (simplify_comparison): Move COMPARISON_P handling out of the
719 loop and restrict the latter part of the loop to scalar_int_modes.
720 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
721 and when considering SUBREG_REGs. Use is_int_mode instead of
722 checking GET_MODE_CLASS against MODE_INT.
724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
725 Alan Hayward <alan.hayward@arm.com>
726 David Sherwood <david.sherwood@arm.com>
728 * combine.c (try_widen_shift_mode): Move check for equal modes to...
729 (simplify_shift_const_1): ...here. Use scalar_int_mode for
730 shift_unit_mode and for modes involved in scalar shifts.
732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
733 Alan Hayward <alan.hayward@arm.com>
734 David Sherwood <david.sherwood@arm.com>
736 * combine.c (force_int_to_mode): New function, split out from...
737 (force_to_mode): ...here. Keep xmode up-to-date and use it
738 instead of GET_MODE (x).
740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
741 Alan Hayward <alan.hayward@arm.com>
742 David Sherwood <david.sherwood@arm.com>
744 * optabs-query.h (extraction_insn::struct_mode): Change type to
745 opt_scalar_int_mode and update comment.
746 (extraction_insn::field_mode): Change type to scalar_int_mode.
747 (extraction_insn::pos_mode): Likewise.
748 * combine.c (make_extraction): Update accordingly.
749 * optabs-query.c (get_traditional_extraction_insn): Likewise.
750 (get_optab_extraction_insn): Likewise.
751 * recog.c (simplify_while_replacing): Likewise.
752 * expmed.c (narrow_bit_field_mem): Change the type of the mode
753 parameter to opt_scalar_int_mode.
755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
756 Alan Hayward <alan.hayward@arm.com>
757 David Sherwood <david.sherwood@arm.com>
759 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
760 to a scalar_int_mode instead of a machine_mode.
761 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
762 (get_best_mode): Return a boolean and use a pointer argument to store
763 the selected mode. Replace the limit mode parameter with a bit limit.
764 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
765 for the values returned by bit_field_mode_iterator::next_mode.
766 (store_bit_field): Update call to get_best_mode.
767 (store_fixed_bit_field): Likewise.
768 (extract_fixed_bit_field): Likewise.
769 * expr.c (optimize_bitfield_assignment_op): Likewise.
770 * fold-const.c (optimize_bit_field_compare): Likewise.
771 (fold_truth_andor_1): Likewise.
772 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
773 Update for new type of m_mode.
774 (get_best_mode): As above.
776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
777 Alan Hayward <alan.hayward@arm.com>
778 David Sherwood <david.sherwood@arm.com>
780 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
781 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
782 (store_bit_field): Check is_a <scalar_int_mode> before calling
783 strict_volatile_bitfield_p.
784 (extract_bit_field): Likewise.
786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
787 Alan Hayward <alan.hayward@arm.com>
788 David Sherwood <david.sherwood@arm.com>
790 * target.def (cstore_mode): Return a scalar_int_mode.
791 * doc/tm.texi: Regenerate.
792 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
793 * targhooks.h (default_cstore_mode): Likewise.
794 * targhooks.c (default_cstore_mode): Likewise, using a forced
796 * expmed.c (emit_cstore): Expect the target of the cstore to be
799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
800 Alan Hayward <alan.hayward@arm.com>
801 David Sherwood <david.sherwood@arm.com>
803 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
805 (niter_desc): Likewise mode.
806 (iv_analyze): Add a mode parameter.
808 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
809 and change its type to scalar_int_mode.
810 * loop-iv.c: Update commentary at head of file.
811 (iv_constant): Pass the mode paraeter before the rtx it describes
812 and change its type to scalar_int_mode. Remove VOIDmode handling.
813 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
814 (iv_extend): Likewise.
815 (shorten_into_mode): Likewise.
816 (iv_add): Use scalar_int_mode.
818 (iv_shift): Likewise.
819 (canonicalize_iv_subregs): Likewise.
820 (get_biv_step_1): Pass the outer_mode parameter before the rtx
821 it describes and change its mode to scalar_int_mode. Also change
822 the type of the returned inner_mode to scalar_int_mode.
823 (get_biv_step): Likewise, turning outer_mode from a pointer
824 into a direct parameter. Update call to get_biv_step_1.
825 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
826 iv_constant and get_biv_step.
827 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
828 and change its type to scalar_int_mode. Don't initialise iv->mode
829 to VOIDmode and remove later checks for its still being VOIDmode.
830 Update calls to iv_analyze_op and iv_analyze_expr. Check
831 is_a <scalar_int_mode> when changing the mode under consideration.
832 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
833 Update call to iv_analyze_expr.
834 (iv_analyze_op): Add a mode parameter. Reject subregs whose
835 inner register is not also a scalar_int_mode. Update call to
837 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
838 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
839 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
840 separate mode class checks. Update calls to iv_analyze. Remove
841 fix-up of VOIDmodes after iv_analyze_biv.
842 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
843 don't have a scalar_int_mode. Update call to biv_p.
845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
846 Alan Hayward <alan.hayward@arm.com>
847 David Sherwood <david.sherwood@arm.com>
849 * cfgexpand.c (convert_debug_memory_address): Use
850 as_a <scalar_int_mode>.
851 * combine.c (expand_compound_operation): Likewise.
852 (make_extraction): Likewise.
853 (change_zero_ext): Likewise.
854 (simplify_comparison): Likewise.
855 * cse.c (cse_insn): Likewise.
856 * dwarf2out.c (minmax_loc_descriptor): Likewise.
857 (mem_loc_descriptor): Likewise.
858 (loc_descriptor): Likewise.
859 * expmed.c (init_expmed_one_mode): Likewise.
860 (synth_mult): Likewise.
861 (emit_store_flag_1): Likewise.
862 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
863 of a comparison with size.
864 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
865 (reduce_to_bit_field_precision): Likewise.
866 * function.c (expand_function_end): Likewise.
867 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
868 * loop-doloop.c (doloop_modify): Likewise.
869 * optabs.c (expand_binop): Likewise.
870 (expand_unop): Likewise.
871 (expand_copysign_absneg): Likewise.
872 (prepare_cmp_insn): Likewise.
873 (maybe_legitimize_operand): Likewise.
874 * recog.c (const_scalar_int_operand): Likewise.
875 * rtlanal.c (get_address_mode): Likewise.
876 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
877 (simplify_cond_clz_ctz): Likewise.
878 * tree-nested.c (get_nl_goto_field): Likewise.
879 * tree.c (build_vector_type_for_mode): Likewise.
880 * var-tracking.c (use_narrower_mode): Likewise.
882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
883 Alan Hayward <alan.hayward@arm.com>
884 David Sherwood <david.sherwood@arm.com>
886 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
887 * builtins.c (expand_builtin_signbit): Use it.
888 * cfgexpand.c (expand_debug_expr): Likewise.
889 * dojump.c (do_jump): Likewise.
890 (do_compare_and_jump): Likewise.
891 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
892 * expmed.c (make_tree): Likewise.
893 * expr.c (expand_expr_real_2): Likewise.
894 (expand_expr_real_1): Likewise.
895 (try_casesi): Likewise.
896 * fold-const-call.c (fold_const_call_ss): Likewise.
897 * fold-const.c (unextend): Likewise.
898 (extract_muldiv_1): Likewise.
899 (fold_single_bit_test): Likewise.
900 (native_encode_int): Likewise.
901 (native_encode_string): Likewise.
902 (native_interpret_int): Likewise.
903 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
904 * internal-fn.c (expand_addsub_overflow): Likewise.
905 (expand_neg_overflow): Likewise.
906 (expand_mul_overflow): Likewise.
907 (expand_arith_overflow): Likewise.
908 * match.pd: Likewise.
909 * stor-layout.c (layout_type): Likewise.
910 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
911 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
912 * tree-ssanames.c (get_range_info): Likewise.
913 * tree-switch-conversion.c (array_value_type) Likewise.
914 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
915 (vect_recog_divmod_pattern): Likewise.
916 (vect_recog_mixed_size_cond_pattern): Likewise.
917 * tree-vrp.c (extract_range_basic): Likewise.
918 (simplify_float_conversion_using_ranges): Likewise.
919 * tree.c (int_fits_type_p): Likewise.
920 * ubsan.c (instrument_bool_enum_load): Likewise.
921 * varasm.c (mergeable_string_section): Likewise.
922 (narrowing_initializer_constant_valid_p): Likewise.
923 (output_constant): Likewise.
925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
926 Alan Hayward <alan.hayward@arm.com>
927 David Sherwood <david.sherwood@arm.com>
929 * machmode.h (NARROWEST_INT_MODE): New macro.
930 * expr.c (alignment_for_piecewise_move): Use it instead of
931 GET_CLASS_NARROWEST_MODE (MODE_INT).
932 (push_block): Likewise.
933 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
935 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
938 Alan Hayward <alan.hayward@arm.com>
939 David Sherwood <david.sherwood@arm.com>
941 * postreload.c (move2add_valid_value_p): Change the type of the
942 mode parameter to scalar_int_mode.
943 (move2add_use_add2_insn): Add a mode parameter and use it instead
945 (move2add_use_add3_insn): Likewise.
946 (reload_cse_move2add): Update accordingly.
948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
949 Alan Hayward <alan.hayward@arm.com>
950 David Sherwood <david.sherwood@arm.com>
952 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
954 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
955 * optabs.c (expand_unop): Likewise.
957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
958 Alan Hayward <alan.hayward@arm.com>
959 David Sherwood <david.sherwood@arm.com>
961 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
962 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
963 (popcount_loc_descriptor): Likewise.
964 (bswap_loc_descriptor): Likewise.
965 (rotate_loc_descriptor): Likewise.
966 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
967 calling the functions above.
969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
970 Alan Hayward <alan.hayward@arm.com>
971 David Sherwood <david.sherwood@arm.com>
973 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
975 (try_combine): Likewise.
976 (simplify_if_then_else): Likewise.
977 * cse.c (cse_insn): Likewise.
978 * dwarf2out.c (mem_loc_descriptor): Likewise.
979 * emit-rtl.c (gen_lowpart_common): Likewise.
980 * simplify-rtx.c (simplify_truncation): Likewise.
981 (simplify_binary_operation_1): Likewise.
982 (simplify_const_relational_operation): Likewise.
983 (simplify_ternary_operation): Likewise.
984 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
987 Alan Hayward <alan.hayward@arm.com>
988 David Sherwood <david.sherwood@arm.com>
990 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
991 * reload.c (push_reload): Likewise.
992 (find_reloads): Likewise.
994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
995 Alan Hayward <alan.hayward@arm.com>
996 David Sherwood <david.sherwood@arm.com>
998 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
999 (make_compound_operation_int): Likewise.
1000 (change_zero_ext): Likewise.
1001 * expr.c (convert_move): Likewise.
1002 (convert_modes): Likewise.
1003 * fwprop.c (forward_propagate_subreg): Likewise.
1004 * loop-iv.c (get_biv_step_1): Likewise.
1005 * optabs.c (widen_operand): Likewise.
1006 * postreload.c (move2add_valid_value_p): Likewise.
1007 * recog.c (simplify_while_replacing): Likewise.
1008 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1009 (simplify_binary_operation_1): Likewise. Remove redundant
1010 mode equality check.
1012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1013 Alan Hayward <alan.hayward@arm.com>
1014 David Sherwood <david.sherwood@arm.com>
1016 * combine.c (combine_simplify_rtx): Add checks for
1017 is_a <scalar_int_mode>.
1018 (simplify_if_then_else): Likewise.
1019 (make_field_assignment): Likewise.
1020 (simplify_comparison): Likewise.
1021 * ifcvt.c (noce_try_bitop): Likewise.
1022 * loop-invariant.c (canonicalize_address_mult): Likewise.
1023 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1026 Alan Hayward <alan.hayward@arm.com>
1027 David Sherwood <david.sherwood@arm.com>
1029 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1030 is_a <scalar_int_mode> instead of != BLKmode.
1032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1033 Alan Hayward <alan.hayward@arm.com>
1034 David Sherwood <david.sherwood@arm.com>
1036 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1037 instead of != VOIDmode.
1038 * combine.c (if_then_else_cond): Likewise.
1039 (change_zero_ext): Likewise.
1040 * dwarf2out.c (mem_loc_descriptor): Likewise.
1041 (loc_descriptor): Likewise.
1042 * rtlanal.c (canonicalize_condition): Likewise.
1043 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1046 Alan Hayward <alan.hayward@arm.com>
1047 David Sherwood <david.sherwood@arm.com>
1049 * simplify-rtx.c (simplify_binary_operation_1): Use
1050 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
1057 (int_traits<unsigned short>) Likewise.
1058 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1059 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1061 * combine.c (set_nonzero_bits_and_sign_copies): Use
1062 is_a <scalar_int_mode>.
1063 (find_split_point): Likewise.
1064 (combine_simplify_rtx): Likewise.
1065 (simplify_logical): Likewise.
1066 (expand_compound_operation): Likewise.
1067 (expand_field_assignment): Likewise.
1068 (make_compound_operation): Likewise.
1069 (extended_count): Likewise.
1070 (change_zero_ext): Likewise.
1071 (simplify_comparison): Likewise.
1072 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1073 (ucompare_loc_descriptor): Likewise.
1074 (minmax_loc_descriptor): Likewise.
1075 (mem_loc_descriptor): Likewise.
1076 (loc_descriptor): Likewise.
1077 * expmed.c (init_expmed_one_mode): Likewise.
1078 * lra-constraints.c (lra_constraint_offset): Likewise.
1079 * optabs.c (prepare_libcall_arg): Likewise.
1080 * postreload.c (move2add_note_store): Likewise.
1081 * reload.c (operands_match_p): Likewise.
1082 * rtl.h (load_extend_op): Likewise.
1083 * rtlhooks.c (gen_lowpart_general): Likewise.
1084 * simplify-rtx.c (simplify_truncation): Likewise.
1085 (simplify_unary_operation_1): Likewise.
1086 (simplify_binary_operation_1): Likewise.
1087 (simplify_const_binary_operation): Likewise.
1088 (simplify_const_relational_operation): Likewise.
1089 (simplify_subreg): Likewise.
1090 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1091 * var-tracking.c (adjust_mems): Likewise.
1092 (prepare_call_arguments): Likewise.
1094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1095 Alan Hayward <alan.hayward@arm.com>
1096 David Sherwood <david.sherwood@arm.com>
1098 * machmode.h (is_int_mode): New fuction.
1099 * combine.c (find_split_point): Use it.
1100 (combine_simplify_rtx): Likewise.
1101 (simplify_if_then_else): Likewise.
1102 (simplify_set): Likewise.
1103 (simplify_shift_const_1): Likewise.
1104 (simplify_comparison): Likewise.
1105 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1106 * cse.c (notreg_cost): Likewise.
1107 (cse_insn): Likewise.
1108 * cselib.c (cselib_lookup_1): Likewise.
1109 * dojump.c (do_jump_1): Likewise.
1110 (do_compare_rtx_and_jump): Likewise.
1111 * dse.c (get_call_args): Likewise.
1112 * dwarf2out.c (rtl_for_decl_init): Likewise.
1113 (native_encode_initializer): Likewise.
1114 * expmed.c (emit_store_flag_1): Likewise.
1115 (emit_store_flag): Likewise.
1116 * expr.c (convert_modes): Likewise.
1117 (store_field): Likewise.
1118 (expand_expr_real_1): Likewise.
1119 * fold-const.c (fold_read_from_constant_string): Likewise.
1120 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1121 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1122 * optabs.c (expand_binop): Likewise.
1123 (expand_unop): Likewise.
1124 (expand_abs_nojump): Likewise.
1125 (expand_one_cmpl_abs_nojump): Likewise.
1126 * simplify-rtx.c (mode_signbit_p): Likewise.
1127 (val_signbit_p): Likewise.
1128 (val_signbit_known_set_p): Likewise.
1129 (val_signbit_known_clear_p): Likewise.
1130 (simplify_relational_operation_1): Likewise.
1131 * tree.c (vector_type_mode): Likewise.
1133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1134 Alan Hayward <alan.hayward@arm.com>
1135 David Sherwood <david.sherwood@arm.com>
1137 * machmode.h (smallest_mode_for_size): Fix formatting.
1138 (smallest_int_mode_for_size): New function.
1139 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1140 instead of smallest_mode_for_size.
1141 * combine.c (make_extraction): Likewise.
1142 * config/arc/arc.c (arc_expand_movmem): Likewise.
1143 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1144 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1145 * config/s390/s390.c (s390_expand_insv): Likewise.
1146 * config/sparc/sparc.c (assign_int_registers): Likewise.
1147 * config/spu/spu.c (spu_function_value): Likewise.
1148 (spu_function_arg): Likewise.
1149 * coverage.c (get_gcov_type): Likewise.
1150 (get_gcov_unsigned_t): Likewise.
1151 * dse.c (find_shift_sequence): Likewise.
1152 * expmed.c (store_bit_field_1): Likewise.
1153 * expr.c (convert_move): Likewise.
1154 (store_field): Likewise.
1155 * internal-fn.c (expand_arith_overflow): Likewise.
1156 * optabs-query.c (get_best_extraction_insn): Likewise.
1157 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1158 * stor-layout.c (layout_type): Likewise.
1159 (initialize_sizetypes): Likewise.
1160 * targhooks.c (default_get_mask_mode): Likewise.
1161 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1164 Alan Hayward <alan.hayward@arm.com>
1165 David Sherwood <david.sherwood@arm.com>
1167 * machmode.h (opt_mode::else_blk): New function.
1168 (int_mode_for_mode): Declare.
1169 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1170 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1172 * cfgexpand.c (expand_debug_expr): Likewise.
1173 * combine.c (gen_lowpart_or_truncate): Likewise.
1174 (gen_lowpart_for_combine): Likewise.
1175 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1176 * config/avr/avr.c (avr_to_int_mode): Likewise.
1177 (avr_out_plus_1): Likewise.
1178 (avr_out_plus): Likewise.
1179 (avr_out_round): Likewise.
1180 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1181 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1182 (s390_expand_vcond): Likewise.
1183 * config/spu/spu.c (spu_split_immediate): Likewise.
1184 (spu_expand_mov): Likewise.
1185 * dse.c (get_stored_val): Likewise.
1186 * expmed.c (store_bit_field_1): Likewise.
1187 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1189 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1190 (extract_low_bits): Likewise.
1191 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1192 handling rather than repeating the check.
1193 (emit_group_store): Likewise.
1194 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1195 * optabs.c (expand_absneg_bit): Likewise.
1196 (expand_copysign_absneg): Likewise.
1197 (expand_copysign_bit): Likewise.
1198 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1199 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1200 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1201 * var-tracking.c (prepare_call_arguments): Likewise.
1202 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1203 int_mode_for_mode instead of mode_for_size.
1204 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1207 Alan Hayward <alan.hayward@arm.com>
1208 David Sherwood <david.sherwood@arm.com>
1210 * machmode.h (int_mode_for_size): New function.
1211 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1212 instead of mode_for_size.
1213 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1215 * combine.c (expand_field_assignment): Use int_mode_for_size
1216 instead of mode_for_size.
1217 (make_extraction): Likewise.
1218 (simplify_shift_const_1): Likewise.
1219 (simplify_comparison): Likewise.
1220 * dojump.c (do_jump): Likewise.
1221 * dwarf2out.c (mem_loc_descriptor): Likewise.
1222 * emit-rtl.c (init_derived_machine_modes): Likewise.
1223 * expmed.c (flip_storage_order): Likewise.
1224 (convert_extracted_bit_field): Likewise.
1225 * expr.c (copy_blkmode_from_reg): Likewise.
1226 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1227 * internal-fn.c (expand_mul_overflow): Likewise.
1228 * lower-subreg.c (simple_move): Likewise.
1229 * optabs-libfuncs.c (init_optabs): Likewise.
1230 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1231 * tree.c (vector_type_mode): Likewise.
1232 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1233 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1234 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1235 * tree-vect-stmts.c (vectorizable_load): Likewise.
1236 (vectorizable_store): Likewise.
1238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * coretypes.h (pod_mode): New type.
1243 (scalar_int_mode_pod): New typedef.
1244 * machmode.h (pod_mode): New class.
1245 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1246 * genmodes.c (emit_mode_int_n): Update accordingly.
1247 * lower-subreg.h (target_lower_subreg): Change type to
1248 scalar_int_mode_pod.
1249 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1250 scalar_int_mode_pod.
1252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1253 Alan Hayward <alan.hayward@arm.com>
1254 David Sherwood <david.sherwood@arm.com>
1256 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1257 machine_mode to scalar_int_mode.
1258 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1259 (rs6000_option_override_internal): Remove cast to int.
1260 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1261 machine_mode to scalar_int_mode.
1262 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1263 (rs6000_option_override_internal): Remove cast to int.
1264 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1265 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1267 * config/s390/s390.c (s390_expand_builtin): Likewise.
1268 * coretypes.h (scalar_int_mode): New type.
1269 (opt_scalar_int_mode): New typedef.
1270 * machmode.h (scalar_int_mode): New class.
1271 (scalar_int_mode::includes_p): New function.
1272 (byte_mode): Change type to scalar_int_mode.
1273 (word_mode): Likewise.
1274 (ptr_mode): Likewise.
1275 * emit-rtl.c (byte_mode): Likewise.
1276 (word_mode): Likewise.
1277 (ptr_mode): Likewise.
1278 (init_derived_machine_modes): Update accordingly.
1279 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1280 and MODE_PARTIAL_INT.
1281 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1282 opt_scalar_int_mode.
1284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1285 Alan Hayward <alan.hayward@arm.com>
1286 David Sherwood <david.sherwood@arm.com>
1288 * target.def (libgcc_floating_mode_supported_p): Take a
1290 * doc/tm.texi: Regenerate.
1291 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1293 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1294 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1298 Alan Hayward <alan.hayward@arm.com>
1299 David Sherwood <david.sherwood@arm.com>
1301 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1302 * doc/tm.texi: Regenerate.
1303 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1304 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1305 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1306 * targhooks.h (default_floatn_mode): Likewise.
1307 * targhooks.c (default_floatn_mode): Likewise.
1308 * tree.c (build_common_tree_nodes): Update accordingly.
1310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1311 Alan Hayward <alan.hayward@arm.com>
1312 David Sherwood <david.sherwood@arm.com>
1314 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1315 (mode_iterator::iterate_p): Likewise.
1316 (mode_iterator::get_wider): Likewise.
1317 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1320 Alan Hayward <alan.hayward@arm.com>
1321 David Sherwood <david.sherwood@arm.com>
1323 * coretypes.h (opt_scalar_float_mode): New typedef.
1324 * machmode.h (float_mode_for_size): New function.
1325 * emit-rtl.c (double_mode): Delete.
1326 (init_emit_once): Use float_mode_for_size.
1327 * stor-layout.c (layout_type): Likewise.
1328 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * output.h (assemble_real): Take a scalar_float_mode.
1335 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1336 * config/arm/arm.md (consttable_4): Likewise.
1337 (consttable_8): Likewise.
1338 (consttable_16): Likewise.
1339 * config/mips/mips.md (consttable_float): Likewise.
1340 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1341 * varasm.c (assemble_real): Take a scalar_float_mode.
1342 (output_constant_pool_2): Update accordingly.
1343 (output_constant): Likewise.
1345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1346 Alan Hayward <alan.hayward@arm.com>
1347 David Sherwood <david.sherwood@arm.com>
1349 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1350 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1351 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1352 (native_encode_real): Likewise.
1353 (native_interpret_real): Likewise.
1354 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1355 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1358 Alan Hayward <alan.hayward@arm.com>
1359 David Sherwood <david.sherwood@arm.com>
1361 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1362 <scalar_float_mode>. Simplify.
1363 (gen_extend_conv_libfunc): Likewise.
1365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1366 Alan Hayward <alan.hayward@arm.com>
1367 David Sherwood <david.sherwood@arm.com>
1369 * coretypes.h (scalar_float_mode): New type.
1370 * machmode.h (mode_traits::from_int): Use machine_mode if
1371 USE_ENUM_MODES is defined.
1372 (is_a): New function.
1374 (dyn_cast): Likewise.
1375 (scalar_float_mode): New class.
1376 (scalar_float_mode::includes_p): New function.
1377 (is_float_mode): Likewise.
1378 * gdbhooks.py (MachineModePrinter): New class.
1379 (build_pretty_printer): Use it for scalar_float_mode.
1380 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1381 (format_helper::format_helper): Turn into a template.
1382 * genmodes.c (get_mode_class): New function.
1383 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1384 or machine_mode if none.
1385 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1386 as_a <scalar_float_mode>.
1387 * dwarf2out.c (mem_loc_descriptor): Likewise.
1388 (insert_float): Likewise.
1389 (add_const_value_attribute): Likewise.
1390 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1391 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1392 (expand_unop): Update accordingly.
1393 (expand_abs_nojump): Likewise.
1394 (expand_copysign_absneg): Take a scalar_float_mode.
1395 (expand_copysign_bit): Likewise.
1396 (expand_copysign): Update accordingly.
1398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1399 Alan Hayward <alan.hayward@arm.com>
1400 David Sherwood <david.sherwood@arm.com>
1402 * coretypes.h (opt_mode): New class.
1403 * machmode.h (opt_mode): Likewise.
1404 (opt_mode::else_void): New function.
1405 (opt_mode::require): Likewise.
1406 (opt_mode::exists): Likewise.
1407 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1408 (GET_MODE_2XWIDER_MODE): Likewise.
1409 (mode_iterator::get_wider): Update accordingly.
1410 (mode_iterator::get_2xwider): Likewise.
1411 (mode_iterator::get_known_wider): Likewise, turning into a template.
1412 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1413 forcing a wider mode to exist.
1414 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1415 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1416 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1417 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1418 * lower-subreg.c (init_lower_subreg): Likewise.
1419 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1420 on the final iteration.
1421 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1422 a wider mode exists before asking for a move pattern.
1423 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1424 forcing a wider mode to exist.
1425 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1426 returning false if no such mode exists.
1427 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1428 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1429 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1430 Avoid checking for a MODE_INT if we already know the mode is not a
1432 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1433 forcing a wider mode to exist.
1434 (expmed_mult_highpart_optab): Likewise.
1435 (expmed_mult_highpart): Likewise.
1436 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1438 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1439 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1440 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1441 * internal-fn.c (expand_mul_overflow): Update use of
1442 GET_MODE_2XWIDER_MODE.
1443 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1444 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1445 GET_MODE_WIDER_MODE.
1446 (convert_plusminus_to_widen): Likewise.
1447 * tree-switch-conversion.c (array_value_type): Likewise.
1448 * var-tracking.c (emit_note_insn_var_location): Likewise.
1449 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1450 Return false inside rather than outside the loop if no wider mode
1452 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1453 and GET_MODE_2XWIDER_MODE
1454 (can_compare_p): Use else_void.
1455 * gdbhooks.py (OptMachineModePrinter): New class.
1456 (build_pretty_printer): Use it for opt_mode.
1458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1459 Alan Hayward <alan.hayward@arm.com>
1460 David Sherwood <david.sherwood@arm.com>
1462 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1463 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1466 Alan Hayward <alan.hayward@arm.com>
1467 David Sherwood <david.sherwood@arm.com>
1469 * machmode.h (mode_traits): New structure.
1470 (get_narrowest_mode): New function.
1471 (mode_iterator::start): Likewise.
1472 (mode_iterator::iterate_p): Likewise.
1473 (mode_iterator::get_wider): Likewise.
1474 (mode_iterator::get_known_wider): Likewise.
1475 (mode_iterator::get_2xwider): Likewise.
1476 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1477 (FOR_EACH_MODE): Likewise.
1478 (FOR_EACH_MODE_FROM): Likewise.
1479 (FOR_EACH_MODE_UNTIL): Likewise.
1480 (FOR_EACH_WIDER_MODE): Likewise.
1481 (FOR_EACH_2XWIDER_MODE): Likewise.
1482 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1483 * combine.c (simplify_comparison): Likewise
1484 * config/i386/i386.c (type_natural_mode): Likewise.
1485 * cse.c (cse_insn): Likewise.
1486 * dse.c (find_shift_sequence): Likewise.
1487 * emit-rtl.c (init_derived_machine_modes): Likewise.
1488 (init_emit_once): Likewise.
1489 * explow.c (hard_function_value): Likewise.
1490 * expmed.c (extract_fixed_bit_field_1): Likewise.
1491 (extract_bit_field_1): Likewise.
1492 (expand_divmod): Likewise.
1493 (emit_store_flag_1): Likewise.
1494 * expr.c (init_expr_target): Likewise.
1495 (convert_move): Likewise.
1496 (alignment_for_piecewise_move): Likewise.
1497 (widest_int_mode_for_size): Likewise.
1498 (emit_block_move_via_movmem): Likewise.
1499 (copy_blkmode_to_reg): Likewise.
1500 (set_storage_via_setmem): Likewise.
1501 (compress_float_constant): Likewise.
1502 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1503 * optabs-query.c (get_best_extraction_insn): Likewise.
1504 * optabs.c (expand_binop): Likewise.
1505 (expand_twoval_unop): Likewise.
1506 (expand_twoval_binop): Likewise.
1507 (widen_leading): Likewise.
1508 (widen_bswap): Likewise.
1509 (expand_parity): Likewise.
1510 (expand_unop): Likewise.
1511 (prepare_cmp_insn): Likewise.
1512 (prepare_float_lib_cmp): Likewise.
1513 (expand_float): Likewise.
1514 (expand_fix): Likewise.
1515 (expand_sfix_optab): Likewise.
1516 * postreload.c (move2add_use_add2_insn): Likewise.
1517 * reg-stack.c (reg_to_stack): Likewise.
1518 * reginfo.c (choose_hard_reg_mode): Likewise.
1519 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1520 * stor-layout.c (mode_for_size): Likewise.
1521 (smallest_mode_for_size): Likewise.
1522 (mode_for_vector): Likewise.
1523 (finish_bitfield_representative): Likewise.
1524 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1525 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1526 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1527 * var-tracking.c (prepare_call_arguments): Likewise.
1529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1530 Alan Hayward <alan.hayward@arm.com>
1531 David Sherwood <david.sherwood@arm.com>
1533 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1534 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1535 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1536 * machmode.h (mode_size): Move earlier in file.
1537 (mode_precision): Likewise.
1538 (mode_inner): Likewise.
1539 (mode_nunits): Likewise.
1540 (mode_unit_size): Likewise.
1541 (unit_unit_precision): Likewise.
1542 (mode_wider): Likewise.
1543 (mode_2xwider): Likewise.
1544 (machine_mode): New class.
1545 (mode_to_bytes): New function.
1546 (mode_to_bits): Likewise.
1547 (mode_to_precision): Likewise.
1548 (mode_to_inner): Likewise.
1549 (mode_to_unit_size): Likewise.
1550 (mode_to_unit_precision): Likewise.
1551 (mode_to_nunits): Likewise.
1552 (GET_MODE_SIZE): Use mode_to_bytes.
1553 (GET_MODE_BITSIZE): Use mode_to_bits.
1554 (GET_MODE_PRECISION): Use mode_to_precision.
1555 (GET_MODE_INNER): Use mode_to_inner.
1556 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1557 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1558 (GET_MODE_NUNITS): Use mode_to_nunits.
1559 * system.h (ALWAYS_INLINE): New macro.
1560 * config/powerpcspe/powerpcspe-c.c
1561 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1562 int for arg1_mode and arg2_mode.
1564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1565 Alan Hayward <alan.hayward@arm.com>
1566 David Sherwood <david.sherwood@arm.com>
1568 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1569 Prefix mode names with E_ in case statements.
1570 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1571 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1572 (aarch64_split_simd_move): Likewise.
1573 (aarch64_gen_storewb_pair): Likewise.
1574 (aarch64_gen_loadwb_pair): Likewise.
1575 (aarch64_gen_store_pair): Likewise.
1576 (aarch64_gen_load_pair): Likewise.
1577 (aarch64_get_condition_code_1): Likewise.
1578 (aarch64_constant_pool_reload_icode): Likewise.
1579 (get_rsqrte_type): Likewise.
1580 (get_rsqrts_type): Likewise.
1581 (get_recpe_type): Likewise.
1582 (get_recps_type): Likewise.
1583 (aarch64_gimplify_va_arg_expr): Likewise.
1584 (aarch64_simd_container_mode): Likewise.
1585 (aarch64_emit_load_exclusive): Likewise.
1586 (aarch64_emit_store_exclusive): Likewise.
1587 (aarch64_expand_compare_and_swap): Likewise.
1588 (aarch64_gen_atomic_cas): Likewise.
1589 (aarch64_emit_bic): Likewise.
1590 (aarch64_emit_atomic_swap): Likewise.
1591 (aarch64_emit_atomic_load_op): Likewise.
1592 (aarch64_evpc_trn): Likewise.
1593 (aarch64_evpc_uzp): Likewise.
1594 (aarch64_evpc_zip): Likewise.
1595 (aarch64_evpc_ext): Likewise.
1596 (aarch64_evpc_rev): Likewise.
1597 (aarch64_evpc_dup): Likewise.
1598 (aarch64_gen_ccmp_first): Likewise.
1599 (aarch64_gen_ccmp_next): Likewise.
1600 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1601 (alpha_emit_xfloating_libcall): Likewise.
1602 (emit_insxl): Likewise.
1603 (alpha_arg_type): Likewise.
1604 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1605 (arc_preferred_simd_mode): Likewise.
1606 (arc_secondary_reload): Likewise.
1607 (get_arc_condition_code): Likewise.
1608 (arc_print_operand): Likewise.
1609 (arc_legitimate_constant_p): Likewise.
1610 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1611 * config/arc/arc.md (casesi_load): Likewise.
1612 (casesi_compact_jump): Likewise.
1613 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1614 (cc_use_register): Likewise.
1615 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1616 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1617 (arm_init_iwmmxt_builtins): Likewise.
1618 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1619 (neon_expand_vector_init): Likewise.
1620 (arm_attr_length_move_neon): Likewise.
1621 (maybe_get_arm_condition_code): Likewise.
1622 (arm_emit_vector_const): Likewise.
1623 (arm_preferred_simd_mode): Likewise.
1624 (arm_output_iwmmxt_tinsr): Likewise.
1625 (thumb1_output_casesi): Likewise.
1626 (thumb2_output_casesi): Likewise.
1627 (arm_emit_load_exclusive): Likewise.
1628 (arm_emit_store_exclusive): Likewise.
1629 (arm_expand_compare_and_swap): Likewise.
1630 (arm_evpc_neon_vuzp): Likewise.
1631 (arm_evpc_neon_vzip): Likewise.
1632 (arm_evpc_neon_vrev): Likewise.
1633 (arm_evpc_neon_vtrn): Likewise.
1634 (arm_evpc_neon_vext): Likewise.
1635 (arm_validize_comparison): Likewise.
1636 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1637 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1638 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1639 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1640 (c6x_preferred_simd_mode): Likewise.
1641 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1642 (epiphany_rtx_costs): Likewise.
1643 * config/epiphany/predicates.md (proper_comparison_operator):
1645 * config/frv/frv.c (condexec_memory_operand): Likewise.
1646 (frv_emit_move): Likewise.
1647 (output_move_single): Likewise.
1648 (output_condmove_single): Likewise.
1649 (frv_hard_regno_mode_ok): Likewise.
1650 (frv_matching_accg_mode): Likewise.
1651 * config/h8300/h8300.c (split_adds_subs): Likewise.
1652 (h8300_rtx_costs): Likewise.
1653 (h8300_print_operand): Likewise.
1654 (compute_mov_length): Likewise.
1655 (output_logical_op): Likewise.
1656 (compute_logical_op_length): Likewise.
1657 (compute_logical_op_cc): Likewise.
1658 (h8300_shift_needs_scratch_p): Likewise.
1659 (output_a_shift): Likewise.
1660 (compute_a_shift_length): Likewise.
1661 (compute_a_shift_cc): Likewise.
1662 (expand_a_rotate): Likewise.
1663 (output_a_rotate): Likewise.
1664 * config/i386/i386.c (classify_argument): Likewise.
1665 (function_arg_advance_32): Likewise.
1666 (function_arg_32): Likewise.
1667 (function_arg_64): Likewise.
1668 (function_value_64): Likewise.
1669 (ix86_gimplify_va_arg): Likewise.
1670 (ix86_legitimate_constant_p): Likewise.
1671 (put_condition_code): Likewise.
1672 (split_double_mode): Likewise.
1673 (ix86_avx256_split_vector_move_misalign): Likewise.
1674 (ix86_expand_vector_logical_operator): Likewise.
1675 (ix86_split_idivmod): Likewise.
1676 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1677 (ix86_build_const_vector): Likewise.
1678 (ix86_build_signbit_mask): Likewise.
1679 (ix86_match_ccmode): Likewise.
1680 (ix86_cc_modes_compatible): Likewise.
1681 (ix86_expand_branch): Likewise.
1682 (ix86_expand_sse_cmp): Likewise.
1683 (ix86_expand_sse_movcc): Likewise.
1684 (ix86_expand_int_sse_cmp): Likewise.
1685 (ix86_expand_vec_perm_vpermi2): Likewise.
1686 (ix86_expand_vec_perm): Likewise.
1687 (ix86_expand_sse_unpack): Likewise.
1688 (ix86_expand_int_addcc): Likewise.
1689 (ix86_split_to_parts): Likewise.
1690 (ix86_vectorize_builtin_gather): Likewise.
1691 (ix86_vectorize_builtin_scatter): Likewise.
1692 (avx_vpermilp_parallel): Likewise.
1693 (inline_memory_move_cost): Likewise.
1694 (ix86_tieable_integer_mode_p): Likewise.
1695 (x86_maybe_negate_const_int): Likewise.
1696 (ix86_expand_vector_init_duplicate): Likewise.
1697 (ix86_expand_vector_init_one_nonzero): Likewise.
1698 (ix86_expand_vector_init_one_var): Likewise.
1699 (ix86_expand_vector_init_concat): Likewise.
1700 (ix86_expand_vector_init_interleave): Likewise.
1701 (ix86_expand_vector_init_general): Likewise.
1702 (ix86_expand_vector_set): Likewise.
1703 (ix86_expand_vector_extract): Likewise.
1704 (emit_reduc_half): Likewise.
1705 (ix86_emit_i387_round): Likewise.
1706 (ix86_mangle_type): Likewise.
1707 (ix86_expand_round_sse4): Likewise.
1708 (expand_vec_perm_blend): Likewise.
1709 (canonicalize_vector_int_perm): Likewise.
1710 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1711 (expand_vec_perm_1): Likewise.
1712 (expand_vec_perm_interleave3): Likewise.
1713 (expand_vec_perm_even_odd_pack): Likewise.
1714 (expand_vec_perm_even_odd_1): Likewise.
1715 (expand_vec_perm_broadcast_1): Likewise.
1716 (ix86_vectorize_vec_perm_const_ok): Likewise.
1717 (ix86_expand_vecop_qihi): Likewise.
1718 (ix86_expand_mul_widen_hilo): Likewise.
1719 (ix86_expand_sse2_abs): Likewise.
1720 (ix86_expand_pextr): Likewise.
1721 (ix86_expand_pinsr): Likewise.
1722 (ix86_preferred_simd_mode): Likewise.
1723 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1724 * config/i386/sse.md (*andnot<mode>3): Likewise.
1725 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1726 (*<code><mode>3): Likewise.
1727 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1728 (ia64_expand_atomic_op): Likewise.
1729 (ia64_arg_type): Likewise.
1730 (ia64_mode_to_int): Likewise.
1731 (ia64_scalar_mode_supported_p): Likewise.
1732 (ia64_vector_mode_supported_p): Likewise.
1733 (expand_vec_perm_broadcast): Likewise.
1734 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1735 (iq2000_function_arg_advance): Likewise.
1736 (iq2000_function_arg): Likewise.
1737 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1738 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1739 (m68k_libcall_value): Likewise.
1740 (m68k_function_value): Likewise.
1741 (sched_attr_op_type): Likewise.
1742 * config/mcore/mcore.c (mcore_output_move): Likewise.
1743 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1745 (microblaze_function_arg): Likewise.
1746 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1747 (mips_print_operand): Likewise.
1748 (mips_mode_ok_for_mov_fmt_p): Likewise.
1749 (mips_vector_mode_supported_p): Likewise.
1750 (mips_preferred_simd_mode): Likewise.
1751 (mips_expand_vpc_loongson_even_odd): Likewise.
1752 (mips_expand_vec_unpack): Likewise.
1753 (mips_expand_vi_broadcast): Likewise.
1754 (mips_expand_vector_init): Likewise.
1755 (mips_expand_vec_reduc): Likewise.
1756 (mips_expand_msa_cmp): Likewise.
1757 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1758 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1759 (cc_flags_for_mode): Likewise.
1760 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1761 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1762 (nds32_output_casesi_pc_relative): Likewise.
1763 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1764 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1765 (nvptx_gen_unpack): Likewise.
1766 (nvptx_gen_pack): Likewise.
1767 (nvptx_gen_shuffle): Likewise.
1768 (nvptx_gen_wcast): Likewise.
1769 (nvptx_preferred_simd_mode): Likewise.
1770 * config/pa/pa.c (pa_secondary_reload): Likewise.
1771 * config/pa/predicates.md (base14_operand): Likewise.
1772 * config/powerpcspe/powerpcspe-c.c
1773 (altivec_resolve_overloaded_builtin): Likewise.
1774 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1776 (rs6000_preferred_simd_mode): Likewise.
1777 (output_vec_const_move): Likewise.
1778 (rs6000_expand_vector_extract): Likewise.
1779 (rs6000_split_vec_extract_var): Likewise.
1780 (reg_offset_addressing_ok_p): Likewise.
1781 (rs6000_legitimate_offset_address_p): Likewise.
1782 (rs6000_legitimize_address): Likewise.
1783 (rs6000_emit_set_const): Likewise.
1784 (rs6000_const_vec): Likewise.
1785 (rs6000_emit_move): Likewise.
1786 (spe_build_register_parallel): Likewise.
1787 (rs6000_darwin64_record_arg_recurse): Likewise.
1788 (swap_selector_for_mode): Likewise.
1789 (spe_init_builtins): Likewise.
1790 (paired_init_builtins): Likewise.
1791 (altivec_init_builtins): Likewise.
1792 (do_load_for_compare): Likewise.
1793 (rs6000_generate_compare): Likewise.
1794 (rs6000_expand_float128_convert): Likewise.
1795 (emit_load_locked): Likewise.
1796 (emit_store_conditional): Likewise.
1797 (rs6000_output_function_epilogue): Likewise.
1798 (rs6000_handle_altivec_attribute): Likewise.
1799 (rs6000_function_value): Likewise.
1800 (emit_fusion_gpr_load): Likewise.
1801 (emit_fusion_p9_load): Likewise.
1802 (emit_fusion_p9_store): Likewise.
1803 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1804 (fusion_gpr_mem_load): Likewise.
1805 (fusion_addis_mem_combo_load): Likewise.
1806 (fusion_addis_mem_combo_store): Likewise.
1807 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1808 (fusion_gpr_mem_load): Likewise.
1809 (fusion_addis_mem_combo_load): Likewise.
1810 (fusion_addis_mem_combo_store): Likewise.
1811 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1813 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1814 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1815 (rs6000_preferred_simd_mode): Likewise.
1816 (output_vec_const_move): Likewise.
1817 (rs6000_expand_vector_extract): Likewise.
1818 (rs6000_split_vec_extract_var): Likewise.
1819 (reg_offset_addressing_ok_p): Likewise.
1820 (rs6000_legitimate_offset_address_p): Likewise.
1821 (rs6000_legitimize_address): Likewise.
1822 (rs6000_emit_set_const): Likewise.
1823 (rs6000_const_vec): Likewise.
1824 (rs6000_emit_move): Likewise.
1825 (rs6000_darwin64_record_arg_recurse): Likewise.
1826 (swap_selector_for_mode): Likewise.
1827 (paired_init_builtins): Likewise.
1828 (altivec_init_builtins): Likewise.
1829 (rs6000_expand_float128_convert): Likewise.
1830 (emit_load_locked): Likewise.
1831 (emit_store_conditional): Likewise.
1832 (rs6000_output_function_epilogue): Likewise.
1833 (rs6000_handle_altivec_attribute): Likewise.
1834 (rs6000_function_value): Likewise.
1835 (emit_fusion_gpr_load): Likewise.
1836 (emit_fusion_p9_load): Likewise.
1837 (emit_fusion_p9_store): Likewise.
1838 * config/rx/rx.c (rx_gen_move_template): Likewise.
1839 (flags_from_mode): Likewise.
1840 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1841 (s390_slb_comparison): Likewise.
1842 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1843 (s390_vector_mode_supported_p): Likewise.
1844 (s390_cc_modes_compatible): Likewise.
1845 (s390_match_ccmode_set): Likewise.
1846 (s390_canonicalize_comparison): Likewise.
1847 (s390_emit_compare_and_swap): Likewise.
1848 (s390_branch_condition_mask): Likewise.
1849 (s390_rtx_costs): Likewise.
1850 (s390_secondary_reload): Likewise.
1851 (__SECONDARY_RELOAD_CASE): Likewise.
1852 (s390_expand_cs): Likewise.
1853 (s390_preferred_simd_mode): Likewise.
1854 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1855 * config/sh/sh.c (sh_print_operand): Likewise.
1856 (dump_table): Likewise.
1857 (sh_secondary_reload): Likewise.
1858 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1859 * config/sh/sh.md (casesi_worker_1): Likewise.
1860 (casesi_worker_2): Likewise.
1861 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1862 (fcc_comparison_operator): Likewise.
1863 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1864 (emit_soft_tfmode_cvt): Likewise.
1865 (sparc_preferred_simd_mode): Likewise.
1866 (output_cbranch): Likewise.
1867 (sparc_print_operand): Likewise.
1868 (sparc_expand_vec_perm_bmask): Likewise.
1869 (vector_init_bshuffle): Likewise.
1870 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1871 (spu_vector_mode_supported_p): Likewise.
1872 (spu_expand_insv): Likewise.
1873 (spu_emit_branch_or_set): Likewise.
1874 (spu_handle_vector_attribute): Likewise.
1875 (spu_builtin_splats): Likewise.
1876 (spu_builtin_extract): Likewise.
1877 (spu_builtin_promote): Likewise.
1878 (spu_expand_sign_extend): Likewise.
1879 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1880 (tilegx_simd_int): Likewise.
1881 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1882 (tilepro_simd_int): Likewise.
1883 * config/v850/v850.c (const_double_split): Likewise.
1884 (v850_print_operand): Likewise.
1885 (ep_memory_offset): Likewise.
1886 * config/vax/vax.c (vax_rtx_costs): Likewise.
1887 (vax_output_int_move): Likewise.
1888 (vax_output_int_add): Likewise.
1889 (vax_output_int_subtract): Likewise.
1890 * config/visium/predicates.md (visium_branch_operator): Likewise.
1891 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1892 (visium_print_operand_address): Likewise.
1893 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1894 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1895 (xtensa_expand_conditional_branch): Likewise.
1896 (xtensa_copy_incoming_a7): Likewise.
1897 (xtensa_output_literal): Likewise.
1898 * dfp.c (decimal_real_maxval): Likewise.
1899 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1902 Alan Hayward <alan.hayward@arm.com>
1903 David Sherwood <david.sherwood@arm.com>
1905 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1906 (mode_nunits_inline): Likewise.
1907 (mode_inner_inline): Likewise.
1908 (mode_unit_size_inline): Likewise.
1909 (mode_unit_precision_inline): Likewise.
1910 (emit_insn_modes_h): Likewise. Also emit a #define of the
1912 (emit_mode_wider): Add an E_ prefix to mode names.
1913 (emit_mode_complex): Likewise.
1914 (emit_mode_inner): Likewise.
1915 (emit_mode_adjustments): Likewise.
1916 (emit_mode_int_n): Likewise.
1917 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1918 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1919 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1920 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1921 (CRC32_BUILTIN, ENTRY): Likewise.
1922 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1923 (aarch64_pop_regs): Likewise.
1924 (aarch64_process_components): Likewise.
1925 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1926 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1927 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1928 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1929 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1930 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1932 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1933 (pcmpistr): Likewise.
1934 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1935 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1936 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1938 * config/rl78/rl78.c (mduc_regs): Likewise.
1939 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1940 (htm_expand_builtin): Likewise.
1941 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1942 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1943 * config/xtensa/xtensa.c (print_operand): Likewise.
1944 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1945 (NUM_MODE_VECTOR_INT): Likewise.
1946 * genoutput.c (null_operand): Likewise.
1947 (output_operand_data): Likewise.
1948 * genrecog.c (print_parameter_value): Likewise.
1949 * lra.c (debug_operand_data): Likewise.
1951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1953 * dwarf2out.c (scompare_loc_descriptor_wide)
1954 (scompare_loc_descriptor_narrow): New functions, split out from...
1955 (scompare_loc_descriptor): ...here.
1956 * expmed.c (emit_store_flag_int): New function, split out from...
1957 (emit_store_flag): ...here.
1959 2017-08-30 Richard Biener <rguenther@suse.de>
1961 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1962 (dwarf2out_early_finish): Move setting of AT_pubnames from
1963 early debug output to early finish.
1965 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1967 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1968 and -mdata-region to the assembler.
1970 2017-08-30 Richard Biener <rguenther@suse.de>
1972 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1974 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1975 (dwarf2out_early_global_decl): Only generate a DIE for the
1976 abstract origin if it doesn't already exist or is a declaration DIE.
1977 (resolve_addr): Do not add the linkage name twice when
1978 generating a stub DIE for the DW_TAG_GNU_call_site target.
1980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1982 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1983 Use machine_mode rather than int for arg1_mode.
1985 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1988 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1989 that the second argument of the built-in functions to unpack
1990 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1991 switch statement instead a lot of if statements.
1992 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1993 Allow 64-bit values to be in Altivec registers as well as
1994 traditional floating point registers.
1995 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1997 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1999 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2000 MAX_REGS_PER_ADDRESS == 1.
2002 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2004 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2005 * config/i386/i386.c (ix86_option_override_internal): Simplify
2006 setting of opts->x_flag_entry.
2008 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2009 Jakub Jelinek <jakub@redhat.com>
2010 Richard Biener <rguenther@suse.de>
2012 PR tree-optimization/81503
2013 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2014 folded constant fits in the target type; reorder tests for clarity.
2016 2017-08-29 Martin Liska <mliska@suse.cz>
2018 * passes.def: Include pass_lower_switch.
2019 * stmt.c (dump_case_nodes): Remove and move to
2020 tree-switch-conversion.
2021 (case_values_threshold): Likewise.
2022 (expand_switch_as_decision_tree_p): Likewise.
2023 (emit_case_decision_tree): Likewise.
2024 (expand_case): Likewise.
2025 (balance_case_nodes): Likewise.
2026 (node_has_low_bound): Likewise.
2027 (node_has_high_bound): Likewise.
2028 (node_is_bounded): Likewise.
2029 (emit_case_nodes): Likewise.
2030 (struct simple_case_node): New struct.
2031 (add_case_node): Remove.
2032 (emit_case_dispatch_table): Use vector instead of case_list.
2033 (reset_out_edges_aux): Remove.
2034 (compute_cases_per_edge): Likewise.
2035 (expand_case): Build list of simple_case_node.
2036 (expand_sjlj_dispatch_table): Use it.
2037 * tree-switch-conversion.c (struct case_node): Moved from
2038 stmt.c and adjusted.
2039 (emit_case_nodes): Likewise.
2040 (node_has_low_bound): Likewise.
2041 (node_has_high_bound): Likewise.
2042 (node_is_bounded): Likewise.
2043 (case_values_threshold): Likewise.
2044 (reset_out_edges_aux): Likewise.
2045 (compute_cases_per_edge): Likewise.
2046 (add_case_node): Likewise.
2047 (dump_case_nodes): Likewise.
2048 (balance_case_nodes): Likewise.
2049 (expand_switch_as_decision_tree_p): Likewise.
2050 (emit_jump): Likewise.
2051 (emit_case_decision_tree): Likewise.
2052 (try_switch_expansion): Likewise.
2053 (do_jump_if_equal): Likewise.
2054 (emit_cmp_and_jump_insns): Likewise.
2055 (fix_phi_operands_for_edge): New function.
2056 (record_phi_operand_mapping): Likewise.
2057 (class pass_lower_switch): New pass.
2058 (pass_lower_switch::execute): New function.
2059 (make_pass_lower_switch): Likewise.
2060 (conditional_probability):
2061 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2062 * tree-pass.h: Add make_pass_lower_switch.
2064 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2067 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2070 2017-08-29 Richard Biener <rguenther@suse.de>
2072 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2073 we do not add a DW_AT_inline attribute twice.
2074 (gen_subprogram_die): Remove code setting DW_AT_inline on
2075 DECL_ABSTRACT_P nodes.
2077 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2079 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2080 calls to internal functions.
2081 (gimplify_modify_expr): Likewise.
2082 * tree-call-cdce.c (use_internal_fn): Likewise.
2083 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2084 (convert_to_divmod): Set the nothrow flag.
2085 * tree-if-conv.c (predicate_mem_writes): Likewise.
2086 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2087 (vectorizable_call): Likewise.
2088 (vectorizable_store): Likewise.
2089 (vectorizable_load): Likewise.
2090 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2091 (vect_recog_mask_conversion_pattern): Likewise.
2093 2017-08-29 Martin Liska <mliska@suse.cz>
2096 * gcc.c (driver_handle_option): Add new argument.
2097 * opts-common.c (handle_option): Pass
2098 target_option_override_hook.
2099 * opts-global.c (lang_handle_option): Add new option.
2100 (set_default_handlers): Add new argument.
2101 (decode_options): Likewise.
2102 * opts.c (target_handle_option): Likewise.
2103 (common_handle_option): Call target_option_override_hook.
2104 * opts.h (struct cl_option_handler_func): Add hook for
2105 target option override.
2106 (struct cl_option_handlers): Likewise.
2107 (set_default_handlers): Add new argument.
2108 (decode_options): Likewise.
2109 (common_handle_option): Likewise.
2110 (target_handle_option): Likewise.
2111 * toplev.c (toplev::main): Pass targetm.target_option.override
2114 2017-08-29 Richard Biener <rguenther@suse.de>
2115 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2117 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2118 life to the active subtree.
2120 2017-08-28 Jeff Law <law@redhat.com>
2122 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2123 derive_equivalences.
2124 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2126 (edge_info::derive_equivalences): New private member function
2128 * tree-ssa-dom.c (class edge_info): Changed from a struct
2129 to a class. Add ctor/dtor, methods and data members.
2130 (edge_info::edge_info): Renamed from allocate_edge_info.
2131 Initialize additional members.
2132 (edge_info::~edge_info): New.
2133 (free_dom_edge_info): Delete the edge info.
2134 (record_edge_info): Use new class & associated member functions.
2135 Tighten forms for testing for edge equivalences.
2136 (record_temporary_equivalences): Iterate over the simple
2137 equivalences rather than assuming there's only one per edge.
2138 (cprop_into_successor_phis): Iterate over the simple
2139 equivalences rather than assuming there's only one per edge.
2140 (optimize_stmt): Use operand_equal_p rather than pointer
2141 equality for mini-DSE code.
2143 2017-08-28 Nathan Sidwell <nathan@acm.org>
2145 * gcc.c (execute): Fold SIGPIPE handling into switch
2146 statement. Adjust internal error message.
2148 2017-08-28 Richard Biener <rguenther@suse.de>
2151 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2152 Do nothing for removed DIEs.
2154 2017-08-28 Richard Biener <rguenther@suse.de>
2156 PR tree-optimization/81977
2157 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2160 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2163 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2164 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2165 using targetm.gen_mem_thread_fence.
2167 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2170 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2171 predicate to register_operand. Reorder operands.
2172 (*btr<mode>): Ditto.
2173 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2174 (*btr<mode>_mask): Ditto.
2176 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2178 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2179 * config/rs6000/xmmintrin.h: New file.
2180 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2182 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2185 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2186 parameter and_insn and return it.
2187 (recombine_lvx_pattern): Insert a copy to ensure availability of
2188 the base register of the copied masking operation at the point of
2189 the instruction replacement.
2190 (recombine_stvx_pattern): Likewise.
2192 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2194 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2195 undocumented switches.
2196 (-mpower9-dform-vector): Likewise.
2197 (-mpower9-dform): Likewise.
2198 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2199 comments to delete references to -mpower9-dform* switches.
2200 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2201 Delete reference to -mpower9-dform* switches, test for
2202 -mpower9-vector instead.
2203 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2204 (OTHER_P9_VECTOR_MASKS): Likewise.
2205 (POWERPC_MASKS): Likewise.
2206 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2207 tests against -mpower9-dform* to -mpower9-vector. Delete code
2208 that checked for -mpower9-dform* consistancy with other options.
2209 Add test for -mpower9-misc to enable other power9 switches.
2210 (rs6000_init_hard_regno_mode_ok): Likewise.
2211 (rs6000_option_override_internal): Likewise.
2212 (rs6000_emit_prologue): Likewise.
2213 (rs6000_emit_epilogue): Likewise.
2214 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2215 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2216 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2218 (emit_fusion_p9_store): Likewise.
2219 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2220 resetting these macros if the assembler does not support ISA 3.0
2222 (TARGET_P9_DFORM_VECTOR): Likewise.
2223 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2224 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2226 2017-08-25 Alan Modra <amodra@gmail.com>
2228 PR rtl-optimization/81747
2229 * cse.c (cse_extended_basic_block): Don't attempt to record
2230 equivalences for degenerate conditional branches that branch
2231 to their fall-through.
2233 2017-08-24 Martin Sebor <msebor@redhat.com>
2236 * gimple-fold.c (size_must_be_zero_p): New function.
2237 (gimple_fold_builtin_memory_op): Call it.
2239 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2241 * config/rs6000/mm_malloc.h: New file.
2243 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2245 PR tree-optimization/81913
2246 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2247 analysis when either IVs in condition can wrap.
2249 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2251 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2252 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2254 2017-08-24 Richard Biener <rguenther@suse.de>
2257 * targhooks.c (default_target_can_inline_p): Properly
2258 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2259 is present and always compare.
2260 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2261 infer -mfpmath=sse from TARGET_SSE_P.
2262 (ix86_can_inline_p): Properly use target_option_default_node when
2263 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2265 2017-08-24 Richard Biener <rguenther@suse.de>
2268 * dwarf2out.c (output_die): Handle flag_generate_offload like
2270 (output_comp_unit): Likewise.
2271 (gen_array_type_die): Likewise.
2272 (dwarf2out_early_finish): Likewise.
2273 (note_variable_value_in_expr): Likewise.
2274 (dwarf2out_finish): Likewise. Adjust assert.
2275 * cgraphunit.c (symbol_table::compile): Move setting of
2276 flag_generate_offload earlier ...
2277 (symbol_table::finalize_compilation_unit): ... here, before
2278 early debug finalization.
2280 2017-08-24 Richard Biener <rguenther@suse.de>
2282 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2283 and ipa-fnsummary.h.
2284 (ix86_can_inline_p): When ix86_fpmath flags do not match
2285 check whether the callee uses FP math at all.
2287 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2290 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2291 instead of TYPE_PRECISION.
2293 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2299 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2301 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2303 (ix86_frame::stack_realign_allocate): New field.
2304 (struct machine_frame_state): Modify comments.
2305 (machine_frame_state::sp_realigned_fp_end): New field.
2306 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2308 (sp_valid_at): Add assertion to assure no attempt to access invalid
2309 offset of a realigned stack.
2310 (fp_valid_at): Likewise.
2311 (choose_baseaddr): Modify comments.
2312 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2313 ix86_expand_prologue.
2314 (ix86_expand_prologue): Modify stack realignment and allocation.
2315 (ix86_expand_epilogue): Modify comments.
2316 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2317 avx2_runtime, avx512f, and avx512f_runtime.
2319 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2321 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2322 (mstackrealign): Do not init to -1.
2323 * config/i386/i386.c (ix86_option_override_internal):
2324 Check opts_set, not opts when setting default value of
2325 opts->x_ix86_force_align_arg_pointer.
2327 2017-08-23 Richard Biener <rguenther@suse.de>
2329 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2330 lang_hooks.decl_printable_name.
2331 * print-rtl-function.c (print_rtx_function): Likewise.
2332 * tree-pretty-print.c (dump_function_header): Likewise.
2334 2017-08-23 Richard Biener <rguenther@suse.de>
2337 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2338 -g0 at compile-time.
2340 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2343 * doc/sourcebuild.texi (Other hardware attributes):
2346 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2349 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2350 Fix single-use check.
2352 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2354 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2356 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2358 * doc/install.texi: Modify to add more details on running selected
2361 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2363 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2364 is combined with -mabi=ms.
2365 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2368 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2370 PR tree-optimization/81488
2371 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2372 and cached_basis fields.
2373 (MAX_SPREAD): New constant.
2374 (alloc_cand_and_find_basis): Initialize new fields.
2375 (clear_visited): New function.
2376 (create_phi_basis_1): Rename from create_phi_basis, set visited
2377 and cached_basis fields.
2378 (create_phi_basis): New wrapper function.
2379 (phi_add_costs_1): Rename from phi_add_costs, add spread
2380 parameter, set visited field, short-circuit when limits reached.
2381 (phi_add_costs): New wrapper function.
2382 (record_phi_increments_1): Rename from record_phi_increments, set
2384 (record_phi_increments): New wrapper function.
2385 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2386 (phi_incr_cost): New wrapper function.
2387 (all_phi_incrs_profitable_1): Rename from
2388 all_phi_incrs_profitable, set visited field.
2389 (all_phi_incrs_profitable): New wrapper function.
2391 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2392 Alan Hayward <alan.hayward@arm.com>
2393 David Sherwood <david.sherwood@arm.com>
2395 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2396 that takes the outer and inner modes.
2397 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2398 comparison as the canonical test for a paradoxical subreg.
2399 * combine.c (simplify_set): Use paradoxical_subreg_p.
2400 (make_extraction): Likewise.
2401 (force_to_mode): Likewise.
2402 (rtx_equal_for_field_assignment_p): Likewise.
2403 (gen_lowpart_for_combine): Likewise.
2404 (simplify_comparison): Likewise.
2405 * cse.c (equiv_constant): Likewise.
2406 * expmed.c (store_bit_field_1): Likewise.
2407 * final.c (alter_subreg): Likewise.
2408 * fwprop.c (propagate_rtx): Likewise.
2409 (forward_propagate_subreg): Likewise.
2410 * ira-conflicts.c (ira_build_conflicts): Likewise.
2411 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2412 * lra-constraints.c (curr_insn_transform): Likewise.
2413 (split_reg): Likewise.
2414 * lra-eliminations.c (move_plus_up): Likewise.
2415 (lra_eliminate_regs_1): Likewise.
2416 * recog.c (general_operand): Likewise.
2417 * ree.c (combine_reaching_defs): Likewise.
2418 * reload.c (push_reload): Likewise.
2419 (find_reloads): Likewise.
2420 * reload1.c (elimination_effects): Likewise.
2421 (compute_reload_subreg_offset): Likewise.
2422 (choose_reload_regs): Likewise.
2423 * rtlanal.c (subreg_lsb_1): Likewise.
2424 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2425 (simplify_subreg): Likewise.
2426 * var-tracking.c (track_loc_p): Likewise.
2427 * emit-rtl.c (byte_lowpart_offset): Likewise.
2428 (paradoxical_subreg_p): Delete out-of-line definition.
2430 2017-08-22 Jeff Law <law@redhat.com>
2432 PR tree-optimization/81741
2433 PR tree-optimization/71947
2434 * tree-ssa-dom.c: Include tree-inline.h.
2435 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2436 equivalences if one is more expensive to compute than the other.
2437 * tree-ssa-scopedtables.h (class const_or_copies): Make
2438 record_const_or_copy_raw method private.
2439 (class avail_exprs_stack): New method simplify_binary_operation.
2440 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2441 avail_exprs_stack::simplify_binary_operation as needed.
2442 (avail_exprs_stack::simplify_binary_operation): New function.
2444 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2446 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2447 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2448 (DOT_SYMBOLS): Likewise.
2449 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2450 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2451 (RS6000_ABI_NAME): Likewise.
2452 (TARGET_CMODEL): Likewise.
2453 (TOC_SECTION_ASM_OP): Likewise.
2454 (SET_CMODEL): New macro.
2455 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2457 2017-08-22 Richard Biener <rguenther@suse.de>
2459 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2460 to allow for free-lang-data replacements similar to verify_type_variant.
2462 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2463 Alan Hayward <alan.hayward@arm.com>
2464 David Sherwood <david.sherwood@arm.com>
2466 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2467 VOIDmode for the LABEL_REF.
2469 2017-08-22 Richard Biener <rguenther@suse.de>
2471 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2473 2017-08-22 Richard Biener <rguenther@suse.de>
2475 * common.opt (feliminate-dwarf2-dups): Ignore.
2476 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2477 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2478 same_die_p_wrap, compute_section_prefix,
2479 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2480 (comdat_symbol_id, comdat_symbol_number): Likewise.
2481 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2483 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2484 (output_die): Mark unreachable path unreachable.
2485 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2486 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2487 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2488 (dwarf2out_early_finish): Likewise.
2490 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2492 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2494 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2497 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2498 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2499 (avr_attribute_table) <io, io_low, address>: Initialize
2500 .decl_required with true.
2502 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2504 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2505 undocumented debugging options.
2506 (-mvsx-scalar-double): Likewise.
2507 (-mallow-df-permute): Likewise.
2508 (-mvectorize-builtins): Likewise.
2509 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2510 (rs6000_builtin_vectorized_function): Likewise.
2511 (rs6000_builtin_md_vectorized_function): Likewise.
2512 (rs6000_opt_vars): Likewise.
2514 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2517 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2518 (*btrq_imm): Rename from *btrq.
2519 (*btcq_imm): Rename from *btcq.
2520 (btsc): New code attribute.
2521 (*<btsc><mode>): New insn pattern.
2522 (*btr<mode>): Ditto.
2523 (*<btsc><mode>_mask): New insn_and_split pattern.
2524 (*btr<mode>_mask): Ditto.
2526 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2527 Alan Hayward <alan.hayward@arm.com>
2528 David Sherwood <david.sherwood@arm.com>
2530 * function.c (pad_below): Simplify padding calculation.
2532 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2533 Alan Hayward <alan.hayward@arm.com>
2534 David Sherwood <david.sherwood@arm.com>
2536 * target.def (function_prologue): Remove frame size argument.
2537 (function_epilogue): Likewise.
2538 * doc/tm.texi: Regenerate.
2539 * final.c (final_start_function): Update call to function_prologue.
2540 (final_end_function): Update call to function_epilogue.
2541 (default_function_pro_epilogue): Remove frame size argument.
2542 * output.h (default_function_pro_epilogue): Likewise.
2543 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2544 (arm_output_function_prologue): Likewise.
2545 * config/frv/frv.c (frv_function_prologue): Likewise.
2546 (frv_function_epilogue): Likewise.
2547 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2548 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2549 (ia64_output_function_epilogue): Likewise.
2550 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2551 (m32r_output_function_epilogue): Likewise.
2552 * config/microblaze/microblaze.c (microblaze_function_prologue)
2553 (microblaze_function_epilogue): Likewise.
2554 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2555 (mips_output_function_epilogue): Likewise.
2556 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2557 (mmix_target_asm_function_epilogue): Likewise.
2558 * config/msp430/msp430.c (msp430_start_function): Likewise.
2559 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2560 (nds32_asm_function_epilogue): Likewise.
2561 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2562 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2563 (pa_output_function_epilogue): Likewise.
2564 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2565 (rs6000_output_function_epilogue): Likewise.
2566 * config/rl78/rl78.c (rl78_start_function): Likewise.
2567 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2568 (rs6000_output_function_epilogue): Likewise.
2569 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2570 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2571 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2572 (sparc_asm_function_epilogue): Likewise.
2574 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2576 * tree.h (type_has_mode_precision_p): New function.
2577 * convert.c (convert_to_integer_1): Use it.
2578 * expr.c (expand_expr_real_2): Likewise.
2579 (expand_expr_real_1): Likewise.
2580 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2581 * match.pd: Likewise.
2582 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2583 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2584 * tree-tailcall.c (process_assignment): Likewise.
2585 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2586 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2587 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2588 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2589 (vectorizable_assignment): Likewise.
2590 (vectorizable_shift): Likewise.
2591 (vectorizable_operation): Likewise.
2592 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2594 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2596 * match.pd: Add pow (C, x) simplification.
2598 2017-08-21 Richard Biener <rguenther@suse.de>
2600 PR tree-optimization/81900
2601 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2602 for blocks with abnormal predecessors.
2603 (compute_antic): Do not set visited flag prematurely.
2605 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2608 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2610 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2612 * stor-layout.h (vector_type_mode): Move to...
2613 * tree.h (vector_type_mode): ...here.
2614 * stor-layout.c (vector_type_mode): Move to...
2615 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2617 2017-08-21 Richard Biener <rguenther@suse.de>
2619 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2620 register_external_die hooks.
2621 (debug_false_tree_charstarstar_uhwistar): Declare.
2622 (debug_nothing_tree_charstar_uhwi): Likewise.
2623 * debug.c (do_nothing_debug_hooks): Adjust.
2624 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2625 (debug_nothing_tree_charstar_uhwi): Likewise.
2626 * dbxout.c (dbx_debug_hooks): Adjust.
2627 (xcoff_debug_hooks): Likewise.
2628 * sdbout.c (sdb_debug_hooks): Likewise.
2629 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2630 * dwarf2out.c (macinfo_label_base): New global.
2631 (dwarf2out_register_external_die): New function for the
2632 register_external_die hook.
2633 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2634 (dwarf2_debug_hooks): Use them.
2635 (dwarf2_lineno_debug_hooks): Adjust.
2636 (struct die_struct): Add with_offset flag.
2637 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2638 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2639 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2640 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2641 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2642 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2643 defining section names for the early LTO debug variants.
2644 (reset_indirect_string): New helper.
2645 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2646 (print_dw_val): Add support for offsetted symbol references.
2647 (get_ultimate_context): Split out from is_cxx.
2648 (is_cxx): Use get_ultimate_context.
2649 (is_fortran): Add decl overload.
2650 (compute_comp_unit_symbol): Split out worker from
2651 compute_section_prefix.
2652 (compute_section_prefix): Call compute_comp_unit_symbol and
2653 set comdat_type_p here.
2654 (output_die): Skip DIE symbol output for the LTO added one.
2655 Handle DIE symbol references with offset.
2656 (output_comp_unit): Guard section name mangling properly.
2657 For LTO debug sections emit a symbol at the section beginning
2658 which we use to refer to its DIEs.
2659 (add_abstract_origin_attribute): For DIEs registered via
2660 dwarf2out_register_external_die directly refer to the early
2661 DIE rather than indirectly through the shadow one we created.
2662 Remove obsolete call to dwarf2out_abstract_function for
2663 non-function/block origins.
2664 (gen_array_type_die): When generating early LTO debug do
2665 not emit DW_AT_string_length.
2666 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2667 late when in LTO. As suggested place a gcc_unreachable for
2668 the DECL_ABSTRACT_P case.
2669 (gen_subprogram_die): Avoid another specification DIE
2670 for early built declarations/definitions for the late LTO case.
2671 (gen_variable_die): Add type references for late duplicated VLA dies
2673 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2674 we have the abstract instance already.
2675 (process_scope_var): Adjust decl DIE contexts in LTO which
2676 first puts them in limbo.
2677 (gen_decl_die): Do not generate type DIEs late apart from
2678 types for VLAs or for decls we do not yet have a DIE. Do not
2679 call dwarf2out_abstract_function late.
2680 (dwarf2out_early_global_decl): Make sure to create DIEs
2681 for abstract instances of a decl first.
2682 (dwarf2out_late_global_decl): Adjust comment.
2683 (output_macinfo_op): With multiple macro sections use
2684 macinfo_label_base to distinguish labels.
2685 (output_macinfo): Likewise. Update macinfo_label_base.
2686 Pass in the line info label.
2687 (note_variable_value_in_expr): When generating LTO resolve
2688 all variable values here by generating DIEs as needed.
2689 (init_sections_and_labels): Add early LTO debug flag parameter
2690 and generate different sections and names if set. Add generation
2691 counter for the labels so we can have multiple of them.
2692 (reset_dies): Helper to allow DIEs to be output multiple times.
2693 (dwarf2out_finish): When outputting DIEs to the fat part of an
2694 LTO object first reset DIEs.
2695 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2696 (modified_type_die): Check for decl_ultimate_origin being self
2698 (gen_type_die_with_usage): Likewise.
2699 (gen_typedef_die): Allow decl_ultimate_origin being self.
2700 (set_decl_abstract_flags): Remove.
2701 (set_block_abstract_flags): Likewise.
2702 (dwarf2out_abstract_function): Treat the early generated DIEs
2703 as the abstract copy and only add DW_AT_inline and
2704 DW_AT_artificial here and call set_decl_origin_self.
2705 If the DIE has an abstract origin don't do anything.
2706 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2707 if we have none yet (Go fails to build one, PR78628).
2708 (variably_modified_type_p): Prevent endless recursion for Ada
2709 cyclic pointer types.
2710 * lto-streamer-in.c: Include debug.h.
2711 (dref_queue): New global.
2712 (lto_read_tree_1): Stream in DIE references.
2713 (lto_input_tree): Register DIE references.
2714 (input_function): Stream DECL_DEBUG_ARGS.
2715 * lto-streamer-out.c: Include debug.h.
2716 (lto_write_tree_1): Output DIE references.
2717 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2718 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2719 (output_function): Stream DECL_DEBUG_ARGS.
2720 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2721 Stream DECL_ABSTRACT_ORIGIN.
2722 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2723 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2724 DECL_CONTEXT for file-scope decls.
2725 * lto-streamer.h (struct dref_entry): Declare.
2726 (dref_queue): Likewise.
2727 * cfgexpand.c (pass_expand::execute): Do not call the
2728 outlining_inline_function hook here.
2729 * lto-wrapper.c (debug_obj): New global.
2730 (tool_cleanup): Unlink it if required.
2731 (debug_objcopy): New function.
2732 (run_gcc): Handle early debug sections in the IL files by
2733 extracting them to separate files, partially linkin them and
2734 feeding the result back as result to the linker.
2735 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2736 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2737 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2738 sections into a separate segment.
2739 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2741 (darwin_asm_dwarf_section): Likewise.
2742 (darwin_asm_output_dwarf_offset): Likewise.
2743 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2745 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2746 Alan Hayward <alan.hayward@arm.com>
2747 David Sherwood <david.sherwood@arm.com>
2749 * read-md.h (md_reader::record_potential_iterator_use): Replace
2750 pointer argument with an rtx and an index.
2751 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2752 (apply_mode_iterator): Likewise.
2753 (apply_code_iterator): Likewise.
2754 (apply_int_iterator): Likewise.
2755 (apply_subst_iterator): Likewise.
2756 (record_iterator_use): Likewise.
2757 (record_attribute_use): Likewise.
2758 (md_reader::record_potential_iterator_use): Likewise. Update calls
2759 to record_iterator_use and apply_iterator.
2760 (iterator_use): Replace ptr with x and index.
2761 (attribute_use): Likewise.
2762 (apply_attribute_uses): Update calls to apply_iterator.
2763 (apply_iterators): Likewise. Update initialization of iterator_use.
2764 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2765 and record_potential_iterator_use.
2766 (rtx_reader::read_rtx_operand): Likewise.
2768 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2769 Alan Hayward <alan.hayward@arm.com>
2770 David Sherwood <david.sherwood@arm.com>
2772 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2775 2017-08-21 Richard Biener <rguenther@suse.de>
2778 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2779 at struct end conservatively when comparing common bases.
2781 2017-08-21 Richard Biener <rguenther@suse.de>
2783 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2784 (mem_ref_in_stmt): Remove.
2785 (determine_max_movement): Use ref index to get at the reference.
2786 (invariantness_dom_walker::before_dom_children): Deal with
2787 lim data already initialized.
2788 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2790 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2792 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2793 (TARGET_ISA_ROUND): Ditto.
2794 (TARGET_ROUND): Ditto.
2795 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2796 * config/i386/i386.md: Ditto.
2797 * config/i386/sse.md: Ditto.
2798 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2799 with OPTION_MASK_ISA_SSE4_1.
2801 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2804 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2805 __builtin_ia32_lzcnt_u16.
2807 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2810 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2811 (rs6000_set_current_function): Rewrite function to use it.
2813 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2816 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2817 and TYPE_WARN_IF_NOT_ALIGN.
2818 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2819 (handle_warn_if_not_align): New.
2820 (place_union_field): Call handle_warn_if_not_align.
2821 (place_field): Call handle_warn_if_not_align.
2822 Copy TYPE_WARN_IF_NOT_ALIGN.
2823 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2824 (layout_type): Likewise.
2825 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2827 (tree_decl_common): Add warn_if_not_align.
2828 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2829 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2830 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2831 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2832 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2833 * doc/extend.texi: Document warn_if_not_aligned attribute.
2834 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2836 2017-08-17 Martin Liska <mliska@suse.cz>
2839 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2840 (get_data_dependence): Use it as pointer type.
2841 (distribute_loop): Likewise.
2843 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2845 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2846 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2847 (p8_vmrgow_<mode>_direct): New define_insn.
2848 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2849 handle endianness for vmrgew and vmrgow permute patterns.
2851 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2853 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2854 * config/rs6000/rs6000-cpus.def: Remove comment.
2855 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2856 (POWERPC_MASKS): Likewise.
2857 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2858 use of TARGET_VSX_TIMODE.
2859 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2860 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2861 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2862 (rs6000_option_override_internal): Remove dead code.
2863 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2864 (rs6000_legitimize_reload_address): Likewise.
2865 (rs6000_legitimate_address_p): Likewise.
2866 (rs6000_opt_masks): Delete "vsx-timode".
2867 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2868 from function comment.
2869 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2870 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2871 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2873 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2874 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2875 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2878 (define_peephole2): Likewise.
2880 2017-08-17 Martin Sebor <msebor@redhat.com>
2883 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2884 past the end of an array.
2885 (test_pp_format): Add test cases.
2887 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2889 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2890 missing ECF_NOTHROW flags.
2892 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2895 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2896 operands residing in integer registers.
2897 (*vsx_le_perm_load_<mode>): Likewise.
2898 (*vsx_le_perm_store_<mode>): Likewise.
2899 (define_peephole2): Add peepholes to optimize the above.
2901 2017-08-17 Marek Polacek <polacek@redhat.com>
2904 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2905 to mimic what shorten_compare did. Change the return type to bool.
2906 (fold_cond_expr_with_comparison): Update call to
2907 operand_equal_for_comparison_p.
2908 (fold_ternary_loc): Likewise.
2910 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2912 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2914 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2916 2017-08-17 Richard Biener <rguenther@suse.de>
2918 * tree-ssa-structalias.c (solve_graph): When propagating
2919 to successors update the graphs succ edges and avoid duplicate work.
2921 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2924 * config/i386/i386.c (ix86_option_override_internal): Save target
2925 specific options after ix86_stack_protector_guard_reg was changed.
2927 2017-08-17 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/81827
2930 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2931 (new_var_info): Initialize it conservatively.
2932 (get_call_vi): Mark register vars.
2933 (new_scalar_tmp_constraint_exp): Likewise.
2934 (handle_rhs_call): Likewise.
2935 (handle_const_call): Likewise.
2936 (create_function_info_for): Likewise.
2937 (solve_constraints): Sort varinfos to separate register from
2938 non-register vars to pack points-to solution bitmaps during
2941 2017-08-17 Marek Polacek <polacek@redhat.com>
2943 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2945 2017-08-17 Richard Biener <rguenther@suse.de>
2947 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2948 to true when overflow is undefined and we saturated the result.
2950 2017-08-17 Alan Modra <amodra@gmail.com>
2953 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2954 Don't use store multiple if only one reg needs saving.
2955 (interesting_frame_related_regno): New function.
2956 (rs6000_frame_related): Don't emit frame info for regs that
2958 (rs6000_emit_epilogue): Likewise.
2960 2017-08-16 Nathan Sidwell <nathan@acm.org>
2962 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2963 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2964 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2965 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2966 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2967 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2968 (verify_type): Adjust for TYPE_BINFO move.
2969 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2970 process TYPE_BINFO directly.
2971 (hash_tree): Likewise.
2972 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2974 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2977 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2979 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2981 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2984 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2985 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2986 Add ix86_binary_operator_ok to insn constraint.
2987 (*iordi_1_bts): Ditto.
2988 (*xordi_1_btc): Ditto.
2989 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2990 Update corresponding peephole2 pattern.
2994 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2996 PR tree-optimization/81832
2997 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2998 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3000 2017-08-16 Marek Polacek <polacek@redhat.com>
3003 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3004 regarding size_zero_node.
3006 2017-08-16 Martin Liska <mliska@suse.cz>
3009 * config.gcc: Respect previously set extra_objs in case
3012 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3014 PR tree-optimization/81835
3015 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3016 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3017 not depend on the phi.
3019 2017-08-16 Alan Modra <amodra@gmail.com>
3021 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3024 2017-08-16 Alan Modra <amodra@gmail.com>
3026 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3027 (save_reg_p): ..into this. Update all callers.
3028 (first_reg_to_save): Simplify.
3030 2017-08-16 Alan Modra <amodra@gmail.com>
3032 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3035 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3039 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3040 constant count if that count is less than 32.
3042 2017-08-15 Nathan Sidwell <nathan@acm.org>
3044 * gcc.c (execute): Emit friendlier message if inferior is killed
3045 by an external cause.
3047 2017-08-15 Richard Biener <rguenther@suse.de>
3049 PR tree-optimization/81790
3050 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3051 CONSTRUCTORs from simplifying and VN.
3053 2017-08-14 Martin Sebor <msebor@redhat.com>
3055 * builtin-attrs.def: Add comments.
3057 2017-08-14 Martin Sebor <msebor@redhat.com>
3060 * doc/extend.texi (attribute nonstring): Document new attribute.
3062 2017-08-14 Martin Sebor <msebor@redhat.com>
3065 * tree-diagnostic.c (default_tree_printer): Handle %G.
3066 * gimple-pretty-print.h (percent_G_format): Declare new function.
3067 * gimple-pretty-print.c (percent_G_format): Define.
3068 * tree-pretty-print.c (percent_K_format): Add argument.
3070 2017-08-14 Martin Sebor <msebor@redhat.com>
3072 PR translation/79998
3073 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3074 Remove a stray space.
3076 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3079 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3080 (*iordi_1_bts): Ditto.
3081 (*xordi_1_btc): Ditto.
3083 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3086 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3087 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3089 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3090 quoted strings, and make more translator-friendly.
3091 (darwin_rs6000_override_options): Likewise.
3092 (rs6000_option_override_internal): Likewise.
3093 (rs6000_return_in_memory): Fix overlong line.
3094 (init_cmulative_args): Use quoted strings, and make more
3095 translator-friendly.
3096 (rs6000_pass_by_reference): Fix overlong line.
3097 (def_builtin): Use quoted strings.
3098 (altivec_expand_predicate_builtin): Use quoted strings, and make
3099 more translator-friendly.
3100 (htm_expand_builtin): Use quoted strings.
3101 (cpu_expand_builtin): Use quoted strings, and make more
3102 translator-friendly.
3103 (altivec_expand_builtin): Likewise.
3104 (paired_expand_predicate_builtin): Likewise.
3105 (rs6000_invalid_builtin): Likewise.
3106 (builtin_function_type): Use quoted strings.
3107 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3108 more translator-friendly.
3109 (rs6000_trampoline_init): Likewise.
3110 (rs6000_handle_altivec_attribute): Likewise.
3111 (rs6000_inner_target_options): Use quoted strings.
3112 (rs6000_disable_incompatible_switches): Likewise.
3113 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3114 strings, and make more translator-friendly.
3115 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3117 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3119 PR tree-optimization/81799
3120 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3121 cond_expr to simple gimple operand.
3123 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3126 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3128 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3132 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3133 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3134 TARGET_GASISR_PROLOGUES.
3135 * config/avr/avr.c (avr_option_override): Same.
3136 (avr_pass_pre_proep::execute): Same.
3138 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3141 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3142 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3144 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3146 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3147 define_insn_and_split. Split to a memory load from 0 in
3148 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3149 using PTR mode iterator.
3150 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3151 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3152 (*add_tp_<mode>): Redefine as define_insn_and_split.
3153 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3154 address space. Merge with *add_tp_x32 using PTR mode iterator.
3155 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3156 Split to an add with a memory load from 0 in
3157 DEFAULT_TLS_SEG_REG address space.
3159 2017-08-12 Andrew Pinski <apinski@cavium.com>
3161 * config/aarch64/aarch64-option-extensions.def (rdma):
3162 Fix feature string to what Linux prints out in /proc/cpuinfo.
3164 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3167 * dwarf2out.c (modified_type_die): For C typedef types that have
3168 an ultimate origin, process the ultimate origin instead of the
3170 (gen_typedef_die): Assert that input DECLs have no ultimate
3172 (gen_type_die_with_usage): For typedef variants that have an
3173 ultimate origin, just call gen_decl_die on the original DECL.
3174 (process_scope_var): Avoid creating DIEs for local typedefs and
3175 concrete static variables.
3177 2017-08-12 Alan Modra <amodra@gmail.com>
3181 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3182 match gnu-user.h startfile.
3183 (ENDFILE_LINUX_SPEC): Similarly.
3185 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3188 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3190 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3192 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3193 * config/aarch64/aarch64.md (mov<mode>): Change.
3194 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3195 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3196 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3198 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3200 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3201 for storage order barriers.
3203 2017-08-11 Martin Liska <mliska@suse.cz>
3206 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3207 variables of void type.
3209 2017-08-11 Martin Liska <mliska@suse.cz>
3211 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3212 TARGET_SUPPORTS_ALIASES.
3213 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3214 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3215 (optimize_weakref): Likewise.
3216 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3217 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3218 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3220 2017-08-11 Martin Liska <mliska@suse.cz>
3223 * config/i386/i386.c (make_resolver_func): Do complete
3224 refactoring of the function.
3226 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3229 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3230 * config/i386/i386.c (ix86_stack_protect_guard): Use
3231 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3232 * doc/invoke.texi (x86 Options): Document
3233 -mstack-protector-guard-symbol= option.
3235 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3237 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3238 * config/i386/i386.c (ix86_split_stack_guard): New function.
3239 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3240 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3241 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3242 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3243 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3244 (split_stack_space_check): Call ix86_split_stack_guard.
3246 2017-08-10 Martin Sebor <msebor@redhat.com>
3248 * print-tree.c (print_node): Print location using the established
3250 Replace spaces with colons.
3251 (debug_raw, debug): Ditto.
3253 2017-08-10 Martin Sebor <msebor@redhat.com>
3256 * pretty-print.c (pp_format): Correct the handling of %s precision.
3258 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3261 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3263 (ix86_finalize_stack_frame_flags): This. Also clear
3264 frame_pointer_needed if -fno-omit-frame-pointer is used without
3266 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3267 with ix86_finalize_stack_frame_flags.
3268 (ix86_expand_epilogue): Likewise.
3269 (ix86_expand_split_stack_prologue): Likewise.
3270 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3272 2017-08-10 Martin Liska <mliska@suse.cz>
3275 * c-attribs.c (handle_target_attribute):
3276 Report warning for an empty string argument of target attribute.
3278 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3281 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3283 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3284 or DECL_NONLOCAL labels.
3285 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3286 or DECL_NONLOCAL labels here.
3288 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3290 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3291 to indicate when early gimple folding has been disabled.
3292 (rs6000_gimple_fold_builtin): Add debug content.
3293 (rs6000_invalid_builtin): Fix whitespace.
3294 (rs6000_expand_builtin): Fix whitespace.
3295 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3297 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3300 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3301 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3303 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3305 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3306 (qdf24xx): Likewise.
3307 * config/aarch64/aarch64.md: Include falkor.md.
3308 * config/aarch64/falkor.md: New.
3310 2017-08-09 Marek Polacek <polacek@redhat.com>
3313 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3314 * diagnostic.c (emit_diagnostic): Add a comment.
3315 (emit_diagnostic_valist): New function.
3317 2017-08-09 Marek Polacek <polacek@redhat.com>
3320 * input.c (make_location): New overload.
3321 * input.h (make_location): Declare.
3323 2017-08-08 Alan Modra <amodra@gmail.com>
3324 H.J. Lu <hongjiu.lu@intel.com>
3327 * gcc.c (NO_PIE_SPEC): Delete.
3328 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3330 (LINK_PIE_SPEC): ..to here.
3331 (LINK_COMMAND_SPEC): Support -no-pie.
3332 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3333 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3335 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3336 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3337 (ENDFILE_CRTEND_SPEC): Similarly.
3339 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3342 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3343 (mstack-protector-guard-offset=): Ditto.
3344 * config/i386/i386.c (ix86_option_override): Handle
3345 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3347 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3348 ix86_stack_protect_guard_offset variables.
3349 (TARGET_STACK_PROTECT_GUARD): Always define.
3350 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3351 and -mstack-protector-guard-offset= options.
3353 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3355 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3356 boundary case for the last candidate.
3358 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3360 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3361 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3363 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3366 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3367 * config/aarch64/aarch64-builtins.c
3368 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3369 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3370 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3372 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3373 Andrew Pinski <pinskia@gmail.com>
3376 * internal-fn.def (XORSIGN): New.
3377 * optabs.def (xorsign_optab): New.
3378 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3379 (convert_expand_mult_copysign): New.
3380 (pass_optimize_widening_mul::execute): Call
3381 convert_expand_mult_copysign.
3383 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3385 PR tree-optimization/81354
3386 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3387 Insert on edges rather than explicitly creating landing pads.
3388 (analyze_candidates_and_replace): Commit edge inserts.
3390 2017-08-08 Richard Biener <rguenther@suse.de>
3393 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3394 (expand_simple_operations): Also look through ADDR_EXPRs with
3395 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3397 2017-08-08 Richard Biener <rguenther@suse.de>
3399 PR tree-optimization/81723
3400 * tree-vect-slp.c (struct bst_traits): New hash traits.
3401 (bst_fail): New global.
3402 (vect_build_slp_tree_2): New worker, split out from ...
3403 (vect_build_slp_tree): ... this now wrapping it with using
3404 bst_fail set to cache SLP tree build fails. Properly handle
3406 (vect_analyze_slp_instance): Allocate and free bst_fail.
3408 2017-08-08 Martin Liska <mliska@suse.cz>
3411 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3412 LABEL_DECLs that can be from a different function.
3414 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3416 PR tree-optimization/81744
3417 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3418 loop's number of iterations.
3420 2017-08-08 Martin Liska <mliska@suse.cz>
3422 * asan.c: Include header files.
3423 * attribs.c (build_decl_attribute_variant): New function moved
3425 (build_type_attribute_qual_variant): Likewise.
3426 (cmp_attrib_identifiers): Likewise.
3427 (simple_cst_list_equal): Likewise.
3428 (omp_declare_simd_clauses_equal): Likewise.
3429 (attribute_value_equal): Likewise.
3430 (comp_type_attributes): Likewise.
3431 (build_type_attribute_variant): Likewise.
3432 (lookup_ident_attribute): Likewise.
3433 (remove_attribute): Likewise.
3434 (merge_attributes): Likewise.
3435 (merge_type_attributes): Likewise.
3436 (merge_decl_attributes): Likewise.
3437 (merge_dllimport_decl_attributes): Likewise.
3438 (handle_dll_attribute): Likewise.
3439 (attribute_list_equal): Likewise.
3440 (attribute_list_contained): Likewise.
3441 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3442 (lookup_attribute_by_prefix): Likewise.
3443 * bb-reorder.c: Include header files.
3444 * builtins.c: Likewise.
3445 * calls.c: Likewise.
3446 * cfgexpand.c: Likewise.
3447 * cgraph.c: Likewise.
3448 * cgraphunit.c: Likewise.
3449 * convert.c: Likewise.
3450 * dwarf2out.c: Likewise.
3451 * final.c: Likewise.
3452 * fold-const.c: Likewise.
3453 * function.c: Likewise.
3454 * gimple-expr.c: Likewise.
3455 * gimple-fold.c: Likewise.
3456 * gimple-pretty-print.c: Likewise.
3457 * gimple.c: Likewise.
3458 * gimplify.c: Likewise.
3459 * hsa-common.c: Likewise.
3460 * hsa-gen.c: Likewise.
3461 * internal-fn.c: Likewise.
3462 * ipa-chkp.c: Likewise.
3463 * ipa-cp.c: Likewise.
3464 * ipa-devirt.c: Likewise.
3465 * ipa-fnsummary.c: Likewise.
3466 * ipa-inline.c: Likewise.
3467 * ipa-visibility.c: Likewise.
3469 * lto-cgraph.c: Likewise.
3470 * omp-expand.c: Likewise.
3471 * omp-general.c: Likewise.
3472 * omp-low.c: Likewise.
3473 * omp-offload.c: Likewise.
3474 * omp-simd-clone.c: Likewise.
3475 * opts-global.c: Likewise.
3476 * passes.c: Likewise.
3477 * predict.c: Likewise.
3478 * sancov.c: Likewise.
3479 * sanopt.c: Likewise.
3480 * symtab.c: Likewise.
3481 * toplev.c: Likewise.
3482 * trans-mem.c: Likewise.
3483 * tree-chkp.c: Likewise.
3484 * tree-eh.c: Likewise.
3485 * tree-into-ssa.c: Likewise.
3486 * tree-object-size.c: Likewise.
3487 * tree-parloops.c: Likewise.
3488 * tree-profile.c: Likewise.
3489 * tree-ssa-ccp.c: Likewise.
3490 * tree-ssa-live.c: Likewise.
3491 * tree-ssa-loop.c: Likewise.
3492 * tree-ssa-sccvn.c: Likewise.
3493 * tree-ssa-structalias.c: Likewise.
3494 * tree-ssa.c: Likewise.
3495 * tree-streamer-in.c: Likewise.
3496 * tree-vectorizer.c: Likewise.
3497 * tree-vrp.c: Likewise.
3499 * ubsan.c: Likewise.
3500 * varasm.c: Likewise.
3501 * varpool.c: Likewise.
3502 * tree.c: Remove functions moved to attribs.[ch].
3504 * config/aarch64/aarch64.c: Add attrs.h header file.
3505 * config/alpha/alpha.c: Likewise.
3506 * config/arc/arc.c: Likewise.
3507 * config/arm/arm.c: Likewise.
3508 * config/avr/avr.c: Likewise.
3509 * config/bfin/bfin.c: Likewise.
3510 * config/c6x/c6x.c: Likewise.
3511 * config/cr16/cr16.c: Likewise.
3512 * config/cris/cris.c: Likewise.
3513 * config/darwin.c: Likewise.
3514 * config/epiphany/epiphany.c: Likewise.
3515 * config/fr30/fr30.c: Likewise.
3516 * config/frv/frv.c: Likewise.
3517 * config/ft32/ft32.c: Likewise.
3518 * config/h8300/h8300.c: Likewise.
3519 * config/i386/winnt.c: Likewise.
3520 * config/ia64/ia64.c: Likewise.
3521 * config/iq2000/iq2000.c: Likewise.
3522 * config/lm32/lm32.c: Likewise.
3523 * config/m32c/m32c.c: Likewise.
3524 * config/m32r/m32r.c: Likewise.
3525 * config/m68k/m68k.c: Likewise.
3526 * config/mcore/mcore.c: Likewise.
3527 * config/microblaze/microblaze.c: Likewise.
3528 * config/mips/mips.c: Likewise.
3529 * config/mmix/mmix.c: Likewise.
3530 * config/mn10300/mn10300.c: Likewise.
3531 * config/moxie/moxie.c: Likewise.
3532 * config/msp430/msp430.c: Likewise.
3533 * config/nds32/nds32-isr.c: Likewise.
3534 * config/nds32/nds32.c: Likewise.
3535 * config/nios2/nios2.c: Likewise.
3536 * config/nvptx/nvptx.c: Likewise.
3537 * config/pa/pa.c: Likewise.
3538 * config/pdp11/pdp11.c: Likewise.
3539 * config/powerpcspe/powerpcspe.c: Likewise.
3540 * config/riscv/riscv.c: Likewise.
3541 * config/rl78/rl78.c: Likewise.
3542 * config/rx/rx.c: Likewise.
3543 * config/s390/s390.c: Likewise.
3544 * config/sh/sh.c: Likewise.
3545 * config/sol2.c: Likewise.
3546 * config/sparc/sparc.c: Likewise.
3547 * config/spu/spu.c: Likewise.
3548 * config/stormy16/stormy16.c: Likewise.
3549 * config/tilegx/tilegx.c: Likewise.
3550 * config/tilepro/tilepro.c: Likewise.
3551 * config/v850/v850.c: Likewise.
3552 * config/vax/vax.c: Likewise.
3553 * config/visium/visium.c: Likewise.
3554 * config/xtensa/xtensa.c: Likewise.
3556 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3559 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3560 constraints since the -mupper-regs-* switches have been
3562 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3563 into a vector from a double word element that was extracted from
3564 another vector, and eliminate extra XXPERMDI instructions.
3565 (vsx_concat_<mode>_2): Likewise.
3566 (vsx_concat_<mode>_3): Likewise.
3567 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3568 concat to allow optimizing inserts from previous extracts.
3570 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3572 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3573 memory reference to a SSP offset in TLS address space.
3574 (ix86_print_operand) <case '@'>: Remove.
3575 (ix86_print_operand_punct_valid_p): Remove '@' code.
3576 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3578 (stack_tls_protect_set_<mode>): Remove.
3579 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3580 (stack_tls_protect_test_<mode>): Remove.
3581 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3583 2017-08-07 Olivier Hainque <hainque@adacore.com>
3586 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3588 2017-08-07 Douglas Rupp <rupp@adacore.com>
3590 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3591 variable was referenced as multidir in command.
3593 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3596 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3599 2017-08-07 Martin Liska <mliska@suse.cz>
3601 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3602 * config/rl78/rl78.c: Add include of attribs.h.
3603 * config/sh/sh.c: Likewise.
3604 * config/v850/v850.c: Likewise.
3606 2017-08-07 Tom de Vries <tom@codesourcery.com>
3609 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3611 2017-08-07 Martin Liska <mliska@suse.cz>
3613 * config/mips/mips.c: Include attribs.h.
3615 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3618 * doc/invoke.texi: Document change in behvaior for -Ofast for
3621 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3623 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3625 (aarch64_pop_regs): Likewise.
3626 (aarch64_gen_load_pair): Likewise.
3627 (aarch64_save_callee_saves): Likewise.
3628 (aarch64_restore_callee_saves): Likewise.
3630 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3632 * config/i386/i386.c: Revert the last change.
3634 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3637 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3639 (ix86_finalize_stack_frame_flags): This. Also clear
3640 frame_pointer_needed if -fno-omit-frame-pointer is used without
3642 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3643 with ix86_finalize_stack_frame_flags.
3644 (ix86_expand_epilogue): Likewise.
3645 (ix86_expand_split_stack_prologue): Likewise.
3647 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3650 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3651 to P_AES for Westmere.
3653 2017-08-07 Jonathan Yong <10walls@gmail.com>
3655 * config/i386/mingw.opt (fset-stack-executable): Removed.
3656 * config/i386/cygming.opt (fset-stack-executable): Moved
3658 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3660 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3662 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3664 2017-08-07 Marek Polacek <polacek@redhat.com>
3667 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3669 2017-08-07 Martin Liska <mliska@suse.cz>
3671 * attribs.h (canonicalize_attr_name): New function.
3672 (cmp_attribs): Move from c-format.c and adjusted.
3673 (is_attribute_p): Moved from tree.h.
3674 * tree-inline.c: Add new includes.
3675 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3676 (private_is_attribute_p): Remove.
3677 (private_lookup_attribute): Likewise.
3678 (private_lookup_attribute_by_prefix): Simplify.
3679 (remove_attribute): Use is_attribute_p.
3680 * tree.h: Remove removed declarations.
3682 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3685 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3686 instead of computing it in the function. Formatting fix.
3687 (expand_case): Don't rely on default_edge being the first edge,
3688 clear it if removing it, pass default_edge to
3689 emit_case_dispatch_table.
3690 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3693 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3695 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3696 insn in the function, emit NOP after the insn.
3698 2017-08-06 Tom de Vries <tom@codesourcery.com>
3700 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3703 2017-08-06 Tom de Vries <tom@codesourcery.com>
3705 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3708 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3710 PR tree-optimization/57371
3711 * match.pd: New pattern.
3713 2017-08-04 Marek Polacek <polacek@redhat.com>
3716 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3717 perform the computation in offset_int.
3719 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3721 PR tree-optimization/81136
3722 * tree-vectorizer.h: Include tree-hash-traits.h.
3723 (vec_base_alignments): New typedef.
3724 (vec_info): Add a base_alignments field.
3725 (vect_record_base_alignments): Declare.
3726 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3728 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3729 (create_data_ref): Add an is_conditional_in_stmt argument.
3730 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3731 the is_conditional_in_stmt field.
3732 (data_ref_loc): Add an is_conditional_in_stmt field.
3733 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3734 (find_data_references_in_stmt): Update call to create_data_ref.
3735 (graphite_find_data_references_in_stmt): Likewise.
3736 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3737 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3738 (vect_record_base_alignment): New function.
3739 (vect_record_base_alignments): Likewise.
3740 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3741 for nested statements even if we fail to compute a misalignment.
3742 Use pooled base alignments for unconditional references.
3743 (vect_find_same_alignment_drs): Compare base addresses instead
3745 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3746 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3748 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3750 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3751 Add an explicit name for the enum. Use auto_vec for slp_instances
3753 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3755 (_bb_vec_info): Add a constructor and destructor.
3756 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3757 (destroy_loop_vec_info): Delete.
3758 (vect_destroy_datarefs): Likewise.
3759 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3760 (vec_info::vec_info): New function.
3761 (vec_info::~vec_info): Likewise.
3762 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3763 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3764 destroy_loop_vec_info.
3765 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3766 (_loop_vec_info::_loop_vec_info): ...this.
3767 (destroy_loop_vec_info): Replace with...
3768 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3769 the stmt_vec_infos. Leave handling of vec_info information to its
3770 destructor. Remove explicit vector releases.
3771 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3772 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3773 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3774 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3775 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3776 (destroy_bb_vec_info): Replace with...
3777 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3778 information to its destructor.
3779 (vect_slp_analyze_bb_1): Use new and delete instead of
3780 new_bb_vec_info and destroy_bb_vec_info.
3781 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3784 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3786 * tree-data-ref.h (subscript): Add access_fn field.
3787 (data_dependence_relation): Add could_be_independent_p.
3788 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3789 (same_access_functions): Move to tree-data-ref.c.
3790 * tree-data-ref.c (ref_contains_union_access_p): New function.
3791 (access_fn_component_p): Likewise.
3792 (access_fn_components_comparable_p): Likewise.
3793 (dr_analyze_indices): Add a reference to access_fn_component_p.
3794 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3796 (constant_access_functions): Likewise.
3797 (add_other_self_distances): Likewise.
3798 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3799 (initialize_data_dependence_relation): Use XCNEW and remove
3800 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3801 of access functions that have the same type. Allow the
3802 subsequence to end with different bases in some circumstances.
3803 Record the chosen access functions in SUB_ACCESS_FN.
3804 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3805 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3806 (subscript_dependence_tester_1): Likewise dra and drb.
3807 (build_classic_dist_vector): Update calls accordingly.
3808 (subscript_dependence_tester): Likewise.
3809 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3810 DDR_COULD_BE_INDEPENDENT_P.
3811 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3812 comp_alias_ddrs instead of may_alias_ddrs.
3813 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3815 (vect_analyze_data_ref_dependence): Use it if
3816 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3817 distance vectors if that fails.
3818 (dependence_distance_ge_vf): New function.
3819 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3820 LOOP_VINFO_MAY_ALIAS_DDRS.
3822 2017-08-04 Richard Biener <rguenther@suse.de>
3825 * fold-const.c (fold_binary_loc): Properly restrict
3826 minus_var0 && minus_var1 case when associating undefined overflow
3829 2017-08-04 Tom de Vries <tom@codesourcery.com>
3831 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3833 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3835 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3836 Don't start diagnostic messages with a capital letter.
3837 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3839 (rs6000_invalid_builtin): Likewise.
3840 (rs6000_trampoline_init): Likewise.
3842 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3845 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3846 after setting changeable df flags.
3848 2017-08-03 Richard Biener <rguenther@suse.de>
3850 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3851 up if the use is in USE - X.
3853 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3855 * toplev.c (dumpfile.h): New include.
3856 (internal_error_reentered): New static function. Use it...
3857 (internal_error_function): ...here to handle reentered internal_error.
3859 2017-08-03 Richard Biener <rguenther@suse.de>
3862 * fold-const.c (split_tree): Add minus_var and minus_con
3863 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3864 here but always use minus_*.
3865 (associate_trees): Assert we never associate with MINUS_EXPR
3866 and NULL first operand. Do not recurse for PLUS_EXPR operands
3867 when associating as MINUS_EXPR either.
3868 (fold_binary_loc): Track minus_var and minus_con.
3870 2017-08-03 Tom de Vries <tom@codesourcery.com>
3873 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3874 ACCEL_COMPILER, apply finish_options on
3875 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3877 2017-08-03 Tom de Vries <tom@codesourcery.com>
3880 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3881 function_entry_patch_area_size > 0.
3883 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3886 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3887 instead of 10??LU, perform unit multiplication in wide_int,
3888 don't change alloc_object_size_limit if the limit is larger
3891 PR tree-optimization/81655
3892 PR tree-optimization/81588
3893 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3894 the case when ranges[i].low and high are 1 for unsigned type with
3898 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3899 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3901 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3903 * tree-vrp.h: Add include guard.
3905 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3908 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3909 (ud2): New insn pattern.
3910 * config/i386/i386.c (ix86_expand_epilogue):
3911 For naked functions, generate ud2 instead of trap insn.
3913 2017-08-02 Marek Polacek <polacek@redhat.com>
3916 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3918 2017-08-02 Tom de Vries <tom@codesourcery.com>
3919 Cesar Philippidis <cesar@codesourcery.com>
3921 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3922 Add missing edge probabilities.
3924 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3926 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3929 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3932 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3933 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3934 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3935 prologue_seq sequences - if any.
3937 2017-08-02 Richard Biener <rguenther@suse.de>
3939 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3940 via vectors if supported, integer extracts via punning if supported
3941 or otherwise vector extracts.
3943 2017-08-02 Richard Biener <rguenther@suse.de>
3945 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3947 (bitmap_insert_into_set): ... this.
3949 2017-08-02 Richard Biener <rguenther@suse.de>
3951 PR tree-optimization/81633
3953 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3955 PR tree-optimization/71752
3956 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3958 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3960 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3961 (machine_function::call_ms2sysv_pad_out): Remove field.
3962 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3963 (ix86_compute_frame_layout): Likewise.
3965 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3968 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3969 attribute with interrupt attribute.
3971 2017-08-01 Andrew Pinski <apinski@cavium.com>
3973 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3974 BIT_INSERT_EXPR's operand 1
3975 to see if the types precision matches.
3977 2017-08-01 Martin Liska <mliska@suse.cz>
3980 * builtins.c (expand_builtin_memcpy_args): Remove.
3981 (expand_builtin_memcpy): Call newly added function
3982 expand_builtin_memory_copy_args.
3983 (expand_builtin_memcpy_with_bounds): Likewise.
3984 (expand_builtin_mempcpy): Remove last argument.
3985 (expand_builtin_mempcpy_with_bounds): Likewise.
3986 (expand_builtin_memory_copy_args): New function created from
3987 expand_builtin_mempcpy_args with small modifications.
3988 (expand_builtin_mempcpy_args): Remove.
3989 (expand_builtin_stpcpy): Remove unused argument.
3990 (expand_builtin): Likewise.
3991 (expand_builtin_with_bounds): Likewise.
3993 2017-08-01 Martin Liska <mliska@suse.cz>
3996 Make mempcpy more optimal (PR middle-end/70140).
3998 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4001 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4002 __builtin_vec_cmpne verify both arguments are compatible vectors
4003 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4004 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4005 move computation of aligned to after checking the argument types.
4009 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4010 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4013 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4014 Doug Rupp <rupp@adacore.com>
4015 Olivier Hainque <hainque@adacore.com>
4017 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4018 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4020 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4021 for TARGET_OS_CPP_BUILTIN.
4022 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4023 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4025 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4026 passing required abi options to the assembler for EABI configurations.
4027 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4028 of .text.hot and .text.unlikely sections for kernel modules when
4029 using ARM style exceptions.
4030 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4031 options. Add EXTRA_CC1_SPEC.
4032 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4034 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4036 (ARM_TARGET2_DWARF_FORMAT): Define.
4037 * config/arm/t-vxworks: Adjust multilib control to removal of the
4038 Diab command line options.
4040 2017-08-01 Martin Liska <mliska@suse.cz>
4042 PR gcov-profile/81561
4043 * gcov.c (unblock): Make unblocking safe as we need to preserve
4044 index correspondence of blocks and block_lists.
4046 2017-08-01 Richard Biener <rguenther@suse.de>
4048 PR tree-optimization/81181
4049 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4050 (compute_antic): ... end of iteration here.
4052 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4054 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4055 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4056 (ftree-slp-vectorize): Likewise.
4057 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4058 can no longer be set independent of flag_tree_loop_vectorize.
4059 * omp-general.c (emp_max_vf): Likewise.
4060 * opts.c (enable_fdo_optimizations): Remove references to
4061 flag_tree_vectorize, these are now implicit.
4062 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4063 and leave it for the options machinery.
4065 2017-08-01 Martin Liska <mliska@suse.cz>
4068 * builtins.c (expand_builtin_memcpy_args): Remove.
4069 (expand_builtin_memcpy): Call newly added function
4070 expand_builtin_memory_copy_args.
4071 (expand_builtin_memcpy_with_bounds): Likewise.
4072 (expand_builtin_mempcpy): Remove last argument.
4073 (expand_builtin_mempcpy_with_bounds): Likewise.
4074 (expand_builtin_memory_copy_args): New function created from
4075 expand_builtin_mempcpy_args with small modifications.
4076 (expand_builtin_mempcpy_args): Remove.
4077 (expand_builtin_stpcpy): Remove unused argument.
4078 (expand_builtin): Likewise.
4079 (expand_builtin_with_bounds): Likewise.
4081 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4084 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4085 print "ds:" only for immediates in generic address space.
4087 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4090 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4091 (ix86_function_ok_for_sibcall): Return false for naked functions.
4093 2017-08-01 Richard Biener <rguenther@suse.de>
4095 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4096 (compute_antic): Seed worklist with exit block predecessors.
4097 * cfganal.c (dfs_find_deadend): For a cycle return the source
4098 of the edge closing it.
4100 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4102 * config/aarch64/aarch64.c
4103 (aarch64_can_const_movi_rtx_p): Move 0 check.
4105 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4107 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4108 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4110 * match.pd: Ditto in address comparison pattern.
4112 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4114 PR tree-optimization/81627
4115 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4116 closed ssa form for store-store chain.
4118 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4120 PR tree-optimization/81620
4121 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4122 for store-store chain.
4124 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4126 PR tree-optimization/81588
4127 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4128 ranges[i].in_p, invert comparison code ccode. For >/>=,
4129 swap rhs1 and rhs2 and comparison code unconditionally,
4130 for </<= don't do that. Don't swap rhs1/rhs2 again if
4131 ranges[i].in_p, instead invert comparison code ccode if
4132 opcode or oe->rank is BIT_IOR_EXPR.
4135 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4136 a direct optab to conversion optab.
4137 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4138 with GET_MODE_INNER as last argument instead of optab_handler.
4139 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4140 vector extraction if possible and optab is available.
4141 * expr.c (store_constructor): Use convert_optab_handler instead
4142 of optab_handler. Use vector initialization from smaller
4143 vectors if possible and optab is available.
4144 * tree-vect-stmts.c (vectorizable_load): Likewise.
4145 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4147 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4148 of vec_init from half-sized vectors with the same element mode.
4149 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4150 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4151 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4152 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4153 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4154 after mode in gen_vec_extract* calls.
4155 (vec_extract<mode>): Renamed to ...
4156 (vec_extract<mode><ssescalarmodelower>): ... this.
4157 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4158 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4159 element mode after mode in gen_vec_init* calls.
4160 (VEC_INIT_HALF_MODE): New mode iterator.
4161 (vec_init<mode>): Renamed to ...
4162 (vec_init<mode><ssescalarmodelower>): ... this.
4163 (vec_init<mode><ssehalfvecmodelower>): New expander.
4164 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4165 (vec_extractv2sfsf): ... this.
4166 (vec_initv2sf): Renamed to ...
4167 (vec_initv2sfsf): ... this.
4168 (vec_extractv2si): Renamed to ...
4169 (vec_extractv2sisi): ... this.
4170 (vec_initv2si): Renamed to ...
4171 (vec_initv2sisi): ... this.
4172 (vec_extractv4hi): Renamed to ...
4173 (vec_extractv4hihi): ... this.
4174 (vec_initv4hi): Renamed to ...
4175 (vec_initv4hihi): ... this.
4176 (vec_extractv8qi): Renamed to ...
4177 (vec_extractv8qiqi): ... this.
4178 (vec_initv8qi): Renamed to ...
4179 (vec_initv8qiqi): ... this.
4180 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
4181 (vec_init<mode>): Renamed to ...
4182 (vec_init<mode><VEC_base_l>): ... this.
4183 (vec_extract<mode>): Renamed to ...
4184 (vec_extract<mode><VEC_base_l>): ... this.
4185 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
4186 (vec_initv2sfsf): ... this.
4187 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
4188 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4189 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
4190 element mode after mode in gen_vec_init* calls.
4191 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4192 (vec_init<mode><Vel>): ... this.
4193 (vec_extract<mode>): Renamed to ...
4194 (vec_extract<mode><Vel>): ... this.
4195 * config/aarch64/iterators.md (Vel): New mode attribute.
4196 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4197 Add element mode after mode in gen_vec_extract* calls.
4198 * config/s390/vector.md (non_vec_l): New mode attribute.
4199 (vec_extract<mode>): Renamed to ...
4200 (vec_extract<mode><non_vec_l>): ... this.
4201 (vec_init<mode>): Renamed to ...
4202 (vec_init<mode><non_vec_l>): ... this.
4203 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4204 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4206 * config/arm/iterators.md (V_elem_l): New mode attribute.
4207 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4208 (vec_extract<mode><V_elem_l>): ... this.
4209 (vec_extractv2di): Renamed to ...
4210 (vec_extractv2didi): ... this.
4211 (vec_init<mode>): Renamed to ...
4212 (vec_init<mode><V_elem_l>): ... this.
4213 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4214 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4215 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4216 Add element mode after gen_vec_extract* calls.
4217 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4218 (vec_init<mode><unitmode>): ... this.
4219 (vec_extract<mode>): Renamed to ...
4220 (vec_extract<mode><unitmode>): ... this.
4221 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4222 (vec_init<mode><unitmode>): ... this.
4223 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4224 (vec_initv2sfsf): ... this.
4225 (vec_extractv2sf): Renamed to ...
4226 (vec_extractv2sfsf): ... this.
4227 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4228 Add element mode after gen_vec_extract* calls.
4229 * config/mips/mips.md (unitmode): New mode iterator.
4230 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4231 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4232 * config/spu/spu.md (inner_l): New mode attribute.
4233 (vec_init<mode>): Renamed to ...
4234 (vec_init<mode><inner_l>): ... this.
4235 (vec_extract<mode>): Renamed to ...
4236 (vec_extract<mode><inner_l>): ... this.
4237 * config/sparc/sparc.md (veltmode): New mode iterator.
4238 (vec_init<VMALL:mode>): Renamed to ...
4239 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4240 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4241 (vec_initv2sisi): ... this.
4242 (vec_initv2sf): Renamed to ...
4243 (vec_initv2sfsf): ... this.
4244 (vec_extractv2sf): Renamed to ...
4245 (vec_extractv2sfsf): ... this.
4246 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4247 (vec_init<mode>): Renamed to ...
4248 (vec_init<mode><VEC_base_l>): ... this.
4249 (vec_extract<mode>): Renamed to ...
4250 (vec_extract<mode><VEC_base_l>): ... this.
4251 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4252 (vec_initv2sfsf): ... this.
4253 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4254 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4255 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4256 gen_vec_init* calls.
4258 2017-08-01 Richard Biener <rguenther@suse.de>
4260 PR tree-optimization/81297
4261 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4262 TREE_OVERFLOW from INTEGER_CSTs.
4264 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4266 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4268 2017-07-31 Carl Love <cel@us.ibm.com>
4270 * config/rs6000/rs6000-c: Add support for built-in functions
4271 vector signed char vec_xl_be (signed long long, signed char *);
4272 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4273 vector signed int vec_xl_be (signed long long, signed int *);
4274 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4275 vector signed long long vec_xl_be (signed long long, signed long long *);
4276 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4277 vector signed short vec_xl_be (signed long long, signed short *);
4278 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4279 vector double vec_xl_be (signed long long, double *);
4280 vector float vec_xl_be (signed long long, float *);
4281 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4282 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4283 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4285 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4286 (altivec_expand_builtin): Add switch statement to call
4287 altivec_expand_xl_be for each builtin.
4288 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4289 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4290 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4291 __builtin_vsx_le_be_v16qi.
4292 * doc/extend.texi: Update the built-in documentation file for the
4293 new built-in functions.
4295 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4298 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4300 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4302 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4304 * config.gcc: Add z14.
4305 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4306 CPU model numbers for z13s and z14.
4307 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4309 * config/s390/s390-opts.h (enum processor_type): Rename
4310 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4311 * config/s390/s390.c (processor_table): Add field for CPU name to
4312 be passed to Binutils.
4313 (s390_asm_output_machine_for_arch): Use the new field in
4314 processor_table for Binutils.
4315 (s390_expand_builtin): Replace arch12 with z14.
4316 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4317 (s390_get_sched_attrmask): Likewise.
4318 (s390_get_unit_mask): Likewise.
4319 * config/s390/s390.opt: Add z14 to processor_type enum.
4321 2017-07-31 Martin Jambor <mjambor@suse.cz>
4324 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4325 regardless of optimization level.
4327 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4328 Martin Liska <mliska@suse.cz>
4330 * predict.def: Remove old comment and adjust probability.
4331 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4334 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4337 * config/i386/i386.c (ix86_function_naked): New function.
4338 (ix86_can_use_return_insn_p): Return false for naked functions.
4339 (ix86_expand_prologue): Skip prologue for naked functions.
4340 (ix86_expand_epilogue): Skip epilogue for naked functions
4341 and emit trap instruction.
4342 (ix86_warn_func_return): New function.
4343 (ix86_attribute_table): Add "naked" attribute specification.
4344 (TARGET_WARN_FUNC_RETURN): Define.
4345 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4347 2017-07-31 Martin Liska <mliska@suse.cz>
4349 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4350 (dump_gimple_bb_header): Always dump BB info.
4351 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4353 2017-07-31 Martin Liska <mliska@suse.cz>
4356 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4357 also with current_function_decl non-null equality.
4359 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4362 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4363 change type to the element type, instead add eltype variable and
4364 use it where we are interested in the element type.
4366 PR tree-optimization/81603
4367 * ipa-polymorphic-call.c
4368 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4369 offset arithmetic in offset_int, bail out if the resulting bit offset
4370 doesn't fit into shwi.
4372 2017-07-31 Martin Liska <mliska@suse.cz>
4374 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4375 (gimplify_save_expr): Fix comment.
4377 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4380 * config/i386/i386.c (ix86_function_arg): Update arguments for
4382 (ix86_compute_frame_layout): Set the initial stack offset to
4383 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4384 INCOMING_FRAME_SP_OFFSET.
4385 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4386 stack before exception handler returns.
4387 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4388 the 'ERROR_CODE' for exception handler.
4390 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4392 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4393 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4394 (ASM_OUTPUT_REG_POP): Ditto.
4395 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4396 instead of asm_fprintf to output pure string.
4398 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4400 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4401 to imported_module_or_decl hook.
4402 (debug_nothing_tree_tree_tree_bool): Remove.
4403 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4404 * debug.c (do_nothing_debug_hooks): Use
4405 debug_nothing_tree_tree_tree_bool_bool instead of
4406 debug_nothing_tree_tree_tree_bool.
4407 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4408 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4409 * sdbout.c (sdb_debug_hooks): Likewise.
4410 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4411 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4413 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4414 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4415 attribute, don't add anything.
4417 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4419 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4420 (fold_build2_stat_loc): Likewise.
4421 (fold_build3_stat_loc): Likewise.
4422 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4423 (fold_build1_loc): Remove macro.
4424 (fold_build2_loc): Likewise.
4425 (fold_build3_loc): Likewise.
4427 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4429 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4430 (gimple_build_debug_bind_source_stat): Likewise.
4431 * gimple.h (gimple_build_debug_bind): Remove macro.
4432 (gimple_build_debug_bind_source): Likewise.
4434 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4436 * bitmap.c (bitmap_alloc): Adjust.
4437 (bitmap_gc_alloc): Likewise.
4438 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4440 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4442 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4443 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4444 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4445 (bitmap_gc_alloc_stat): Likewise.
4446 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4448 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4450 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4451 * rtl.h (shallow_copy_rtx): Remove macro.
4453 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4455 * emit-rtl.c (gen_raw_REG): Adjust.
4456 * gengenrtl.c (gendef): Likewise.
4457 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4458 * rtl.h (rtx_alloc): Remove macro.
4460 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4462 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4463 (build_tree_list_stat): Likewise.
4464 * tree.h (build_tree_list): Remove macro.
4465 (build_tree_list_vec): Likewise.
4467 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4469 * tree.c (make_vector_stat): Remove _stat from name.
4470 (build_vector_stat): Likewise.
4471 * tree.h (make_vector_stat): Remove macro.
4472 (build_vector_stat): Likewise.
4474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4476 * tree.h (build_var_debug_value): Remove prototype.
4478 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4480 * tree.c (tree_cons_stat): Remove _stat from name.
4481 * tree.h (tree_cons): Remove macro.
4483 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4485 * tree.c (build_vl_exp_stat): Remove _stat from name.
4486 * tree.h (build_vl_exp): Remove macro.
4488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4490 * tree.c (build_decl_stat): Remove _stat from name.
4491 * tree.h (build_decl): Remove macro.
4493 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4495 * gimple.c (gimple_build_with_ops_stat): Adjust.
4496 (gimple_alloc_stat): Remove _stat from name.
4497 * gimple.h (gimple_alloc): Remove macro.
4499 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4501 * tree.c (make_tree_vec_stat): Remove _stat from name.
4502 (grow_tree_vec_stat): Likewise.
4503 * tree.h (make_tree_vec_stat): Adjust prototype.
4504 (grow_tree_vec_stat): Likewise.
4505 (make_tree_vec): Remove macro.
4506 (grow_tree_vec): Likewise.
4508 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4510 * fold-const.c (fold_build1_stat_loc): Adjust.
4511 (fold_build2_stat_loc): Likewise.
4512 (fold_build3_stat_loc): Likewise.
4513 * tree.c (build0_stat): Remove _stat from name.
4514 (build1_stat): Likewise.
4515 (build2_stat): Likewise.
4516 (build3_stat): Likewise.
4517 (build4_stat): Likewise.
4518 (build5_stat): Likewise.
4519 * tree.h (build1_loc): Remove macro, and rename _stat function
4521 (build2_loc): Likewise.
4522 (build3_loc): Likewise.
4523 (build4_loc): Likewise.
4524 (build5_loc): Likewise.
4526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4528 * tree.c (make_int_cst_stat): Remove _stat from name.
4529 * tree.h (make_int_cst_stat): Adjust prototype.
4530 (make_int_cst): Remove macro.
4532 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4534 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4535 * tree.h (make_tree_binfo_stat): Adjust prototype.
4536 (make_tree_binfo): Remove.
4538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4540 * tree.c (copy_node_stat): Rename to copy_node.
4541 (build_distinct_type_copy): Adjust.
4542 * tree.h (copy_node_stat): Adjust prototype.
4543 (copy_node): Remove macro.
4545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4547 * tree.c (make_node_stat): rename to make_node.
4548 (build_tree_list_stat): Adjust.
4549 (build0_stat): Likewise.
4550 (build2_stat): Likewise.
4551 (build3_stat): Likewise.
4552 (build4_stat): Likewise.
4553 (build5_stat): Likewise.
4554 (build_decl_stat): Likewise.
4555 * tree.h (make_node_stat): Adjust prototype.
4556 (make_node): remove macro.
4558 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4560 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4561 (PPC_FEATURE2_SCV): Likewise.
4562 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4564 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4566 * config/aarch64/aarch64.c
4567 (aarch64_internal_mov_immediate): Add new special pattern.
4568 * config/aarch64/aarch64.md (*movdi_aarch64):
4569 Add reg/32bit const mov case.
4571 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4572 Richard Sandiford <richard.sandiford@linaro.org>
4574 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4575 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4576 Add integer and movi cases.
4577 (movi-split-hf-df-sf split, fp16): New.
4578 (enabled): Added TARGET_FP_F16INST.
4579 * config/aarch64/iterators.md (GPF_HF): New.
4580 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4582 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4584 * config/aarch64/aarch64.c
4585 (aarch64_simd_container_mode): Add prototype.
4586 (aarch64_expand_mov_immediate): Add HI support.
4587 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4588 (aarch64_can_const_movi_rtx_p): New.
4589 (aarch64_preferred_reload_class):
4590 Remove restrictions of using FP registers for certain SIMD operations.
4591 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4592 (aarch64_valid_floating_const): Add integer move validation.
4593 (aarch64_simd_imm_scalar_p): Remove.
4594 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4595 (aarch64_legitimate_constant_p): Expand list of supported cases.
4596 * config/aarch64/aarch64-protos.h
4597 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4598 (aarch64_reinterpret_float_as_int): New.
4599 (aarch64_simd_imm_scalar_p): Remove.
4600 * config/aarch64/constraints.md (Uvi): New.
4601 (Dd): Split into Ds and new Dd.
4602 * config/aarch64/aarch64.md (*movsi_aarch64):
4604 (*movdi_aarch64): Add SIMD mov case.
4606 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4608 * tree-predcom.c: (struct chain): Handle store-store chain in which
4609 stores for elimination only store loop invariant values.
4610 (execute_pred_commoning_chain): Ditto.
4611 (prepare_initializers_chain_store_elim): Ditto.
4612 (prepare_finalizers): Ditto.
4613 (is_inv_store_elimination_chain): New function.
4614 (initialize_root_vars_store_elim_1): New function.
4616 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4618 * tree-predcom.c: Revise general description of the pass.
4619 (enum chain_type): New enum type for store elimination.
4620 (struct chain): New field supporting store elimination.
4621 (struct component): Ditto.
4622 (dump_chain): Dump store-stores chain.
4623 (release_chain): Release resources.
4624 (split_data_refs_to_components): Compute and create component
4625 contains only stores for elimination.
4626 (get_chain_last_ref_at): New function.
4627 (make_invariant_chain): Initialization.
4628 (make_rooted_chain): Specify chain type in parameter and record it.
4629 (add_looparound_copies): Skip for store-stores chain.
4630 (determine_roots_comp): Compute type of chain and pass it to
4632 (initialize_root_vars_store_elim_2): New function.
4633 (finalize_eliminated_stores): New function.
4634 (remove_stmt): Handle store for elimination.
4635 (execute_pred_commoning_chain): Execute predictive commoning on
4637 (determine_unroll_factor): Skip unroll for store-stores chain.
4638 (prepare_initializers_chain_store_elim): New function.
4639 (prepare_initializers_chain): Hanlde store-store chain.
4640 (prepare_finalizers_chain, prepare_finalizers): New function.
4641 (tree_predictive_commoning_loop): Return integer value indicating
4642 if loop is unrolled or lcssa form is corrupted.
4643 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4645 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4647 * tree-predcom.c (initialize_root): Delete.
4648 (execute_pred_commoning_chain): Initialize root vars and replace
4649 reference of non-combined chain directly, rather than call above
4652 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4654 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4655 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4657 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4659 * tree-predcom.c (struct chain): New field init_seq.
4660 (release_chain): Release init_seq.
4661 (prepare_initializers_chain): Record intialization stmts in above
4663 (insert_init_seqs): New function.
4664 (tree_predictive_commoning_loop): Call insert_init_seqs.
4666 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4668 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4670 2017-07-28 Richard Biener <rguenther@suse.de>
4672 * match.pd: Remove superfluous :c.
4673 * genmatch.c (simplify::id): Add member.
4674 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4676 (current_id): New global.
4677 (dt_node::parent): Move from ...
4678 (dt_operand::parent): ... here. Add for_id member.
4679 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4680 (decision_tree::find_node): Relax order requirement when
4681 merging DT_TRUE nodes to ones inbetween the current simplify
4682 and the one we try to merge with. Add diagnostic whenever
4683 we need to enforce pattern order by not merging.
4684 (decision_tree::insert): Set current_id.
4685 (decision_tree::print_node): Dump parent node and for_id.
4686 (parser::last_id): Add member.
4687 (parser::push_simplify): Assign unique id.
4688 (parser::parser): Initialize last_id.
4690 2017-07-28 Martin Liska <mliska@suse.cz>
4693 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4694 gimple_build_debug_bind.
4696 2017-07-28 Richard Biener <rguenther@suse.de>
4698 PR tree-optimization/81502
4699 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4701 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4703 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4704 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4705 for BIT_FIELD_REF args.
4706 * fold-const.c (make_bit_field_ref): Likewise.
4707 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4709 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4712 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4713 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4714 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4715 Or it into SANITIZER_UNDEFINED.
4716 * ubsan.c: Include gimple-fold.h and varasm.h.
4717 (ubsan_expand_ptr_ifn): New function.
4718 (instrument_pointer_overflow): New function.
4719 (maybe_instrument_pointer_overflow): New function.
4720 (instrument_object_size): Formatting fix.
4721 (pass_ubsan::execute): Call instrument_pointer_overflow
4722 and maybe_instrument_pointer_overflow.
4723 * internal-fn.c (expand_UBSAN_PTR): New function.
4724 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4725 * sanitizer.def (__ubsan_handle_pointer_overflow,
4726 __ubsan_handle_pointer_overflow_abort): New builtins.
4727 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4728 * internal-fn.def (UBSAN_PTR): New internal function.
4729 * opts.c (sanitizer_opts): Add pointer-overflow.
4730 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4731 * fold-const.c (build_range_check): Compute pointer range check in
4732 integral type if pointer arithmetics would be needed. Formatting
4735 2017-07-28 Martin Liska <mliska@suse.cz>
4738 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4739 parameters that are of a variable-length.
4741 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4743 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4745 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4746 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4747 (CRT_CALL_STATIC_FUNCTION): Likewise.
4748 (ASM_DEFAULT_SPEC): New define.
4749 (ASM_SPEC32): Likewise.
4750 (ASM_SPEC64): Likewise.
4751 (ASM_SPEC_COMMON): Likewise.
4752 (ASM_SPEC): Likewise.
4753 (INVALID_64BIT): Likewise.
4754 (LINK_OS_DEFAULT_SPEC): Likewise.
4755 (LINK_OS_SPEC32): Likewise.
4756 (LINK_OS_SPEC64): Likewise.
4757 (POWERPC_LINUX): Likewise.
4758 (PTRDIFF_TYPE): Likewise.
4759 (RESTORE_FP_PREFIX): Likewise.
4760 (RESTORE_FP_SUFFIX): Likewise.
4761 (SAVE_FP_PREFIX): Likewise.
4762 (SAVE_FP_SUFFIX): Likewise.
4763 (SIZE_TYPE): Likewise.
4764 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4765 (TARGET_64BIT): Likewise.
4766 (TARGET_64BIT): Likewise.
4767 (TARGET_AIX): Likewise.
4768 (WCHAR_TYPE_SIZE): Likewise.
4769 (WCHAR_TYPE): Undefine.
4770 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4771 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4772 (CPP_OS_RTEMS_SPEC): Delete.
4773 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4774 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4776 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4778 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4780 PR tree-optimization/81578
4781 * tree-parloops.c (build_new_reduction): Bail out if
4782 reduction_code isn't one of the standard OpenMP reductions.
4783 Move the details printing after that decision.
4785 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4787 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4788 related to reload_in_progress.
4789 (splat_input_operand): Likewise.
4790 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4792 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4794 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4795 (TARGET_INSTANTIATE_DECLS): Likewise.
4796 (legitimate_indexed_address_p): Delete reload_in_progress code.
4797 (rs6000_debug_legitimate_address_p): Likewise.
4798 (rs6000_eliminate_indexed_memrefs): Likewise.
4799 (rs6000_emit_le_vsx_store): Likewise.
4800 (rs6000_emit_move_si_sf_subreg): Likewise.
4801 (rs6000_emit_move): Likewise.
4802 (register_to_reg_type): Likewise.
4803 (rs6000_pre_atomic_barrier): Likewise.
4804 (rs6000_machopic_legitimize_pic_address): Likewise.
4805 (rs6000_allocate_stack_temp): Likewise.
4806 (rs6000_address_for_fpconvert): Likewise.
4807 (rs6000_address_for_altivec): Likewise.
4808 (rs6000_secondary_memory_needed_rtx): Delete function.
4809 (rs6000_check_sdmode): Likewise.
4810 (rs6000_alloc_sdmode_stack_slot): Likewise.
4811 (rs6000_instantiate_decls): Likewise.
4812 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4813 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4814 Delete reload_in_progress.
4815 (*vec_reload_and_plus_<mptrsize>): Likewise.
4816 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4817 (vsx_div_v2di): Likewise.
4818 (vsx_udiv_v2di): Likewise.
4820 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4822 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4823 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4824 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4825 (rs6000_debug_reg_global): Delete print of LRA status.
4826 (rs6000_option_override_internal): Delete dead LRA related code.
4827 (rs6000_lra_p): Delete function.
4828 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4830 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4832 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4833 * config/riscv/rtems.h: New file.
4835 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4836 Sudakshina Das <sudi.das@arm.com>
4838 * config/aarch64/aarch64.md
4839 (define_split for and<mode>3nr_compare): Move
4840 non aarch64_logical_operand to a register.
4841 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4842 register immediate operand to a register.
4843 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4845 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4848 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4850 2017-07-27 Richard Biener <rguenther@suse.de>
4852 PR tree-optimization/81573
4853 PR tree-optimization/81494
4854 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4855 multi defuse cycle case.
4857 2017-07-27 Richard Biener <rguenther@suse.de>
4859 PR tree-optimization/81571
4860 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4863 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4865 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4866 earlier and only if MASK_FPU is set. Adjust formatting.
4868 2017-07-27 Martin Liska <mliska@suse.cz>
4870 * opt-functions.awk: Add validation of value of Init.
4871 * optc-gen.awk: Pass new argument.
4873 2017-07-27 Martin Liska <mliska@suse.cz>
4875 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4876 Fix wrong condition.
4878 2017-07-27 Martin Liska <mliska@suse.cz>
4880 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4881 BBs and edges seen by autoFDO.
4883 2017-07-27 Richard Biener <rguenther@suse.de>
4885 PR tree-optimization/81502
4886 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4887 with incompatible but same sized type.
4888 (execute_update_addresses_taken): Likewise.
4890 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4892 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4893 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4895 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4898 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4899 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4900 Change s_operand to memory_operand.
4902 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4904 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4905 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4906 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4907 Emit instructions rather than returning an expression. Handle TFmode
4908 and KFmode by casting to TImode.
4909 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4910 (rs6000_emit_le_vsx_store): Likewise.
4911 * config/rs6000/vsx.md (VSX_TI): New iterator.
4912 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4913 (*vsx_le_undo_permute_<mode>): Likewise.
4914 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4915 emit the split sequence.
4916 (*vsx_le_perm_store_<mode>): Likewise.
4918 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4920 PR tree-optimization/81555
4921 PR tree-optimization/81556
4922 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4923 if true, force CHANGED for the recursive invocation.
4924 (reassociate_bb): Remember original length of ops array, pass
4925 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4927 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4928 attributes for noipa attribute. For naked attribute use
4929 lookup_attribute first before lookup_attribute_spec.
4930 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4932 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4933 for functions with noipa attribute.
4934 (cgraph_externally_visible_p): Return true for functions with noipa
4936 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4937 for functions with noipa attribute.
4938 * doc/extend.texi: Document noipa function attribute.
4939 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4940 also for functions with noipa attribute.
4941 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4943 2017-07-26 Andrew Pinski <apinski@cavium.com>
4945 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4946 vec_unalign_load_cost and vec_unalign_store_cost.
4948 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4950 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4951 -mvsx-small-integer option.
4952 (ISA_3_0_MASKS_IEEE): Likewise.
4953 (OTHER_VSX_VECTOR_MASKS): Likewise.
4954 (POWERPC_MASKS): Likewise.
4955 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4956 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4957 code, only testing for DImode being allowed in non-VSX floating
4959 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4960 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4962 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4963 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4964 TARGET_P8_VECTOR test.
4965 (rs6000_secondary_reload_simple_move): Likewise.
4966 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4967 since TARGET_P9_VECTOR was already tested.
4968 (rs6000_opt_masks): Remove -mvsx-small-integer.
4969 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4970 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4972 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4973 test for TARGET_VEXTRACTUB was used, and that uses
4975 (p9 extract splitter): Likewise.
4976 (vsx_extract_<mode>_di_p9): Likewise.
4977 (vsx_extract_<mode>_store_p9): Likewise.
4978 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4979 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4980 the elimination of TARGET_VSX_SMALL_INTEGER.
4981 (vsx_extract_<mode>_p8): Likewise.
4982 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4983 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4984 (vsx_set_<mode>_p9): Likewise.
4985 (vsx_set_v4sf_p9): Likewise.
4986 (vsx_set_v4sf_p9_zero): Likewise.
4987 (vsx_insert_extract_v4sf_p9): Likewise.
4988 (vsx_insert_extract_v4sf_p9_2): Likewise.
4989 * config/rs6000/rs6000.md (sign extend splitter): Change
4990 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4991 (floatsi<mode>2_lfiwax_mem): Likewise.
4992 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4993 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4994 since a test for TARGET_P9_VECTOR was used.
4995 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4996 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4997 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4998 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4999 TARGET_P8_VECTOR test.
5000 (fix_trunc<mode>si2_stfiwx): Likewise.
5001 (fix_trunc<mode>si2_internal): Likewise.
5002 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5003 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5005 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5006 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5007 TARGET_P8_VECTOR test.
5008 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5009 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5010 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5012 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5013 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5014 since a test for TARGET_P9_VECTOR was used.
5015 (splitter for loading small constants): Likewise.
5017 2017-07-26 Andrew Pinski <apinski@cavium.com>
5019 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5022 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5025 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5026 (fp_valid_at): Likewise.
5028 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5030 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5031 (qdf24xx_addrcost_table): Likewise.
5032 (cortexa57_tunings): Update to use generic_branch_cost.
5033 (cortexa72_tunings): Likewise.
5034 (cortexa73_tunings): Likewise.
5035 (qdf24xx_tunings): Likewise.
5037 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5039 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5040 (thunderx2t99_branch_cost): Likewise.
5041 (cortexa35_tunings): Update to use generic_branch_cost.
5042 (cortexa53_tunings): Likewise.
5043 (cortexa57_tunings): Likewise.
5044 (cortexa72_tunings): Likewise.
5045 (cortexa73_tunings): Likewise.
5046 (thunderx2t99_tunings): Likewise.
5048 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5050 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5051 (sparc_option_override): Honour MASK_FSMULD.
5052 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5053 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5054 * config/sparc/sparc.opt (mfsmuld): New option.
5055 * doc/invoke.texi (mfsmuld): Document option.
5057 2017-07-26 Marek Polacek <polacek@redhat.com>
5060 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5062 2017-07-26 Richard Biener <rguenther@suse.de>
5064 * gimple-match-head.c (do_valueize): Return OP if valueize
5066 (get_def): New helper to get at the def stmt of a SSA name
5068 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5069 do_valueize to get at the def stmt.
5070 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5072 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5075 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5077 2017-07-26 Martin Liska <mliska@suse.cz>
5080 * function.c (expand_function_start): Make expansion of
5081 nonlocal_goto_save_area after parm_birth_insn.
5083 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5085 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5086 from all CPU target flags enable members.
5088 2017-07-26 Richard Biener <rguenther@suse.de>
5090 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5091 (decision_tree::gen): Make 'type' const.
5092 (write_predicate): Likewise.
5094 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5096 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5097 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5098 (rs6000_option_override_internal): Likewise.
5099 (rs6000_expand_vector_set): Likewise.
5100 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5101 (TARGET_UPPER_REGS_SF): Likewise.
5102 (TARGET_UPPER_REGS_DI): Likewise.
5103 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5104 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5105 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5106 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5107 (Splitters for DI constants in Altivec registers): Likewise.
5108 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5109 (vsx_set_v4sf_p9): Likewise.
5110 (vsx_set_v4sf_p9_zero): Likewise.
5111 (vsx_insert_extract_v4sf_p9): Likewise.
5112 (vsx_insert_extract_v4sf_p9_2): Likewise.
5114 2017-07-25 Carl Love <cel@us.ibm.com>
5116 * doc/extend.texi: Update the built-in documentation file for the
5117 existing built-in functions
5118 vector signed char vec_cnttz (vector signed char);
5119 vector unsigned char vec_cnttz (vector unsigned char);
5120 vector signed short vec_cnttz (vector signed short);
5121 vector unsigned short vec_cnttz (vector unsigned short);
5122 vector signed int vec_cnttz (vector signed int);
5123 vector unsigned int vec_cnttz (vector unsigned int);
5124 vector signed long long vec_cnttz (vector signed long long);
5125 vector unsigned long long vec_cnttz (vector unsigned long long);
5127 2017-07-25 Andrew Pinski <apinski@cavium.com>
5129 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5130 accesses where the use is for the first operand of a BIT_INSERT.
5132 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5135 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5136 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5138 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5140 * config/i386/gstabs.h: Delete.
5141 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5143 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5145 * config/i386/i386.c (ix86_decompose_address): Do not check for
5146 register RTX when looking at index_reg or base_reg.
5147 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5149 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5151 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5152 to update EH info here.
5154 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5156 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5158 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5160 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5162 2017-07-25 Torsten Duwe <duwe@suse.de>
5164 * common.opt: Introduce -fpatchable-function-entry
5165 command line option, and its variables function_entry_patch_area_size
5166 and function_entry_patch_area_start.
5167 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5168 including a two-value parser.
5169 * target.def (print_patchable_function_entry): New target hook.
5170 * targhooks.h (default_print_patchable_function_entry): New function.
5171 * targhooks.c (default_print_patchable_function_entry): Likewise.
5172 * toplev.c (process_options): Switch off IPA-RA if
5173 patchable function entries are being generated.
5174 * varasm.c (assemble_start_function): Look at the
5175 patchable-function-entry command line switch and current
5176 function attributes and maybe generate NOP instructions by
5177 calling the print_patchable_function_entry hook.
5178 * doc/extend.texi: Document patchable_function_entry attribute.
5179 * doc/invoke.texi: Document -fpatchable_function_entry
5180 command line option.
5181 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5183 * doc/tm.texi: Re-generate.
5185 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5188 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5189 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5191 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5193 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5195 2017-07-25 Richard Biener <rguenther@suse.de>
5197 PR tree-optimization/81455
5198 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5199 not walk in cycles when looking for guards.
5201 2017-07-25 Richard Biener <rguenther@suse.de>
5203 PR tree-optimization/81529
5204 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5205 when optimizing backedge uses.
5207 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5209 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5211 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5212 to dl_section_ref. On AIX, append an expression to subtract
5213 the size of the section length to dl_section_ref.
5215 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5217 * configure.ac: If any of the config.* scripts fail, exit 1.
5218 * configure: Regenerate.
5220 2017-07-25 Richard Biener <rguenther@suse.de>
5223 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5224 of immediate uses, be more verbose on errors.
5226 2017-07-25 Richard Biener <rguenther@suse.de>
5228 PR tree-optimization/81510
5229 * tree-vect-loop.c (vect_is_simple_reduction): When the
5230 reduction stmt is not inside the loop bail out.
5232 2017-07-25 Richard Biener <rguenther@suse.de>
5234 PR tree-optimization/81303
5235 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5236 profitability check against LOOP_VINFO_NITERSM1.
5238 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5240 * domwalk.c (cmp_bb_postorder): Simplify.
5241 (sort_bbs_postorder): New function. Use it...
5242 (dom_walker::walk): ...here to optimize common cases.
5244 2017-07-25 Martin Liska <mliska@suse.cz>
5247 * ipa-visibility.c (function_and_variable_visibility): Make the
5248 redirection just on target that supports aliasing.
5249 Fix GNU coding style.
5251 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5254 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5256 * config/arm/rtems.h: Likewise.
5257 * config/bfin/rtems.h: Likewise.
5258 * config/i386/rtemself.h: Likewise.
5259 * config/lm32/rtems.h: Likewise.
5260 * config/m32c/rtems.h: Likewise.
5261 * config/m68k/rtemself.h: Likewise.
5262 * config/microblaze/rtems.h: Likewise.
5263 * config/mips/rtems.h: Likewise.
5264 * config/moxie/rtems.h: Likewise.
5265 * config/nios2/rtems.h: Likewise.
5266 * config/powerpcspe/rtems.h: Likewise.
5267 * config/rs6000/rtems.h: Likewise.
5268 * config/rtems.h: Likewise.
5269 * config/sh/rtems.h: Likewise.
5270 * config/sh/rtemself.h: Likewise.
5271 * config/sparc/rtemself.h: Likewise.
5273 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5276 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5277 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5278 * tree-ssa-structalias.c (alias_get_name): Same.
5280 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5283 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5284 instructions if no du chain is found.
5286 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5288 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5290 2017-07-25 Richard Biener <rguenther@suse.de>
5293 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5296 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5298 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5300 (ISA_2_7_MASKS_SERVER): Likewise.
5301 (ISA_3_0_MASKS_IEEE): Likewise.
5302 (OTHER_P8_VECTOR_MASKS): Likewise.
5303 (OTHER_VSX_VECTOR_MASKS): Likewise.
5304 (POWERPC_MASKS): Likewise.
5305 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5306 duplicate list of options.
5307 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5308 explicit -mupper-regs options.
5309 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5310 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5311 alias for -mupper-regs-df.
5312 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5313 (rs6000_init_hard_regno_mode_ok): Likewise.
5314 (rs6000_option_override_internal): Likewise.
5315 (rs6000_opt_masks): Likewise.
5316 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5317 options in terms of whether -mvsx or -mpower8-vector was used.
5318 (TARGET_UPPER_REGS_DI): Likewise.
5319 (TARGET_UPPER_REGS_SF): Likewise.
5320 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5321 -mupper-regs-* options.
5323 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5325 * passes.c (emergency_dump_function): Print some empty lines and a
5326 header before the RTL dump.
5328 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5330 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5332 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5335 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5336 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5338 2017-07-24 Carl Love <cel@us.ibm.com>
5340 * config/rs6000/rs6000-c.c: Add support for built-in functions
5341 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5342 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5343 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5344 vec_extract_fp_from_shortl): Add defines for the two builtins.
5345 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5346 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5348 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5349 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5350 * doc/extend.texi: Update the built-in documentation file for the
5351 new built-in function.
5353 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5356 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5358 * doc/generic.texi: Likewise.
5359 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5360 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5362 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5364 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5365 (aarch64_mls_elt_merge<mode>): Likewise.
5367 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5369 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5370 having __cxa_atexit.
5372 2017-07-23 Michael Collison <michael.collison@arm.com>
5374 * config/arm/arm.c (arm_option_override): Deprecate
5375 use of -mstructure-size-boundary.
5376 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5377 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5379 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5382 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5383 Reduce cost estimate for direct moves.
5385 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5388 * config/i386/i386.c (ix86_option_override_internal): Disable
5389 BMI, BMI2 and TBM instructions for -m16.
5391 2017-07-21 Carl Love <cel@us.ibm.com>
5393 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5394 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5395 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5396 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5397 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5398 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5399 VMULOSW): New enum "unspec" values.
5400 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5401 altivec_vmulosw): New patterns.
5402 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5403 VMULOSW): Add definitions.
5405 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5407 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5408 (qdf24xx): Likewise.
5409 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5410 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5411 (AARCH64_FL_V8_1): Renumber.
5412 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5413 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5414 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5415 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5416 rdma to feature modifiers list.
5418 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5421 * ipa-visibility (function_and_variable_visibility): Convert
5422 recursive PLT call to direct call if appropriate.
5424 2017-07-21 Andrew Pinski <apinski@cavium.com>
5426 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5427 operand 1 to see if the types precision matches.
5428 * fold-const.c (operand_equal_p): Likewise.
5430 2017-07-21 Richard Biener <rguenther@suse.de>
5432 PR tree-optimization/81303
5433 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5434 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5435 (vect_peeling_hash_get_lowest_cost): Adjust.
5436 (vect_enhance_data_refs_alignment): Likewise. Use
5437 vect_get_peeling_costs_all_drs to compute the penalty for no
5438 peeling to match up costs.
5440 2017-07-21 Richard Biener <rguenther@suse.de>
5442 PR tree-optimization/81500
5443 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5444 we didn't identify a reduction path.
5446 2017-07-21 Tom de Vries <tom@codesourcery.com>
5447 Cesar Philippidis <cesar@codesourcery.com>
5449 PR gcov-profile/81442
5450 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5453 2017-07-21 Tom de Vries <tom@codesourcery.com>
5456 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5458 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5459 nvptx_override_options_after_change.
5461 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5463 * dwarf2out.c (output_file_names): Avoid double testing for
5466 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5468 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5470 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5472 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5474 (try_crossjump_to_edge): Do not punt on partitioned functions.
5476 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5478 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5479 Put all BBs reachable only via paths crossing cold region to cold
5481 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5483 2016-07-21 Richard Biener <rguenther@suse.de>
5485 PR tree-optimization/81303
5486 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5487 into account prologue and epilogue iterations when raising
5488 min_profitable_iters to sth at least covering one vector iteration.
5490 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5492 * config/arm/arm.c (arm_test_cpu_arch_dat):
5495 2017-07-20 Nathan Sidwell <nathan@acm.org>
5497 Remove TYPE_METHODS.
5498 * tree.h (TYPE_METHODS): Delete.
5499 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5500 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5501 (dbxout_type_methods): Scan TYPE_FIELDS.
5502 (dbxout_type): Don't check TYPE_METHODS here.
5503 * function.c (use_register_for_decl): Always ignore register for
5504 class types when not optimizing.
5505 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5506 * tree.c (free_lang_data_in_type): Stitch out member functions and
5507 templates from TYPE_FIELDS.
5508 (build_distinct_type_copy, verify_type_variant,
5509 verify_type): Member fns are on TYPE_FIELDS.
5510 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5511 * tree-pretty-print.c (dump_generic_node): Likewise.
5513 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5516 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5517 V2TImode and V4TImode.
5518 (ix86_expand_vector_extract): Likewise.
5519 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5520 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5521 (ssescalarmode): Handle V4TImode and V2TImode.
5522 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5523 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5524 (VEXTRACTI128_MODE): New mode iterator.
5525 (splitter for *vec_extractv?ti first element): New.
5526 (VEC_INIT_MODE): New mode iterator.
5527 (vec_init<mode>): Consolidate 3 expanders into one using
5528 VEC_INIT_MODE mode iterator.
5530 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5532 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5533 non_spilled_static_chain_regno_p.
5535 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5537 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5539 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5541 * bb-reorder.c (connect_traces): Allow copying of blocks within
5544 2017-07-20 Richard Biener <rguenther@suse.de>
5546 * gimple.h (gimple_phi_result): Add gphi * overload.
5547 (gimple_phi_result_ptr): Likewise.
5548 (gimple_phi_arg): Likewise. Adjust index assert to only
5549 allow actual argument accesses rather than all slots available
5551 (gimple_phi_arg_def): Add gphi * overload.
5552 * tree-phinodes.c (make_phi_node): Initialize only actual
5554 (resize_phi_node): Clear memory not covered by old node,
5555 do not initialize excess argument slots.
5556 (reserve_phi_args_for_new_edge): Initialize new argument slot
5559 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5561 PR tree-optimization/81388
5563 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5565 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5566 by removing computation of may_be_zero.
5568 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5569 Tom de Vries <tom@codesourcery.com>
5572 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5573 when gimple level profile disagrees with what RTL expander did.
5575 2017-07-20 Richard Biener <rguenther@suse.de>
5577 PR tree-optimization/61171
5578 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5579 (vect_analyze_stmt): Add slp instance parameter.
5580 (vectorizable_reduction): Likewise.
5581 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5582 (vect_is_simple_reduction): Deal with chains not detected
5583 as SLP reduction chain, specifically not properly associated
5584 chains containing a mix of plus/minus.
5585 (get_reduction_op): Remove.
5586 (get_initial_defs_for_reduction): Simplify, pass in whether
5587 this is a reduction chain, pass in the SLP node for the PHIs.
5588 (vect_create_epilog_for_reduction): Get the SLP instance as
5590 (vectorizable_reduction): Get the SLP instance as arg.
5591 During analysis remember the SLP node with the PHIs in the
5592 instance. Simplify getting at the vectorized reduction PHIs.
5593 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5594 through SLP instance.
5595 (vect_slp_analyze_operations): Likewise.
5596 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5597 (vect_transform_stmt): Likewise.
5599 2017-07-20 Tom de Vries <tom@codesourcery.com>
5601 PR tree-optimization/81489
5602 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5603 read of phi arg location to before loop that modifies phi.
5605 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5607 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5610 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5613 * except.c (execute): Fix ordering issue.
5615 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5617 PR rtl-optimization/81423
5618 * combine.c (make_compound_operation_int): Don't try to optimize
5619 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5621 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5623 PR rtl-optimization/81423
5624 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5625 with a constant that is -1 in the truncated to mode.
5627 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5629 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5630 (determine_unlikely_bbs): ... here.
5631 * predict.h (propagate_unlikely_bbs_forward): Declare.
5632 * cfgexpand.c (pass_expand::execute): Use it.
5633 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5635 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5636 propagate_unlikely_bbs_forward.
5638 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5641 * except.c (maybe_add_nop_after_section_switch): New function.
5644 2017-07-19 Tom de Vries <tom@codesourcery.com>
5646 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5648 2017-07-19 Tom de Vries <tom@codesourcery.com>
5650 * gimple.h (gimple_phi_arg): Make assert more strict.
5652 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5654 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5655 * config/rs6000/mmintrin.h: New file.
5656 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5658 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5660 PR tree-optimization/81346
5661 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5663 2017-07-19 Tom de Vries <tom@codesourcery.com>
5665 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5667 2017-07-19 Tom de Vries <tom@codesourcery.com>
5669 * config/nvptx/nvptx-modes.def: Add V2DImode.
5670 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5671 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5672 (nvptx_output_mov_insn): Handle lack of mov.b128.
5673 (nvptx_print_operand): Handle 'H' and 'L' codes.
5674 (nvptx_vector_mode_supported): Allow V2DImode.
5675 (nvptx_preferred_simd_mode): New function.
5676 (nvptx_data_alignment): New function.
5677 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5678 nvptx_preferred_simd_mode.
5679 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5681 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5683 2017-07-19 Tom de Vries <tom@codesourcery.com>
5685 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5686 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5687 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5688 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5689 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5690 (mov<VECIM>_insn): New define_insn.
5691 (define_expand "mov<VECIM>): New define_expand.
5693 2017-07-19 Tom de Vries <tom@codesourcery.com>
5695 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5697 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5699 PR tree-optimization/81346
5700 * fold-const.h (fold_div_compare, range_check_type): Declare.
5701 * fold-const.c (range_check_type): New function.
5702 (build_range_check): Use range_check_type.
5703 (fold_div_compare): No longer static, rewritten into
5704 a match.pd helper function.
5705 (fold_comparison): Don't call fold_div_compare here.
5706 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5709 2017-07-19 Nathan Sidwell <nathan@acm.org>
5711 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5712 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5713 * tree.c (find_decls_types_r, verify_type): Use
5714 TYPE_{MIN,MAX}_VALUE_RAW.
5715 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5716 (hash_tree): Likewise.
5717 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5719 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5722 2017-07-18 Tom de Vries <tom@codesourcery.com>
5725 * omp-expand.c (expand_omp_for_static_chunk): Handle
5726 equal-argument loop exit phi.
5728 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5731 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5732 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5733 operand 2 predicate.
5734 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5735 operand 2 predicate.
5736 (ror,rol -> rorx splitters): Use const_int_operand as
5737 operand 2 predicate.
5739 2017-06-18 Richard Biener <rguenther@suse.de>
5741 PR tree-optimization/81410
5742 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5743 the gap in the ! slp_perm SLP case after each group.
5745 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5748 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5751 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5754 * predict.c (set_even_probabilities): Cleanup; do not affect
5755 probabilities that are already known.
5756 (combine_predictions_for_bb): Call even when count is set.
5758 2017-07-18 Nathan Sidwell <nathan@acm.org>
5760 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5763 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5766 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5767 optimization for loop niter analysis.
5769 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5772 * config/avr/avr.c (avr_optimize_casesi): Don't use
5773 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5775 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5777 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5778 body_cost_vec from _vect_peel_extended_info.
5779 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5780 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5783 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5785 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5787 2017-07-18 Richard Biener <rguenther@suse.de>
5789 PR tree-optimization/80620
5790 PR tree-optimization/81403
5791 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5792 info when re-using a VN table entry.
5794 2017-07-18 Richard Biener <rguenther@suse.de>
5796 PR tree-optimization/81418
5797 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5798 vectype_in. Verify that with lane-reducing reduction operations
5799 we have a single def-use cycle.
5801 2017-07-17 Carl Love <cel@us.ibm.com>
5803 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5805 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5806 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5807 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5808 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5809 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5810 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5811 VMULOSW): New enum "unspec" values.
5812 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5813 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5814 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5815 altivec_vmulosw): New patterns.
5816 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5817 VMULOSW): Add definitions.
5818 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5820 * config/alpha/alpha.c: Include predict.h.
5822 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5824 * tree-vrp.c (compare_assert_loc): Fix comparison function
5825 to return predictable results.
5827 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5829 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5832 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5833 * doc/invoke.texi (mexpand-adddi): Update text.
5835 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5837 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5838 that also clobbers the CC register. The old expand code is moved
5840 (*arc_clzsi2): ... here.
5841 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5842 the CC register. The old expand code is moved to ...
5843 (arc_ctzsi2): ... here.
5845 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5847 * config/arc/arc.opt (mindexed-loads): Use initial value
5848 TARGET_INDEXED_LOADS_DEFAULT.
5849 (mauto-modify-reg): Use initial value
5850 TARGET_AUTO_MODIFY_REG_DEFAULT.
5851 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5852 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5853 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5854 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5856 2017-07-17 Martin Liska <mliska@suse.cz>
5859 * opts.c (finish_options): Do not allow -fgnu-tm
5860 w/ -fsanitize={kernel-,}address. Say sorry.
5862 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5865 * tree-loop-distribution.c (classify_partition): Only assert on
5866 numer of iterations.
5867 (merge_dep_scc_partitions): Delete prameter. Update function call.
5868 (distribute_loop): Remove code handling loop with unknown niters.
5869 (pass_loop_distribution::execute): Skip loop with unknown niters.
5871 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5874 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5875 function sort_partitions_by_post_order.
5877 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5879 PR tree-optimization/81374
5880 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5881 the max index of basic blocks, rather than number of basic blocks.
5883 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5885 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5887 (arc_legitimate_pic_operand_p): Likewise.
5888 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5890 (arc_needs_pcl_p): Likewise.
5891 (arc_legitimate_pc_offset_p): Likewise.
5892 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5893 function is also used in constrains.md.
5894 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5895 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5896 PLUS. Only return true/false in known cases, otherwise assert.
5897 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5898 is already called in arc_legitimate_constant_p.
5899 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5901 (LEGITIMATE_PIC_OPERAND_P): Use
5902 arc_raw_symbolic_reference_mentioned_p function.
5903 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5908 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5909 Andrew Burgess <andrew.burgess@embecosm.com>
5911 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5912 (arc_return_address_register): New function.
5913 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5914 (arc_handle_fndecl_attribute): Add naked attribute.
5915 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5916 (TARGET_WARN_FUNC_RETURN): Likewise.
5917 (arc_allocate_stack_slots_for_args): New function.
5918 (arc_warn_func_return): Likewise.
5919 (machine_function): Change type fn_type.
5920 (arc_compute_function_type): Consider new naked function type,
5921 change function return type.
5922 (arc_must_save_register): Adapt to handle new
5923 arc_compute_function_type's return type.
5924 (arc_expand_prologue): Likewise.
5925 (arc_expand_epilogue): Likewise.
5926 (arc_return_address_regs): Delete.
5927 (arc_return_address_register): New function.
5928 (arc_epilogue_uses): Use above function.
5929 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5930 (arc_function_type): Change encoding, add naked type.
5931 (ARC_INTERRUPT_P): Change to handle the new encoding.
5932 (ARC_FAST_INTERRUPT_P): Likewise.
5933 (ARC_NORMAL_P): Define.
5934 (ARC_NAKED_P): Likewise.
5935 (arc_compute_function_type): Delete prototype.
5936 * config/arc/arc.md (in_ret_delay_slot): Use
5937 arc_return_address_register function.
5938 (simple_return): Likewise.
5939 (p_return_i): Likewise.
5941 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5943 PR tree-optimization/81428
5944 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5945 can't be built for those types.
5947 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5949 Remove stuff dead since r239246.
5951 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5952 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5953 (avr_inform_devices): Remove dead stuff.
5955 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5957 * config/arm/arm_neon.h: Fix softp typo.
5959 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5961 PR tree-optimization/81365
5962 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5963 aggregate moves onto bb predecessor edges, make sure there are no
5964 loads that could alias the lhs in between the start of bb and the
5967 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5970 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5971 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5972 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5974 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5976 PR tree-optimization/81396
5977 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5978 (init_symbolic_number): Initialize it to 1.
5979 (perform_symbolic_merge): Add n_ops from both operands into the new
5981 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5982 without base_addr as useless if they need more than one operation.
5983 (bswap_replace): Handle !bswap case for NULL base_addr.
5985 2017-07-17 Tom de Vries <tom@codesourcery.com>
5988 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5991 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5993 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5994 conditional builtin define __FIX_LEON3FT_B2BST.
5996 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5998 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5999 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6002 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6004 PR rtl-optimization/81424
6005 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6006 to remove potential trapping from operands if -fnon-call-exceptions.
6008 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6010 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6011 profile_proability for scalling.
6012 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6014 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6016 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6018 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6020 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6021 fixpoint arithmetics.
6023 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6025 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6026 fixpoint arithmetics.
6028 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6030 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6031 fixpoint arithmetics.
6033 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6035 * profile-count.h (profile_probability::from_reg_br_prob_note,
6036 profile_probability::to_reg_br_prob_note): New functions.
6037 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6038 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6039 * predict.c (probability_reliable_p): Update.
6040 (edge_probability_reliable_p): Update.
6041 (br_prob_note_reliable_p): Update.
6042 (invert_br_probabilities): Update.
6043 (add_reg_br_prob_note): New function.
6044 (combine_predictions_for_insn): Update.
6045 * asan.c (asan_clear_shadow): Update.
6046 * cfgbuild.c (compute_outgoing_frequencies): Update.
6047 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6048 (update_br_prob_note): Update.
6049 (rtl_verify_edges): Update.
6050 (purge_dead_edges): Update.
6051 (fixup_reorder_chain): Update.
6052 * emit-rtl.c (try_split): Update.
6053 * ifcvt.c (cond_exec_process_insns): Update.
6054 (cond_exec_process_if_block): Update.
6055 (dead_or_predicable): Update.
6056 * internal-fn.c (expand_addsub_overflow): Update.
6057 (expand_neg_overflow): Update.
6058 (expand_mul_overflow): Update.
6059 * loop-doloop.c (doloop_modify): Update.
6060 * loop-unroll.c (compare_and_jump_seq): Update.
6061 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6062 * predict.h: Update.
6063 * reorg.c (mostly_true_jump): Update.
6065 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6066 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6067 * config/arc/arc.c: (emit_unlikely_jump): Update.
6068 * config/arm/arm.c: (emit_unlikely_jump): Update.
6069 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6070 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6071 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6072 (ix86_print_operand): Update.
6073 (ix86_split_fp_branch): Update.
6074 (predict_jump): Update.
6075 * config/ia64/ia64.c (ia64_print_operand): Update.
6076 * config/mmix/mmix.c (mmix_print_operand): Update.
6077 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6078 (rs6000_expand_split_stack_prologue): Update.
6079 * config/rs6000/rs6000.c: Update.
6080 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6081 (s390_expand_vec_movstr): Update.
6082 (s390_expand_cs_tdsi): Update.
6083 (s390_expand_split_stack_prologue): Update.
6084 * config/sh/sh.c (sh_print_operand): Update.
6085 (expand_cbranchsi4): Update.
6086 (expand_cbranchdi4): Update.
6087 * config/sparc/sparc.c (output_v9branch): Update.
6088 * config/spu/spu.c (get_branch_target): Update.
6089 (ea_load_store_inline): Update.
6090 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6091 * config/tilepro/tilepro.c: Update.
6093 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6095 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6096 (gimplify_save_expr): Likewise.
6098 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6100 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6102 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6104 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6106 * timevar.def (TV_IPA_FNSUMMARY): Define.
6108 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6110 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6111 to back store errata sensitive sequence from being generated.
6112 (sqrtdf2_fix): Likewise.
6114 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6116 * tree-ssa-threadupdate.c (compute_path_counts,
6117 update_joiner_offpath_counts): Use profile_probability.
6119 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
6122 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6124 * config/arm/arm-c.c (arm_cpu_builtins): Define
6125 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6127 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
6129 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6130 array entries to represent __ieee128 versions of the
6131 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6132 scalar_extract_sig, and scalar_insert_exp built-in functions.
6133 (altivec_resolve_overloaded_builtin): Add special case handling
6134 for the __builtin_scalar_insert_exp function, as represented by
6135 the P9V_BUILTIN_VEC_VSIEDP constant.
6136 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6137 exponent support for __ieee128 argument.
6138 (VSESQP): Add scalar extract signature support for __ieee128
6140 (VSTDCNQP): Add scalar test negative support for __ieee128
6142 (VSIEQP): Add scalar insert exponent support for __int128 argument
6143 with __ieee128 result.
6144 (VSIEQPF): Add scalar insert exponent support for __ieee128
6145 argument with __ieee128 result.
6146 (VSTDCQP): Add scalar test data class support for __ieee128
6148 (VSTDCNQP): Add overload support for scalar test negative with
6150 (VSTDCQP): Add overload support for scalar test data class
6152 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6154 (UNSPEC_VSX_SIEXPQP): New constant.
6155 (xsxexpqp): New insn for VSX scalar extract exponent quad
6157 (xsxsigqp): New insn for VSX scalar extract significand quad
6159 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6160 precision with floating point argument.
6161 (xststdcqp): New expand for VSX scalar test data class quad
6163 (xststdcnegqp): New expand for VSX scalar test negative quad
6165 (xststdcqp): New insn to match expansions for VSX scalar test data
6166 class quad precision and VSX scalar test negative quad precision.
6167 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6168 special case operand checking to enforce that second operand of
6169 VSX scalar test data class with quad precision argument is a 7-bit
6171 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6172 prototypes and descriptions of __ieee128 versions of
6173 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6174 scalar_test_data_class, and scalar_test_neg built-in functions.
6176 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6178 PR tree-optimization/81162
6179 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6180 replace a negate with an add.
6182 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6184 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6186 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6188 * config/arm/arm-c.c (arm_cpu_builtins): Define
6189 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6191 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6193 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6194 (armv8-r): Set ARM Cortex-R52 as default CPU.
6195 * config/arm/arm-tables.opt: Regenerate.
6196 * config/arm/arm-tune.md: Regenerate.
6197 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6199 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6200 extension for -mcpu=cortex-r52.
6202 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6204 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6205 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6206 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6207 (fp-armv8): Define it as FP_ARMv8 only.
6208 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6209 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6211 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6212 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6213 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6214 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6215 than TARGET_FPU_ARMV8.
6216 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6217 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6218 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6220 * config/arm/neon.md (neon_vrint): Likewise.
6221 (neon_vcvt): Likewise.
6222 (neon_<fmaxmin_op><mode>): Likewise.
6223 (<fmaxmin><mode>3): Likewise.
6224 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6225 * config/arm/predicates.md (arm_cond_move_operator): Check against
6226 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6228 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6230 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6233 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6235 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6236 loop in comment with memset.
6238 2017-07-14 Martin Liska <mliska@suse.cz>
6240 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6241 * dwarf2out.c (is_java): Remove the function.
6242 (output_pubname): Remove usage of the function.
6243 (lower_bound_default): Remove usage of DW_LANG_Java.
6244 (gen_compile_unit_die): Likewise.
6245 * gcc.c: Remove compiler defaults for .java and .zip files.
6246 * gimple-expr.c (remove_suffix): Change as there's no longer
6247 extension than 4-letter one.
6248 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6249 (gimplify_save_expr): Likewise.
6250 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6251 as it's possible even for other languages than Java.
6252 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6253 * lto-opts.c (lto_write_options): Remove reference to Java.
6254 * opts.c (strip_off_ending): Update file extension handling.
6255 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6256 * tree-eh.c (lower_resx): Likewise.
6257 * tree.c (free_lang_data_in_type): Remove dead code.
6258 (find_decls_types_r): Likewise.
6259 (build_common_builtin_nodes): Remove Java from a comment.
6260 (verify_type): Remove dead code.
6261 * varasm.c (assemble_external): Remove Java from a comment.
6263 2017-07-14 Martin Liska <mliska@suse.cz>
6265 * opts.c (finish_options): Add quotes.
6266 (common_handle_option): Likewise.
6268 2017-07-14 Martin Liska <mliska@suse.cz>
6270 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6271 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6273 * dwarf2out.c (lower_bound_default): Do not handle
6275 (gen_compile_unit_die): Likewise.
6276 * gcc.c: Remove default extension binding for GNU Pascal.
6277 * stmt.c: Remove Pascal language from a comment.
6278 * xcoffout.c: Likewise.
6280 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6283 * diagnostic-show-locus.c (fixit_cmp): New function.
6284 (layout::layout): Sort m_fixit_hints.
6285 (column_range::column_range): Assert that the values are valid.
6286 (struct char_span): New struct.
6287 (correction::overwrite): New method.
6288 (struct source_line): New struct.
6289 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6290 calls in terms of classes source_line and char_span, and
6291 correction::overwrite.
6292 (selftest::test_overlapped_fixit_printing_2): New function.
6293 (selftest::diagnostic_show_locus_c_tests): Call it.
6295 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6297 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6298 early if there is no lhs.
6300 2017-07-13 Martin Liska <mliska@suse.cz>
6302 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6303 (gen_reference_type_die): Likewise.
6304 * stor-layout.c: Remove Pascal-related comment.
6306 2017-07-13 Martin Liska <mliska@suse.cz>
6308 * opts.c (finish_options): Add quotes to error messages.
6309 (parse_sanitizer_options): Likewise.
6311 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6313 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6315 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6317 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6319 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6321 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6323 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6325 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6328 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6329 provides the hardware capability bits, define the macro
6330 __BUILTIN_CPU_SUPPORTS__.
6331 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6332 if GLIBC does not provide the hardware capability bits. Add a
6333 gcc_unreachable call if the built-in cpu function is neither
6334 __builtin_cpu_is nor __builtin_cpu_supports.
6335 (rs6000_get_function_versions_dispatcher): Change the warning
6336 that an old GLIBC is used which does not export the capability
6337 bits to be an error.
6338 * doc/extend.texi (target_clones attribute): Document the
6339 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6340 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6341 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6342 the macros defined by GCC if the newer GLIBC is available.
6344 2017-07-12 Jeff Law <law@redhat.com>
6346 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6347 remaining includes slightly.
6348 * config/riscv/riscv-builtins.c: Include profile-count.h.
6350 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6353 * config/avr/avr.c (avr_set_current_function): In diagnostic
6354 messages: Quote keywords and (parts of) identifiers.
6355 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6358 2017-07-12 Carl Love <cel@us.ibm.com>
6360 * config/rs6000/rs6000-c.c: Add support for built-in functions
6361 vector bool char vec_revb (vector bool char);
6362 vector bool short vec_revb (vector short char);
6363 vector bool int vec_revb (vector bool int);
6364 vector bool long long vec_revb (vector bool long long);
6365 * doc/extend.texi: Update the built-in documentation file for the
6366 new built-in functions.
6368 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6370 * config/s390/s390.md: Remove movcc splitter.
6372 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6374 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6375 load/store on condition.
6377 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6380 * config/avr/avr.c (avr_encode_section_info)
6381 [progmem && !TREE_READONLY]: Error if progmem object needs
6384 2017-07-11 Michael Collison <michael.collison@arm.com>
6386 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6389 2017-07-11 Carl Love <cel@us.ibm.com>
6391 * config/rs6000/rs6000-c.c: Add support for builtins
6392 vector unsigned int vec_parity_lsbb (vector signed int);
6393 vector unsigned int vec_parity_lsbb (vector unsigned int);
6394 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6395 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6396 vector unsigned long long vec_parity_lsbb (vector signed long long);
6397 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6398 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6399 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6400 * doc/extend.texi: Update the built-in documentation file for the
6401 new built-in functions.
6403 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6405 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6406 (layout::m_primary_loc): New field.
6407 (layout::layout): Initialize new field. Move location filtering
6408 logic from here to...
6409 (layout::maybe_add_location_range): ...this new method. Add
6410 support for filtering to just the lines already specified by other
6412 (layout::will_show_line_p): New method.
6413 (gcc_rich_location::add_location_if_nearby): New method.
6414 (selftest::test_add_location_if_nearby): New test function.
6415 (selftest::diagnostic_show_locus_c_tests): Call it.
6416 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6419 2017-07-11 Tom de Vries <tom@codesourcery.com>
6421 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6422 (bb_first_real_insn): New function.
6423 (nvptx_single): Add extra initialization of broadcasted condition
6426 2017-07-11 Nathan Sidwell <nathan@acm.org>
6428 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6430 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6432 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6433 Binutils doc as TEXI will mess them up.
6434 * doc/invoke.texi (AVR Options): Same here.
6436 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6438 * config/sparc/sparc.opt (mfix-ut700): New option.
6439 (mfix-gr712rc): Likewise.
6440 (sparc_fix_b2bst): New variable.
6441 * doc/invoke.texi (SPARC options): Document them.
6442 (ARM options): Fix warnings.
6443 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6444 instructions to prevent sequences that can trigger the store-store
6445 errata for certain LEON3FT processors.
6446 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6447 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6448 * config/sparc/sparc.md (fix_b2bst): New attribute.
6449 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6451 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6454 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6456 (*rsqrtsf2_sse): Ditto.
6458 (div<mode>3): Macroize insn from divdf3 and divsf3
6459 using MODEF mode iterator.
6461 2017-07-10 Martin Sebor <msebor@redhat.com>
6463 PR tree-optimization/80397
6464 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6465 instead of testing for equality to INTEGER_TYPE.
6467 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6469 * config.gcc: Remove uclibc from arc target spec.
6471 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6473 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6475 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6478 * lto-wrapper.c (remove_option): New function.
6479 (merge_and_complain): Merge PIC/PIE options more realistically.
6481 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6483 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6487 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6488 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6489 * config.in: Regenerate.
6490 * configure: Regenerate.
6491 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6492 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6493 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6494 (TARGET_GASISR_PROLOGUES): ...target mask.
6495 * common/config/avr/avr-common.c
6496 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6497 Set -mgas-isr-prologues.
6498 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6499 INSERT_PASS_BEFORE for it.
6500 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6501 * config/avr/avr.c (avr_option_override)
6502 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6503 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6504 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6505 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6506 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6508 (make_avr_pass_pre_proep): New function.
6509 (emit_push_sfr) <treg>: Add argument to function and use it
6511 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6512 and set machine->gasisr.yes.
6513 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6514 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6515 __gcc_isr.n_pushed to .L__stack_usage.
6516 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6517 (avr_asm_final_postscan_insn): ...this new static function.
6518 * config/avr/avr.h (machine_function)
6519 <is_no_gccisr, use_L__stack_usage>: New fields.
6520 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6521 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6522 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6523 (gasisr, *gasisr): New expander and insn.
6524 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6525 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6526 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6528 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6530 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6533 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6535 Move jump-tables out of .text again.
6538 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6539 (ASM_OUTPUT_ADDR_VEC): New function.
6540 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6541 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6542 INSN_ADDRESSes as asm comment.
6543 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6544 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6545 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6546 * config/avr/avr.md (*tablejump): Adjust comment.
6547 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6548 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6550 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6551 (avr_output_addr_vec): New proto.
6552 (avr_log_t) <insn_addresses>: New field.
6554 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6557 * config/i386/i386.c (ix86_function_arg_advance): Set
6558 outgoing_args_on_stack to true if there are outgoing arguments
6560 (ix86_function_arg): Likewise.
6561 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6562 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6563 * config/i386/i386.h (machine_function): Add
6564 outgoing_args_on_stack.
6566 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6568 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6570 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6572 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6574 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6575 (REAL_H): Remove $(MACHMODE_H).
6576 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6578 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6579 $(MACHMODE_H) and double-int.h.
6580 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6582 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6585 2017-07-07 Andrew Pinski <apinski@cavium.com>
6587 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6588 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6590 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6592 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6593 Add warning if GCC was not configured to link against a GLIBC that
6594 exports the hardware capability bits.
6595 (make_resolver_func): Make resolver function private and not a
6596 COMDAT function. Create the name with clone_function_name instead
6597 of make_unique_name.
6600 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6601 correct operand in doing the split.
6603 2017-07-07 Carl Love <cel@us.ibm.com>
6605 * config/rs6000/rs6000-c: Add support for built-in function
6606 vector unsigned short vec_pack_to_short_fp32 (vector float,
6608 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6609 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6610 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6611 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6612 (convert_4f32_8i16): Add define_expand.
6613 * doc/extend.texi: Update the built-in documentation file for the
6614 new built-in function.
6616 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6618 * config/sparc/m8.md: New file.
6619 * config/sparc/sparc.md: Include m8.md.
6621 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6623 * config/sparc/sparc.opt: New option -mvis4b.
6624 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6625 (sparc_option_override): Handle VIS4B.
6626 (enum sparc_builtins): Define
6627 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6628 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6629 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6630 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6631 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6632 (check_constant_argument): New function.
6633 (sparc_vis_init_builtins): Define builtins
6634 __builtin_vis_dictunpack{8,16,32},
6635 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6636 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6637 __builtin_vis_fpcmpde{8,16,32}shl and
6638 __builtin_vis_fpcmpur{8,16,32}shl.
6639 (sparc_expand_builtin): Check that the constant operands to
6640 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6641 constant and in range.
6642 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6644 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6645 (SPARC_IMM5_P): Likewise.
6646 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6647 (enabled): Handle vis4b.
6648 (UNSPEC_DICTUNPACK): New unspec.
6649 (UNSPEC_FPCMPSHL): Likewise.
6650 (UNSPEC_FPUCMPSHL): Likewise.
6651 (UNSPEC_FPCMPDESHL): Likewise.
6652 (UNSPEC_FPCMPURSHL): Likewise.
6653 (cpu_feature): New CPU feature `vis4b'.
6654 (dictunpack{8,16,32}): New insns.
6655 (FPCSMODE): New mode iterator.
6656 (fpcscond): New code iterator.
6657 (fpcsucond): Likewise.
6658 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6659 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6660 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6661 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6662 * config/sparc/constraints.md: Define constraints `q' for unsigned
6663 2-bit integer constants and `t' for unsigned 5-bit integer
6665 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6667 (imm5_operand_dictunpack16): Likewise.
6668 (imm5_operand_dictunpack32): Likewise.
6669 (imm2_operand): Likewise.
6670 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6671 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6672 ditunpack* and fpcmp*shl builtins.
6674 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6676 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6677 * config.in: Add HAVE_AS_SPARC6 define.
6678 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6680 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6682 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6683 (CPP_CPU_SPEC): Handle m8.
6684 (ASM_CPU_SPEC): Likewise.
6685 * config/sparc/sparc-opts.h (enum processor_type): Add
6687 * config/sparc/sparc.c (m8_costs): New struct.
6688 (sparc_option_override): Handle TARGET_CPU_m8.
6689 (sparc32_initialize_trampoline): Likewise.
6690 (sparc64_initialize_trampoline): Likewise.
6691 (sparc_issue_rate): Likewise.
6692 (sparc_register_move_cost): Likewise.
6693 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6694 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6695 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6696 (CPP_CPU_SPEC): Handle M8.
6697 (ASM_CPU_SPEC): Likewise.
6698 (AS_M8_FLAG): Define.
6699 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6700 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6701 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6703 * configure: Regenerate.
6704 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6707 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6709 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6711 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6712 ("*movdi_insn_sp32"): Do not set v3pipe.
6713 ("*movsi_insn"): Likewise.
6714 ("*movdi_insn_sp64"): Likewise.
6715 ("*movsf_insn"): Likewise.
6716 ("*movdf_insn_sp32"): Likewise.
6717 ("*movdf_insn_sp64"): Likewise.
6718 ("*zero_extendsidi2_insn_sp64"): Likewise.
6719 ("*sign_extendsidi2_insn"): Likewise.
6720 ("*mov<VM32:mode>_insn"): Likewise.
6721 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6722 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6723 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6724 ("<vlop:code><VL:mode>3"): Likewise.
6725 ("*not_<vlop:code><VL:mode>3"): Likewise.
6726 ("*nand<VL:mode>_vis"): Likewise.
6727 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6728 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6729 ("one_cmpl<VL:mode>2"): Likewise.
6730 ("faligndata<VM64:mode>_vis"): Likewise.
6731 ("alignaddrsi_vis"): Likewise.
6732 ("alignaddrdi_vis"): Likweise.
6733 ("alignaddrlsi_vis"): Likewise.
6734 ("alignaddrldi_vis"): Likewise.
6735 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6736 ("bmaskdi_vis"): Likewise.
6737 ("bmasksi_vis"): Likewise.
6738 ("bshuffle<VM64:mode>_vis"): Likewise.
6739 ("cmask8<P:mode>_vis"): Likewise.
6740 ("cmask16<P:mode>_vis"): Likewise.
6741 ("cmask32<P:mode>_vis"): Likewise.
6742 ("pdistn<P:mode>_vis"): Likewise.
6743 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6745 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6747 * config/sparc/sparc.md ("subtype"): New insn attribute.
6748 ("*wrgsr_sp64"): Set insn subtype.
6749 ("*rdgsr_sp64"): Likewise.
6750 ("alignaddrsi_vis"): Likewise.
6751 ("alignaddrdi_vis"): Likewise.
6752 ("alignaddrlsi_vis"): Likewise.
6753 ("alignaddrldi_vis"): Likewise.
6754 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6755 ("fexpand_vis"): Likewise.
6756 ("fpmerge_vis"): Likewise.
6757 ("faligndata<VM64:mode>_vis"): Likewise.
6758 ("bshuffle<VM64:mode>_vis"): Likewise.
6759 ("cmask8<P:mode>_vis"): Likewise.
6760 ("cmask16<P:mode>_vis"): Likewise.
6761 ("cmask32<P:mode>_vis"): Likewise.
6762 ("fchksm16_vis"): Likewise.
6763 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6764 ("fmean16_vis"): Likewise.
6765 ("fp<plusminus_insn>64_vis"): Likewise.
6766 ("<plusminus_insn>v8qi3"): Likewise.
6767 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6768 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6769 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6770 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6771 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6772 ("*movqi_insn"): Likewise.
6773 ("*movhi_insn"): Likewise.
6774 ("*movsi_insn"): Likewise.
6775 ("movsi_pic_gotdata_op"): Likewise.
6776 ("*movdi_insn_sp32"): Likewise.
6777 ("*movdi_insn_sp64"): Likewise.
6778 ("movdi_pic_gotdata_op"): Likewise.
6779 ("*movsf_insn"): Likewise.
6780 ("*movdf_insn_sp32"): Likewise.
6781 ("*movdf_insn_sp64"): Likewise.
6782 ("*zero_extendhisi2_insn"): Likewise.
6783 ("*zero_extendqihi2_insn"): Likewise.
6784 ("*zero_extendqisi2_insn"): Likewise.
6785 ("*zero_extendqidi2_insn"): Likewise.
6786 ("*zero_extendhidi2_insn"): Likewise.
6787 ("*zero_extendsidi2_insn_sp64"): Likewise.
6788 ("ldfsr"): Likewise.
6789 ("prefetch_64"): Likewise.
6790 ("prefetch_32"): Likewise.
6791 ("tie_ld32"): Likewise.
6792 ("tie_ld64"): Likewise.
6793 ("*tldo_ldub_sp32"): Likewise.
6794 ("*tldo_ldub1_sp32"): Likewise.
6795 ("*tldo_ldub2_sp32"): Likewise.
6796 ("*tldo_ldub_sp64"): Likewise.
6797 ("*tldo_ldub1_sp64"): Likewise.
6798 ("*tldo_ldub2_sp64"): Likewise.
6799 ("*tldo_ldub3_sp64"): Likewise.
6800 ("*tldo_lduh_sp32"): Likewise.
6801 ("*tldo_lduh1_sp32"): Likewise.
6802 ("*tldo_lduh_sp64"): Likewise.
6803 ("*tldo_lduh1_sp64"): Likewise.
6804 ("*tldo_lduh2_sp64"): Likewise.
6805 ("*tldo_lduw_sp32"): Likewise.
6806 ("*tldo_lduw_sp64"): Likewise.
6807 ("*tldo_lduw1_sp64"): Likewise.
6808 ("*tldo_ldx_sp64"): Likewise.
6809 ("*mov<VM32:mode>_insn"): Likewise.
6810 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6811 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6813 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6815 * config/sparc/sparc.md ("type"): New insn type viscmp.
6816 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6818 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6819 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6820 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6821 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6823 ("n7_vis_logical_11cycle"): Likewise.
6824 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6825 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6826 * config/sparc/niagara.md ("niag_vis"): Likewise.
6827 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6828 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6830 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6832 * config/sparc/sparc.md: New instruction type `bmask'.
6833 (bmaskdi_vis): Use the `bmask' type.
6834 (bmasksi_vis): Likewise.
6835 * config/sparc/ultra3.md (us3_array): Likewise.
6836 * config/sparc/niagara7.md (n7_array): Likewise.
6837 * config/sparc/niagara4.md (n4_array): Likewise.
6838 * config/sparc/niagara2.md (niag2_vis): Likewise.
6839 (niag3_vis): Likewise.
6840 * config/sparc/niagara.md (niag_vis): Likewise.
6842 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6844 * ipa-comdats.c: Remove optimize check from gate.
6845 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6846 for functions not optimized.
6847 (ipa_fn_summary_read): Skip optimize check.
6848 (ipa_fn_summary_write): Likewise.
6849 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6851 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6853 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6854 (check_callers): Give up on caller which is not optimized.
6855 (inline_small_functions): Likewise.
6856 (ipa_inline): Do not give up when not optimizing.
6857 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6858 away unoptimizes cdtors.
6859 (whole_program_function_and_variable_visibility): Do
6860 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6861 * ipa.c (process_references): Do not check optimize.
6862 (symbol_table::remove_unreachable_nodes): Update optimize check.
6863 (set_writeonly_bit): Update optimize check.
6864 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6865 (pass_ipa_single_use::gate): Remove.
6867 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6869 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6870 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6871 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6872 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6873 permute_load, permute_store, adjust_extract, adjust_splat,
6874 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6875 replace_swap_with_copy, dump_swap_insn_table,
6876 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6877 recombine_lvx_pattern, recombine_stvx_pattern,
6878 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6879 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6880 to file rs6000-p8swap.c.
6881 * config/rs6000/rs6000-p8swap.c: New file.
6882 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6883 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6884 and rs6000*-*-* targets.
6886 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6888 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6890 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6892 * lto-wrapper.c (merge_and_complain): Do not merge
6893 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6894 fsigned_zeros, ftrapping_math, fwrapv.
6895 (append_compiler_options): Do not track these options.
6896 (append_linker_options): Likewie
6898 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6900 * cgraphunit.c (cgraph_node::finalize_function): When
6901 !flag_toplevel_reorde set no_reorder flag.
6902 (varpool_node::finalize_decl): Likewise.
6903 (symbol_table::compile): Drop no toplevel reorder path.
6905 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6907 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6908 edges; zero probability is not better than uninitialized.
6910 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6912 * asan.h (asan_sanitize_allocas_p): Declare.
6913 * asan.c (asan_sanitize_allocas_p): New function.
6914 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6915 (handle_builtin_alloca): Likewise.
6916 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6917 if !asan_sanitize_allocas_p.
6918 * params.def (asan-instrument-allocas): Add new option.
6919 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6920 * opts.c (common_handle_option): Disable allocas sanitization for
6923 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6925 * asan.c: Include gimple-fold.h.
6926 (get_last_alloca_addr): New function.
6927 (handle_builtin_stackrestore): Likewise.
6928 (handle_builtin_alloca): Likewise.
6929 (asan_emit_allocas_unpoison): Likewise.
6930 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6931 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6932 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6933 (instrument_builtin_call): Pass gimple iterator to
6934 get_mem_refs_of_builtin_call.
6935 (last_alloca_addr): New global.
6936 * asan.h (asan_emit_allocas_unpoison): Declare.
6937 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6938 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6939 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6940 if function calls alloca.
6941 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6942 * gimple-fold.h (replace_call_with_value): Declare.
6943 * internal-fn.c: Include asan.h.
6944 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6945 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6947 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6949 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6950 (C_SELFTEST_FLAGS): New.
6951 (CPP_SELFTEST_FLAGS): New.
6952 (SELFTEST_DEPS): New, from deps of s-selftest.
6953 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6954 (CPP_SELFTEST_DEPS): New.
6955 (selftest): Add dependency on s-selftest-c++.
6956 (s-selftest): Rename to...
6957 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6958 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6959 than SELFTEST_FLAGS.
6960 (selftest-gdb): Rename to...
6961 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6963 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6964 (selftest-valgrind): Rename to...
6965 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6967 (selftest-valgrind): Reintroduce as an alias for
6968 selftest-c-valgrind.
6969 (s-selftest-c++): New.
6970 (selftest-c++-gdb): New.
6971 (selftest-c++-valgrind): New.
6973 2017-07-06 Olivier Hainque <hainque@adacore.com>
6975 * gcc.c (process_command): When deciding if undefined variables
6976 should be ignored when processing specs, accept "gcc -v" as well.
6978 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6980 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6981 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6983 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6985 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6986 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6987 * config/arm/arm-tables.opt: Regenerate.
6988 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6990 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6992 2017-07-06 Carl Love <cel@us.ibm.com>
6994 * ChangeLog: Clean up from mid air collision
6996 2017-07-06 Carl Love <cel@us.ibm.com>
6998 * config/rs6000/rs6000-c.c: Add support for built-in functions
6999 vector signed int vec_subc (vector signed int, vector signed int);
7000 vector signed __int128 vec_subc (vector signed __int128,
7001 vector signed __int128);
7002 vector unsigned __int128 vec_subc (vector unsigned __int128,
7003 vector unsigned __int128);
7004 vector signed int vec_sube (vector signed int, vector signed int,
7006 vector unsigned int vec_sube (vector unsigned int,
7007 vector unsigned int,
7008 vector unsigned int);
7009 vector signed __int128 vec_sube (vector signed __int128,
7010 vector signed __int128,
7011 vector signed__int128);
7012 vector unsigned __int128 vec_sube (vector unsigned __int128,
7013 vector unsigned __int128,
7014 vector unsigned __int128);
7015 vector signed int vec_subec (vector signed int, vector signed int,
7017 vector unsigned int vec_subec (vector unsigned int,
7018 vector unsigned int,
7019 vector unsigned int);
7020 vector signed __int128 vec_subec (vector signed __int128,
7021 vector signed __int128,
7022 vector signed__int128);
7023 vector unsigned __int128 vec_subec (vector unsigned __int128,
7024 vector unsigned __int128,
7025 vector unsigned __int128);
7026 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7027 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7028 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7029 BU_ALTIVEC_OVERLOAD_X definitions.
7030 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7031 * doc/extend.texi: Update the built-in documentation file for the new
7034 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7037 * diagnostic-show-locus.c (layout::layout): Use start and finish
7038 spelling location for the start and finish of each range.
7039 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7040 Add unused aspect param.
7041 * input.c (expand_location_1): Add "aspect" param, and use it
7042 to access the correct part of the location.
7043 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7045 (expand_location_to_spelling_point): Likewise.
7046 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7047 param, and pass it to expand_location_1.
7049 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7051 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7052 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7053 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7054 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7055 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7056 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7057 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7058 _mm_maskz_getmant_ss): New intrinsics.
7059 (__builtin_ia32_getexpss128_mask): Changed to ...
7060 __builtin_ia32_getexpss128_round ... this.
7061 (__builtin_ia32_getexpsd128_mask): Changed to ...
7062 __builtin_ia32_getexpsd128_round ... this.
7063 * config/i386/i386-builtin-types.def
7064 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7065 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7066 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7067 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7068 __builtin_ia32_getmantss_mask_round): New builtins.
7069 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7070 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7071 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7072 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7073 * config/i386/sse.md
7074 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7075 avx512f_sgetexp<mode><mask_scalar_name>
7076 <round_saeonly_scalar_name> ... this.
7077 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7078 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7079 vgetexp<ssescalarmodesuffix>
7080 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7081 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7082 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7083 avx512f_vgetmant<mode><mask_scalar_name>
7084 <round_saeonly_scalar_name> ... this.
7085 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7086 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7087 vgetmant<ssescalarmodesuffix>
7088 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7089 %0<mask_scalar_operand4>, %1, %2
7090 <round_saeonly_scalar_mask_op4>, %3} ... this.
7091 * config/i386/subst.md (mask_scalar_operand4,
7092 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7093 round_saeonly_scalar_nimm_predicate): New subst attributes.
7095 2017-07-06 Julia Koval <julia.koval@intel.com>
7097 * config/i386/i386.c (ix86_erase_embedded_rounding):
7098 Remove code for old rounding pattern.
7100 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7102 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7104 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7106 * doc/sourcebuild.texi (Test Directives, Variants of
7107 dg-require-support): Add documentation for dg-require-stack-check.
7109 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7111 * config/i386/subst.md (mask_scalar, round_scalar,
7112 round_saeonly_scalar): New meta-templates.
7113 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7114 round_scalar_mask_operand3, round_scalar_mask_op3,
7115 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7116 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7117 round_saeonly_scalar_constraint,
7118 round_saeonly_scalar_prefix): New subst attribute.
7119 * config/i386/sse.md
7120 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7121 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7122 <round_scalar_name> ... this.
7123 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7124 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7125 <round_scalar_name> ... this.
7126 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7127 <sse>_vm<code><mode>3<mask_scalar_name>
7128 <round_saeonly_scalar_name> ... this.
7129 (v<plusminus_mnemonic><ssescalarmodesuffix>
7130 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7131 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7132 v<plusminus_mnemonic><ssescalarmodesuffix>
7133 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7134 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7135 (v<multdiv_mnemonic><ssescalarmodesuffix>
7136 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7137 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7138 v<multdiv_mnemonic><ssescalarmodesuffix>
7139 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7140 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7141 (v<maxmin_float><ssescalarmodesuffix>
7142 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7143 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7144 v<maxmin_float><ssescalarmodesuffix>
7145 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7146 %0<mask_scalar_operand3>, %1, %<iptr>2
7147 <round_saeonly_scalar_mask_op3>} ... this.
7149 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7151 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7152 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7154 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7155 Alan Hayward <alan.hayward@arm.com>
7156 David Sherwood <david.sherwood@arm.com>
7158 * combine.c (simplify_if_then_else): Remove "enum" before
7160 * compare-elim.c (can_eliminate_compare): Likewise.
7161 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7163 (aarch64_lookup_simd_builtin_type): Likewise.
7164 (aarch64_simd_builtin_type): Likewise.
7165 (aarch64_init_simd_builtin_types): Likewise.
7166 (aarch64_simd_expand_args): Likewise.
7167 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7169 (aarch64_reverse_mask): Likewise.
7170 (aarch64_simd_emit_reg_reg_move): Likewise.
7171 (aarch64_gen_adjusted_ldpstp): Likewise.
7172 (aarch64_ccmp_mode_to_code): Likewise.
7173 (aarch64_operands_ok_for_ldpstp): Likewise.
7174 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7175 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7177 (aarch64_min_divisions_for_recip_mul): Likewise.
7178 (aarch64_reassociation_width): Likewise.
7179 (aarch64_get_condition_code_1): Likewise.
7180 (aarch64_simd_emit_reg_reg_move): Likewise.
7181 (aarch64_simd_attr_length_rglist): Likewise.
7182 (aarch64_reverse_mask): Likewise.
7183 (aarch64_operands_ok_for_ldpstp): Likewise.
7184 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7185 (aarch64_gen_adjusted_ldpstp): Likewise.
7186 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7188 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7189 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7190 (arm_lookup_simd_builtin_type): Likewise.
7191 (arm_simd_builtin_type): Likewise.
7192 (arm_init_simd_builtin_types): Likewise.
7193 (arm_expand_builtin_args): Likewise.
7194 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7195 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7196 (ft32_setup_incoming_varargs): Likewise.
7197 (ft32_function_arg): Likewise.
7198 (ft32_function_arg_advance): Likewise.
7199 (ft32_pass_by_reference): Likewise.
7200 (ft32_arg_partial_bytes): Likewise.
7201 (ft32_valid_pointer_mode): Likewise.
7202 (ft32_addr_space_pointer_mode): Likewise.
7203 (ft32_addr_space_legitimate_address_p): Likewise.
7204 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7206 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7207 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7208 (iamcu_alignment): Likewise.
7209 (canonicalize_vector_int_perm): Likewise.
7210 (ix86_noce_conversion_profitable_p): Likewise.
7211 (ix86_mpx_bound_mode): Likewise.
7212 (ix86_operands_ok_for_move_multiple): Likewise.
7213 * config/microblaze/microblaze-protos.h
7214 (microblaze_expand_conditional_branch_reg): Likewise.
7215 * config/microblaze/microblaze.c
7216 (microblaze_expand_conditional_branch_reg): Likewise.
7217 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7219 (rs6000_reassociation_width): Likewise.
7220 (rs6000_invalid_binary_op): Likewise.
7221 (fusion_p9_p): Likewise.
7222 (emit_fusion_p9_load): Likewise.
7223 (emit_fusion_p9_store): Likewise.
7224 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7226 (riscv_hard_regno_mode_ok_p): Likewise.
7227 (riscv_address_insns): Likewise.
7228 (riscv_split_symbol): Likewise.
7229 (riscv_legitimize_move): Likewise.
7230 (riscv_function_value): Likewise.
7231 (riscv_hard_regno_nregs): Likewise.
7232 (riscv_expand_builtin): Likewise.
7233 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7234 (riscv_build_integer): Likewise.
7235 (riscv_split_integer): Likewise.
7236 (riscv_legitimate_constant_p): Likewise.
7237 (riscv_cannot_force_const_mem): Likewise.
7238 (riscv_regno_mode_ok_for_base_p): Likewise.
7239 (riscv_valid_base_register_p): Likewise.
7240 (riscv_valid_offset_p): Likewise.
7241 (riscv_valid_lo_sum_p): Likewise.
7242 (riscv_classify_address): Likewise.
7243 (riscv_legitimate_address_p): Likewise.
7244 (riscv_address_insns): Likewise.
7245 (riscv_load_store_insns): Likewise.
7246 (riscv_force_binary): Likewise.
7247 (riscv_split_symbol): Likewise.
7248 (riscv_force_address): Likewise.
7249 (riscv_legitimize_address): Likewise.
7250 (riscv_move_integer): Likewise.
7251 (riscv_legitimize_const_move): Likewise.
7252 (riscv_legitimize_move): Likewise.
7253 (riscv_address_cost): Likewise.
7254 (riscv_subword): Likewise.
7255 (riscv_output_move): Likewise.
7256 (riscv_canonicalize_int_order_test): Likewise.
7257 (riscv_emit_int_order_test): Likewise.
7258 (riscv_function_arg_boundary): Likewise.
7259 (riscv_pass_mode_in_fpr_p): Likewise.
7260 (riscv_pass_fpr_single): Likewise.
7261 (riscv_pass_fpr_pair): Likewise.
7262 (riscv_get_arg_info): Likewise.
7263 (riscv_function_arg): Likewise.
7264 (riscv_function_arg_advance): Likewise.
7265 (riscv_arg_partial_bytes): Likewise.
7266 (riscv_function_value): Likewise.
7267 (riscv_pass_by_reference): Likewise.
7268 (riscv_setup_incoming_varargs): Likewise.
7269 (riscv_print_operand): Likewise.
7270 (riscv_elf_select_rtx_section): Likewise.
7271 (riscv_save_restore_reg): Likewise.
7272 (riscv_for_each_saved_reg): Likewise.
7273 (riscv_register_move_cost): Likewise.
7274 (riscv_hard_regno_mode_ok_p): Likewise.
7275 (riscv_hard_regno_nregs): Likewise.
7276 (riscv_class_max_nregs): Likewise.
7277 (riscv_memory_move_cost): Likewise.
7278 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7279 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7280 (rl78_addr_space_address_mode): Likewise.
7281 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7283 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7284 (rs6000_reassociation_width): Likewise.
7285 (rs6000_invalid_binary_op): Likewise.
7286 (fusion_p9_p): Likewise.
7287 (emit_fusion_p9_load): Likewise.
7288 (emit_fusion_p9_store): Likewise.
7289 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7290 (ok_for_simple_move_operands): Likewise.
7291 (ok_for_simple_move_strict_operands): Likewise.
7292 (ok_for_simple_arith_logic_operands): Likewise.
7293 (visium_legitimize_reload_address): Likewise.
7294 (visium_select_cc_mode): Likewise.
7295 (output_cbranch): Likewise.
7296 (visium_split_double_move): Likewise.
7297 (visium_expand_copysign): Likewise.
7298 (visium_expand_int_cstore): Likewise.
7299 (visium_expand_fp_cstore): Likewise.
7300 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7301 (visium_function_arg): Likewise.
7302 (visium_function_arg_advance): Likewise.
7303 (visium_libcall_value): Likewise.
7304 (visium_setup_incoming_varargs): Likewise.
7305 (visium_legitimate_constant_p): Likewise.
7306 (visium_legitimate_address_p): Likewise.
7307 (visium_legitimize_address): Likewise.
7308 (visium_secondary_reload): Likewise.
7309 (visium_register_move_cost): Likewise.
7310 (visium_memory_move_cost): Likewise.
7311 (prepare_move_operands): Likewise.
7312 (ok_for_simple_move_operands): Likewise.
7313 (ok_for_simple_move_strict_operands): Likewise.
7314 (ok_for_simple_arith_logic_operands): Likewise.
7315 (visium_function_value_1): Likewise.
7316 (rtx_ok_for_offset_p): Likewise.
7317 (visium_legitimize_reload_address): Likewise.
7318 (visium_split_double_move): Likewise.
7319 (visium_expand_copysign): Likewise.
7320 (visium_expand_int_cstore): Likewise.
7321 (visium_expand_fp_cstore): Likewise.
7322 (visium_split_cstore): Likewise.
7323 (visium_select_cc_mode): Likewise.
7324 (visium_split_cbranch): Likewise.
7325 (output_cbranch): Likewise.
7326 (visium_print_operand_address): Likewise.
7327 * expmed.c (flip_storage_order): Likewise.
7328 * expmed.h (emit_cstore): Likewise.
7329 (flip_storage_order): Likewise.
7330 * genrecog.c (validate_pattern): Likewise.
7331 * hsa-gen.c (gen_hsa_addr): Likewise.
7332 * internal-fn.c (expand_arith_overflow): Likewise.
7333 * ira-color.c (allocno_copy_cost_saving): Likewise.
7334 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7335 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7336 (process_invariant_for_inheritance): Likewise.
7337 * lra-eliminations.c (move_plus_up): Likewise.
7338 * omp-low.c (lower_oacc_reductions): Likewise.
7339 * simplify-rtx.c (simplify_subreg): Likewise.
7340 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7341 (TARGET_CHKP_BOUND_MODE): Likewise..
7342 * targhooks.c (default_chkp_bound_mode): Likewise.
7343 (default_setup_incoming_vararg_bounds): Likewise.
7344 * targhooks.h (default_chkp_bound_mode): Likewise.
7345 (default_setup_incoming_vararg_bounds): Likewise.
7346 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7347 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7348 (have_whole_vector_shift): Likewise.
7349 * tree-vect-stmts.c (vectorizable_load): Likewise.
7350 * doc/tm.texi: Regenerate.
7352 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7354 Graceful degrade if Binutils PR21472 is not available.
7357 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7358 .rodata in flash test fails.
7359 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7360 * confgure: Regenerate.
7361 * config.in: Regenerate.
7362 * config/avr/avr.c (avr_asm_named_section)
7363 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7364 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7365 (avr_asm_init_sections): Same.
7367 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7369 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7370 (fma<VH:mode>4_intrinsic): Likewise.
7371 (*fmsub<VCVTF:mode>4): Likewise.
7372 (*fmsub<VH:mode>4_intrinsic): Likewise.
7374 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7377 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7378 Don't depend on "optimize > 0".
7379 (out_movhi_r_mr, out_movqi_mr_r): Same.
7380 (out_movhi_mr_r, out_movqi_r_mr): Same.
7381 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7382 io_address_operand on "optimize > 0".
7384 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7386 * tree-loop-distribution.c: Add general explanantion on the pass.
7387 (generate_loops_for_partition): Mark distributed loop.
7388 (pg_add_dependence_edges): New parameter. Handle alias data
7389 dependence specially and record it in the parameter if asked.
7390 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7391 (init_partition_graph_vertices, add_partition_graph_edge): New.
7392 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7393 (free_partition_graph_vdata, build_partition_graph): New.
7394 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7395 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7396 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7397 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7398 (version_for_distribution_p, finalize_partitions): New.
7399 (distribute_loop): Handle alias data dependence specially. Factor
7400 out loop fusion code as functions and call these functions.
7402 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7404 * tree-loop-distribution.c (classify_partition): New parameter and
7405 better handle reduction statement.
7406 (rdg_build_partitions): Revise comment.
7407 (distribute_loop): Compute statements in all partitions and pass it
7408 to classify_partition.
7410 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7412 * tree-loop-distribution.c (enum partition_type): New.
7413 (struct partition): New field type.
7414 (partition_merge_into): Add parameter. Update partition type.
7415 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7416 (build_rdg_partition_for_vertex): Compute partition type.
7417 (rdg_build_partitions): Dump partition type.
7418 (distribute_loop): Update calls to partition_merge_into.
7420 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7422 * tree-loop-distribution.c (struct ddr_hasher): New.
7423 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7425 (classify_partition): Call get_data_dependence.
7426 (pg_add_dependence_edges): Ditto.
7427 (distribute_loop): Release data dependence hash table.
7429 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7431 * tree-loop-distribution.c (ref_base_address): Delete.
7432 (similar_memory_accesses): Rename ...
7433 (share_memory_accesses): ... to this. Check if partitions access
7434 the same memory reference.
7435 (distribute_loop): Call share_memory_accesses.
7437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7439 * tree-loop-distribution.c (struct partition): New field recording
7441 (partition_alloc, partition_free): Init and release data refs.
7442 (partition_merge_into): Merge data refs.
7443 (build_rdg_partition_for_vertex): Collect data refs for partition.
7444 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7446 (distribute_loop): Remve data refs from vertice data of partition
7449 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7451 * tree-loop-distribution.c (params.h): Include header file.
7452 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7453 (datarefs_vec): New global var.
7454 (create_rdg_vertices): Use datarefs_vec directly.
7455 (free_rdg): Don't free data references.
7456 (build_rdg): Update use. Don't free data references.
7457 (distribute_loop): Compute global variable for data references.
7458 Bail out if there are too many data references.
7460 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7462 * tree-loop-distribution.c (loop_nest): New global var.
7463 (build_rdg): Use loop directly, rather than loop nest.
7464 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7466 (distribute_loop): Compute global variable loop nest. Update use.
7468 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7470 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7471 (partition_merge_into): New parameter. Dump reason for fusion.
7472 (distribute_loop): Update use of partition_merge_into.
7474 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7476 * tree-loop-distribution.c (bb_top_order_index): New.
7477 (bb_top_order_index_size, bb_top_order_cmp): New.
7478 (stmts_from_loop): Use topological order.
7479 (pass_loop_distribution::execute): Compute and release topological
7480 order for basic blocks.
7482 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7484 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7487 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7489 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7490 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7491 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7492 * internal-fn.def (LOOP_DIST_ALIAS): New.
7493 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7494 (fold_loop_internal_call): ... this.
7495 (vect_loop_dist_alias_call): New function.
7496 (set_uid_loop_bbs): Call fold_loop_internal_call.
7497 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7500 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7503 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7504 Require dead FLAGS_REG at the beginning of a peephole.
7506 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7509 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7510 arguments in the call to __builtin_ia32_sbb_u32.
7511 (_subborrow_u64): Swap _X and _Y arguments in the call to
7512 __builtin_ia32_sbb_u64.
7514 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7517 * tree-vrp.c (compare_assert_loc): Turn into a function template
7518 with stable template parameter. Only test if a->e is NULL,
7519 !a->e == !b->e has been verified already. Use e == NULL or
7520 e != NULL instead of e or ! e tests. If stable is true, don't use
7521 iterative_hash_expr, on the other side allow a or b or both NULL
7522 and sort the NULLs last.
7523 (process_assert_insertions): Sort using compare_assert_loc<false>
7524 instead of compare_assert_loc, later sort using
7525 compare_assert_loc<true> before calling process_assert_insertions_for
7526 in a loop. Use break instead of continue once seen NULL pointer.
7528 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7530 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7531 Cortex-R7 and Cortex-R8 processors.
7533 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7535 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7536 uninitialized while src is not.
7538 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7540 * common/config/arm/arm-common.c: Adjust include path for
7542 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7543 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7544 (arm-cpu-data.h): Likewise.
7545 (arm-cpu-cdata.h): Likewise.
7546 * config/arm/arm-cpu.h: Delete.
7547 * config/arm/arm-cpu-cdata.h: Delete.
7548 * config/arm/arm-cpu-data.h: Delete.
7550 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7552 * config/arm/arm-cpus.in (cortex-a55): New.
7553 (cortex-a75): Likewise.
7554 (cortex-a75.cortex-a55): Likewise.
7555 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7557 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7558 * config/arm/arm-cpu-cdata.h: Regenerate.
7559 * config/arm/arm-cpu-data.h: Regenerate.
7560 * config/arm/arm-cpu.h: Regenerate.
7561 * config/arm/arm-tables.opt: Regenerate.
7562 * config/arm/arm-tune.md: Regenerate.
7564 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7566 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7568 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7570 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7573 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7575 PR tree-optimization/81292
7576 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7577 full_string_p, also call adjust_related_strinfos if the adjustment
7578 is simple, otherwise invalidate related strinfos.
7580 2017-07-04 Martin Liska <mliska@suse.cz>
7583 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7584 newly created variable as DECL_IGNORED_P.
7586 2017-07-04 Martin Liska <mliska@suse.cz>
7589 * ipa-inline.c (inline_small_functions):
7590 Use xstrdup_for_dump.
7592 2017-07-04 Tom de Vries <tom@codesourcery.com>
7594 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7596 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7599 * config/darwin.c (darwin_function_switched_text_sections):
7602 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7604 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7606 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7608 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7610 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7612 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7613 min_profitable_iters, and th as inclusive lower bounds.
7614 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7615 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7616 for min_profitable_iters and min_profitable_estimate.
7617 (vect_transform_loop): Treat th as an inclusive lower bound.
7618 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7620 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7623 * config/darwin.c (darwin_function_switched_text_sections):
7624 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7625 in two pieces, and suppress the use of buf.
7627 2017-07-03 Nathan Sidwell <nathan@acm.org>
7629 * hash-table.h (hash_table_mod1): Fix indentation.
7631 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7634 * predict.c (force_edge_cold): Be more careful about propagation
7636 * profile-count.h (profile_probability::guessed,
7637 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7639 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7641 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7643 * doc/invoke.texi (rcpc architecture extension): Document it.
7645 2017-07-03 Richard Biener <rguenther@suse.de>
7647 PR tree-optimization/60510
7648 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7649 the scalar reduction PHI and use it.
7650 (vectorizable_reduction): Properly guard the single_defuse_cycle
7651 path for non-SLP reduction chains where we cannot use it.
7652 Rework reduc_def/index and vector type deduction. Rework
7653 vector operand gathering during reduction op code-gen.
7654 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7655 chains dissolve the chain and leave it to non-SLP reduction
7658 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7660 * tree-data-ref.h (dr_alignment): Declare.
7661 * tree-data-ref.c (dr_alignment): New function.
7662 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7663 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7665 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7667 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7669 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7670 and base_misalignment fields.
7671 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7672 * tree-data-ref.c: Include builtins.h.
7673 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7674 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7675 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7676 * tree-vect-data-refs.c: Include tree-cfg.h.
7677 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7678 fields instead of calculating an alignment here.
7679 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7680 innermost_loop_behavior fields.
7682 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7684 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7686 (DR_STEP_ALIGNMENT): New macro.
7687 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7688 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7689 (create_data_ref): Print it.
7690 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7691 to tell whether the step preserves vector (mis)alignment.
7692 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7693 Move the check for an integer step and generalise to all INTEGER_CST.
7694 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7695 Print the outer step alignment.
7697 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7699 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7700 with offset_alignment.
7701 (DR_ALIGNED_TO): Delete.
7702 (DR_OFFSET_ALIGNMENT): New macro.
7703 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7704 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7705 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7706 (create_data_ref): Likewise.
7707 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7708 (vect_analyze_data_refs): Likewise.
7709 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7710 creating dummy innermost behavior.
7712 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7714 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7715 with a "innermost_loop_behavior *" and refeence tree.
7716 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7717 (create_data_ref): Update call accordingly.
7718 * tree-predcom.c (find_looparound_phi): Likewise.
7720 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7722 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7723 fields with dr_wrt_vec_loop.
7724 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7725 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7726 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7727 (vect_dr_behavior): New function.
7728 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7729 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7730 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7731 track whether the step preserves the misalignment.
7732 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7733 Use vect_dr_behavior.
7734 (vect_setup_realignment): Update call accordingly.
7735 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7736 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7737 call to vect_create_addr_base_for_vector_ref.
7738 (vect_create_cond_for_align_checks): Likewise.
7739 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7740 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7741 (vect_recog_mask_conversion_pattern): Likewise.
7742 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7743 (new_stmt_vec_info): Remove redundant zeroing.
7745 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7747 * common/config/arm/arm-common.c (arm_be8_option): New function.
7748 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7749 (ISA_ARMv6): Add isa_bit_be8.
7750 * config/arm/arm.h (arm_be8_option): Add prototype.
7751 (BE8_SPEC_FUNCTION): New define.
7752 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7753 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7754 (mlittle-endian): Similarly.
7755 (mbe8, mbe32): New options.
7756 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7757 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7759 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7761 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7763 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7765 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7766 (cleanup_tree_cfg_bb): Use it.
7767 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7769 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7771 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7774 * loop-doloop.c (add_test): Update profile.
7776 2017-07-03 Martin Liska <mliska@suse.cz>
7779 * sanopt.c (rewrite_usage_of_param): New function.
7780 (sanitize_rewrite_addressable_params): Likewise.
7781 (pass_sanopt::execute): Call rewrite_usage_of_param.
7783 2017-07-03 Richard Biener <rguenther@suse.de>
7785 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7786 back to using VIEW_CONVERT_EXPR.
7788 2017-07-03 Martin Liska <mliska@suse.cz>
7791 * doc/extend.texi: Document when a resolver function is
7792 generated for target_clones.
7794 2017-07-03 Martin Liska <mliska@suse.cz>
7796 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7797 and shadow memory of auto variables which are subject of
7798 use-after-scope sanitization.
7799 (asan_expand_mark_ifn): Add do set only when is_poison.
7801 2016-07-03 Richard Biener <rguenther@suse.de>
7803 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7805 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7806 (vectorizable_reduction): Perform reduction PHI creation when
7807 visiting a reduction PHI and adjust and simplify code generation
7808 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7809 (vect_transform_loop): Visit reduction PHIs.
7810 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7811 defs into the SLP tree.
7812 (vect_build_slp_tree): Reduction defs terminate the recursion.
7813 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7815 (vect_get_vec_defs_for_stmt_copy): Export.
7816 (vect_get_vec_defs): Likewise.
7817 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7819 (vect_get_vec_defs_for_stmt_copy): Declare.
7820 (vect_get_vec_defs): Likewise.
7822 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7824 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7825 parameter with a "loop" parameter and use it instead of the
7826 loop containing DR_STMT. Don't check simple_iv when doing
7827 BB analysis. Describe the two analysis modes in the comment.
7829 2017-07-03 Tom de Vries <tom@codesourcery.com>
7831 PR tree-optimization/69468
7832 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7833 (find_same_succ_bb): Handle ignore_edge_flags.
7835 2017-07-03 Tom de Vries <tom@codesourcery.com>
7837 PR tree-optimization/81192
7838 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7840 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7842 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7844 2017-07-03 Tom de Vries <tom@codesourcery.com>
7846 PR tree-optimization/81192
7847 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7848 BB_SAME_SUCC (bb) == NULL.
7850 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7852 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7855 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7857 * dumpfile.c: Include profile-count.h
7858 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7860 (insert_cond_bb): Update profile.
7861 * tree-cfg.h (insert_cond_bb): Update prototype.
7862 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7863 * tree-dump.c: Do not include tree-cfg.
7865 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7867 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7869 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7871 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7876 * tree-complex.c (expand_complex_div_wide): update profile.
7878 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7879 Alan Hayward <alan.hayward@arm.com>
7880 David Sherwood <david.sherwood@arm.com>
7882 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7883 (CORETYPES_H): New define.
7884 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7885 (insn-modes-inline.h, s-modes-inline-h): New rules.
7886 (generated_files): Add insn-modes-inline.h.
7887 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7888 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7889 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7890 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7891 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7892 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7893 (build/gencodes.o, build/genconditions.o): Likewise.
7894 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7895 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7896 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7897 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7898 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7899 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7900 * coretypes.h: Include everything up to real.h for generators.
7901 Include insn-modes.h first. Include wide-int-print.h after
7902 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7903 * machmode.h: Don't include insn-modes.h here.
7904 * function-tests.c: Remove includes of signop.h, machmode.h,
7905 double-int.h and wide-int.h.
7907 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7909 * optc-save-gen.awk: Likewise.
7910 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7911 * godump.c: Remove include of wide-int-print.h.
7912 * pretty-print.h: Likewise.
7913 * wide-int-print.cc: Likewise.
7914 * wide-int.cc: Likewise.
7915 * hash-map-tests.c: Remove include of signop.h.
7916 * hash-set-tests.c: Likewise.
7917 * rtl-tests.c: Likewise.
7918 * mkconfig.sh: Remove include of machmode.h.
7919 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7921 (emit_insn_modes_inline_h): ...this new function. Emit the code
7922 into an insn-modes-inline.h header file, adding appropriate
7923 include guards and end comments.
7924 (emit_insn_modes_c_header): Remove include of machmode.h.
7925 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7927 (main): Handle -i flag and call emit_insn_modes_inline_h when
7930 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7932 * tree-ssa-strlen.c (strinfo): Rename the length field to
7933 nonzero_chars. Add a full_string_p field.
7934 (compare_nonzero_chars, zero_length_string_p): New functions.
7935 (get_addr_stridx): Add an offset_out parameter.
7936 Use compare_nonzero_chars.
7937 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7938 (new_strinfo): Update after above changes to strinfo.
7939 (set_endptr_and_length): Set full_string_p.
7940 (get_string_length): Update after above changes to strinfo.
7941 (unshare_strinfo): Update call to new_strinfo.
7942 (maybe_invalidate): Likewise.
7943 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7944 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7945 as a uhwi instead of an shwi. Update after above changes to
7946 strinfo and new_strinfo.
7947 (zero_length_string): Assert that chainsi contains full strings.
7948 Use zero_length_string_p. Update call to new_strinfo.
7949 (adjust_related_strinfos): Update after above changes to strinfo.
7950 Copy full_string_p from origsi.
7951 (adjust_last_stmt): Use zero_length_string_p.
7952 (handle_builtin_strlen): Update after above changes to strinfo and
7953 new_strinfo. Install the lhs as the string length if the previous
7954 entry didn't describe a full string.
7955 (handle_builtin_strchr): Update after above changes to strinfo
7957 (handle_builtin_strcpy): Likewise.
7958 (handle_builtin_strcat): Likewise.
7959 (handle_builtin_malloc): Likewise.
7960 (handle_pointer_plus): Likewise.
7961 (handle_builtin_memcpy): Likewise. Track nonzero characters
7962 that aren't necessarily followed by a nul terminator.
7963 (handle_char_store): Likewise.
7965 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7967 PR tree-optimization/80769
7968 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7969 for malloc and calloc. Document the new invariant that all related
7970 strinfos have delayed lengths or none do.
7971 (verify_related_strinfos): Move earlier in file.
7972 (set_endptr_and_length): New function, split out from...
7973 (get_string_length): ...here. Also set the lengths of related
7975 (zero_length_string): Assert that chainsi has known (rather than
7977 (adjust_related_strinfos): Likewise.
7979 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7981 PR tree-optimization/81136
7982 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7983 assert that two references with the same misalignment have the same
7984 compile-time misalignment if those compile-time misalignments
7987 2017-07-01 Andi Kleen <ak@linux.intel.com>
7989 * print-tree.c (print_node): Print all attributes.
7991 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7993 * cfg.c (scale_bbs_frequencies): New function.
7994 * cfg.h (scale_bbs_frequencies): Declare it.
7995 * cfgloopanal.c (single_likely_exit): Cleanup.
7996 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7998 (scale_loop_profile): Likewise.
7999 (loop_version): Likewise.
8000 (create_empty_loop_on_edge): Update.
8001 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8002 scale_loop_frequencies, scale_loop_profile, loopify,
8003 loop_version): Update prototypes.
8004 * modulo-sched.c (sms_schedule): Update.
8005 * predict.c (unlikely_executed_edge_p): Also check probability.
8006 (probably_never_executed_edge_p): Fix typo.
8007 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8008 * tree-parloops.c (gen_parallel_loop): Update.
8009 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8010 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8011 * tree-ssa-loop-split.c (split_loop): Update.
8012 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8013 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8014 (vect_loop_versioning): Update.
8015 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8017 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8019 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8021 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8023 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8024 to keep profile consistent.
8026 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8028 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8029 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8030 * profile-count.h (max_safe_multiplier): Make unsigned.
8031 (profile_count::guessed_zero): New.
8033 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8035 * bb-reorder.c (fix_up_crossing_landing_pad,
8036 fix_crossing_conditional_branches): Use make_single_succ_edge
8037 to keep profile consistent.
8039 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8041 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8044 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8047 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8048 the right scopes, make sure cond_jump isn't preserved between multiple
8049 iterations. Search for fallthru edge whenever there are 3+ edges and
8050 use find_fallthru_edge for it.
8052 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8054 Patch by Alexander Monakov <amonakov@ispras.ru>
8055 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8056 probabilities consistently.
8058 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8060 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8061 emit_cmp_and_jump_insns.
8063 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8066 * tree-inline.c (expand_call_inline): Combine profile statuses.
8068 2017-06-30 Andrew Pinski <apinski@cavium.com>
8070 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8071 fold_stmt returned true.
8073 2017-06-30 Nathan Sidwell <nathan@acm.org>
8075 * ggc.h (empty_string): Delete.
8076 * cfgexpand.c (expand_asm_stmt): Use plain "".
8077 * optabs.c (expand_asm_memory_barrier): Likewise.
8078 * stringpool.c (empty_string): Delete.
8079 (digit_vector, digit_string): Delete.
8080 (ggc_alloc_string): Use plain "", don't optimize single digit
8081 strings. Use ggc_alloc_atomic.
8083 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8085 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8086 comparison set and one other set, use the cost of the non-comparison
8089 2017-06-30 Nathan Sidwell <nathan@acm.org>
8091 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8094 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8096 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8097 loops. Remove now unneeded calls to gimple_switch_set_label() that
8098 just set removed labels to NULL_TREE.
8100 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8102 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8103 (set_range_info): ...here. Only call set_range_info_raw if domain
8105 (set_nonzero_bits): Call set_range_info_raw.
8106 * tree-ssanames.h (set_range_info_raw): New.
8108 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8111 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8112 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8113 of nonimmediate_operand and <store_mask_constraint> instead of m
8114 for the input operand. For V8FI iterator, always split if input
8115 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8116 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8117 <store_mask_predicate> instead of register_operand and
8118 <store_mask_constraint> instead of v for the input operand. Make
8119 sure both operands aren't MEMs for if not <mask_applied>.
8121 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8123 * lto-wrapper.c (copy_file) Close both file descriptors before
8126 2017-06-30 Martin Liska <mliska@suse.cz>
8129 * multiple_target.c (create_dispatcher_calls): Make ifunc
8130 also for function that don't have calls or are not referenced.
8132 2017-06-30 Richard Biener <rguenther@suse.de>
8134 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8135 analyze the first scalar stmt. Move vector type computation
8136 for the BB case here from ...
8137 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8138 live operation processing in the SLP case properly.
8140 2017-06-30 Richard Biener <rguenther@suse.de>
8142 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8144 2017-06-30 Martin Liska <mliska@suse.cz>
8147 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8148 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8150 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8152 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8155 2017-06-30 Richard Biener <rguenther@suse.de>
8157 PR tree-optimization/81249
8158 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8159 condition reduction result to original scalar type.
8161 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8163 * profile-count.h (enum profile_quality): Fix typos and whitespace
8166 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8168 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8169 type for branch probabilities.
8171 2017-06-29 Julian Brown <julian@codesourcery.com>
8172 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8174 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8175 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8176 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8177 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8179 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8181 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8182 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8183 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8184 CC usage from generic code to here.
8185 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8186 CC usage into the target macros.
8188 2017-06-29 Maya Rashish <coypu@sdf.org>
8190 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8193 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8195 * arm/arm-builtins.c: Include profile-count.h
8196 * except.c (sjlj_emit_function_enter): Use
8197 profile_probability::unlikely.
8199 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8201 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8202 and tocrel_offset be pointer args rather than implicitly using
8204 (legitimate_constant_pool_address_p, rs6000_emit_move,
8205 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8206 tocrel_offset and use in toc_relative_expr_p call.
8207 (print_operand, print_operand_address): Use static tocrel_base_oac
8208 and tocrel_offset_oac.
8209 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8212 2017-06-29 Maya Rashish <coypu@sdf.org>
8214 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8216 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8218 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8219 objects, take into account only the alignment of 'op0' and 'mode1' if
8222 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8224 * ccmp.c (ccmp_tree_comparison_p): New function.
8225 (ccmp_candidate_p): Update to use above function.
8226 (get_compare_parts): New function.
8227 (expand_ccmp_next): Update to use new functions.
8228 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8230 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8231 take mode as argument.
8232 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8233 * expr.c (expand_expr_real_1): Pass mode as argument.
8235 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8237 * combine.c (combine_instructions): Print insns to dump_file, together
8240 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8242 * asan.c (asan_emit_stack_protection): Update.
8243 (create_cond_insert_point): Update.
8244 * auto-profile.c (afdo_propagate_circuit): Update.
8245 * basic-block.h (struct edge_def): Turn probability to
8246 profile_probability.
8247 (EDGE_FREQUENCY): Update.
8248 * bb-reorder.c (find_traces_1_round): Update.
8249 (better_edge_p): Update.
8250 (sanitize_hot_paths): Update.
8251 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8252 (make_single_succ_edge): Update.
8253 (check_bb_profile): Update.
8254 (dump_edge_info): Update.
8255 (update_bb_profile_for_threading): Update.
8256 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8257 probabilitycount to 0.
8258 * cfgbuild.c (compute_outgoing_frequencies): Update.
8259 * cfgcleanup.c (try_forward_edges): Update.
8260 (outgoing_edges_match): Update.
8261 (try_crossjump_to_edge): Update.
8262 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8263 (expand_gimple_tailcall): Update.
8264 (construct_init_block): Use make_single_succ_edge.
8265 (construct_exit_block): Use make_single_succ_edge.
8266 * cfghooks.c (verify_flow_info): Update.
8267 (redirect_edge_succ_nodup): Update.
8268 (split_edge): Update.
8269 (account_profile_record): Update.
8270 * cfgloopanal.c (single_likely_exit): Update.
8271 * cfgloopmanip.c (scale_loop_profile): Update.
8272 (set_zero_probability): Remove.
8273 (duplicate_loop_to_header_edge): Update.
8274 * cfgloopmanip.h (loop_version): Update prototype.
8275 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8276 (force_nonfallthru_and_redirect): Update.
8277 (update_br_prob_note): Update.
8278 (rtl_verify_edges): Update.
8279 (purge_dead_edges): Update.
8280 (rtl_lv_add_condition_to_bb): Update.
8281 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8282 * cgraphunit.c (init_lowered_empty_function): Update.
8283 (cgraph_node::expand_thunk): Update.
8284 * cilk-common.c: Include profile-count.h
8285 * dojump.c (inv): Remove.
8286 (jumpifnot): Update.
8287 (jumpifnot_1): Update.
8288 (do_jump_1): Update.
8290 (do_jump_by_parts_greater_rtx): Update.
8291 (do_compare_rtx_and_jump): Update.
8292 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8293 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8294 * dwarf2cfi.c: Include profile-count.h
8295 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8296 (sjlj_emit_dispatch_table): Likewise.
8297 * explow.c: Include profile-count.h
8298 * expmed.c (emit_store_flag_force): Update.
8299 (do_cmp_and_jump): Update.
8300 * expr.c (compare_by_pieces_d::generate): Update.
8301 (compare_by_pieces_d::finish_mode): Update.
8302 (emit_block_move_via_loop): Update.
8303 (store_expr_with_bounds): Update.
8304 (store_constructor): Update.
8305 (expand_expr_real_2): Update.
8306 (expand_expr_real_1): Update.
8307 * expr.h (try_casesi, try_tablejump): Update prototypes.
8308 * gimple-pretty-print.c (dump_probability): Update.
8309 (dump_profile): New.
8310 (dump_gimple_label): Update.
8311 (dump_gimple_bb_header): Update.
8312 * graph.c (draw_cfg_node_succ_edges): Update.
8313 * hsa-gen.c (convert_switch_statements): Update.
8314 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8315 (find_if_case_1): Update.
8316 (find_if_case_2): Update.
8317 * internal-fn.c (expand_arith_overflow_result_store): Update.
8318 (expand_addsub_overflow): Update.
8319 (expand_neg_overflow): Update.
8320 (expand_mul_overflow): Update.
8321 (expand_vector_ubsan_overflow): Update.
8322 * ipa-cp.c (good_cloning_opportunity_p): Update.
8323 * ipa-split.c (split_function): Use make_single_succ_edge.
8324 * ipa-utils.c (ipa_merge_profiles): Update.
8325 * loop-doloop.c (add_test): Update.
8326 (doloop_modify): Update.
8327 * loop-unroll.c (compare_and_jump_seq): Update.
8328 (unroll_loop_runtime_iterations): Update.
8329 * lra-constraints.c (lra_inheritance): Update.
8330 * lto-streamer-in.c (input_cfg): Update.
8331 * lto-streamer-out.c (output_cfg): Update.
8332 * mcf.c (adjust_cfg_counts): Update.
8333 * modulo-sched.c (sms_schedule): Update.
8334 * omp-expand.c (expand_omp_for_init_counts): Update.
8335 (extract_omp_for_update_vars): Update.
8336 (expand_omp_ordered_sink): Update.
8337 (expand_omp_for_ordered_loops): Update.
8338 (expand_omp_for_generic): Update.
8339 (expand_omp_for_static_nochunk): Update.
8340 (expand_omp_for_static_chunk): Update.
8341 (expand_cilk_for): Update.
8342 (expand_omp_simd): Update.
8343 (expand_omp_taskloop_for_outer): Update.
8344 (expand_omp_taskloop_for_inner): Update.
8345 * omp-simd-clone.c (simd_clone_adjust): Update.
8346 * optabs.c (expand_doubleword_shift): Update.
8347 (expand_abs): Update.
8348 (emit_cmp_and_jump_insn_1): Update.
8349 (expand_compare_and_swap_loop): Update.
8350 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8351 * predict.c (predictable_edge_p): Update.
8352 (edge_probability_reliable_p): Update.
8353 (set_even_probabilities): Update.
8354 (combine_predictions_for_insn): Update.
8355 (combine_predictions_for_bb): Update.
8356 (propagate_freq): Update.
8357 (estimate_bb_frequencies): Update.
8358 (force_edge_cold): Update.
8359 * profile-count.c (profile_count::dump): Add missing space into dump.
8360 (profile_count::debug): Add newline.
8361 (profile_count::differs_from_p): Explicitly convert to unsigned.
8362 (profile_count::stream_in): Update.
8363 (profile_probability::dump): New member function.
8364 (profile_probability::debug): New member function.
8365 (profile_probability::differs_from_p): New member function.
8366 (profile_probability::differs_lot_from_p): New member function.
8367 (profile_probability::stream_in): New member function.
8368 (profile_probability::stream_out): New member function.
8369 * profile-count.h (profile_count_quality): Rename to ...
8370 (profile_quality): ... this one.
8371 (profile_probability): New.
8372 (profile_count): Update.
8373 * profile.c (compute_branch_probabilities): Update.
8374 * recog.c (peep2_attempt): Update.
8375 * sched-ebb.c (schedule_ebbs): Update.
8376 * sched-rgn.c (find_single_block_region): Update.
8377 (compute_dom_prob_ps): Update.
8378 (schedule_region): Update.
8379 * sel-sched-ir.c (compute_succs_info): Update.
8380 * stmt.c (struct case_node): Update.
8381 (do_jump_if_equal): Update.
8382 (get_outgoing_edge_probs): Update.
8383 (conditional_probability): Update.
8384 (emit_case_dispatch_table): Update.
8385 (expand_case): Update.
8386 (expand_sjlj_dispatch_table): Update.
8387 (emit_case_nodes): Update.
8388 * targhooks.c: Update.
8389 * tracer.c (better_p): Update.
8390 (find_best_successor): Update.
8391 * trans-mem.c (expand_transaction): Update.
8392 * tree-call-cdce.c: Update.
8393 * tree-cfg.c (gimple_split_edge): Upate.
8394 (move_sese_region_to_fn): Upate.
8395 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8396 * tree-eh.c (lower_resx): Upate.
8397 (cleanup_empty_eh_move_lp): Upate.
8398 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8399 * tree-inline.c (copy_edges_for_bb): Update.
8400 (copy_cfg_body): Update.
8401 * tree-parloops.c (gen_parallel_loop): Update.
8402 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8403 (gimple_gen_time_profiler): Update.
8404 * tree-ssa-dce.c (remove_dead_stmt): Update.
8405 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8406 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8407 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8408 (unloop_loops): Update.
8409 (try_peel_loop): Update.
8410 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8411 * tree-ssa-loop-split.c (connect_loops): Update.
8412 (split_loop): Update.
8413 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8414 (hoist_guard): Update.
8415 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8416 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8417 (value_replacement): Update.
8418 * tree-ssa-reassoc.c (branch_fixup): Update.
8419 * tree-ssa-tail-merge.c (replace_block_by): Update.
8420 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8421 (create_edge_and_update_destination_phis): Update.
8422 (compute_path_counts): Update.
8423 (recompute_probabilities): Update.
8424 (update_joiner_offpath_counts): Update.
8425 (freqs_to_counts_path): Update.
8426 (duplicate_thread_path): Update.
8427 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8428 (struct switch_conv_info): Update.
8429 (gen_inbound_check): Update.
8430 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8431 (vect_do_peeling): Update.
8432 (vect_loop_versioning): Update.
8433 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8434 (optimize_mask_stores): Update.
8435 * ubsan.c (ubsan_expand_null_ifn): Update.
8436 * value-prof.c (gimple_divmod_fixed_value): Update.
8437 (gimple_divmod_fixed_value_transform): Update.
8438 (gimple_mod_pow2): Update.
8439 (gimple_mod_pow2_value_transform): Update.
8440 (gimple_mod_subtract): Update.
8441 (gimple_mod_subtract_transform): Update.
8442 (gimple_ic): Update.
8443 (gimple_stringop_fixed_value): Update.
8444 (gimple_stringops_transform): Update.
8445 * value-prof.h: Update.
8447 2017-06-29 Carl Love <cel@us.ibm.com>
8449 * config/rs6000/rs6000-c.c: Add support for built-in functions
8450 vector signed int vec_signed (vector float);
8451 vector signed long long vec_signed (vector double);
8452 vector signed int vec_signed2 (vector double, vector double);
8453 vector signed int vec_signede (vector double);
8454 vector signed int vec_signedo (vector double);
8455 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8456 instruction generator.
8457 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8458 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8459 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8461 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8462 vunsignede_v2df): Add define_expands.
8463 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8464 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8465 VEC_UNSIGNEDO): Add definitions.
8466 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8467 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8468 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8469 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8470 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8471 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8472 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8473 vec_unsignede, vec_unsignedo): Add builtin defines.
8474 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8476 * doc/extend.texi: Update the built-in documentation file for the
8477 new built-in functions.
8479 2017-06-29 Richard Biener <rguenther@suse.de>
8481 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8482 reduction chains to LOOP_VINFO_REDUCTIONS.
8483 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8484 SLP reductions after processing reduction chains.
8486 2017-06-29 Nathan Sidwell <nathan@acm.org>
8488 * builtins.c (fold_builtin_FUNCTION): Use
8489 lang_hooks.decl_printable_name.
8491 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8494 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8495 with only one label.
8496 * stmt.c (expand_case): Assert NCASES is greater than one.
8498 2017-06-29 Richard Biener <rguenther@suse.de>
8500 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8502 (group_case_labels): Likewise.
8503 (find_taken_edge): Push sanity checking on val to workers...
8504 (find_taken_edge_cond_expr): ... here
8505 (find_taken_edge_switch_expr): ... and here, handle cases
8506 with just a default label.
8507 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8508 (group_case_labels): Likewise.
8509 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8510 group_case_labels does anything cleanup the CFG again.
8512 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8514 PR tree-optimization/81196
8515 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8516 exit condition comparing two IVs.
8518 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8520 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8521 profile to the dummy entry at the end of the list of architectures.
8522 * config/arm/arm-cpu-cdata.h: Regenerated.
8524 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8525 Michael Collison <michael.collison@arm.com>
8528 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8530 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8531 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8532 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8533 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8534 of shift when the shift amount is masked with constant equal to
8535 the size of the mode.
8536 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8539 2017-06-29 Martin Liska <mliska@suse.cz>
8541 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8543 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8546 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8549 (edge_growth_cache_entry::edge_growth_cache_entry): New
8551 (reset_edge_growth_cache): Update to use constructor.
8553 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8555 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8556 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8557 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8559 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8561 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8562 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8564 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8566 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8567 (*-linux-uclibc*): Add t-uclibc tmake_file.
8568 * config/t-musl: New.
8569 * config/t-uclibc: New.
8571 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8573 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8575 (gen_comm_data): Emit architectural setting of arch_prof.
8576 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8578 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8579 (armv8-m.base, armv8-m.main): Likewise.
8580 * arm-protos.h (arm_build_target): Add profile field.
8581 (arch_option): Likewise.
8582 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8584 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8585 arm_active_target.profile.
8587 2017-06-28 Richard Biener <rguenther@suse.de>
8590 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8591 TYPE_OVERFLOW_WRAPS.
8592 * match.pd (negate_expr_p): Likewise.
8593 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8594 fold_build2, not fold_binary.
8596 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8598 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8599 Convert memory address to Pmode.
8600 (aarch64_print_operand): Assert MEM operands are always Pmode.
8602 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8605 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8606 Remove redundant if.
8607 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8608 * config/arm/aarch-common-protos.h
8609 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8610 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8612 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8615 * multiple_target.c (create_dispatcher_calls): Set the default
8616 clone to be static, not public.
8618 2017-06-28 Richard Biener <rguenther@suse.de>
8620 * tree-vect-loop.c (vectorizable_reduction): Move special
8621 cond reduction IV var creation ...
8622 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8623 parameter. Use STMT_VINFO_VECTYPE.
8624 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8627 2017-06-28 Martin Liska <mliska@suse.cz>
8630 * ipa-visibility.c (non_local_p): Handle visibility.
8632 2017-06-28 Martin Liska <mliska@suse.cz>
8635 * common.opt: Add IntegerRange to various options.
8636 * opt-functions.awk (integer_range_info): New function.
8637 * optc-gen.awk: Add integer_range_info to cl_options struct.
8638 * opts-common.c (decode_cmdline_option): Handle
8639 CL_ERR_INT_RANGE_ARG.
8640 (cmdline_handle_error): Likewise.
8641 * opts.c (print_filtered_help): Show valid interval in
8642 when --help is provided.
8643 * opts.h (struct cl_option): Add range_min and range_max fields.
8644 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8646 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8648 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8649 (x * C EQ/NE y * C): New transformation.
8651 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8653 * genmultilib (combination_space): Accept '+' in option names.
8655 2017-06-28 Martin Liska <mliska@suse.cz>
8658 * asan.c (instrument_derefs): Bail out inner references
8659 that are hard register variables.
8661 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8664 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8665 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8667 2017-06-28 Richard Biener <rguenther@suse.de>
8669 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8670 (vect_get_slp_defs): Adjust.
8671 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8673 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8675 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8676 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8677 (vectorizable_reduction): Adjust.
8678 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8680 (vect_get_slp_defs): Likewise.
8681 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8682 (vectorizable_bswap): Adjust.
8683 (vectorizable_call): Likewise.
8684 (vectorizable_conversion): Likewise.
8685 (vectorizable_assignment): Likewise.
8686 (vectorizable_shift): Likewise.
8687 (vectorizable_operation): Likewise.
8688 (vectorizable_store): Likewise.
8689 (vectorizable_condition): Likewise.
8690 (vectorizable_comparison): Likewise.
8692 2017-06-28 Michael Collison <michael.collison@arm.com>
8695 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8697 (arm_gen_movmemqi): Removed unused variable 'i'.
8698 Convert 'for' loop into 'while' loop.
8699 (arm_expand_prologue): Remove last unnecessary set of insn.
8700 (thumb_pop): Remove unused variable 'pushed_words'.
8701 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8703 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8705 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8706 * config/s390/s390-protos.h: Add prototype of
8707 s390_rel_address_ok_p.
8708 * config/s390/s390.c (s390_got_symbol): New function.
8709 (s390_rel_address_ok_p): New function.
8710 (legitimize_pic_address): Use s390_rel_address_ok_p.
8711 (s390_load_got): Use s390_got_symbol.
8712 (s390_option_override): Issue error if
8713 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8714 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8716 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8718 2017-06-27 Andrew Pinski <apinski@cavium.com>
8720 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8721 (X * copysign (1.0, X)): New pattern.
8722 (X * copysign (1.0, -X)): New pattern.
8723 (copysign (-1.0, CST)): New pattern.
8725 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8727 * genmultilib (combination_space): Remove variable.
8728 Validate reuse rules against regular expression for any sequence
8729 of multilib options in any order.
8731 2017-06-27 Michael Collison <michael.collison@arm.com>
8733 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8734 call aarch64_split_simd_combine.
8735 * (aarch64_combine_internal<mode>): Delete pattern.
8736 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8737 Allow register and subreg operands.
8739 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8741 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8742 specific need, just fallback on defaults.
8743 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8745 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8746 Olivier Hainque <hainque@adacore.com>
8748 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8750 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8751 targets. Pick a default if no particular attempt applied.
8752 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8755 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8757 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8758 (x86_64-wrs-vxworks7): Likewise.
8760 2017-06-27 Marek Polacek <polacek@redhat.com>
8763 * ubsan.c (instrument_null): Check get_base_address's result for null.
8765 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8767 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8769 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8771 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8772 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8773 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8774 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8776 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8777 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8778 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8779 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8780 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8781 BUILT_IN_FEUPDATEENV): New builtins.
8782 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8783 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8784 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8785 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8787 (builtin_structptr_types): Adjust size.
8788 * tree.c (builtin_structptr_types): Add four entries.
8790 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8791 Olivier Hainque <hainque@adacore.com>
8793 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8794 (TLS_SYM): New local macro, forcing reference to __tls__ on
8795 link command lines for VxWorks 7 RTPs, triggering initialization
8797 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8798 OS features TLS support, true for RTPs on VxWorks 7.
8799 * config/vxworks.c (vxworks_override_options): Setup emutls
8802 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8804 * predict.c (test_prediction_value_range): Use -1U instead of -1
8805 to avoid narrowing conversion warning.
8806 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8807 to avoid narrowing conversion warning.
8808 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8810 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8812 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8814 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8815 64bit configurations.
8816 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8817 (SIZE_TYPE): Likewise.
8818 * config/vxworks.c (vxworks_emutls_var_fields): Use
8819 long_unsigned_type_node instead of unsigned_type_node as the offset
8820 field type, which is "pointer" mode in emutls.c.
8822 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8825 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8828 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8829 gimple_vuse copying separately from gimple_vdef copying.
8831 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8833 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8836 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8837 Olivier Hainque <hainque@adacore.com>
8839 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8840 for all vxworks7 targets.
8841 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8842 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8843 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8844 variations for VX6/VX7 and 32/64bits later on in ...
8845 (VXWORKS_LIB_SPEC): Leverage new macros.
8846 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8847 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8849 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8851 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8852 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8854 2017-06-26 Carl Love <cel@us.ibm.com>
8856 * config/rs6000/rs6000-c.c: Add support for built-in functions
8857 vector bool char vec_reve (vector bool char);
8858 vector signed char vec_reve (vector signed char);
8859 vector unsigned char vec_reve (vector unsigned char);
8860 vector bool int vec_reve (vector bool int);
8861 vector signed int vec_reve (vector signed int);
8862 vector unsigned int vec_reve (vector unsigned int);
8863 vector bool long long vec_reve (vector bool long long);
8864 vector signed long long vec_reve (vector signed long long);
8865 vector unsigned long long vec_reve (vector unsigned long long);
8866 vector bool short vec_reve (vector bool short);
8867 vector signed short vec_reve (vector signed short);
8868 vector double vec_reve (vector double);
8869 vector float vec_reve (vector float);
8870 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8871 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8872 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8873 (altivec_vreve): New pattern.
8874 * config/rs6000/altivec.h (vec_reve): New define.
8875 * doc/extend.texi (vec_rev): Update the built-in documentation file
8876 for the new built-in functions.
8878 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8880 PR tree-optimization/71815
8881 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8883 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8885 (slsr_process_phi): Likewise.
8886 (replace_uncond_cands_and_profitable_phis): Don't replace a
8887 multiply candidate with a stride of 1 (copy or cast).
8888 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8890 (lowest_cost_path): Likewise.
8891 (total_savings): Likewise.
8893 2017-06-26 Richard Biener <rguenther@suse.de>
8896 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8897 Use def_builtin_pure for all gather builtins.
8899 2017-06-26 Richard Biener <rguenther@suse.de>
8901 PR tree-optimization/81203
8902 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8905 2017-06-26 Marek Polacek <polacek@redhat.com>
8908 * doc/invoke.texi: Document -Wmultistatement-macros.
8910 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8912 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8913 arm_neon_ok_no_float_abi effective target.
8915 2017-06-26 Richard Biener <rguenther@suse.de>
8917 PR tree-optimization/80928
8918 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8919 (copy_bbs): Set BB_DUPLICATED flag early.
8920 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8922 (execute_on_shrinking_pred): Likewise.
8923 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8924 BB_DUPLICATED blocks.
8925 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8926 iterate over all PHIs considering removal of *gsi.
8928 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8930 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8933 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8935 * config/rs6000/rs6000-string.c: (expand_block_clear,
8936 do_load_for_compare, select_block_compare_mode,
8937 compute_current_alignment, expand_block_compare,
8938 expand_strncmp_align_check, expand_strn_compare,
8939 expand_block_move, rs6000_output_load_multiple)
8940 Move functions related to string/block move/compare
8942 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8943 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8944 for this function which is now used in two files.
8945 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8946 * config.gcc: Add rs6000-string.o to extra_objs for
8947 targets powerpc*-*-* and rs6000*-*-*.
8949 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8952 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8953 32-bit, since indexed is not valid for DImode.
8954 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8955 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8956 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8957 (define_peephole2 for Altivec d-form store): Likewise.
8960 * multiple_target.c (create_dispatcher_calls): Only create the
8961 dispatcher call if the function is the default clone of a
8964 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8967 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8968 a call, force the call to not be a tail call.
8970 2017-06-23 Jeff Law <law@redhat.com>
8972 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8975 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8977 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8978 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8979 handling for early expansion of vector shifts (sl,sr,sra,rl).
8980 (builtin_function_type): Add vector shift right instructions
8981 to the unsigned argument list.
8983 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8985 rtl-optimizatoin/79286
8986 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8987 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8988 trap. PIC register plus a const unspec without offset can never trap.
8990 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8992 * tree.h (builtin_structptr_type): New type.
8993 (builtin_structptr_types): Declare new array.
8994 * tree.c (builtin_structptr_types): New array.
8995 (free_lang_data, build_common_tree_nodes): Use it.
8997 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9000 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9003 2017-06-22 Matt Turner <mattst88@gmail.com>
9005 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9006 Lake models to skylake case. Assume skylake for unknown
9007 models with clflushopt.
9009 2017-06-22 Jeff Law <law@redhat.com>
9011 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9012 frame sizes that do not satisfy aarch64_uimm12_shift.
9014 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9016 * profile-count.h (apply_probability,
9017 apply_scale, probability_in): Fix checks for zero.
9019 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9021 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9022 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9024 2016-06-22 Richard Biener <rguenther@suse.de>
9026 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9027 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9028 REDUC_MAX_EXPR support.
9029 (vectorizable_reduction): Likewise.
9030 (vect_create_epilog_for_reduction): Likewise.
9032 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9034 * match.pd (A / (1 << B) -> A >> B): New.
9035 * generic-match-head.c: Include optabs-tree.h.
9036 * gimple-match-head.c: Likewise.
9037 * optabs-tree.h (target_supports_op_p): New.
9038 * optabs-tree.c (target_supports_op_p): New.
9040 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9042 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9043 $gcc_cv_ld --help output.
9044 (gcc_cv_ld_demangle): Likewise.
9045 (gcc_cv_ld_eh_frame_hdr): Likewise.
9046 (gcc_cv_ld_pie): Likewise.
9047 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9048 (gcc_cv_ld_buildid): Likewise.
9049 (gcc_cv_ld_sysroot): Likewise.
9050 (ld_bndplt_support): Likewise.
9051 (ld_pushpopstate_support): Likewise.
9052 * configure: Regenerate.
9053 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9055 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9058 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9059 operands indexes to avoid gap between operands and match_dups.
9061 2017-06-21 Andrew Pinski <apinski@cavium.com>
9063 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9064 Increment Arith_shift and Arith_shift_reg by 1.
9065 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9067 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9068 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9069 (aarch64_strip_extend): Add new argument and test for it.
9070 (aarch64_cheap_mult_shift_p): New function.
9071 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9072 add a cost if it is true.
9073 Update calls to aarch64_strip_extend.
9074 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9076 2017-06-21 Andrew Pinski <apinski@cavium.com>
9078 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9080 (thunderxt88): Likewise.
9081 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9082 (thunderx_prefetch_tune): New variable.
9083 (thunderx2t99_prefetch_tune): Update for the correct values.
9084 (thunderxt88_tunings): New variable.
9085 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9086 generic_prefetch_tune.
9087 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9089 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9091 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9092 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9093 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9094 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9095 (aarch64_atomic_cas<mode>, GPI): Likewise.
9097 2017-06-21 Martin Liska <mliska@suse.cz>
9099 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9100 statements on cold and hot labels.
9101 * predict.c (tree_estimate_probability_bb): Remove the
9102 prediction from this place.
9104 2017-06-21 Martin Liska <mliska@suse.cz>
9106 PR tree-optimization/79489
9107 * gimplify.c (maybe_add_early_return_predict_stmt): New
9109 (gimplify_return_expr): Call the function.
9110 * predict.c (tree_estimate_probability_bb): Remove handling
9112 * predict.def: Update comment about early return predictor.
9113 * gimple-predict.h (is_gimple_predict): New function.
9114 * predict.def: Change default value of early return to 66.
9115 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9117 * passes.def: Put pass_strip_predict_hints to the beginning of
9120 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9122 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9123 FUNCTION_DECL declarations.
9124 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9126 (dwaf2out_decl): Likewise.
9127 * godump.c (go_early_global_decl): Skip call to the real debug hook
9128 for FUNCTION_DECL declarations.
9129 * passes.c (rest_of_decl_compilation): Skip call to the
9130 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9131 -fdump-go-spec is passed.
9133 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9135 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9136 Reorder for compactness.
9137 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9138 (def_builtin_pure, def_builtin_pure2): New functions.
9139 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9141 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9143 * match.pd (nop_convert): New predicate.
9144 ((A +- CST1) +- CST2): Allow some NOP conversions.
9146 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9149 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9150 with ctors/dtors if GOVD_SHARED is set.
9152 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9154 * config/aarch64/aarch64.md (movti_aarch64):
9155 Emit mov rather than orr.
9156 (movtf_aarch64): Likewise.
9157 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9158 Emit mov rather than orr.
9160 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9162 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9163 Swap alternatives, make integer dup more expensive.
9165 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9167 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9168 Return true for non-tls symbols.
9170 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9172 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9173 (cortex-a75): Likewise.
9174 (cortex-a75.cortex-a55): Likewise.
9175 * config/aarch64/aarch64-tune.md: Regenerate.
9176 * doc/invoke.texi (-mtune): Document new values for -mtune.
9178 2017-06-21 Tom de Vries <tom@codesourcery.com>
9180 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9182 (Effective-Target Keywords, Other attributes): Suggest using
9183 dg-add-options stack_size feature to get stack limit in stack_size
9184 effective target documentation.
9186 2017-06-21 Julian Brown <julian@codesourcery.com>
9187 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9189 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9190 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9191 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9193 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9194 attribute type list for neon_multiply.
9195 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9196 attribute type list for neon_multiply.
9197 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9198 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9199 attribute type list for neon_multiply.
9200 * config/arm/types.md (crypto_pmull): Add.
9201 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9202 attribute type list.
9204 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9206 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9209 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9211 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9212 to make sure not to dereference a NULL cost_classes_ptr pointer.
9214 2017-06-20 Carl Love <cel@us.ibm.com>
9216 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9217 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9218 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9219 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9220 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9221 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9222 VMULOSW): New enum "unspec" values.
9223 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9224 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9225 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9226 altivec_vmulosw): New patterns.
9227 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9228 VMULOSW): Add definitions.
9230 2017-06-20 Julia Koval <julia.koval@intel.com>
9232 * config/i386/i386.c: Fix rounding expand for new pattern.
9233 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9235 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9237 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9238 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9240 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9242 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9245 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9247 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9248 architecture, then by implementer ID.
9249 * config/aarch64/aarch64-tune.md: Regenerate.
9251 2017-06-20 Richard Biener <rguenther@suse.de>
9254 * fold-const.c (split_tree): Fold to type before negating.
9256 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9258 * diagnostic-show-locus.c
9259 (selftest::test_fixit_deletion_affecting_newline): New function.
9260 (selftest::diagnostic_show_locus_c_tests): Call it.
9262 2017-06-20 Andreas Schwab <schwab@suse.de>
9265 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9268 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9270 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9271 __ARM_FEATURE_COPROC according to support.
9273 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9275 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9276 Rewritten to avoid overflow for > 32-bit pointers.
9279 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9280 by removing enum keyword.
9281 (ubsan_type_descriptor): Likewise. Formatting fix.
9284 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9285 splitter): Require TARGET_SSE2 in the condition.
9287 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9290 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9291 for doing vector set of SFmode on ISA 3.0.
9292 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9293 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9295 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9296 SFmode value into a V4SF variable that was extracted from another
9297 V4SF variable without converting the element to double precision
9298 and back to single precision vector format.
9299 (vsx_insert_extract_v4sf_p9_2): Likewise.
9301 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9303 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9304 in UWHI to avoid undefined overflow.
9307 * ubsan.h (enum ubsan_encode_value_phase): New.
9308 (ubsan_encode_value): Change second argument to
9309 enum ubsan_encode_value_phase with default value of
9310 UBSAN_ENCODE_VALUE_GENERIC.
9311 * ubsan.c (ubsan_encode_value): Change second argument to
9312 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9313 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9314 create_tmp_var_raw instead of create_tmp_var and use a
9316 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9317 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9318 ubsan_encode_value callers.
9321 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9322 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9323 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9325 2017-06-19 Richard Biener <rguenther@suse.de>
9328 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9329 estimates if we changed anything.
9331 2017-06-19 Richard Biener <rguenther@suse.de>
9333 PR tree-optimization/80887
9334 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9335 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9336 simplified lookups, then reset mprts_hook.
9337 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9339 (try_to_simplify): Likewise.
9341 2017-06-19 Martin Liska <mliska@suse.cz>
9344 * gimplify.c (gimplify_switch_expr):
9345 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9347 2017-06-19 Martin Liska <mliska@suse.cz>
9349 * doc/install.texi: Document that PGO runs in 4 stages.
9351 2017-06-19 Martin Liska <mliska@suse.cz>
9354 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9355 to dispatcher function name.
9356 * multiple_target.c (replace_function_decl): New function.
9357 (create_dispatcher_calls): Redirect both edges and references.
9359 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9361 * profile-count.c (profile_count::dump): Dump quality.
9362 (profile_count::differs_from_p): Update for unsigned val.
9363 * profile-count.h (profile_count_quality): New enum.
9364 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9366 2017-06-19 Richard Biener <rguenther@suse.de>
9368 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9369 struct function as arg.
9370 (estimate_numbers_of_iterations): Export overload with loop arg.
9371 (free_numbers_of_iterations_estimates_loop): Use an overload of
9372 free_numbers_of_iterations_estimates instead.
9373 * tree-cfg.c (remove_bb): Adjust.
9374 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9375 * tree-parloops.c (gen_parallel_loop): Likewise.
9376 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9378 (tree_unroll_loops_completely): Likewise.
9379 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9380 Use an overload instead and export.
9381 (estimated_loop_iterations): Adjust.
9382 (max_loop_iterations): Likewise.
9383 (likely_max_loop_iterations): Likewise.
9384 (estimate_numbers_of_iterations): Take struct function as arg
9386 (loop_exits_before_overflow): Adjust.
9387 (free_numbers_of_iterations_estimates_loop): Use an overload.
9388 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9389 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9391 2017-06-19 Richard Biener <rguenther@suse.de>
9394 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9395 RANGE_EXPR conservatively.
9397 2017-06-16 Carl Love <cel@us.ibm.com>
9399 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9400 definitions for vec_float, vec_float2, vec_floato,
9401 vec_floate built-ins.
9402 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9403 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9405 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9406 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9407 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9408 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9409 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9410 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9411 vec_floato): Add builtin defines.
9412 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9413 Update the built-in documentation file for the new built-in
9416 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9418 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9419 (mthumb): Mark as the negative of -marm.
9421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9423 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9425 (ARM Options, -mtune): Document that this accepts the same
9426 extension options as -mcpu.
9427 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9431 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9432 permitted extensions.
9434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9436 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9437 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9438 (armv8-m.main): Add option +nodsp.
9439 * config/arm/arm-cpu-cdata.h: Regenerated.
9441 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9443 * config/arm/t-fuchsia: New file.
9444 * config.gcc (arm*-*-fuchsia*): Use it.
9446 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9448 * config/arm/t-symbian: Rewrite for new option infrastructure.
9450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9452 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9453 (MULTILIB_REQUIRED): Likewise.
9455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9457 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9458 (MULTILIB_RESUE): Likewise.
9459 (MULTILIB_MATCHES): Likewise.
9460 (MULTLIB_REQUIRED): Likewise.
9462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9464 * config/arm/t-rtems: Rewrite for new option framework.
9466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9468 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9469 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9470 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9471 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9472 * config/arm/t-multilib: ... here.
9473 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9474 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9475 armv7-a and armv8*-a when A-profile libraries have not been built.
9476 * config/arm/t-rmprofile: Rewrite.
9478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9480 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9481 with a backslash. Remove the backslash after substituting unescaped
9483 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9487 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9488 the driver to pass through the default setting of -mfloat-abi.
9489 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9491 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9492 (all_feat_combs): New rule.
9493 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9495 * config/arm/t-aprofile: Rewrite.
9497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9499 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9500 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9501 fpu is not specified by the user/command-line.
9502 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9503 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9504 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9505 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9506 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9507 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9509 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9511 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9512 * config/arm/t-arm-elf: Rewritten.
9514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9516 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9517 have some floating-point instructions.
9518 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9519 (TARGET_MAYBE_HARD_FLOAT): New macro.
9520 * config/arm/arm-builtins.c (arm_init_builtins): Use
9521 TARGET_MAYBE_HARD_FLOAT.
9522 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9526 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9527 (configargs.h): Include it.
9528 (arm_print_hint_for_fpu_option): New function.
9529 (arm_parse_fpu_option): New function.
9530 (candidate_extension): New class.
9531 (arm_canon_for_multilib): New function.
9532 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9533 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9534 (ARCH_CANONICAL_SPECS): New macro.
9535 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9539 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9540 are set after handling multilib fragments. Set target_cpu_default2
9543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9545 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9547 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9548 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9550 (check_arch): Likewise.
9551 * config/arm/arm.c (arm_configure_build_target): Handle
9552 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9553 options in the default.
9555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9557 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9558 when an option is an alias of another.
9559 * config/arm/parsecpu.awk (optalias): New parser token.
9560 (gen_comm_data): Mark non-alias options as such. Emit entries
9561 for extension aliases.
9562 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9563 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9564 (armv6kz, armv6zk, armv6t2): Likewise.
9565 (armv7): Make vfpv3-d16 an alias.
9566 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9568 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9569 Sort in canonical order.
9570 (armv8-a): Sort in canonical order.
9571 (armv8.1-a, armv8.2-a): Likewise.
9572 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9574 (cortex-a9): Sort in canonical order.
9575 * config/arm/arm.c (selftests.h): Include it.
9576 (arm_test_cpu_arch_data): New function.
9577 (arm_run_self_tests): New function.
9578 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9579 (targetm): Move declaration to the end of the file.
9580 * arm-cpu-cdata.h: Regenerated.
9582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9584 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9585 call to target_mode_check describing the type of option passed.
9586 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9587 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9588 arm_parse_cpu_option_name to match parameters against list of
9590 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9591 arm_arch_core_flags data structure.
9592 * config/arm/arm-cpu_cdata.h: Regenerated.
9594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9596 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9598 (arm_print_hint_for_cpu_option): Likewise.
9599 (arm_print_hint_for_arch_option): Likewise.
9600 (arm_parse_cpu_option_name): Likewise.
9601 (arm_parse_arch_option_name): Likewise.
9602 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9603 of entries in the all_fpus list.
9604 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9605 (arm_parse_cpu_option_name): Declare.
9606 (arm_parse_arch_option_name): Declare.
9607 (arm_parse_option_features): Declare.
9608 (arm_intialize_isa): Declare.
9609 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9611 (gen_comm_data): ... here. Make definitions non-static.
9612 * config/arm/arm-cpu-data.h: Regenerated.
9613 * config/arm/arm-cpu-cdata.h: Regenerated.
9615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9617 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9618 (cpu_arch_extension): New structure.
9619 (cpu_arch_option, arch_option, cpu_option): New structures.
9620 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9622 (gen_data): Generate new format data tables.
9623 * config/arm/arm.c (cpu_tune): New structure.
9624 (cpu_option, processors): Delete.
9625 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9626 (arm_print_hint_for_cpu_option): ... this and ...
9627 (arm_print_hint_for_arch_option): ... this.
9628 (arm_parse_arch_cpu_name): Delete. Replace with ...
9629 (arm_parse_cpu_option_name): ... this and ...
9630 (arm_parse_arch_option_name): ... this.
9631 (arm_unrecognized_feature): Change type of target parameter to
9633 (arm_parse_arch_cpu_features): Delete. Replace with ...
9634 (arm_parse_option_features): ... this.
9635 (arm_configure_build_target): Rework to use new configuration data
9637 (arm_print_tune_info): Rework for new configuration data tables.
9638 * config/arm/arm-cpu-data.h: Regenerated.
9639 * config/arm/arm-cpu.h: Regenerated.
9641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9643 * Makefile.in (OBJS): Move sbitmap.o from here ...
9644 (OBJS-libcommon): ... to here.
9646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9648 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9649 (ISA_ALL_CRYPTO): New macro.
9650 (ISA_ALL_SIMD): New macro
9651 (ISA_ALL_FP): New macro.
9652 * config/arm/arm.c (fpu_bitlist): Update initializer.
9653 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9655 (arm9e): Add fpu. Add option for nofp
9656 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9657 (arm926ej-s, arm1026ej-s): Likewise.
9658 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9659 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9660 neon-fp16, neon-vfpv4, nofp and nosimd.
9661 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9662 (cortex-a8): Add fpu. Add option for nofp.
9663 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9664 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9665 (cortex-r4f): Add fpu.
9666 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9667 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9669 (cortex-r8): Likewise.
9670 (cortex-m4): Add fpu. Add option for nofp.
9671 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9672 (cortex-a17.cortex-a7): Likewise.
9673 (cortex-a32): Add fpu. Add options for crypto and nofp.
9674 (cortex-a35, cortex-a53): Likewise.
9675 (cortex-a57): Add fpu. Add option for crypto.
9676 (cortex-a72, cortex-a73): Likewise.
9677 (exynos-m1): Likewise.
9678 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9679 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9680 (cortex-m33): Add fpu. Add option for nofp.
9681 * config/arm/arm-cpu-cdata.h: Regenerated
9682 * config/arm/arm-cpu-data.h: Regenerated.
9684 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9686 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9687 (armv5te, armv5tej): Likewise.
9688 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9689 (armv7): Add options fp and vfpv3-d16.
9690 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9691 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9693 (armv7ve): Likewise.
9694 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9695 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9696 (armv8-a): Add nocrypto option.
9697 (armv8.1-a, armv8.2-a): Likewise.
9698 (armv8-m.main): add options fp, fp.dp and nofp.
9700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9702 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9704 (armv8-a+crc): Delete.
9705 (armv8.1-a): Add options simd, crypto and nofp.
9706 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9707 (armv8.2-a+fp16): Delete.
9708 (armv8-m.main): Add option dsp.
9709 (armv8-m.main+dsp): Delete.
9710 (cortex-a8): Add fpu. Add nofp option.
9711 (cortex-a9): Add fpu. Add nofp and nosimd options.
9712 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9713 link to main cpu and architecture data structures.
9714 (gen_comm_data): Only put isa attributes from the main architecture
9716 (option): New statement for architecture and CPU entries.
9717 * arm.c (struct cpu_option): New structure.
9718 (struct processors): Add entry for options.
9719 (arm_unrecognized_feature): New function.
9720 (arm_parse_arch_cpu_name): Ignore any characters after the first
9722 (arm_parse_arch_cpu_feature): New function.
9723 (arm_configure_build_target): Separate out any CPU and architecture
9724 features and parse separately. Don't error out if -mfpu=auto is
9725 used with only an architecture string.
9726 (arm_print_asm_arch_directives): New function.
9727 (arm_file_start): Call it.
9728 * config/arm/arm-cpu-cdata.h: Regenerated.
9729 * config/arm/arm-cpu-data.h: Likewise.
9730 * config/arm/arm-tables.opt: Likewise.
9732 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9734 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9735 assembler when it is not -mfpu=auto.
9737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9739 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9740 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9741 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9742 (ASM_CPU_SPEC): Rewrite.
9743 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9744 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9745 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9746 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9747 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9748 copied string is NUL-terminated. Also strip any characters prefixed
9750 (arm_rewrite_selected_arch): New function.
9751 (arm_rewrite_march): New function.
9753 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9755 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9756 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9757 (march, mcpu, mtune): Convert to string-based options.
9758 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9759 (arm_parse_arch_cpu_name): New function.
9760 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9761 identify selected architecture or CPU.
9762 (arm_option_save): New function.
9763 (TARGET_OPTION_SAVE): Redefine.
9764 (arm_option_restore): Restore string options.
9765 (arm_option_print): Print string options.
9767 2017-06-16 Martin Sebor <msebor@redhat.com>
9769 PR tree-optimization/80933
9770 PR tree-optimization/80934
9771 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9772 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9773 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9774 (gimple_fold_builtin): Call them.
9776 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9778 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9779 as unlikely; update profile.
9781 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9783 * predict.c (force_edge_cold): Handle declaring edges impossible
9786 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9788 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9790 (try_unroll_loop_completely): Fix reporting.
9792 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9794 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9796 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9799 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9800 if given a non-constant argument for an intrinsic which requires a
9803 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9805 * profile.c (compare_freqs): New function.
9806 (branch_prob): Sort edge list.
9807 (find_spanning_tree): Assume that the list is priority sorted.
9809 2017-06-16 Richard Biener <rguenther@suse.de>
9811 PR tree-optimization/81090
9812 * passes.def (pass_record_bounds): Remove.
9813 * tree-pass.h (make_pass_record_bounds): Likewise.
9814 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9815 make_pass_record_bounds): Likewise.
9816 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9817 not free niter estimates at the beginning but at the end.
9818 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9820 2017-06-16 Richard Biener <rguenther@suse.de>
9822 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9823 initializer to workaround ICE in host GCC 4.8.
9825 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9827 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9829 (clone_inlined_nodes): Update.
9831 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9833 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9834 prefetch settings, and enable prefetching by default at -O3.
9836 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9838 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9839 Set flag_prefetch_loop_arrays according to tuning data.
9841 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9843 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9845 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9846 [Unrelated to main purpose of the patch] Place the pointer field last
9847 to enable type checking errors when tune structure are wrongly merged.
9848 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9849 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9850 (thunderx2t99_prefetch_tune): New tune constants.
9851 (tune_params *_tunings): Update all tunings (no functional change).
9852 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9853 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9854 from tunings structures.
9856 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9859 * ubsan.c (instrument_null): Add T argument, use it instead
9860 of computing it based on IS_LHS.
9861 (instrument_object_size): Likewise.
9862 (pass_ubsan::execute): Adjust instrument_null and
9863 instrument_object_size callers to pass gimple_get_lhs or
9864 gimple_assign_rhs1 result to it. Use instrument_null instead of
9865 calling get_base_address and instrument_mem_ref. Handle
9866 aggregate call arguments for object-size sanitization.
9868 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9870 PR tree-optimization/81089
9871 * tree-vrp.c (is_masked_range_test): Validate operands of
9874 2017-06-15 Martin Sebor <msebor@redhat.com>
9877 * dumpfile.c (dump_register): Avoid calling memset to initialize
9878 a class with a default ctor.
9879 * gcc.c (struct compiler): Remove const qualification.
9880 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9881 * hash-table.h: Ditto.
9882 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9884 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9885 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9887 * params.h (struct param_info): Make struct members non-const.
9888 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9889 with default initialization.
9890 * vec.h (vec_copy_construct, vec_default_construct): New helper
9892 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9893 with vec_copy_construct.
9894 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9895 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9896 * doc/invoke.texi (-Wclass-memaccess): Document.
9898 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9900 * emit-rtl.h (is_leaf): Update comment about local
9903 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9906 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9907 for a variable to have a section before checking if the section has a
9909 Set section to.persistent if persistent attribute is set.
9910 Warn if .persistent attribute is used on an automatic variable.
9912 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9914 PR rtl-optimization/80474
9915 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9917 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9919 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9922 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9924 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9925 (casesi): Emit bounds checking as RTL.
9926 (casesi_internal_mips16_<mode>): Remove bounds checking.
9928 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9930 * config/xtensa/xtensa.c (xtensa_option_override): Append
9931 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9932 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9933 xtensa_doloop_hooks): Define unconditionally.
9934 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9936 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9937 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9938 for it in xtensa_option_override.
9939 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9940 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9942 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9944 * doc/cppopts.texi: Document '-' special value to -MF.
9946 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9948 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9949 (cortex_a53_fconst): Likewise.
9950 (cortex_a53_fpmul): Likewise.
9951 (cortex_a53_f_load_64): Likewise.
9952 (cortex_a53_f_load_many): Likewise.
9953 (cortex_a53_advsimd_alu): Likewise.
9954 (cortex_a53_advsimd_alu_q): Likewise.
9955 (cortex_a53_advsimd_mul): Likewise.
9956 (cortex_a53_advsimd_mul_q): Likewise.
9957 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9958 Add missing fmul, r2f_cvt and fconst cases.
9960 2017-06-14 Richard Biener <rguenther@suse.de>
9963 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9965 (fold_binary_loc): When associating do not treat pre-existing
9966 TREE_OVERFLOW on literal constants as a reason to allow
9967 TREE_OVERFLOW on associated literal constants.
9969 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9971 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9972 (MASK_FEATURES): New macro.
9973 * config/sparc/sparc.c (sparc_option_override): Remove the special
9974 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9976 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9978 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9979 a division of 0 if non-call exceptions are enabled.
9981 2017-06-14 Andrew Pinski <apinski@cavium.com>
9982 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9985 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9986 Improve vector initialization code gen for only variable case.
9988 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9990 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9992 2017-06-14 Richard Biener <rguenther@suse.de>
9994 PR tree-optimization/81083
9995 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9998 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10000 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10001 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10002 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10003 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10004 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10005 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10007 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10009 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10010 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10012 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10014 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10016 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10018 * config/rs6000/t-rtems: Don't handle SPE.
10020 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10022 * config/rs6000/t-linux: Don't handle SPE.
10024 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10026 * config/rs6000/eabispe.h: Delete file.
10028 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10030 * config/rs6000/t-spe: Delete file.
10032 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10034 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10035 (rs6000_legitimate_offset_address_p): Return false for anything in
10036 V2SImode or V2SFmode.
10038 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10040 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10041 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10042 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10044 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10045 (rs6000_legitimate_offset_address_p): Ditto.
10046 (rs6000_emit_move): Ditto.
10047 (rs6000_init_builtins): Remove V4HI_type_node.
10049 2017-06-13 Martin Liska <mliska@suse.cz>
10052 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10053 (gate_asan): Likewise.
10054 * asan.h (asan_no_sanitize_address_p): Remove the function.
10055 (sanitize_flags_p): New function.
10056 * builtins.def: Fix coding style.
10057 * common.opt: Use renamed enum value.
10058 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10059 * doc/extend.texi: Document no_sanitize attribute.
10060 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10061 to SANITIZE_UNDEFINED_NONDEFAULT.
10062 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10063 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10064 Use sanitize_flags_p.
10065 * gimplify.c (gimplify_function_tree): Likewise.
10066 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10067 * opts.c (parse_no_sanitize_attribute): New function.
10068 (common_handle_option): Use renamed enum value.
10069 * opts.h (parse_no_sanitize_attribute): Declare.
10070 * tree.c (sanitize_flags_p): New function.
10071 * tree.h: Declared here.
10072 * tsan.c: Use sanitize_flags_p.
10073 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10074 (instrument_mem_ref): Likewise.
10075 (instrument_bool_enum_load): Likewise.
10076 (do_ubsan_in_current_function): Remove the function.
10077 (pass_ubsan::execute): Use sanitize_flags_p.
10078 * ubsan.h: Remove do_ubsan_in_current_function
10079 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10080 (dump_function_to_file): Use it here.
10082 2017-06-13 Martin Jambor <mjambor@suse.cz>
10084 PR tree-optimization/80803
10085 PR tree-optimization/81063
10086 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10087 (propagate_subaccesses_across_link): Enqueue subtree whenever
10088 necessary instead of relying on the caller.
10090 2017-06-13 Martin Jambor <mjambor@suse.cz>
10092 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10093 that have a first_link.
10094 (sort_and_splice_var_accesses): Do not check first_link before
10096 (subtree_mark_written_and_enqueue): Likewise.
10097 (propagate_all_subaccesses): Likewise and do not stop at first
10098 parent with a first_link.
10100 2017-06-13 Martin Jambor <mjambor@suse.cz>
10102 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10105 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10107 * match.pd: New pattern.
10109 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10111 * tree-vrp.c (is_masked_range_test): New function.
10112 (register_edge_assert_for): Determine ranges for
10115 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10117 PR tree-optimization/67328
10118 * fold-const.c (maskable_range_p): New function.
10119 (build_range_check): Generate bittests if possible.
10121 2017-06-13 Martin Liska <mliska@suse.cz>
10123 * gimple-pretty-print.c (dump_probability): Add new argument.
10124 (dump_edge_probability): Dump both probability and count.
10125 (dump_gimple_label): Likewise.
10126 (dump_gimple_bb_header): Likewise.
10128 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10131 * config/avr/avr-devices.c: Fix indentation.
10132 * config/avr/gen-avr-mmcu-specs.c: Dito.
10134 2017-06-13 Richard Biener <rguenther@suse.de>
10136 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10137 instead get vector type from stmt_info.
10138 (vectorizable_reduction): Adjust. Remove dead code.
10140 2017-06-13 Richard Biener <rguenther@suse.de>
10142 PR middle-end/81065
10143 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10144 case of C * (x * C2 + C3).
10145 (fold_addr_of_array_ref_difference): Properly fold index difference.
10147 2017-06-12 David S. Miller <davem@davemloft.net>
10150 * config/sparc/sparc.md (return expander): Emit frame blockage if
10151 function uses alloca.
10153 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10155 * combine.c (make_field_assignment): Check len rather than the mode
10156 precision when calling force_to_mode.
10158 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10160 Support multilibs and devices that see flash in RAM address range.
10163 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10164 (avr_mcu_t) <flash_pm_offset>: New field.
10165 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10166 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10167 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10168 (AVR_TINY_PM_OFFSET): Remove macro.
10169 * config/avr/avr.opt (-mshort-calls): New option.
10170 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10171 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10172 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10173 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10174 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10175 instead of avr_arch->have_jmp_call.
10176 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10177 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10178 avr_arch->flash_pm_offset to define.
10179 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10180 new field flash_pm_offset. Add entry for avrxmega3.
10181 (avr_texinfo): Add entry for avrxmega3.
10182 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10183 attiny212, attiny214,
10184 attiny412, attiny414, attiny416, attiny417,
10185 attiny814, attiny816, attiny817,
10186 attiny1614, attiny1616, attiny1617,
10187 attiny3214, attiny3216, attiny3217.
10188 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10189 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10190 (avr_print_operand_address) [AVR_TINY]: Same.
10191 (avr_asm_init_sections) <readonly_data_section>: Only patch
10192 callback if avr_arch->flash_pm_offset = 0.
10193 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10194 for rodata if avr_arch->flash_pm_offset != 0.
10195 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10196 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10197 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10198 (m_options): Append opt_rcall.
10199 (m_dirnames): Append dir_rcall.
10200 * config/avr/t-multilib: Regenerate.
10202 * configure.ac [target=avr]: Check whether avrxmega3 default
10203 linker description file works as needed.
10204 * configure: Regenerate.
10205 * doc/avr-mmcu.texi: Regenerate.
10206 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10207 <__AVR_ARCH__>: Document avrxmega3 and 103.
10208 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10209 <__AVR_SHORT_CALLS__>: Document it.
10210 <__AVR_PM_BASE_ADDRESS__>: Document it.
10211 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10212 (AVR Variable Attributes) <progmem>: Document this is
10213 not needed for avrxmega3.
10214 (AVR Named Address Spaces) <__flash>: Dito.
10216 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10218 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10220 2017-06-12 Doug Rupp <rupp@adacore.com>
10222 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10223 Append vxworks-stdint.h to the tm_file list.
10224 * config/vxworks-stdint.h: New file.
10226 2017-06-12 Martin Liska <mliska@suse.cz>
10228 PR tree-optimization/81041
10229 * tree-profile.c (gimple_gen_ic_func_profiler):
10230 Create an extra BB in profile-generate
10231 (gimple_gen_time_profiler): Likewise.
10233 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10235 PR tree-optimization/81003
10236 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10237 (update_range_test): Use it instead of force_gimple_operand_gsi.
10239 2017-06-12 Richard Biener <rguenther@suse.de>
10241 PR tree-optimization/81053
10242 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10243 with backedge value not defined in loop. Simplify def stmt
10246 2017-06-11 Tom de Vries <tom@codesourcery.com>
10249 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10251 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10252 nvptx_cannot_force_const_mem.
10254 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10256 * opts.c (finish_options): Move test for flag_split_stack after
10257 it has been initialized.
10259 2017-06-11 Jason Merrill <jason@redhat.com>
10261 * tree.h (id_equal): New.
10262 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10263 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10264 instead of strcmp of IDENTIFIER_POINTER.
10266 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10268 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10269 (mark_all_inlined_calls_cdtor): Fix formating.
10270 (inline_transform): Rescale profile before inlining.
10272 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10274 * cgraph.h (cgraph_edge::clone): Update prototype.
10275 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10276 (cgraph_node::create_clone): Update.
10277 (cgraph_node::create_version_clone): Update.
10278 * tree-inline.c (copy_bb): Update.
10279 (expand_call_inline): Update.
10281 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10283 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10284 factored out from ...
10285 (rs6000_emit_prologue): ... here.
10287 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10289 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10290 factored out from ...
10291 (rs6000_emit_prologue): ... here.
10293 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10295 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10297 (handle_missing_profiles): Fix computation of tp_first_run.
10298 (counts_to_freqs): Do not touch freqs when count is 0.
10300 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10302 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10305 2017-06-10 Tom de Vries <tom@codesourcery.com>
10307 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10308 attributes): Document signal effective target.
10310 2017-06-10 Tom de Vries <tom@codesourcery.com>
10312 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10313 Document effective target stack_size.
10315 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10317 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10318 to the edit_context if they can be auto-applied.
10320 2017-06-9 Ian Lance Taylor <iant@golang.org>
10322 * opts.c (finish_options): If -fsplit-stack, disable implicit
10323 -forder-blocks-and-partition.
10324 * doc/invoke.texi (Optimize Options): Document that when using
10325 -fsplit-stack -forder-blocks-and-partition is not implicitly
10328 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10330 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10331 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10332 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10333 * builtins.def (abort, trap, unreachable): Declare cold.
10334 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10335 * tree-core.h (ECF_COLD): New.
10336 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10337 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10339 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10341 * predict.c (unlikely_executed_stmt_p): Cleanup.
10343 2017-06-09 Richard Biener <rguenther@suse.de>
10345 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10346 model if the ref is always written to.
10348 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10350 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10352 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10354 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10357 2017-06-09 Tom de Vries <tom@codesourcery.com>
10360 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10361 "target cannot support label values" when encountering LABEL_REF.
10363 2017-06-09 Martin Liska <mliska@suse.cz>
10365 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10366 (gimple_gen_ic_func_profiler): Emit direct comparison
10367 of __gcov_indirect_call_callee with NULL.
10368 (gimple_gen_time_profiler): Change probability from
10369 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10371 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10373 * profile.c (edge_gcov_counts): Turn to pointer.
10374 (compute_branch_probabilities, compute_branch_probabilities): Update.
10375 (branch_prob): Do not clear edge_gcov_count.
10376 * profile.h (edge_gcov_counts): Turn to pointer.
10377 (edge_gcov_count): Update.
10379 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10381 * gimple.h (gimple_check_failed): Mark cold.
10383 2017-06-09 Richard Biener <rguenther@suse.de>
10385 PR tree-optimization/66623
10386 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10387 refactor check_reduction into two parts, properly computing
10388 whether we have to check reduction validity for outer loop
10391 2017-06-09 Richard Biener <rguenther@suse.de>
10393 PR tree-optimization/79483
10394 * graphite-scop-detection.c (order): New global.
10395 (get_order): Compute bb to order mapping that satisfies code
10396 generation constraints.
10397 (cmp_pbbs): New helper.
10398 (build_scops): Start domwalk at entry block, sort generated
10401 2017-06-09 Richard Biener <rguenther@suse.de>
10403 PR middle-end/81007
10404 * ipa-polymorphic-call.c
10405 (ipa_polymorphic_call_context::restrict_to_inner_class):
10406 Skip FIELD_DECLs with error_mark_node type.
10407 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10410 2017-06-09 Martin Liska <mliska@suse.cz>
10412 * predict.c (struct branch_predictor): New struct.
10413 (test_prediction_value_range): New test.
10414 (predict_c_tests): New function.
10415 * selftest-run-tests.c (selftest::run_tests): Run the function.
10416 * selftest.h: Declare new tests.
10418 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10421 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10422 gen_add3_insn did not fail.
10423 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10424 r0, construct that number in a temporary reg and add that reg to r0.
10425 If asked to put the result in r0 as well, fail.
10427 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10429 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10430 for early expansion of vec_eqv.
10432 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10434 PR middle-end/81005
10435 * ubsan.c (instrument_null): Avoid pointless code temporary.
10436 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10438 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10442 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10445 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10448 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10450 * predict.c (unlikely_executed_edge_p): Move ahead.
10451 (probably_never_executed_edge_p): Use it.
10453 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10455 PR middle-end/79988
10456 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10457 gimple_call_builtin_p call.
10459 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10461 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10462 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10463 rtl_check_failed_type2, rtl_check_failed_code1,
10464 rtl_check_failed_code2, rtl_check_failed_code_mode,
10465 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10466 rtvec_check_failed_bounds, rtl_check_failed_flag,
10467 _fatal_insn_not_found, _fatal_insn): Likewise.
10468 * tree.h (tree_contains_struct_check_failed,
10469 tree_check_failed, tree_not_check_failed,
10470 tree_class_check_failed, tree_range_check_failed,
10471 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10472 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10473 tree_operand_check_failed, omp_clause_check_failed,
10474 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10477 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10479 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10480 flag_branch_probabilities.
10481 * ipa-inline.c (edge_badness): Likewise.
10482 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10483 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10484 * predict.c (maybe_hot_frequency_p): Likewise.
10485 (probably_never_executed): Likewise.
10486 * sched-ebb.c (schedule_ebbs): Likewise.
10487 * sched-rgn.c (find_single_block_region): Likewise.
10488 * tracer.c (tail_duplicate): Likewise.
10490 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10492 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10493 longer requires x_flag_profile_use.
10495 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10497 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10498 instead of flag_reorder_blocks_and_partition.
10499 * dbxout.c (dbxout_function_end): Likewise.
10500 * dwarf2out.c (gen_subprogram_die): Likewise.
10501 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10502 * hw-doloop.c (reorg_loops): Likewise.
10503 * varasm.c (assemble_start_function,
10504 assemble_end_function): Likewise.
10505 (decide_function_section): Do not check for
10506 flag_reorder_blocks_and_partition.
10508 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10510 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10512 (chkp_get_hard_register_fake_addr_expr): Ditto.
10513 (chkp_build_addr_expr): Add check for hard reg case.
10514 (chkp_parse_array_and_component_ref): Ditto.
10515 (chkp_find_bounds_1): Ditto.
10516 (chkp_process_stmt): Don't generate bounds store for
10519 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10521 * predict.c (maybe_hot_bb_p): Do not check profile status.
10522 (maybe_hot_edge_p): Likewise.
10523 (probably_never_executed): Check for zero counts even if profile
10525 (unlikely_executed_edge_p): New function.
10526 (unlikely_executed_stmt_p): New function.
10527 (unlikely_executed_bb_p): New function.
10528 (set_even_probabilities): Use unlikely predicates.
10529 (combine_predictions_for_bb): Likewise.
10530 (predict_paths_for_bb): Likewise.
10531 (predict_paths_leading_to_edge): Likewise.
10532 (determine_unlikely_bbs): New function.
10533 (estimate_bb_frequencies): Use it.
10534 (compute_function_frequency): Use zero counts even if profile is
10536 * profile-count.h: Fix typo.
10538 2017-08-08 Julia Koval <julia.koval@intel.com>
10540 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10541 _mm512_mask_cvtsepi16_storeu_epi8,
10542 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10543 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10544 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10545 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10546 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10547 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10548 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10549 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10550 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10551 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10552 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10553 __builtin_ia32_pmovuswb256mem_mask,
10554 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10555 __builtin_ia32_pmovwb512mem_mask): New builtins.
10557 2017-08-08 Julia Koval <julia.koval@intel.com>
10559 PR target/73350,80862
10560 * config/i386/subst.md (round): Fix round pattern.
10561 * config/i386/i386.c (ix86_erase_embedded_rounding):
10562 Fix erasing rounding for the fixed pattern.
10564 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10566 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10568 2017-06-08 Martin Liska <mliska@suse.cz>
10570 PR gcov-profile/80911
10571 * gcov.c (block_info::block_info): New constructor.
10573 2017-06-07 Carl Love <cel@us.ibm.com>
10575 * config/rs6000/rs6000-c: The return type of the following
10576 built-in functions was implemented as int not long long. Fix sign
10577 of return value for the unsigned version of vec_mulo and vec_mule.
10578 vector unsigned long long vec_bperm (vector unsigned long long,
10579 vector unsigned char)
10580 vector signed long long vec_mule (vector signed int,
10582 vector unsigned long long vec_mule (vector unsigned int,
10583 vector unsigned int)
10584 vector signed long long vec_mulo (vector signed int,
10586 vector unsigned long long vec_mulo (vector unsigned int,
10587 vector unsigned int)
10588 * doc/extend.texi: Fix the documentation for the built-in
10591 2017-06-07 Carl Love <cel@us.ibm.com>
10594 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10597 2017-06-07 Carl Love <cel@us.ibm.com>
10599 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10600 support, Generate doublehv for signed int/float for BE case only.
10602 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10604 * doc/invoke.texi (mcx16): Rewrite.
10606 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10608 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10609 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10610 *mov<mode>_softfloat, and an anonymous splitter): Use
10611 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10615 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10617 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10618 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10619 (rs6000_debug_reg_global): Adjust.
10620 (rs6000_init_hard_regno_mode_ok): Adjust.
10621 (rs6000_dbx_register_number): Adjust.
10622 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10623 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10624 Remove SPE_ACC and SPEFSCR.
10625 (REG_ALLOC_ORDER): Ditto.
10626 (FRAME_POINTER_REGNUM): Change to 111.
10627 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10628 (REG_CLASS_NAMES): Ditto.
10629 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10630 (REGISTER_NAMES): Ditto.
10631 (ADDITIONAL_REG_NAMES): Ditto.
10632 (rs6000_reg_names): Ditto.
10633 * config/rs6000/rs6000.md: Renumber some register number
10636 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10638 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10640 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10641 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10643 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10644 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10645 Delete the SPE high registers.
10646 (REG_ALLOC_ORDER): Ditto.
10647 (enum reg_class): Remove SPE_HIGH_REGS.
10648 (REG_CLASS_NAMES): Ditto.
10649 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10650 (REGISTER_NAMES): Ditto.
10651 (rs6000_reg_names): Ditto.
10652 * doc/tm.texi.in: Remove SPE as example.
10653 * doc/tm.texi: Regenerate.
10655 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10657 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10658 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10659 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10660 * config/rs6000/rs6000.md (type): Remove "brinc".
10662 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10664 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10665 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10666 * config/rs6000/linuxspe.h: Delete file.
10667 * config/rs6000/rs6000.md: Don't include spe.md.
10668 * config/rs6000/spe.h: Delete file.
10669 * config/rs6000/spe.md: Delete file.
10670 * config/rs6000/t-rs6000: Remove spe.md.
10672 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10674 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10675 (reg_or_none500mem_operand): Delete.
10676 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10677 instead of reg_or_none500mem_operand.
10679 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10681 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10682 handling of SPE flags.
10683 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10685 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10687 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10689 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10690 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10691 paired_divv2sf3): Similar.
10692 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10693 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10694 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10696 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10697 Rename the paired_* instruction patterns.
10698 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10700 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10701 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10702 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10703 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10704 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10705 PAIRED_VECTOR_MODE.
10706 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10707 (spe_func_has_64bit_regs_p): Delete.
10708 (spe_expand_predicate_builtin): Delete.
10709 (spe_expand_evsel_builtin): Delete.
10710 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10711 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10712 (invalid_e500_subreg): Delete.
10713 (rs6000_legitimize_address): Always force_reg op2 as well, for
10714 paired single memory accesses.
10715 (rs6000_member_type_forces_blk): Delete.
10716 (rs6000_spe_function_arg): Delete.
10717 (rs6000_expand_unop_builtin): Delete SPE handling.
10718 (rs6000_expand_binop_builtin): Ditto.
10719 (spe_expand_stv_builtin): Delete.
10720 (bdesc_2arg_spe): Delete.
10721 (spe_expand_builtin): Delete.
10722 (spe_expand_predicate_builtin): Delete.
10723 (spe_expand_evsel_builtin): Delete.
10724 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10725 (spe_init_builtins): Delete.
10726 (spe_func_has_64bit_regs_p): Delete.
10727 (savres_routine_name): Delete "info" parameter. Adjust callers.
10728 (rs6000_emit_stack_reset): Ditto.
10729 (rs6000_dwarf_register_span): Delete.
10730 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10731 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10732 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10734 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10736 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10737 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10738 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10739 mulv2sf3, divv2sf3): Delete expanders.
10741 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10743 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10745 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10747 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10748 * config/rs6000/rs6000.c: Ditto.
10750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10752 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10753 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10754 comparison_operator.
10756 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10758 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10759 * config/rs6000/rs6000.opt: Ditto.
10760 * config/rs6000/t-rtems: Ditto.
10762 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10764 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10765 TARGET_E500_SINGLE by 0, simplify.
10766 * config/rs6000/rs6000.c: Ditto.
10767 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10768 (spe_build_register_parallel): Delete.
10769 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10770 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10771 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10772 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10773 (E500_CONVERT): Delete.
10774 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10776 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10778 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10779 * config/rs6000/dfp.md: Ditto.
10780 (negdd2, *negdd2_fpr): Merge.
10781 (absdd2, *absdd2_fpr): Merge.
10782 (negtd2, *negtd2_fpr): Merge.
10783 (abstd2, *abstd2_fpr): Merge.
10784 * config/rs6000/e500.h: Delete file.
10785 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10786 TARGET_FPRS by 1 and simplify.
10787 * config/rs6000/rs6000-c.c: Ditto.
10788 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10789 TARGET_DF_SPE by 0.
10790 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10792 * config/rs6000/rs6000.md: Ditto.
10793 (floatdidf2, *floatdidf2_fpr): Merge.
10794 (move_from_CR_gt_bit): Delete.
10795 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10796 (E500_CR_IOR_COMPARE): Delete.
10797 (All patterns that require !TARGET_FPRS): Delete.
10798 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10800 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10802 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10804 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10806 * graphds.c (add_edge): Intitialize edge's attached data.
10807 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10808 pointer parameter. Call pointed function on each edge during
10809 graph traversing. Skip traversing the edge when the function
10811 (graphds_dfs, graphds_scc): Ditto.
10812 (for_each_edge): New parameter. Pass the new parameter to callback
10814 * graphds.h (skip_edge_callback): New function pointer type.
10815 (graphds_dfs, graphds_scc): New function pointer parameter.
10816 (graphds_edge_callback, for_each_edge): New parameter.
10818 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10820 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10821 out code checking if runtime alias check is possible to below ...
10822 Call the new function.
10823 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10824 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10826 2017-06-07 Marek Polacek <polacek@redhat.com>
10829 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10830 TYPE_OVERFLOW_WRAPS check.
10832 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10834 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10835 if versioning is required.
10836 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10837 peeling with the check for versioning.
10839 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10841 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10842 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10843 Set true to new parameter if new ssa variable is defined.
10844 (vect_gen_vector_loop_niters): Refactor. Set range information
10845 for the new vector loop bound variable.
10846 (vect_do_peeling): Ditto.
10848 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10850 * tree-affine.c (ssa.h): Include header file.
10851 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10852 has wrapping overflow behavior.
10854 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10856 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10858 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10860 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10861 (tree_to_aff_combination): ... here.
10863 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10865 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10866 reg_pressure model function.
10867 (ivopts_global_cost_for_size): Delete.
10868 (determine_set_costs, iv_ca_recount_cost): Call new model function
10869 ivopts_estimate_reg_pressure.
10871 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10873 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10874 expensive than udiv. Remove floating point cases from mod.
10876 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10878 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10879 Increase idiv cost.
10881 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10883 * config/aarch64/aarch64.md
10884 (copysignsf3): Fix mask generation.
10886 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10888 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10890 (class dump_manager): Add register_dumps method.
10891 * dumpfile.c: Include langhooks.h.
10892 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10893 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10894 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10895 (dump_manager::dump_register): Start with 512 entries instead of 32.
10896 (dump_manager::register_dumps): New method.
10897 * toplev.c (general_init): Instead of invoking register_dumps
10898 langhook, invoke register_dumps method on the dump manager.
10899 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10902 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10904 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10905 Say that # requires an associated define_split to exist, and that
10906 the define_split must be suitable for use after register allocation.
10908 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10910 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10911 (compute_outgoing_frequencies): Also initialize zero counts.
10912 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10913 around loops; preserve more of profile when nothing changes.
10915 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10917 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10919 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10920 * config/arm/arm-cpu-cdata.h: Regenerate.
10921 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10922 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10923 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10924 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10925 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10927 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10929 * config/arm/t-rmprofile: Likewise.
10930 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10932 2017-06-06 David S. Miller <davem@davemloft.net>
10935 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10936 blockage if function uses alloca.
10938 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10940 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10941 New "uid" fields to hold pretty-print IDs of group and ref.
10942 Memory references are now identified as <group_id>:<ref_id>
10943 instead of using [random] addresses.
10944 (dump_mem_details): Simplify, no functional change.
10945 (dump_mem_ref): Simplify and make output more concise.
10946 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10947 (find_or_create_group): Initialize group uid.
10948 (record_ref): Initialize ref uid. Improve debug output.
10949 (prune_group_by_reuse, should_issue_prefetch_p,)
10950 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10951 (mark_nontemporal_store, determine_loop_nest_reuse):
10952 Improve debug output.
10954 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10956 * dbgcnt.def (prefetch): New debug counter.
10957 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10958 (schedule_prefetches): Stop issueing prefetches if debug counter
10961 2017-06-06 Tom de Vries <tom@codesourcery.com>
10963 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10964 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10966 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10968 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10969 Use aarch64_reg_or_zero predicate for operand 4.
10970 (aarch64_compare_and_swap<mode> define_insn_and_split):
10971 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10972 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10974 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10976 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10977 (arm_compute_save_core_reg_mask): This.
10978 (thumb1_compute_save_reg_mask): Rename into ...
10979 (thumb1_compute_save_core_reg_mask): This.
10980 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10981 (arm_compute_frame_layout): Likewise.
10983 2017-06-06 Richard Biener <rguenther@suse.de>
10985 PR tree-optimization/80974
10986 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10987 keep or clear leaders SSA info.
10989 2017-06-06 Tom de Vries <tom@codesourcery.com>
10991 * config/nvptx/nvptx.c (split_mode_p): New function.
10992 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10994 2017-06-06 Tom de Vries <tom@codesourcery.com>
10996 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10998 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11001 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11004 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11006 * shrink-wrap.c (handle_simple_exit): Update profile.
11007 (try_shrink_wrapping): Upate profile.
11009 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11011 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11012 (tree_guess_outgoing_edge_probabilities): New.
11013 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11014 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11016 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11018 * ipa-split.c (split_function): Initialize return bb profile.
11020 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11022 * profile.c (compute_branch_probabilities): Also initialize
11023 EXIT_BLOCK profile.
11025 2017-06-06 Richard Biener <rguenther@suse.de>
11027 PR tree-optimization/80928
11028 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11029 (vect_analyze_loop_operations): Properly guard analysis for
11031 (vect_transform_loop): Likewise.
11032 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11033 (vect_model_induction_cost): Do not cost for pure SLP.
11034 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11035 of induction in inner loop vectorization.
11036 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11037 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11038 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11040 (vect_analyze_slp_cost_1): Cost induction.
11041 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11042 (vect_get_slp_vect_defs): Likewise.
11043 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11044 (vect_transform_stmt): Handle SLP reductions.
11045 * tree-vectorizer.h (vectorizable_induction): Adjust.
11047 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11049 * config/rs6000/rs6000.c (make_resolver_func): Update
11050 init_lowered_empty_function call.
11052 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11054 * doc/invoke.texi: Document the -fprofile-abs-path option.
11055 * common.opt (fprofile-abs-path): New option.
11056 * gcov-io.h (gcov_write_filename): Declare.
11057 * gcov-io.c (gcov_write_filename): New function.
11058 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11059 * profile.c (output_location): Likewise.
11061 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11063 * shring-wrap.c: Revert accidental commit.
11065 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11067 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11069 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11071 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11073 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11074 profile in callgraph edge.
11075 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11076 (apply_scale): Likewise.
11077 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11078 Also scale profile when inlining function with zero profile.
11079 (initialize_cfun): Update exit block profile even when it is zero.
11080 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11081 when profile is read.
11083 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11085 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11086 (CLONE_*): New constants to define the processors we can generate
11087 code for with the target_clone attribute.
11088 (rs6000_clone_map): New array to identify which clone processors
11089 the current program is running on.
11090 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11091 target_clone attribute.
11092 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11093 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11094 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11095 (cpu_expand_builtin): Add support for target_clone attribute.
11096 (rs6000_valid_attribute_p): Allow "default" attribute.
11097 (get_decl_name): New debug function to simplify printing the
11098 current function name in debugging statements.
11099 (rs6000_clone_priority): New functions to support the target_clone
11100 attribute, and be able to generate code to switch between ISA 2.05
11101 through ISA 3.0 (power6 through power9).
11102 (rs6000_compare_version_priority): Likewise.
11103 (rs6000_get_function_versions_dispatcher): Likewise.
11104 (make_resolver_func): Likewise.
11105 (add_condition_to_bb): Likewise.
11106 (dispatch_function_versions): Likewise.
11107 (rs6000_generate_version_dispatcher_body): Likewise.
11108 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11109 (fusion_gpr_load_p): Fix a spacing issue.
11110 * doc/extend.texi (Common Function Attributes): Document that the
11111 PowerPC supports the target_clone attribute.
11113 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11115 * config/arm/arm.h: explain F symbol found in description of ARM
11116 register allocation in its legend.
11118 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11120 * config/mips/frame-header-opt.c: Include profile-count.h.
11121 * config/riscv/riscv.c: Include profile-count.h
11123 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11125 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11127 (sm_set_flag_if_changed): Add bbs field.
11128 (execute_sm_if_changed_flag_set): Pass BBS.
11129 (execute_sm): Update.
11131 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11133 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11136 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11138 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11139 (peephole2): New peephole2 to emit the above.
11140 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11142 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11144 * config/aarch64/aarch64.c (define_peephole2 above
11145 *sub_<shift>_<mode>): New peephole.
11147 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11149 * config/i386/i386.c (make_resolver_func): Update.
11150 * Makefile.in: Add profile-count.h and profile-count.o
11151 * auto-profile.c (afdo_indirect_call): Update to new API.
11152 (afdo_set_bb_count): Update.
11153 (afdo_propagate_edge): Update.
11154 (afdo_propagate_circuit): Update.
11155 (afdo_calculate_branch_prob): Update.
11156 (afdo_annotate_cfg): Update.
11157 * basic-block.h: Include profile-count.h
11158 (struct edge_def): Turn count to profile_count.
11159 (struct basic_block_def): Likewie.
11160 (REG_BR_PROB_BASE): Move to profile-count.h
11161 (RDIV): Move to profile-count.h
11162 * bb-reorder.c (max_entry_count): Turn to profile_count.
11163 (find_traces): Update.
11164 (rotate_loop):Update.
11165 (connect_traces):Update.
11166 (sanitize_hot_paths):Update.
11167 * bt-load.c (migrate_btr_defs): Update.
11168 * cfg.c (RDIV): Remove.
11169 (init_flow): Use alloc_block.
11170 (alloc_block): Uninitialize count.
11171 (unchecked_make_edge): Uninitialize count.
11172 (check_bb_profile): Update.
11173 (dump_edge_info): Update.
11174 (dump_bb_info): Update.
11175 (update_bb_profile_for_threading): Update.
11176 (scale_bbs_frequencies_int): Update.
11177 (scale_bbs_frequencies_gcov_type): Update.
11178 (scale_bbs_frequencies_profile_count): New.
11179 * cfg.h (update_bb_profile_for_threading): Update.
11180 (scale_bbs_frequencies_profile_count): Declare.
11181 * cfgbuild.c (compute_outgoing_frequencies): Update.
11182 (find_many_sub_basic_blocks): Update.
11183 * cfgcleanup.c (try_forward_edges): Update.
11184 (try_crossjump_to_edge): Update.
11185 * cfgexpand.c (expand_gimple_tailcall): Update.
11186 (construct_exit_block): Update.
11187 * cfghooks.c (verify_flow_info): Update.
11188 (dump_bb_for_graph): Update.
11189 (split_edge): Update.
11190 (make_forwarder_block): Update.
11191 (duplicate_block): Update.
11192 (account_profile_record): Update.
11193 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11194 (get_estimated_loop_iterations): Update.
11195 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11196 (single_likely_exit): Update.
11197 * cfgloopmanip.c (scale_loop_profile): Update.
11199 (set_zero_probability): Update.
11200 (lv_adjust_loop_entry_edge): Update.
11201 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11202 (purge_dead_edges): Update.
11203 (rtl_account_profile_record): Update.
11204 * cgraph.c (cgraph_node::create): Uninitialize count.
11205 (symbol_table::create_edge): Uninitialize count.
11206 (cgraph_update_edges_for_call_stmt_node): Update.
11207 (cgraph_edge::dump_edge_flags): Update.
11208 (cgraph_node::dump): Update.
11209 (cgraph_edge::maybe_hot_p): Update.
11210 * cgraph.h: Include profile-count.h
11211 (create_clone), create_edge, create_indirect_edge): Update.
11212 (cgraph_node): Turn count to profile_count.
11213 (cgraph_edge0: Likewise.
11214 (make_speculative, clone): Update.
11215 (create_edge): Update.
11216 (init_lowered_empty_function): Update.
11217 * cgraphclones.c (cgraph_edge::clone): Update.
11218 (duplicate_thunk_for_node): Update.
11219 (cgraph_node::create_clone): Update.
11220 * cgraphunit.c (cgraph_node::analyze): Update.
11221 (cgraph_node::expand_thunk): Update.
11222 * final.c (dump_basic_block_info): Update.
11223 * gimple-streamer-in.c (input_bb): Update.
11224 * gimple-streamer-out.c (output_bb): Update.
11225 * graphite.c (print_global_statistics): Update.
11226 (print_graphite_scop_statistics): Update.
11227 * hsa-brig.c: Include basic-block.h.
11228 * hsa-dump.c: Include basic-block.h.
11229 * hsa-gen.c (T sum_slice): Update.
11230 (convert_switch_statements):Update.
11231 * hsa-regalloc.c: Include basic-block.h.
11232 * ipa-chkp.c (chkp_produce_thunks): Update.
11233 * ipa-cp.c (struct caller_statistics): Update.
11234 (init_caller_stats): Update.
11235 (gather_caller_stats): Update.
11236 (ipcp_cloning_candidate_p): Update.
11237 (good_cloning_opportunity_p): Update.
11238 (get_info_about_necessary_edges): Update.
11239 (dump_profile_updates): Update.
11240 (update_profiling_info): Update.
11241 (update_specialized_profile): Update.
11242 (perhaps_add_new_callers): Update.
11243 (decide_about_value): Update.
11244 (ipa_cp_c_finalize): Update.
11245 * ipa-devirt.c (struct odr_type_warn_count): Update.
11246 (struct decl_warn_count): Update.
11247 (struct final_warning_record): Update.
11248 (possible_polymorphic_call_targets): Update.
11249 (ipa_devirt): Update.
11250 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11251 * ipa-icf.c (sem_function::merge): Update.
11252 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11253 * ipa-inline.c (compute_uninlined_call_time): Update.
11254 (compute_inlined_call_time): Update.
11255 (want_inline_small_function_p): Update.
11256 (want_inline_self_recursive_call_p): Update.
11257 (edge_badness): Update.
11258 (lookup_recursive_calls): Update.
11259 (recursive_inlining): Update.
11260 (inline_small_functions): Update.
11261 (dump_overall_stats): Update.
11262 (dump_inline_stats): Update.
11263 * ipa-profile.c (ipa_profile_generate_summary): Update.
11264 (ipa_propagate_frequency): Update.
11265 (ipa_profile): Update.
11266 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11267 * ipa-utils.c (ipa_merge_profiles): Update.
11268 * loop-doloop.c (doloop_modify): Update.
11269 * loop-unroll.c (report_unroll): Update.
11270 (unroll_loop_runtime_iterations): Update.
11271 * lto-cgraph.c (lto_output_edge): Update.
11272 (lto_output_node): Update.
11273 (input_node): Update.
11274 (input_edge): Update.
11275 (merge_profile_summaries): Update.
11276 * lto-streamer-in.c (input_cfg): Update.
11277 * lto-streamer-out.c (output_cfg): Update.
11278 * mcf.c (create_fixup_graph): Update.
11279 (adjust_cfg_counts): Update.
11280 (sum_edge_counts): Update.
11281 * modulo-sched.c (sms_schedule): Update.
11282 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11283 * predict.c (maybe_hot_count_p): Update.
11284 (probably_never_executed): Update.
11285 (dump_prediction): Update.
11286 (combine_predictions_for_bb): Update.
11287 (propagate_freq): Update.
11288 (handle_missing_profiles): Update.
11289 (counts_to_freqs): Update.
11290 (rebuild_frequencies): Update.
11291 (force_edge_cold): Update.
11292 * predict.h: Include profile-count.h
11293 (maybe_hot_count_p, counts_to_freqs): UPdate.
11294 * print-rtl-function.c: Do not include cfg.h
11295 * print-rtl.c: Include basic-block.h
11296 * profile-count.c: New file.
11297 * profile-count.h: New file.
11298 * profile.c (is_edge_inconsistent): Update.
11299 (correct_negative_edge_counts): Update.
11300 (is_inconsistent): Update.
11301 (set_bb_counts): Update.
11302 (read_profile_edge_counts): Update.
11303 (compute_frequency_overlap): Update.
11304 (compute_branch_probabilities): Update; Initialize and deinitialize
11306 (branch_prob): Update.
11307 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11308 (edge_gcov_count): New.
11309 (bb_gcov_count): New.
11310 * shrink-wrap.c (try_shrink_wrapping): Update.
11311 * tracer.c (better_p): Update.
11312 * trans-mem.c (expand_transaction): Update.
11313 (ipa_tm_insert_irr_call): Update.
11314 (ipa_tm_insert_gettmclone_call): Update.
11315 * tree-call-cdce.c: Update.
11316 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11317 (gimple_duplicate_sese_tail): Update.
11318 (gimple_account_profile_record): Update.
11319 (execute_fixup_cfg): Update.
11320 * tree-inline.c (copy_bb): Update.
11321 (copy_edges_for_bb): Update.
11322 (initialize_cfun): Update.
11323 (freqs_to_counts): Update.
11324 (copy_cfg_body): Update.
11325 (expand_call_inline): Update.
11326 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11327 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11328 (try_unroll_loop_completely): Update.
11329 (try_peel_loop): Update.
11330 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11331 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11332 * tree-ssa-loop-split.c (connect_loops): Update.
11333 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11334 * tree-ssa-reassoc.c (branch_fixup): Update.
11335 * tree-ssa-tail-merge.c (replace_block_by): Update.
11336 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11337 (compute_path_counts): Update.
11338 (update_profile): Update.
11339 (recompute_probabilities): Update.
11340 (update_joiner_offpath_counts): Update.
11341 (estimated_freqs_path): Update.
11342 (freqs_to_counts_path): Update.
11343 (clear_counts_path): Update.
11344 (ssa_fix_duplicate_block_edges): Update.
11345 (duplicate_thread_path): Update.
11346 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11347 (struct switch_conv_info): Update.
11348 * tree-tailcall.c (decrease_profile): Update.
11349 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11350 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11351 * value-prof.c (check_counter): Update.
11352 (gimple_divmod_fixed_value): Update.
11353 (gimple_mod_pow2): Update.
11354 (gimple_mod_subtract): Update.
11355 (gimple_ic_transform): Update.
11356 (gimple_stringop_fixed_value): Update.
11357 * value-prof.h (gimple_ic): Update.
11359 2017-06-02 Carl Love <cel@us.ibm.com>
11361 * config/rs6000/rs6000-c: Add support for built-in functions
11362 vector double vec_doublee (vector signed int);
11363 vector double vec_doublee (vector unsigned int);
11364 vector double vec_doublee (vector float);
11365 vector double vec_doubleh (vector signed int);
11366 vector double vec_doubleh (vector unsigned int);
11367 vector double vec_doubleh (vector float);
11368 vector double vec_doublel (vector signed int);
11369 vector double vec_doublel (vector unsigned int);
11370 vector double vec_doublel (vector float);
11371 vector double vec_doubleo (vector signed int);
11372 vector double vec_doubleo (vector unsigned int);
11373 vector double vec_doubleo (vector float);.
11374 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11375 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11377 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11378 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11379 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11381 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11382 vec_doublel, vec_doubleh.
11383 * doc/extend.texi: Update the built-in documentation file for the
11384 new built-in functions.
11386 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11389 * ipa-inline-analysis.c (free_growth_caches): Set
11390 edge_removal_hook_holder to NULL after removing it.
11392 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11394 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11395 comparision with zero.
11397 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11398 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11399 for early expansion of vec_min and vec_max builtins.
11400 (builtin_function_type): Add min/max unsigned variants to those
11401 identified as having unsigned arguments.
11403 2017-06-02 Olivier Hainque <hainque@adacore.com>
11405 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11407 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11409 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11410 Use VALL_F16 iterator rather than VALL.
11412 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11414 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11415 Emit CBNZ inside loop when doing a strong exchange and comparing
11416 against zero. Generate the CC flags after the loop.
11418 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11420 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11421 (dl_section_ref): New.
11422 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11423 On AIX, append an expression to subtract the size of the
11424 section length to dl_section_ref.
11426 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11428 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11429 for early expansion of vector absolute builtins.
11431 2017-06-02 Richard Biener <rguenther@suse.de>
11433 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11434 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11436 2017-06-02 Richard Biener <rguenther@suse.de>
11438 PR tree-optimization/80948
11439 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11440 stmt order as well.
11442 2017-06-02 Richard Biener <rguenther@suse.de>
11444 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11446 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11447 for inductions as relevant.
11449 2017-06-02 Richard Biener <rguenther@suse.de>
11451 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11452 (vectorizable_induction): ... this. Remove dead code.
11454 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11456 * builtins. (expand_builtin_alloca): Remove second parameter and
11457 infer its value from the first parameter instead.
11458 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11460 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11462 PR rtl-optimization/80903
11463 * loop-doloop.c (add_test): Unshare sequence.
11465 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11467 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11469 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11471 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11473 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11474 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11475 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11476 (xlogue_layout::get_stub_rtx): Make static.
11477 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11478 (xlogue_layout::compute_stub_managed_regs): Rename to...
11479 (xlogue_layout::count_stub_managed_regs): ...this.
11480 (xlogue_layout::is_stub_managed_reg): New function.
11481 (xlogue_layout::m_stub_names): Rename to...
11482 (xlogue_layout::s_stub_names): ...this, make static.
11483 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11484 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11485 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11486 xlogue_layout::s_stub_names): Instantiate statics.
11487 (stub_managed_regs): Remove.
11488 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11489 (disable_call_ms2sysv_xlogues): Rename to...
11490 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11491 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11493 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11494 change after reload_completed.
11495 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11497 (ix86_expand_prologue): Likewise.
11498 (ix86_expand_epilogue): Likewise.
11499 (ix86_expand_split_stack_prologue): Likewise.
11500 (ix86_compute_frame_layout): Remove frame parameter ...
11501 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11502 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11504 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11505 (ix86_frame): Move from here ...
11506 * config/i386/i386.h (ix86_frame): ... to here.
11507 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11508 complete ix86_frame data structure instead. Remove some_ld_name.
11510 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11512 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11513 symbols that hold a DECL_VALUE_EXPR.
11515 2017-06-01 Martin Jambor <mjambor@suse.cz>
11517 PR tree-optimization/80898
11518 * tree-sra.c (process_subtree_disqualification): Removed.
11519 (disqualify_candidate): Do not acll
11520 process_subtree_disqualification.
11521 (subtree_mark_written_and_enqueue): New function.
11522 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11523 RHS has been disqualified and re-queue LHS if necessary. Apart
11524 from that, ignore disqualified RHS.
11526 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11528 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11529 address fetch for z10 or later.
11531 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11533 * config/arc/arc.md (tst_movb): Add guard when splitting.
11535 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11537 * config/arc/arc.c (arc_can_eliminate): Test against
11538 arc_frame_pointer_needed.
11540 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11542 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11543 to prevent store reordering.
11544 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11545 (type): Add block type.
11546 (stack_tie): Define special instruction to be used in
11549 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11551 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11552 constraint. It is not valid for the pattern.
11553 (noncommutative_binary_comparison): Likewise.
11555 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11557 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11560 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11562 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11563 be used by the reg-alloc.
11565 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11567 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11568 reg-alloc when having mul64 or mul32x16 instructions.
11569 (mulsidi3): Likewise.
11570 (umulsidi3): Likewise.
11571 (mulsi32x16): New pattern.
11572 (mulsi64): Likewise.
11573 (mulsidi64): Likewise.
11574 (umulsidi64): Likewise.
11575 (MUL32x16_REG): Define.
11576 (mul64_600): Use MUL32x16_REG.
11577 (mac64_600): Likewise.
11578 (umul64_600): Likewise.
11579 (umac64_600): Likewise.
11581 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11583 * config/arc/arc.md (mulsi3_700): Make it commutative.
11585 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11587 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11589 (*sign_extendsidi2_insn): Likewise for movstosw.
11591 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11593 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11594 the type of the input discriminant value. Convert the
11595 discriminant value of signedness vary.
11597 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11599 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11600 Add to -Wall section.
11602 2017-06-01 Richard Biener <rguenther@suse.de>
11604 PR middle-end/66313
11605 * fold-const.c (fold_plusminus_mult_expr): If the factored
11606 factor may be zero use a wrapping type for the inner operation.
11607 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11608 and handle moved defs.
11609 (process_assignment): Properly guard the unary op case. Return a
11610 tri-state indicating that moving the stmt before the call may allow
11611 to continue. Pass through to_move.
11612 (find_tail_calls): Handle moving unrelated defs before
11615 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11618 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11619 splitter result in the canonical way.
11621 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11623 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11624 also for 32bit target. Update insn attributes.
11625 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11627 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11629 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11630 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11631 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11632 (_mm_maskz_min_ss): New intrinsics.
11634 2017-05-31 Martin Liska <mliska@suse.cz>
11636 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11637 Change comment style to one we normally use.
11638 (vectorizable_reduction): Likewise.
11639 (vectorizable_induction): Likewise.
11640 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11641 (vectorizable_call): Likewise.
11642 (vectorizable_simd_clone_call): Likewise.
11643 (vectorizable_conversion): Likewise.
11644 (vectorizable_assignment): Likewise.
11645 (vectorizable_shift): Likewise.
11646 (vectorizable_operation): Likewise.
11647 (vectorizable_store): Likewise.
11648 (vectorizable_load): Likewise.
11649 * tree-vectorizer.h: Likewise.
11651 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11653 * passes.c (emergency_dump_function): New.
11654 * tree-pass.h (emergency_dump_function): Declare.
11655 * plugin.c (plugins_internal_error_function): Remove.
11656 * plugin.h (plugins_internal_error_function): Remove declaration.
11657 * toplev.c (internal_error_function): New static function. Use it...
11658 (general_init): ...here.
11660 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11662 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11663 (arc_rtx_costs): Add costs for new patterns.
11664 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11665 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11667 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11669 * tree-ssa-strlen.c (get_next_strinfo): New function.
11670 (get_stridx_plus_constant): Use it.
11671 (zero_length_string): Likewise.
11672 (adjust_related_strinfos): Likewise.
11673 (adjust_last_stmt): Likewise.
11675 2017-05-31 Richard Biener <rguenther@suse.de>
11678 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11679 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11681 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11683 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11684 loop_vinfo argument and use of dependence distance vectors.
11685 Check instead whether the two references differ only in their
11686 initial value and assume that they have the same alignment if the
11687 difference is a multiple of the vector alignment.
11688 (vect_analyze_data_refs_alignment): Update call accordingly.
11690 2017-05-31 Martin Liska <mliska@suse.cz>
11693 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11695 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11697 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11698 (create_intersect_range_checks): Move from ...
11699 * tree-data-ref.c (create_intersect_range_checks_index)
11700 (create_intersect_range_checks): ... to here.
11701 (create_runtime_alias_checks): New function factored from ...
11702 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11703 here. Call above function.
11704 * tree-data-ref.h (create_runtime_alias_checks): New function.
11706 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11708 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11709 segment length for dr_b and compute it in wide_int.
11711 2017-05-31 Richard Biener <rguenther@suse.de>
11713 PR tree-optimization/80906
11714 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11715 and pass through iv_map.
11716 (copy_bb_and_scalar_dependences): Adjust.
11717 (translate_pending_phi_nodes): Likewise.
11718 (copy_loop_close_phi_args): Handle code-generating IVs instead
11721 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11723 * diagnostic-color.c (color_dict): Add "type-diff".
11724 (parse_gcc_colors): Update comment.
11725 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11726 -fdiagnostics-show-template-tree and -fno-elide-type.
11727 (GCC_COLORS): Add type-diff to example.
11729 (-fdiagnostics-show-template-tree): New.
11730 (-fno-elide-type): New.
11731 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11732 the pp_format_decoder callback. Call any m_format_postprocessor's
11734 (pretty_printer::pretty_printer): Initialize
11735 m_format_postprocessor.
11736 (pretty_printer::~pretty_printer): Delete any
11737 m_format_postprocessor.
11738 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11739 (class format_postprocessor): New class.
11740 (struct pretty_printer::format_decoder): Document the new parameters.
11741 (struct pretty_printer::m_format_postprocessor): New field.
11742 * tree-diagnostic.c (default_tree_printer): Update for new
11743 bool and const char ** params.
11744 * tree-diagnostic.h (default_tree_printer): Likewise.
11746 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11748 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11749 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11750 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11751 rs6000_gen_cell_microcode code.
11752 (rs6000_final_prescan_insn): Delete.
11753 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11754 "warn-cell-microcode" entries.
11755 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11756 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11757 throughout. Change cc_reg_not_micro_cr0_operand to
11758 cc_reg_not_cr0_operand throughout.
11759 (*extendhi<mode>2_noload): Delete.
11760 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11761 (mwarn-cell-microcode): Delete.
11762 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11763 -mgen-cell-microcode and -mwarn-cell-microcode.
11765 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11768 * config/i386/constraints.md (Yd): New constraint.
11770 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11771 and (?Yd, r) alternatives. Update insn attributes.
11772 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11773 and (?*Yd, r) alternatives. Update insn attributes.
11774 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11776 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11778 * gimplify.c (gimplify_modify_expr): Don't create a
11779 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11782 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11784 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11786 2017-05-30 Richard Biener <rguenther@suse.de>
11788 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11789 and reduc_def fields.
11790 (STMT_VINFO_REDUC_TYPE): New define.
11791 (STMT_VINFO_REDUC_DEF): Likewise.
11792 (vect_force_simple_reduction): Adjust prototype.
11793 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11794 (vect_is_simple_reduction): Remove check_reduction argument.
11795 (vect_force_simple_reduction): Adjust and set
11796 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11797 (vectorizable_reduction): Do not re-do reduction analysis
11798 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11799 * tree-parloops.c (gather_scalar_reductions): Adjust.
11801 2017-05-30 Richard Biener <rguenther@suse.de>
11803 PR middle-end/80901
11804 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11807 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11809 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11810 Introduce unknown_misalignment parameter and remove vf.
11811 (vect_peeling_hash_get_lowest_cost):
11812 Pass unknown_misalignment parameter.
11813 (vect_enhance_data_refs_alignment):
11814 Fix unsupportable data ref treatment.
11816 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11818 * tree-vect-data-refs.c (vect_get_data_access_cost):
11819 Workaround for SLP handling.
11820 (vect_enhance_data_refs_alignment):
11821 Compute costs for doing no peeling at all, compare to the best
11822 peeling costs so far and avoid peeling if cheaper.
11824 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11826 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11827 Return peeling info and set costs to zero for unlimited cost
11829 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11830 with unknown misalignment. Compute and costs for unknown
11831 misalignment, compare them to the costs for known misalignment
11832 and choose the cheapest for peeling.
11834 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11836 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11837 (vect_get_peeling_costs_all_drs): Create function.
11838 (vect_peeling_hash_get_lowest_cost):
11839 Use vect_get_peeling_costs_all_drs.
11840 (vect_peeling_supportable): Create function.
11841 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11843 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11845 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11846 DR_HAS_NEGATIVE_STEP.
11847 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11848 (vect_enhance_data_refs_alignment): Use.
11849 (vect_duplicate_ssa_name_ptr_info): Use.
11850 * tree-vectorizer.h (dr_misalignment): Use.
11851 (known_alignment_for_access_p): Use.
11853 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11856 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11858 (has_section_name): New function.
11860 2017-05-30 Martin Liska <mliska@suse.cz>
11863 * auto-profile.c (get_function_decl_from_block): Fix
11866 2017-05-30 Richard Biener <rguenther@suse.de>
11868 PR middle-end/80876
11869 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11871 2017-05-30 Martin Liska <mliska@suse.cz>
11873 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11874 * dumpfile.h (struct dump_file_info): Remove ctors.
11876 2017-05-30 Martin Liska <mliska@suse.cz>
11878 * predict.def: Fix GNU coding style.
11880 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11882 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11883 Mark 'to' argument with ATTRIBUTE_UNUSED.
11885 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11887 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11888 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11889 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11892 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11894 * doc/install.texi (Options specification): Restore entry of
11895 --enable-sjlj-exceptions.
11897 2017-05-27 Michael Eager <eager@eagercon.com>
11900 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11902 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11904 * config/microblaze/microblaze.h
11905 (FIXED_REGISTERS): Update in macro.
11906 (CALL_USED_REGISTERS): Update in macro.
11908 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11910 * doc/install.texi: Add links to macOS binary distributions.
11912 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11916 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11918 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11920 2017-05-26 Martin Liska <mliska@suse.cz>
11922 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11924 2017-05-26 Martin Liska <mliska@suse.cz>
11926 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11927 always leading ';; '.
11928 (dump_bb_info): Likewise.
11929 (brief_dump_cfg): Likewise.
11930 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11931 * dumpfile.c: Remove usage of TDF_VERBOSE.
11932 * dumpfile.h (enum dump_kind): Likewise.
11933 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11934 * print-tree.c (debug_verbose): Remove.
11935 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11936 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11937 * tree-diagnostic.c (default_tree_printer): Replace
11938 TDF_DIAGNOSTIC with TDF_SLIM.
11940 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11942 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11943 in parameter loop, rather than loop_vinfo.
11944 (create_intersect_range_checks): Ditto.
11945 (vect_create_cond_for_alias_checks): Update call to above functions.
11947 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11949 PR tree-optimization/80815
11950 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11951 for merging runtime alias checks. Handle negative DR_STEPs.
11953 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11955 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11957 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11958 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11959 out code pruning runtime alias checks.
11960 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11961 factored out from above.
11962 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11964 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11966 (prune_runtime_alias_test_list): New decalaration.
11968 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11970 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11971 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11972 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11973 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11974 (operator==, comp_dr_with_seg_len_pair): Ditto.
11975 (vect_prune_runtime_alias_test_list): Ditto.
11977 2017-05-26 Martin Liska <mliska@suse.cz>
11980 * params.def: Bound partial-inlining-entry-probability param.
11982 2017-05-26 Marek Polacek <polacek@redhat.com>
11985 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11988 2017-05-26 Richard Biener <rguenther@suse.de>
11990 PR tree-optimization/80842
11991 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11994 2017-05-26 Richard Biener <rguenther@suse.de>
11996 PR tree-optimization/80844
11997 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11999 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12001 * doc/md.texi (Machine Constraints): Update x86 family
12002 machine constraints section to match 'config/i386/constraints.md'.
12004 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12006 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12008 2017-05-25 Nathan Sidwell <nathan@acm.org>
12010 * doc/invoke.texi (--enable-languages): Update documentation.
12012 2017-05-25 Martin Liska <mliska@suse.cz>
12014 * dumpfile.c: Add TDF_FOLDING.
12015 * dumpfile.h (enum dump_kind): Likewise.
12016 * genmatch.c (dt_simplify::gen_1): Use it.
12018 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12020 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12022 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12024 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12025 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12027 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12029 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12030 * match.pd (X == C): Rewrite it here.
12031 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12032 with_certain_nonzero_bits2): New predicates.
12033 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12035 2017-05-24 Nathan Sidwell <nathan@acm.org>
12037 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12040 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12043 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12045 * config/powerpcspe: New port. Files are copied from the rs6000
12046 port, with "rs6000" in filenames replaced by "powerpcspe".
12048 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12050 PR rtl-optimization/80754
12051 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12053 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12055 * config/sparc/sparc.md (length): Return the correct value for -mflat
12056 sibcalls to match output_sibcall.
12058 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12062 * config/rs6000/rs6000.c (struct machine_function): Add new field
12064 (rs6000_get_separate_components): Init that field, use it.
12065 (rs6000_components_for_bb): Use the field.
12067 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12069 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12071 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12073 PR middle-end/80823
12074 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12076 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12079 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12080 address against address_operand predicate.
12081 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12083 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12085 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12086 parameters passed indirectly.
12088 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12090 * config/i386/i386.md (*movdi_internal): Remove SSE4
12091 alternative 18 (?r, *v). Update insn attributes.
12092 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12093 Update insn attributes.
12094 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12095 Update insn attributes.
12096 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12097 alternative 1 (r, v). Remove isa attribute.
12098 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12099 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12100 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12102 2017-05-23 Tom de Vries <tom@codesourcery.com>
12104 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12107 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12109 * cgraphunit.c (symbol_table::process_new_functions): Update.
12110 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12111 (inline_generate_summary): Rename to ...
12112 (ipa_fn_summary_generate): ... this one.
12113 (inline_read_summary): Rename to ...
12114 (ipa_fn_summary_read): ... this one.
12115 (inline_write_summary): Rename to ...
12116 (ipa_fn_summary_write): ... this one.
12117 (inline_free_summary): Rename to ...
12118 (ipa_free_fn_summary): ... this one.
12119 (pass_data_local_fn_summary, pass_local_fn_summary,
12120 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12121 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12122 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12123 make_pass_ipa_fn_summary): New.
12124 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12125 inline_write_summary, inline_free_summary): Remove.
12126 (ipa_free_fn_summary) : New.
12127 * ipa-inline.c (ipa_inline): Update.
12128 (pass_ipa_inline): Do not generate summaries.
12129 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12131 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12132 and add pass_ipa_fn_summary.
12133 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12135 (make_pass_inline_parameters): Remove.
12137 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12139 * omp-low.c (struct omp_context): Remove "default_kind" member.
12142 * omp-offload.c (execute_oacc_device_lower): Remove the
12143 parallelism dimensions function attributes for unparallelized
12144 OpenACC kernels constructs.
12146 2017-05-23 Martin Liska <mliska@suse.cz>
12148 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12150 (cgraph_edge::make_speculative): Likewise.
12151 (cgraph_edge::resolve_speculation): Likewise.
12152 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12153 (cgraph_node::dump): Likewise.
12154 * cgraph.h: Likewise.
12155 * cgraphunit.c (analyze_functions): Likewise.
12156 (symbol_table::compile): Likewise.
12157 * ipa-cp.c (print_all_lattices): Likewise.
12158 (determine_versionability): Likewise.
12159 (initialize_node_lattices): Likewise.
12160 (ipcp_verify_propagated_values): Likewise.
12161 (estimate_local_effects): Likewise.
12162 (update_profiling_info): Likewise.
12163 (create_specialized_node): Likewise.
12164 (perhaps_add_new_callers): Likewise.
12165 (decide_about_value): Likewise.
12166 (decide_whether_version_node): Likewise.
12167 (identify_dead_nodes): Likewise.
12168 (ipcp_store_bits_results): Likewise.
12169 * ipa-devirt.c (dump_targets): Likewise.
12170 (ipa_devirt): Likewise.
12171 * ipa-icf.c (sem_item::dump): Likewise.
12172 (sem_function::equals): Likewise.
12173 (sem_variable::equals): Likewise.
12174 (sem_item_optimizer::read_section): Likewise.
12175 (sem_item_optimizer::execute): Likewise.
12176 (congruence_class::dump): Likewise.
12177 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12178 (dump_inline_summary): Likewise.
12179 (estimate_node_size_and_time): Likewise.
12180 (inline_analyze_function): Likewise.
12181 * ipa-inline-transform.c (inline_call): Likewise.
12182 * ipa-inline.c (report_inline_failed_reason): Likewise.
12183 (want_early_inline_function_p): Likewise.
12184 (edge_badness): Likewise.
12185 (update_edge_key): Likewise.
12186 (inline_small_functions): Likewise.
12187 * ipa-profile.c (ipa_profile): Likewise.
12188 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12189 (ipa_make_edge_direct_to_target): Likewise.
12190 (remove_described_reference): Likewise.
12191 (ipa_impossible_devirt_target): Likewise.
12192 (propagate_controlled_uses): Likewise.
12193 (ipa_print_node_params): Likewise.
12194 (ipcp_transform_function): Likewise.
12195 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12196 (propagate_pure_const): Likewise.
12197 * ipa-reference.c (generate_summary): Likewise.
12198 (read_write_all_from_decl): Likewise.
12199 (propagate): Likewise.
12200 (ipa_reference_read_optimization_summary): Likewise.
12201 * ipa-utils.c (ipa_merge_profiles): Likewise.
12202 * ipa.c (walk_polymorphic_call_targets): Likewise.
12203 (symbol_table::remove_unreachable_nodes): Likewise.
12204 (ipa_single_use): Likewise.
12205 * passes.c (execute_todo): Likewise.
12206 * predict.c (drop_profile): Likewise.
12207 * symtab.c (symtab_node::get_dump_name): New function.
12208 (symtab_node::dump_name): Likewise.
12209 (symtab_node::dump_asm_name): Likewise.
12210 (symtab_node::dump_references): Likewise.
12211 (symtab_node::dump_referring): Likewise.
12212 (symtab_node::dump_base): Likewise.
12213 (symtab_node::debug_symtab): Likewise.
12214 * tree-sra.c (convert_callers_for_node): Likewise.
12215 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12216 * value-prof.c (init_node_map): Likewise.
12218 2017-05-23 Martin Liska <mliska@suse.cz>
12220 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12221 and symtab_node::debug_symtab to symbol_table::debug.
12222 * cgraphunit.c (analyze_functions): Use the renamed function.
12223 (symbol_table::compile): Likewise.
12224 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12225 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12226 * passes.c (execute_todo): Likewise.
12227 * symtab.c (symbol_table::dump): New function.
12228 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12230 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12232 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12233 that nonconst implies exec.
12235 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12237 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12238 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12239 (inline_edge_summary_vec): Turn into ...
12240 (ipa_call_summaries): ... this one.
12241 (redirect_to_unreachable, edge_set_predicate,
12242 evaluate_properties_for_edge, inline_summary_alloc,
12243 reset_ipa_call_summary, reset_inline_summary,
12244 inline_summary_t::duplicate): Update.
12245 (inline_edge_duplication_hook): Turn to ...
12246 (ipa_call_summary_t::duplicate): ... this one.
12247 (inline_edge_removal_hook): Turn to ...
12248 (ipa_call_summary_t::remove): ... this one.
12249 (dump_inline_edge_summary): Turn to ...
12250 (dump_ipa_call_summary): ... this one.
12251 (estimate_function_body_sizes): Update.
12252 (inline_update_callee_summaries): Update.
12253 (remap_edge_change_prob): Update.
12254 (remap_edge_summaries): Update.
12255 (inline_merge_summary): Update.
12256 (do_estimate_edge_time): Update.
12257 (inline_generate_summary): Update.
12258 (inline_read_section): Update.
12259 (inline_read_summary): Update.
12260 (inline_free_summary): Update.
12261 * ipa-inline.c (can_inline_edge_p): Update.
12262 (compute_inlined_call_time): Update.
12263 (want_inline_small_function_p): Update.
12264 (edge_badness): Update.
12265 (early_inliner): Update.
12266 * ipa-inline.h (inline_edge_summary): Turn to ...
12267 (ipa_call_summary): ... this one.
12268 (ipa_call_summary_t): New class.
12269 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12270 (ipa_call_summaries): New.
12271 (inline_edge_summary): Remove.
12272 (estimate_edge_growth): Update.
12273 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12274 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12275 * ipa-split.c (execute_split_functions): Update.
12276 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12278 2017-05-23 Tom de Vries <tom@codesourcery.com>
12280 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12281 attributes): Document rdrand effective target.
12283 2017-05-23 Tom de Vries <tom@codesourcery.com>
12285 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12286 attributes): Sort alphabetically.
12288 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12290 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12292 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12295 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12296 V2DF/V2DI splat into two separate patterns, one that handles
12297 registers, and the other that only handles memory. Drop support
12298 for splatting from a GPR on ISA 2.07 and then splitting the
12299 splat into direct move and splat.
12300 (vsx_splat_<mode>_reg): Likewise.
12301 (vsx_splat_<mode>_mem): Likewise.
12303 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12305 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12307 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12309 PR middle-end/80809
12310 * omp-low.c (finish_taskreg_remap): New function.
12311 (finish_taskreg_scan): If unit size of ctx->record_type
12312 is non-constant, unshare the size expression and replace
12313 decls in it with possible outer var refs.
12315 PR middle-end/80809
12316 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12317 GOVD_SHARED rather than GOVD_PRIVATE with it.
12318 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12319 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12321 PR middle-end/80853
12322 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12323 as last argument to build_outer_var_ref for pointer bases of array
12324 section reductions.
12326 2017-05-19 Martin Sebor <msebor@redhat.com>
12328 * print-tree.c (print_node): Print DECL_READ_P flag.
12330 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12332 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12333 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12334 * cgraph.c: Likewise.
12335 * cgraphunit.c: Likewise.
12336 * gengtype.c: Likewise.
12337 * ipa-cp.c: Likewise.
12338 * ipa-devirt.c: Likewise.
12339 * ipa-icf.c: Likewise.
12340 * ipa-predicate.c: Likewise.
12341 * ipa-profile.c: Likewise.
12342 * ipa-prop.c: Likewise.
12343 * ipa-split.c: Likewise.
12345 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12346 edge_predicate_pool, dump_inline_hints,
12347 inline_summary::account_size_time, redirect_to_unreachable,
12348 edge_set_predicate, set_hint_predicate,
12349 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12350 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12351 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12352 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12353 ipa_call_summary_t::remove, initialize_growth_caches,
12354 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12355 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12356 mark_modified, unmodified_parm_1, unmodified_parm,
12357 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12358 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12359 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12360 will_be_nonconstant_predicate, record_modified_bb_info,
12361 get_minimal_bb, record_modified, param_change_prob,
12362 phi_result_unknown_predicate, predicate_for_phi_result,
12363 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12364 estimate_function_body_sizes, compute_inline_parameters,
12365 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12366 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12367 inline_update_callee_summaries, remap_edge_change_prob,
12368 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12369 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12370 inline_analyze_function, inline_summary_t::insert,
12371 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12372 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12373 inline_free_summary): Move to ipa-fnsummary.h
12374 (predicate_t): Remove.
12375 * ipa-fnsummary.c: New file.
12376 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12377 (enum inline_hints_vals, inline_hints, agg_position_info,
12378 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12379 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12380 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12381 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12382 inline_read_summary, inline_write_summary, inline_free_summary,
12383 inline_analyze_function, initialize_inline_failed,
12384 inline_merge_summary, inline_update_overall_summary,
12385 compute_inline_parameters): Move to ipa-fnsummary.h
12386 * ipa-fnsummary.h: New file.
12387 * ipa-inline-transform.h: Include ipa-inline.h.
12388 * ipa-inline.c: LIkewise.
12390 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12392 * ipa-inline.c (edge_badness): Use inlined_time instead of
12393 inline_summaries->get.
12395 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12397 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12399 2017-05-22 Nathan Sidwell <nathan@acm.org>
12401 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12402 (fdump-lang): Document 'raw' option.
12403 * dumpfile.h (TDI_tu): Delete.
12404 * dumpfile.c (dump_files): Remove translation-unit.
12405 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12407 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12409 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12410 command option from $(AWK) call.
12411 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12412 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12413 [FORMAT]: Remove handling of variable.
12414 * config/avr/t-multilib: Regenerate.
12416 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12418 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12420 (dump_inline_summary): Do not print self_time.
12421 (estimate_function_body_sizes): Do not set self_time.
12422 (compute_inline_parameters): Likewise.
12423 (inline_read_section, inline_write_summary): Do not stream self_time.
12424 * ipa-inline.h (inline_summary): Drop self_time.
12426 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12428 * ipa-inline-analysis.c (account_size_time): Rename to ...
12429 (inline_summary::account_size_time): ... this one.
12430 (reset_ipa_call_summary): Turn to ...
12431 (ipa_call_summary::reset): ... this one.
12432 (reset_inline_summary): Turn to ...
12433 (inline_summary::reset): ... this one.
12434 (inline_summary_t::remove): Update.
12435 (inline_summary_t::duplicate): Update.
12436 (ipa_call_summary_t::remove): Update.
12437 (dump_inline_summary): Update.
12438 (estimate_function_body_sizes): Update.
12439 (compute_inline_parameters): Update.
12440 (estimate_node_size_and_time): Update.
12441 (inline_merge_summary): Update.
12442 (inline_update_overall_summary): Update.
12443 (inline_read_section): Update.
12444 (inline_write_summary): Update.
12445 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12446 add account_size_time and reset member functions.
12447 (ipa_call_summary): Add reset function.
12448 * ipa-predicate.h (predicate::operator &): Constify.
12450 2017-05-22 Richard Biener <rguenther@suse.de>
12452 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12454 2017-05-19 Jason Merrill <jason@redhat.com>
12456 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12458 2017-05-19 Marek Polacek <polacek@redhat.com>
12461 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12462 TYPE_OVERFLOW_WRAPS checks.
12464 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12466 * tree-core.h (enum omp_clause_default_kind): Add
12467 "OMP_CLAUSE_DEFAULT_PRESENT".
12468 * tree-pretty-print.c (dump_omp_clause): Handle it.
12469 * gimplify.c (enum gimplify_omp_var_data): Add
12470 "GOVD_MAP_FORCE_PRESENT".
12471 (gimplify_adjust_omp_clauses_1): Map it to
12472 "GOMP_MAP_FORCE_PRESENT".
12473 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12475 * gimplify.c (oacc_default_clause): Clarify.
12477 2017-05-19 Nathan Sidwell <nathan@acm.org>
12479 LANG_HOOK_REGISTER_DUMPS
12480 * toplev.c (general_init): Call register dump lang hook.
12481 * doc/invoke.texi: Document -fdump-lang option family.
12482 * dumpfile.c (dump_files): Remove class dump here.
12483 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12484 * dumpfile.h (tree_dump_index): Remove TDI_class.
12485 * langhooks-def.h (lhd_register_dumps): Declare.
12486 (LANG_HOOKS_REGISTER_DUMPS): Define.
12487 (LANG_HOOKS_INITIALIZER): Add it.
12488 * langhooks.c (lhd_register_dumps): Define.
12489 * langhooks.h (struct lang_hooks): Add register_dumps.
12491 2017-05-19 Nathan Sidwell <nathan@acm.org>
12493 * context.h (context::set_passes): New.
12494 * context.c (context::context): Do not create pass manager.
12495 * toplev.c (general_init): Create pass manager here.
12497 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12499 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12500 use this splitter if two add or or instructions would also work for
12501 the constant we want to generate.
12503 2017-05-19 Richard Biener <rguenther@suse.de>
12506 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12507 predicate evaluation.
12509 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12511 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12513 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12516 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12518 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12519 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12520 (inline_edge_summary_vec): Turn into ...
12521 (ipa_call_summaries): ... this one.
12522 (redirect_to_unreachable, edge_set_predicate,
12523 evaluate_properties_for_edge, inline_summary_alloc,
12524 reset_ipa_call_summary, reset_inline_summary,
12525 inline_summary_t::duplicate): Update.
12526 (inline_edge_duplication_hook): Turn to ...
12527 (ipa_call_summary_t::duplicate): ... this one.
12528 (inline_edge_removal_hook): Turn to ...
12529 (ipa_call_summary_t::remove): ... this one.
12530 (dump_inline_edge_summary): Turn to ...
12531 (dump_ipa_call_summary): ... this one.
12532 (estimate_function_body_sizes): Update.
12533 (inline_update_callee_summaries): Update.
12534 (remap_edge_change_prob): Update.
12535 (remap_edge_summaries): Update.
12536 (inline_merge_summary): Update.
12537 (do_estimate_edge_time): Update.
12538 (inline_generate_summary): Update.
12539 (inline_read_section): Update.
12540 (inline_read_summary): Update.
12541 (inline_free_summary): Update.
12542 * ipa-inline.c (can_inline_edge_p): Update.
12543 (compute_inlined_call_time): Update.
12544 (want_inline_small_function_p): Update.
12545 (edge_badness): Update.
12546 (early_inliner): Update.
12547 * ipa-inline.h (inline_edge_summary): Turn to ...
12548 (ipa_call_summary): ... this one.
12549 (ipa_call_summary_t): New class.
12550 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12551 (ipa_call_summaries): New.
12552 (inline_edge_summary): Remove.
12553 (estimate_edge_growth): Update.
12554 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12555 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12556 * ipa-split.c (execute_split_functions): Update.
12557 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12559 2017-05-19 Richard Biener <rguenther@suse.de>
12561 PR middle-end/80764
12562 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12564 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12566 * config/rs6000/rs6000.c (struct machine_function): Add field
12567 fpr_is_wrapped_separately.
12568 (rs6000_get_separate_components): Use 64 components. Handle the
12569 new FPR components.
12570 (rs6000_components_for_bb): Handle the FPR components.
12571 (rs6000_emit_prologue_components): Handle the FPR components.
12572 (rs6000_emit_epilogue_components): Handle the FPR components.
12573 (rs6000_set_handled_components): Handle the FPR components.
12574 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12575 that are already separately shrink-wrapped.
12576 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12577 that are already separately shrink-wrapped.
12579 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12582 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12585 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12586 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12587 cases where the register allocator uses a move and an offsettable
12588 memory operation to/from a FPR register on ISA 2.06/2.07.
12589 (define_peephole2 for Altivec d-form store): Likewise.
12591 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12594 * config/i386/mmx.md (*mov<mode>_internal): Enable
12595 alternatives 11, 12, 13 and 14 also for 32bit targets.
12596 Remove alternatives 15, 16, 17 and 18.
12597 * config/i386/sse.md (vec_concatv2di): Change
12598 alternative (!x, *y) to (x, ?!*Yn).
12600 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12602 * dumpfile.h (enum dump_kind): Remove stray comma.
12604 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12606 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12607 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12608 predicate::num_conditions
12609 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12610 (CHANGED): turn into predicate::changed.
12611 (agg_position_info): Move to ipa-predicate.h
12612 (add_condition, predicate::add_clause, predicate::operator &=,
12613 predicate::or_with, predicate::evaluate, predicate::probability,
12614 dump_condition, dump_clause, predicate::dump,
12615 predicate::remap_after_duplication, predicate::remap_after_inlining,
12616 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12617 (evaluate_conditions_for_known_args): Update.
12618 (set_cond_stmt_execution_predicate): Update.
12619 * ipa-inline.h: Include ipa-predicate.h
12620 (condition, inline_param_summary, conditions, agg_position_info,
12621 predicate): Move to ipa-predicate.h
12622 * ipa-predicate.c: New file.
12623 * ipa-predicate.h: New file.
12625 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12627 * final.c (leaf_function_p): Check we are not in a sequence.
12629 2017-05-18 Martin Liska <mliska@suse.cz>
12631 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12632 * dumpfile.c (dump_register): Use new enum dump_kind.
12633 (get_dump_file_name): Likewise.
12634 (dump_enable_all): Likewise.
12635 (dump_switch_p_1): Likewise.
12636 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12637 * dumpfile.h (enum dump_kind): New enum type.
12638 (struct dump_file_info): Create constructor and
12639 format fields and comments.
12640 * passes.c (pass_manager::register_one_dump_file):
12642 * statistics.c (statistics_early_init): Likewise.
12643 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12644 TDF_TREE with TDF_SLIM.
12645 (gather_memory_references_ref): Likewise.
12647 2017-05-18 Martin Liska <mliska@suse.cz>
12649 * vec.h (struct vnull): Use it.
12651 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12653 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12654 (true_predicate, false_predicate, true_predicate_p,
12655 false_predicate_p): Remove.
12656 (single_cond_predicate, not_inlined_predicate): Turn to member function
12658 (add_condition): Update.
12659 (add_clause): Turn to...
12660 (predicate::add_clause): ... this one; update; allow passing NULL
12662 (and_predicates): Turn to ...
12663 (predicate::operator &=): ... this one.
12664 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12665 (or_predicates): Turn to ...
12666 (predicate::or_with): ... this one.
12667 (evaluate_predicate): Turn to ...
12668 (predicate::evaluate): ... this one.
12669 (predicate_probability): Turn to ...
12670 (predicate::probability): ... this one.
12671 (dump_condition): Update.
12672 (dump_predicate): Turn to ...
12673 (predicate::dump): ... this one.
12674 (account_size_time): Update.
12675 (edge_set_predicate): Update.
12676 (set_hint_predicate): UPdate.
12677 (evaluate_conditions_for_known_args): Update.
12678 (evaluate_properties_for_edge): Update.
12679 (remap_predicate_after_duplication): Turn to...
12680 (predicate::remap_after_duplication): ... this one.
12681 (remap_hint_predicate_after_duplication): Update.
12682 (inline_summary_t::duplicate): UPdate.
12683 (dump_inline_edge_summary): Update.
12684 (dump_inline_summary): Update.
12685 (set_cond_stmt_execution_predicate): Update.
12686 (set_switch_stmt_execution_predicate): Update.
12687 (compute_bb_predicates): Update.
12688 (will_be_nonconstant_expr_predicate): Update.
12689 (will_be_nonconstant_predicate): Update.
12690 (phi_result_unknown_predicate): Update.
12691 (predicate_for_phi_result): Update.
12692 (array_index_predicate): Update.
12693 (estimate_function_body_sizes): Update.
12694 (estimate_node_size_and_time): Update.
12695 (estimate_ipcp_clone_size_and_time): Update.
12696 (remap_predicate): Rename to ...
12697 (predicate::remap_after_inlining): ... this one.
12698 (remap_hint_predicate): Update.
12699 (inline_merge_summary): Update.
12700 (inline_update_overall_summary): Update.
12701 (estimate_size_after_inlining): Update.
12702 (read_predicate): Rename to ...
12703 (predicate::stream_in): ... this one.
12704 (read_inline_edge_summary): Update.
12705 (write_predicate): Rename to ...
12706 (predicate::stream_out): ... this one.
12707 (write_inline_edge_summary): Update.
12708 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12709 (clause_t): Turn to uint32_t
12710 (predicate): Turn to class; implement constructor and operators
12712 (size_time_entry): Update.
12713 (inline_summary): Update.
12714 (inline_edge_summary): Update.
12716 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12718 * fold-const.c (fold_binary_loc): Move transformation...
12719 * match.pd (C - X CMP X): ... here.
12721 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12723 * config/sparc/sparc.c (sparc_option_override): Set function
12724 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12725 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12727 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12729 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12731 2017-05-18 Marek Polacek <polacek@redhat.com>
12734 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12735 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12736 gimple_assign_load_p.
12738 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12740 PR middle-end/80692
12741 * real.c (do_compare): Give decimal_do_compare preference over
12742 comparing just the signs.
12744 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12746 * doc/md.texi (Canonicalization of Instructions): Describe the
12747 canonical form of instructions that inherently set a condition
12750 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12752 PR middle-end/80775
12753 * tree-cfg.c: Move deletion of unreachable case statements to after
12754 the merging of consecutive case labels.
12756 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12758 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12759 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12760 restoring of callee-saved registers.
12762 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12764 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12765 * config/visium/visium.c (single_set_and_flags): Likewise.
12766 * config/visium/visium.md (Substitutions): Likewise.
12768 2017-05-17 Martin Liska <mliska@suse.cz>
12770 * cfg.c: Introduce dump_flags_t type and
12771 use it instead of int type.
12773 * cfghooks.c: Likewise.
12774 * cfghooks.h (struct cfg_hooks): Likewise.
12775 * cfgrtl.c: Likewise.
12776 * cfgrtl.h: Likewise.
12777 * cgraph.c (cgraph_node::get_body): Likewise.
12778 * coretypes.h: Likewise.
12779 * domwalk.c: Likewise.
12780 * domwalk.h: Likewise.
12781 * dumpfile.c (struct dump_option_value_info): Likewise.
12782 (dump_enable_all): Likewise.
12783 (dump_switch_p_1): Likewise.
12784 (opt_info_switch_p): Likewise.
12785 * dumpfile.h (enum tree_dump_index): Likewise.
12786 (struct dump_file_info): Likewise.
12787 * genemit.c: Likewise.
12788 * generic-match-head.c: Likewise.
12789 * gengtype.c (open_base_files): Likewise.
12790 * gimple-pretty-print.c: Likewise.
12791 * gimple-pretty-print.h: Likewise.
12792 * graph.c (print_graph_cfg): Likewise.
12793 * graphite-scop-detection.c (dot_all_sese): Likewise.
12794 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12795 * loop-unroll.c (report_unroll): Likewise.
12796 * passes.c (pass_manager::register_one_dump_file): Likewise.
12797 * print-tree.c: Likewise.
12798 * statistics.c: Likewise.
12799 * tree-cfg.c: Likewise.
12800 * tree-cfg.h: Likewise.
12801 * tree-dfa.c: Likewise.
12802 * tree-dfa.h: Likewise.
12803 * tree-dump.c (dump_function): Likewise.
12804 * tree-dump.h (struct dump_info): Likewise.
12805 * tree-pretty-print.c: Likewise.
12806 * tree-pretty-print.h: Likewise.
12807 * tree-ssa-live.c: Likewise.
12808 * tree-ssa-live.h: Likewise.
12809 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12810 * tree-vect-loop.c: Likewise.
12811 * tree-vect-slp.c: Likewise.
12813 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12814 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12816 PR tree-optimization/80457
12817 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12818 of all arguments to a statement as scalar_to_vec operations.
12819 (vectorizable_call): Adjust call to vect_model_simple_cost for
12821 (vectorizable_conversion): Likewise.
12822 (vectorizable_assignment): Likewise.
12823 (vectorizable_shift): Likewise.
12824 (vectorizable_operation): Likewise.
12825 (vectorizable_comparison): Likewise.
12826 (vect_is_simple_cond): Record the def types for operands.
12827 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12828 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12829 for statement argument count.
12831 2017-05-16 Carl Love <cel@us.ibm.com>
12833 * config/rs6000/rs6000-c: Add support for built-in functions
12834 vector unsigned long long vec_bperm (vector unsigned long long,
12835 vector unsigned char)
12836 vector signed long long vec_mule (vector signed int,
12838 vector unsigned long long vec_mule (vector unsigned int,
12839 vector unsigned int)
12840 vector signed long long vec_mulo (vector signed int,
12842 vector unsigned long long vec_mulo (vector unsigned int,
12843 vector unsigned int)
12844 vector signed char vec_sldw (vector signed char,
12845 vector signed char,
12847 vector unsigned char vec_sldw (vector unsigned char,
12848 vector unsigned char,
12850 vector signed short vec_sldw (vector signed short,
12851 vector signed short,
12853 vector unsigned short vec_sldw (vector unsigned short,
12854 vector unsigned short,
12856 vector signed int vec_sldw (vector signed int,
12859 vector unsigned int vec_sldw (vector unsigned int,
12860 vector unsigned int,
12862 vector signed long long vec_sldw (vector signed long long,
12863 vector signed long long,
12865 vector unsigned long long vec_sldw (vector unsigned long long,
12866 vector unsigned long long,
12868 * config/rs6000/rs6000-c: Add support for built-in functions
12869 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12870 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12871 * doc/extend.texi: Update the built-in documentation for the
12872 new built-in functions.
12874 2017-05-16 Marek Polacek <polacek@redhat.com>
12878 * tree.c (save_expr): Don't fold the expression.
12880 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12882 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12883 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12884 and (?*y,m). Update insn attributes.
12886 2017-05-16 Martin Liska <mliska@suse.cz>
12888 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12889 flags argument of print_gimple_stmt, print_gimple_expr,
12890 print_generic_stmt and print_generic_expr.
12891 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12892 * coretypes.h: Likewise.
12893 * except.c (dump_eh_tree): Likewise.
12894 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12895 * gimple-pretty-print.h: Likewise.
12896 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12897 (backprop::push_to_worklist): Likewise.
12898 (backprop::pop_from_worklist): Likewise.
12899 (backprop::process_use): Likewise.
12900 (backprop::intersect_uses): Likewise.
12901 (note_replacement): Likewise.
12902 * gimple-ssa-store-merging.c
12903 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12904 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12905 (pass_store_merging::execute): Likewise.
12906 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12907 (ssa_base_cand_dump_callback): Likewise.
12908 (dump_incr_vec): Likewise.
12909 (replace_refs): Likewise.
12910 (replace_mult_candidate): Likewise.
12911 (create_add_on_incoming_edge): Likewise.
12912 (create_phi_basis): Likewise.
12913 (insert_initializers): Likewise.
12914 (all_phi_incrs_profitable): Likewise.
12915 (introduce_cast_before_cand): Likewise.
12916 (replace_one_candidate): Likewise.
12917 * gimplify.c (gimplify_expr): Likewise.
12918 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12919 (set_rename): Likewise.
12920 (rename_uses): Likewise.
12921 (copy_loop_phi_nodes): Likewise.
12922 (add_close_phis_to_merge_points): Likewise.
12923 (copy_loop_close_phi_args): Likewise.
12924 (copy_cond_phi_args): Likewise.
12925 (graphite_copy_stmts_from_block): Likewise.
12926 (translate_pending_phi_nodes): Likewise.
12927 * graphite-poly.c (print_pdr): Likewise.
12928 (dump_gbb_cases): Likewise.
12929 (dump_gbb_conditions): Likewise.
12930 (print_scop_params): Likewise.
12931 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12932 (build_cross_bb_scalars_use): Likewise.
12933 (gather_bbs::before_dom_children): Likewise.
12934 * hsa-dump.c (dump_hsa_immed): Likewise.
12935 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12936 (get_replacement_map): Likewise.
12937 * ipa-inline-analysis.c (dump_condition): Likewise.
12938 (estimate_function_body_sizes): Likewise.
12939 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12940 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12941 * ipa-prop.c (ipa_dump_param): Likewise.
12942 (ipa_print_node_jump_functions_for_edge): Likewise.
12943 (ipa_modify_call_arguments): Likewise.
12944 (ipa_modify_expr): Likewise.
12945 (ipa_dump_param_adjustments): Likewise.
12946 (ipa_dump_agg_replacement_values): Likewise.
12947 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12948 * ipa-pure-const.c (check_stmt): Likewise.
12949 (pass_nothrow::execute): Likewise.
12950 * ipa-split.c (execute_split_functions): Likewise.
12951 * omp-offload.c (dump_oacc_loop_part): Likewise.
12952 (dump_oacc_loop): Likewise.
12953 * trans-mem.c (tm_log_emit): Likewise.
12954 (tm_memopt_accumulate_memops): Likewise.
12955 (dump_tm_memopt_set): Likewise.
12956 (dump_tm_memopt_transform): Likewise.
12957 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12958 (print_loop): Likewise.
12959 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12960 (chkp_gather_checks_info): Likewise.
12961 (chkp_get_check_result): Likewise.
12962 (chkp_remove_check_if_pass): Likewise.
12963 (chkp_use_outer_bounds_if_possible): Likewise.
12964 (chkp_reduce_bounds_lifetime): Likewise.
12965 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12966 (chkp_mark_completed_bounds): Likewise.
12967 (chkp_register_incomplete_bounds): Likewise.
12968 (chkp_mark_invalid_bounds): Likewise.
12969 (chkp_maybe_copy_and_register_bounds): Likewise.
12970 (chkp_build_returned_bound): Likewise.
12971 (chkp_get_bound_for_parm): Likewise.
12972 (chkp_build_bndldx): Likewise.
12973 (chkp_get_bounds_by_definition): Likewise.
12974 (chkp_generate_extern_var_bounds): Likewise.
12975 (chkp_get_bounds_for_decl_addr): Likewise.
12976 * tree-chrec.c (chrec_apply): Likewise.
12977 * tree-data-ref.c (dump_data_reference): Likewise.
12978 (dump_subscript): Likewise.
12979 (dump_data_dependence_relation): Likewise.
12980 (analyze_overlapping_iterations): Likewise.
12981 * tree-inline.c (expand_call_inline): Likewise.
12982 (tree_function_versioning): Likewise.
12983 * tree-into-ssa.c (dump_defs_stack): Likewise.
12984 (dump_currdefs): Likewise.
12985 (dump_names_replaced_by): Likewise.
12986 (dump_update_ssa): Likewise.
12987 (update_ssa): Likewise.
12988 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12989 * tree-parloops.c (build_new_reduction): Likewise.
12990 (try_create_reduction_list): Likewise.
12991 (ref_conflicts_with_region): Likewise.
12992 (oacc_entry_exit_ok_1): Likewise.
12993 (oacc_entry_exit_single_gang): Likewise.
12994 * tree-pretty-print.h: Likewise.
12995 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12996 (get_scalar_evolution): Likewise.
12997 (add_to_evolution): Likewise.
12998 (get_loop_exit_condition): Likewise.
12999 (analyze_evolution_in_loop): Likewise.
13000 (analyze_initial_condition): Likewise.
13001 (analyze_scalar_evolution): Likewise.
13002 (instantiate_scev): Likewise.
13003 (number_of_latch_executions): Likewise.
13004 (gather_chrec_stats): Likewise.
13005 (final_value_replacement_loop): Likewise.
13006 (scev_const_prop): Likewise.
13007 * tree-sra.c (dump_access): Likewise.
13008 (disqualify_candidate): Likewise.
13009 (create_access): Likewise.
13010 (reject): Likewise.
13011 (maybe_add_sra_candidate): Likewise.
13012 (create_access_replacement): Likewise.
13013 (analyze_access_subtree): Likewise.
13014 (analyze_all_variable_accesses): Likewise.
13015 (sra_modify_assign): Likewise.
13016 (initialize_constant_pool_replacements): Likewise.
13017 (find_param_candidates): Likewise.
13018 (decide_one_param_reduction): Likewise.
13019 (replace_removed_params_ssa_names): Likewise.
13020 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13021 * tree-ssa-copy.c (dump_copy_of): Likewise.
13022 (copy_prop_visit_cond_stmt): Likewise.
13023 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13024 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13025 (record_equivalences_from_stmt): Likewise.
13026 * tree-ssa-dse.c (compute_trims): Likewise.
13027 (delete_dead_call): Likewise.
13028 (delete_dead_assignment): Likewise.
13029 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13030 (forward_propagate_into_cond): Likewise.
13031 (pass_forwprop::execute): Likewise.
13032 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13033 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13035 (move_computations_worker): Likewise.
13036 (execute_sm): Likewise.
13037 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13038 (remove_exits_and_undefined_stmts): Likewise.
13039 (remove_redundant_iv_tests): Likewise.
13040 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13041 (adjust_iv_update_pos): Likewise.
13042 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13043 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13044 (value_replacement): Likewise.
13045 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13046 * tree-ssa-pre.c (print_pre_expr): Likewise.
13047 (get_representative_for): Likewise.
13048 (create_expression_by_pieces): Likewise.
13049 (insert_into_preds_of_block): Likewise.
13050 (eliminate_insert): Likewise.
13051 (eliminate_dom_walker::before_dom_children): Likewise.
13052 (eliminate): Likewise.
13053 (remove_dead_inserted_code): Likewise.
13054 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13055 * tree-ssa-reassoc.c (get_rank): Likewise.
13056 (eliminate_duplicate_pair): Likewise.
13057 (eliminate_plus_minus_pair): Likewise.
13058 (eliminate_not_pairs): Likewise.
13059 (undistribute_ops_list): Likewise.
13060 (eliminate_redundant_comparison): Likewise.
13061 (update_range_test): Likewise.
13062 (optimize_range_tests_var_bound): Likewise.
13063 (optimize_vec_cond_expr): Likewise.
13064 (rewrite_expr_tree): Likewise.
13065 (rewrite_expr_tree_parallel): Likewise.
13066 (linearize_expr): Likewise.
13067 (break_up_subtract): Likewise.
13068 (linearize_expr_tree): Likewise.
13069 (attempt_builtin_powi): Likewise.
13070 (attempt_builtin_copysign): Likewise.
13071 (transform_stmt_to_copy): Likewise.
13072 (transform_stmt_to_multiply): Likewise.
13073 (dump_ops_vector): Likewise.
13074 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13075 (print_scc): Likewise.
13076 (set_ssa_val_to): Likewise.
13077 (visit_reference_op_store): Likewise.
13078 (visit_use): Likewise.
13079 (sccvn_dom_walker::before_dom_children): Likewise.
13080 (run_scc_vn): Likewise.
13081 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13083 (expr_hash_elt::print): Likewise.
13084 (const_and_copies::pop_to_marker): Likewise.
13085 (const_and_copies::record_const_or_copy_raw): Likewise.
13086 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13087 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13088 (dump_predicates): Likewise.
13089 (find_uninit_use): Likewise.
13090 (warn_uninitialized_phi): Likewise.
13091 (pass_late_warn_uninitialized::execute): Likewise.
13092 * tree-ssa.c (verify_vssa): Likewise.
13093 (verify_ssa): Likewise.
13094 (maybe_optimize_var): Likewise.
13095 * tree-vrp.c (dump_value_range): Likewise.
13096 (dump_all_value_ranges): Likewise.
13097 (dump_asserts_for): Likewise.
13098 (register_edge_assert_for_2): Likewise.
13099 (vrp_visit_cond_stmt): Likewise.
13100 (vrp_visit_switch_stmt): Likewise.
13101 (vrp_visit_stmt): Likewise.
13102 (vrp_visit_phi_node): Likewise.
13103 (simplify_cond_using_ranges_1): Likewise.
13104 (fold_predicate_in): Likewise.
13105 (evrp_dom_walker::before_dom_children): Likewise.
13106 (evrp_dom_walker::push_value_range): Likewise.
13107 (evrp_dom_walker::pop_value_range): Likewise.
13108 (execute_early_vrp): Likewise.
13110 2017-05-16 Richard Biener <rguenther@suse.de>
13112 * dwarf2out.c (loc_list_from_tree_1): Do not create
13113 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13115 2017-05-16 Richard Biener <rguenther@suse.de>
13117 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13119 (note_variable_value_in_expr): If we resolved the decl ref
13120 do not push to the stack.
13122 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13124 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13125 operations in fast-math mode.
13126 (vaddq_f16): Likewise.
13127 (vmul_f16): Likewise.
13128 (vmulq_f16): Likewise.
13129 (vsub_f16): Likewise.
13130 (vsubq_f16): Likewise.
13131 * config/arm/neon.md (add<mode>3): New.
13133 (fma:<VH:mode>3): New. Also remove outdated comment.
13136 2017-05-16 Martin Liska <mliska@suse.cz>
13140 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13141 (odr_types_equivalent_p): Likewise.
13143 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13145 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13147 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13150 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13151 non-interunit SSE move alternatives with '?'.
13152 (zero-extendsidi peephole2): New peephole to skip intermediate
13153 general register in SSE zero-extend sequence.
13155 2017-05-15 Jeff Law <law@redhat.com>
13157 * reorg.c (relax_delay_slots): Create a new variable to hold
13158 the temporary target rather than clobbering TARGET_LABEL.
13160 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13161 missing argument to extract_bit_field call.
13162 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13164 2017-05-15 Martin Liska <mliska@suse.cz>
13167 * gcc.c (process_command): Do not allow empty argument of -o option.
13169 2017-05-15 Renlin Li <renlin.li@arm.com>
13171 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13172 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13173 * config/aarch64/constraints.md (Usf): Add long call check.
13174 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13175 (call_value): Likewise.
13176 (sibcall): Likewise.
13177 (sibcall_value): Likewise.
13179 (call_value_insn): New.
13180 (sibcall_insn): Update rtx pattern.
13181 (sibcall_value_insn): Likewise.
13182 (call_internal): Remove.
13183 (call_value_internal): Likewise.
13184 (sibcall_internal): Likewise.
13185 (sibcall_value_internal): Likewise.
13186 (call_reg): Likewise.
13187 (call_symbol): Likewise.
13188 (call_value_reg): Likewise.
13189 (call_value_symbol): Likewise.
13191 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13194 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13196 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13198 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13199 compatible with CCGOCmode and with CCZmode.
13201 2017-05-14 Martin Sebor <msebor@redhat.com>
13203 PR middle-end/77671
13204 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13205 (gimple_fold_builtin_snprintf): Same.
13206 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13207 (gimple_fold_builtin_snprintf): Same.
13208 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13209 of character types.
13210 (is_call_safe): New function.
13211 (try_substitute_return_value): Call it.
13212 (try_simplify_call): New function.
13213 (pass_sprintf_length::handle_gimple_call): Call it.
13215 2017-05-14 Martin Sebor <msebor@redhat.com>
13217 PR middle-end/80669
13218 * builtins.c (expand_builtin_stpncpy): Simplify.
13220 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13222 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13223 * config/i386/i386.h
13224 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13225 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13226 (struct machine_function): Add new members call_ms2sysv,
13227 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13228 (struct machine_frame_state): New fields sp_realigned and
13229 sp_realigned_offset.
13230 * config/i386/i386.c
13231 (enum xlogue_stub): New enum.
13232 (enum xlogue_stub_sets): New enum.
13233 (class xlogue_layout): New class.
13234 (struct ix86_frame): New fields stack_realign_allocate_offset,
13235 stack_realign_offset and outlined_save_offset. Modify comments to
13236 detail stack layout when using out-of-line stubs.
13237 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13238 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13239 -mcall-ms2sysv-xlogues.
13240 (stub_managed_regs): New static variable.
13241 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13242 registers managed by out-of-line stub.
13243 (disable_call_ms2sysv_xlogues): New function.
13244 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13245 m->call_ms2sysv when appropriate and compute frame layout for
13247 (sp_valid_at, fp_valid_at): New inline functions.
13248 (choose_basereg): New function.
13249 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13251 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13252 Use align parameter of choose_baseaddr to generated aligned SSE movs
13254 (pro_epilogue_adjust_stack): Modify to track
13255 machine_frame_state::sp_realigned.
13256 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13257 (ix86_nsaved_sseregs): Likewise.
13258 (ix86_emit_save_regs): Likewise.
13259 (ix86_emit_save_regs_using_mov): Likewise.
13260 (ix86_emit_save_sse_regs_using_mov): Likewise.
13261 (get_scratch_register_on_entry): Likewise.
13262 (gen_frame_set): New function.
13263 (gen_frame_load): Likewise.
13264 (gen_frame_store): Likewise.
13265 (emit_outlined_ms2sysv_save): Likewise.
13266 (emit_outlined_ms2sysv_restore): Likewise.
13267 (ix86_expand_prologue): Modify stack re-alignment code and call
13268 emit_outlined_ms2sysv_save when appropriate.
13269 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13270 parameter rtx_insn *insn, which allows the function to be used to only
13271 generate the notes.
13272 (ix86_expand_epilogue): Modify validity checks of frame and stack
13273 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13274 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13275 * config/i386/predicates.md
13276 (save_multiple): New predicate.
13277 (restore_multiple): Likewise.
13278 * config/i386/sse.md
13279 (save_multiple<mode>): New pattern.
13280 (save_multiple_realign<mode>): Likewise.
13281 (restore_multiple<mode>): Likewise.
13282 (restore_multiple_and_return<mode>): Likewise.
13283 (restore_multiple_leave_return<mode>): Likewise.
13284 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13286 2017-05-14 Julia Koval <julia.koval@intel.com>
13288 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13289 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13290 (__builtin_ia32_xsetbv): New builtins.
13291 * config/i386/i386.c (ix86_expand_special_args_builtin):
13293 (ix86_expand_builtin): Special expand for new intrinsics.
13294 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13295 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13296 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13298 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13300 * cfganal.c (inverted_post_order_compute): Change argument type
13302 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13303 * df-core.c (rest_of_handle_df_initialize): Adjust.
13304 (rest_of_handle_df_finish): Likewise.
13305 (df_analyze_1): Likewise.
13306 (df_analyze): Likewise.
13307 (loop_inverted_post_order_compute): Change argument to be a vec *.
13308 (df_analyze_loop): Adjust.
13309 (df_get_n_blocks): Likewise.
13310 (df_get_postorder): Likewise.
13311 * df.h (struct df_d): Change field to be a vec.
13312 * lcm.c (compute_laterin): Adjust.
13313 (compute_available): Likewise.
13314 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13315 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13316 * tree-ssa-pre.c (compute_antic): Likewise.
13318 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13320 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13321 (depth_first_search::depth_first_search): Change structure init
13322 function to this constructor.
13323 (depth_first_search::add_bb): Rename function to this member.
13324 (depth_first_search::execute): Likewise.
13325 (flow_dfs_compute_reverse_finish): Adjust.
13327 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13329 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13330 (longest_simple_path): Likewise.
13331 * shrink-wrap.c (spread_components): Likewise.
13332 (disqualify_problematic_components): Likewise.
13333 (emit_common_heads_for_components): Likewise.
13334 (emit_common_tails_for_components): Likewise.
13335 (insert_prologue_epilogue_for_components): Likewise.
13337 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13339 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13342 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13344 * df-core.c (df_set_blocks): Start using auto_bitmap.
13345 (df_compact_blocks): Likewise.
13346 * df-problems.c (df_rd_confluence_n): Likewise.
13347 * df-scan.c (df_insn_rescan_all): Likewise.
13348 (df_process_deferred_rescans): Likewise.
13349 (df_update_entry_block_defs): Likewise.
13350 (df_update_exit_block_uses): Likewise.
13351 (df_entry_block_bitmap_verify): Likewise.
13352 (df_exit_block_bitmap_verify): Likewise.
13353 (df_scan_verify): Likewise.
13354 * lra-constraints.c (lra_constraints): Likewise.
13355 (undo_optional_reloads): Likewise.
13356 (lra_undo_inheritance): Likewise.
13357 * lra-remat.c (calculate_gen_cands): Likewise.
13358 (do_remat): Likewise.
13359 * lra-spills.c (assign_spill_hard_regs): Likewise.
13360 (spill_pseudos): Likewise.
13361 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13362 (bitmap_set_subtract_values): Likewise.
13364 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13366 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13367 management with auto_bitmap.
13368 (fix_inter_tick): Likewise.
13369 (fix_recovery_deps): Likewise.
13370 * ira.c (add_store_equivs): Likewise.
13371 (find_moveable_pseudos): Likewise.
13372 (split_live_ranges_for_shrink_wrap): Likewise.
13373 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13374 (rtx_reuse_manager::seen_def_p): Likewise.
13375 (rtx_reuse_manager::set_seen_def): Likewise.
13376 * print-rtl.h (class rtx_reuse_manager): Likewise.
13378 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13380 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13382 (migrate_btr_def): Likewise.
13383 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13384 * df-core.c (loop_post_order_compute): Likewise.
13385 (loop_inverted_post_order_compute): Likewise.
13386 * hsa-common.h: Likewise.
13387 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13388 * init-regs.c (initialize_uninitialized_regs): Likewise.
13389 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13390 (inline_small_functions): Likewise.
13391 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13392 * ira.c (combine_and_move_insns): Likewise.
13393 (build_insn_chain): Likewise.
13394 * loop-invariant.c (find_invariants): Likewise.
13395 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13396 * predict.c (tree_predict_by_opcode): Likewise.
13397 (predict_paths_leading_to): Likewise.
13398 (predict_paths_leading_to_edge): Likewise.
13399 (estimate_loops_at_level): Likewise.
13400 (estimate_loops): Likewise.
13401 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13402 (spread_components): Likewise.
13403 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13404 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13405 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13406 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13407 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13408 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13409 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13410 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13411 (mark_threaded_blocks): Likewise.
13412 (thread_through_all_blocks): Likewise.
13413 * tree-ssa.c (verify_ssa): Likewise.
13414 (execute_update_addresses_taken): Likewise.
13415 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13417 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13419 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13421 (post_order_compute): Likewise.
13422 (inverted_post_order_compute): Likewise.
13423 (pre_and_rev_post_order_compute_fn): Likewise.
13425 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13427 * genrecog.c (int_set::int_set): Explicitly construct our
13428 auto_vec base class.
13429 * vec.h (auto_vec::auto_vec): New constructor.
13431 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13433 * bitmap.h (class auto_bitmap): New constructor taking
13434 bitmap_obstack * argument.
13436 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13438 * bitmap.h (class auto_bitmap): Change type of m_bits to
13439 bitmap_head, and adjust ctor / dtor and member operators.
13441 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13443 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13444 when returned register mode doesn't match original mode.
13446 2017-05-12 Jeff Law <law@redhat.com>
13447 Jakub Jelinek <jakub@redhat.com>
13449 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13450 we look for cc setter after the compare-elim changes.
13451 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13452 within the vector to match what compare-elim now expects.
13453 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13454 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13456 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13457 after the compare-elim changes.
13458 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13459 the vector to match what compare-elim now expects.
13460 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13461 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13462 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13463 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13464 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13466 * config/visium/visium.c (single_set_and_flags): Fix where
13467 we look for cc setter after the compare-elim changes.
13468 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13469 with the vector to match what compare-elim now expects.
13470 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13471 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13472 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13473 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13474 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13475 (neg<mode>2_insn_set_overflow): Likewise.
13477 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13479 PR middle-end/79794
13480 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13481 maybe_expand_insn call, set ops[0].target. If still set after call,
13482 set alt_rtl. Add extra arg to recursive calls.
13483 (extract_bit_field): Add alt_rtl argument. Pass to
13485 * expmed.h (extract_bit_field): Fix prototype.
13486 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13487 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13488 to extract_bit_field_calls.
13489 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13490 Pass alt_rtl to extract_bit_field calls.
13491 * calls.c (store_unaligned_arguments_into_psuedos)
13492 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13493 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13495 * optabs.h (struct expand_operand): Add target bitfield.
13497 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13499 * compare-elim.c (try_eliminate_compare): Canonicalize
13500 operation with embedded compare to
13501 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13502 (set (reg) (operation)].
13504 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13506 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13509 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13510 cost of adding a carry flag for ADC instruction.
13511 [case MINUS]: Ignore the cost of subtracting a carry flag
13512 for SBB instruction.
13514 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13516 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13518 * config/rs6000/bmiintrin.h: New file.
13519 * config/rs6000/bmi2intrin.h: New file.
13520 * config/rs6000/x86intrin.h: New file.
13522 2017-05-12 Jeff Law <law@redhat.com>
13524 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13527 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13529 PR middle-end/80707
13530 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13532 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13534 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13535 early expansion of vector divide builtins.
13536 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13537 builtins identified as having unsigned arguments.
13539 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13541 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13542 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13543 expansion of vector logical operations (and, andc, or, xor,
13546 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13548 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13549 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13551 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13553 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13554 early GIMPLE expansion of vector multiplies.
13556 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13558 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13559 TARGET_HAVE_MOVT conditional.
13560 (movt splitter): Likewise.
13562 2017-05-12 Richard Biener <rguenther@suse.de>
13564 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13565 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13566 Fold all stmts not inplace.
13568 2017-05-12 Richard Biener <rguenther@suse.de>
13570 PR tree-optimization/80713
13571 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13572 inserted_exprs bit for not removed stmts.
13574 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13576 PR middle-end/69921
13577 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13578 parallelized" attribute for parallelized OpenACC kernels.
13579 * omp-offload.c (execute_oacc_device_lower): Use it.
13581 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13582 Set "oacc kernels" attribute.
13583 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13584 parameter. Adjust all users.
13585 (oacc_fn_attrib_kernels_p): Remove function.
13586 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13587 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13588 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13589 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13590 assert "oacc kernels" attribute is set.
13592 2017-05-11 Carl Love <cel@us.ibm.com>
13594 * config/rs6000/rs6000-c: Add support for built-in functions
13595 vector unsigned char vec_popcnt (vector signed char)
13596 vector unsigned char vec_popcnt (vector unsigned char)
13597 vector unsigned short vec_popcnt (vector signed short)
13598 vector unsigned short vec_popcnt (vector unsigned short)
13599 vector unsigned int vec_popcnt (vector signed int)
13600 vector unsigned int vec_popcnt (vector unsigned int)
13601 vector unsigned long long vec_popcnt (vector signed long long)
13602 vector unsigned long long vec_popcnt (vector unsigned long long)
13603 vector signed long long vec_slo (vector signed long long,
13604 vector signed char)
13605 vector signed long long vec_slo (vector signed long long,
13606 vector unsigned char)
13607 vector unsigned long long vec_slo (vector unsigned long long,
13608 vector signed char)
13609 vector unsigned long long vec_slo (vector unsigned long long,
13610 vector unsigned char)
13611 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13612 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13613 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13614 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13615 * doc/extend.texi: Update the built-in documentation file for the
13616 new built-in functions.
13618 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13620 * attribs.h (sorted_attr_string): Move machine independent
13621 functions for target clone support from the i386 port to common
13622 code. Rename ix86_function_versions to common_function_versions.
13623 Rename make_name to make_unique_name.
13624 (common_function_versions): Likewise.
13625 (make_unique_name): Likewise.
13626 (make_dispatcher_decl): Likewise.
13627 (is_function_default_version): Likewise.
13628 * attribs.c (attr_strcmp): Likewise.
13629 (sorted_attr_string): Likewise.
13630 (common_function_versions): Likewise.
13631 (make_unique_name): Likewise.
13632 (make_dispatcher_decl): Likewise.
13633 (is_function_default_version): Likewise.
13634 * config/i386/i386.c (attr_strcmp): Likewise.
13635 (sorted_attr_string): Likewise.
13636 (ix86_function_versions): Likewise.
13637 (make_name): Likewise.
13638 (make_dispatcher_decl): Likewise.
13639 (is_function_default_version): Likewise.
13640 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13642 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13645 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13646 Account for direct move costs for vec_construct of integer
13649 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13652 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13653 (UNSPEC_STX_ATOMIC): Ditto.
13654 (loaddi_via_sse): New insn.
13655 (storedi_via_sse): Ditto.
13656 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13657 Update corresponding peephole2 patterns.
13658 (atomic_storedi_fpu): Ditto.
13660 2017-05-11 Julia Koval <julia.koval@intel.com>
13662 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13663 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13665 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13666 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13667 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13669 2017-05-11 Nathan Sidwell <nathan@acm.org>
13671 * graphite-poly.c: Include dumpfile.h.
13673 * dumpfle.h (dump_function): Declare here ...
13674 * tree-dump.h (dump_function): ... not here.
13675 * dumpfile.c: #include tree-cfg.h.
13676 (dump_function): Move here from ...
13677 * tree-dump.c (dump_function): ... here.
13678 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13679 * graphite-poly.c: Don't include tree-dump.h.
13680 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13681 * print-tree.c: Likewise.
13682 * stor-layout.c: Likewise.
13683 * tree-nested.c: Likewise.
13685 * dumpfile.c (dump_start): Use TDF_FLAGS.
13686 (dump_enable_all): Fix TDF_KIND check thinko.
13688 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13690 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13691 array entries to represent two legal parameterizations of the
13692 overloaded __builtin_cmpb function, as represented by the
13693 P6_OV_BUILTIN_CMPB constant.
13694 (altivec_resolve_overloaded_builtin): Add special case handling
13695 for the __builtin_cmpb function, as represented by the
13696 P6_OV_BUILTIN_CMPB constant.
13697 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13698 (BU_P6_64BIT_2): New macro.
13699 (BU_P6_OVERLOAD_2): New macro
13700 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13701 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13702 (CMPB): Add overload support to represent both 32-bit and 64-bit
13703 compare-bytes function.
13704 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13705 support for TARGET_CMPB.
13706 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13707 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13708 documentation of the __builtin_cmpb overloaded built-in function.
13710 2017-05-11 Richard Biener <rguenther@suse.de>
13712 PR tree-optimization/80705
13713 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13714 bases are not vectorizable.
13716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13718 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13719 when counting register pressure.
13721 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13723 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13724 (iv_ca_more_deps): Renamed to ...
13725 (iv_ca_compare_deps): ... this.
13726 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13728 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13730 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13732 (determine_group_iv_costs): ... here.
13733 (find_inv_vars_cb): Record inv var if it's not recorded before.
13735 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13737 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13738 (get_shiftadd_cost): Ditto.
13740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13742 * tree-ssa-address.c: Include header file.
13743 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13745 (add_to_parts): Refactor.
13746 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13747 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13750 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13752 PR tree-optimization/53090
13753 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13755 (extract_cond_operands): Detect condition with IV on both sides
13756 and return COMP_IV_EXPR_2.
13757 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13758 (rewrite_use_compare): Simplify by removing call to function
13759 extract_cond_operands.
13761 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13763 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13764 (extract_cond_operands): Detect condition comparing against non-
13765 invariant bound and return appropriate enum value.
13766 (find_interesting_uses_cond): Update use of extract_cond_operands.
13767 Handle its return value accordingly.
13768 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13772 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13773 nonlinear iv_use computation in loop invariant sensitive way.
13775 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13777 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13778 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13782 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13783 (dump_cand): Support iv_cand.inv_exprs.
13784 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13786 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13791 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13793 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13794 as local function. Include necessary header files.
13795 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13797 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13799 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13803 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13804 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13805 RSHIFT_EXPR and BIT_NOT_EXPR.
13807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13809 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13810 (adjust_setup_cost): New parameter supporting round up adjustment.
13811 (struct address_cost_data): Delete.
13812 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13813 (split_address_cost, ptr_difference_cost): Delete.
13814 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13815 (struct ainc_cost_data): New struct.
13816 (get_address_cost_ainc): New function.
13817 (get_address_cost, get_computation_cost): Reimplement.
13818 (determine_group_iv_cost_address): Record inv_expr for all uses of
13820 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13821 (iv_ca_has_deps): Reimplemented to ...
13822 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13824 (iv_ca_extend): Call iv_ca_more_deps.
13826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13828 * tree-ssa-address.c (struct mem_address): Move to header file.
13829 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13830 * tree-ssa-address.h (struct mem_address): Move from C file.
13831 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13833 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13835 * tree-affine.h (aff_combination_type): New interface.
13836 (aff_combination_zero_p): Remove static.
13837 (aff_combination_const_p): New interface.
13838 (aff_combination_singleton_var_p): New interfaces.
13840 2017-05-11 Richard Biener <rguenther@suse.de>
13842 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13843 Skip unreachable blocks and destinations.
13844 (eliminate): Move stmt removal and fixup ...
13845 (fini_eliminate): ... here. Skip inserted exprs.
13846 (pass_pre::execute): Move fini_pre after fini_eliminate.
13847 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13848 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13849 PRE to get rid of dead code that has invalid SSA form and
13850 split critical edges again.
13852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13854 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13856 2017-05-11 Richard Biener <rguenther@suse.de>
13858 * passes.c (execute_function_todo): Verify loops if they are
13859 said to be up-to-date.
13860 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13861 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13863 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13866 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13867 handle calling assemble_external ourself.
13870 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13871 modes with zero size. Enhance comment.
13873 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13875 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13876 built-ins for vec_xl and vec_xst with short and char pointer
13879 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13881 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13882 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13883 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13884 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13885 (_mm_maskz_min_round_ss): New intrinsics.
13886 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13887 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13888 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13889 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13890 (__builtin_ia32_minss_mask_round): New builtins.
13891 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13892 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13893 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13895 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13896 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13898 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13901 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13903 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13904 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13905 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13906 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13907 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13908 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13909 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13910 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13911 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13912 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13913 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13914 (__builtin_ia32_mulss_mask_round): New builtins.
13915 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13916 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13917 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13919 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13920 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13922 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13925 2017-05-10 Julia Koval <julia.koval@intel.com>
13927 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13928 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13929 (_mm256_setr_m128i): New intrinsics.
13931 2017-05-10 Julia Koval <julia.koval@intel.com>
13933 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13934 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13935 (_mm_maskz_rcp14_ss): New intrinsics.
13936 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13937 (__builtin_ia32_rcp14ss_mask): New builtins.
13938 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13940 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13942 PR tree-optimization/51513
13943 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13944 (assert_unreachable_fallthru_edge_p): Use it.
13945 (group_case_labels_stmt): Likewise.
13946 * tree-cfg.h: Prototype it.
13947 * stmt.c: Include cfghooks.h and tree-cfg.h.
13948 (emit_case_dispatch_table) <gap_label>: New local variable.
13949 Use it to fill dispatch table gaps.
13950 Test for default_label before updating probabilities.
13951 (expand_case) <default_label>: Remove unneeded initialization.
13952 Test for unreachable default case statement and remove its edge.
13953 Set default_label accordingly.
13954 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13956 2017-05-10 Carl Love <cel@us.ibm.com>
13958 * config/rs6000/rs6000-c: Add support for built-in functions
13959 vector signed char vec_neg (vector signed char)
13960 vector signed short int vec_neg (vector short int)
13961 vector signed int vec_neg (vector signed int)
13962 vector signed long long vec_neg (vector signed long long)
13963 vector float vec_neg (vector float)
13964 vector double vec_neg (vector double)
13965 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13967 * config/rs6000/altivec.h: Add define for vec_neg
13968 * doc/extend.texi: Update the built-in documentation for the
13969 new built-in functions.
13971 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13973 PR tree-optimization/77644
13974 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13976 2017-05-10 Nathan Sidwell <nathan@acm.org>
13978 * dumpfile.h (TDI_lang_all): New.
13979 (TDF_KIND): New. Renumber others
13980 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13982 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13984 (get_dump_file_name): Adjust suffix generation.
13985 (dump_enable_all): Use TDF_KIND.
13986 * doc/invoke.texi (-fdump-lang-all): Document.
13988 * dumpfile.h: Tabify.
13990 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13993 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13994 Move member access before delete.
13996 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13998 * tree-inline.c (expand_call_inline): Split block at stmt
14001 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14004 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14005 are now unused after splitting mov{sf,sd}_hardfloat.
14006 (f32_lr2): Likewise.
14007 (f32_lm): Likewise.
14008 (f32_lm2): Likewise.
14009 (f32_li): Likewise.
14010 (f32_li2): Likewise.
14011 (f32_lv): Likewise.
14012 (f32_sr): Likewise.
14013 (f32_sr2): Likewise.
14014 (f32_sm): Likewise.
14015 (f32_sm2): Likewise.
14016 (f32_si): Likewise.
14017 (f32_si2): Likewise.
14018 (f32_sv): Likewise.
14019 (f32_dm): Likewise.
14020 (f32_vsx): Likewise.
14021 (f32_av): Likewise.
14022 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14023 For movsf, order stores so the VSX stores occur before the GPR
14024 store which encourages the register allocator to use a traditional
14025 FPR instead of a GPR. For movsd, order the stores so that the GPR
14026 store comes before the VSX stores to allow the power6 to work.
14027 This is due to the power6 not having a 32-bit integer store
14028 instruction from a FPR.
14029 (movsf_hardfloat): Likewise.
14030 (movsd_hardfloat): Likewise.
14032 2017-05-09 Martin Sebor <msebor@redhat.com>
14034 PR translation/80280
14035 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14038 PR translation/80280
14039 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14040 data member added in r247778.
14041 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14043 2017-05-09 Nathan Sidwell <nathan@acm.org>
14045 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14047 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14050 2017-05-09 Marek Polacek <polacek@redhat.com>
14052 * doc/invoke.texi: Fix typo.
14054 2017-05-09 Richard Biener <rguenther@suse.de>
14056 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14057 (vrp_val_is_min): Likewise.
14058 (set_value_range_to_value): Likewise.
14059 (set_value_range_to_nonnegative): Likewise.
14060 (gimple_assign_nonzero_p): Likewise.
14061 (gimple_stmt_nonzero_p): Likewise.
14062 (vrp_int_const_binop): Likewise. Remove unreachable case.
14063 (adjust_range_with_scev): Adjust comments.
14064 (compare_range_with_value): Likewise.
14065 (extract_range_from_phi_node): Likewise.
14066 (test_for_singularity): Likewise.
14068 2017-05-09 Richard Biener <rguenther@suse.de>
14070 * tree-vrp.c (get_single_symbol): Add assert that we don't
14071 get overflowed constants as invariant part.
14072 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14073 checks. Use wi::cmp instead of recursing for integer constants.
14074 (compare_values): Just ignore whether we assumed undefined
14075 overflow instead of failing the compare.
14076 (extract_range_for_var_from_comparison_expr): Add comment before the
14077 TREE_NO_WARNING sets.
14078 (test_for_singularity): Likewise.
14079 (extract_range_from_comparison): Do not disable optimization
14080 when we assumed undefined overflow.
14081 (extract_range_basic): Remove init of unused var.
14083 2017-05-09 Richard Biener <rguenther@suse.de>
14085 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14086 (extract_range_from_multiplicative_op_1): Adjust.
14087 (extract_range_from_binary_expr_1): Use int_const_binop.
14089 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14092 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14093 rs6000_store_data_bypass_p in seven define_bypass directives and
14094 in several comments.
14095 * config/rs6000/rs6000-protos.h: Add prototype for
14096 rs6000_store_data_bypass_p function.
14097 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14098 function implements slightly different (rs6000-specific) semantics
14099 than store_data_bypass_p, returning false rather than aborting
14100 with assertion error when arguments do not satisfy the
14101 requirements of store data bypass.
14102 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14103 rs6000_store_data_bypass_p.
14105 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14107 * config/xtensa/xtensa-protos.h
14108 (xtensa_initial_elimination_offset): New declaration.
14109 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14110 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14111 macro definition, add case for FRAME_POINTER_REGNUM when
14112 FRAME_GROWS_DOWNWARD.
14113 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14114 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14115 xtensa_initial_elimination_offset.
14117 2017-05-08 Nathan Sidwell <nathan@acm.org>
14119 * doc/invoke.texi: Alphabetize -fdump options.
14121 2017-05-08 Martin Sebor <msebor@redhat.com>
14123 PR translation/80280
14124 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14126 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14128 * target.def (compute_frame_layout): New optional target hook.
14129 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14130 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14131 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14133 * reload1.c (verify_initial_elim_offsets): Likewise.
14134 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14135 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14136 (arm_get_frame_offsets): Split up into this ...
14137 (arm_compute_frame_layout): ... and this function.
14139 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14141 * config/aarch64/constraints.md (Usa): New constraint.
14142 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14144 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14146 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14147 with_multilib_list after it has been checked.
14149 2017-05-08 Richard Biener <rguenther@suse.de>
14151 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14152 (bitmap_set_subtract_values): Likewise.
14154 2017-05-08 Richard Biener <rguenther@suse.de>
14156 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14157 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14159 (gimple_stmt_nonzero_warnv_p): Rename to ...
14160 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14162 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14163 (extract_range_basic): Adjust, do not disable propagation on
14164 strict overflow sensitive simplification.
14165 (vrp_visit_cond_stmt): Likewise.
14167 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14169 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14170 body size unconditionally.
14172 2017-05-07 Jeff Law <law@redhat.com>
14175 2017-05-06 Jeff Law <law@redhat.com>
14176 PR tree-optimization/78496
14177 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14180 PR tree-optimization/78496
14181 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14182 (simplify_stmt_using_ranges): Call it.
14183 (vrp_dom_walker::before_dom_children): Extract equivalences
14184 from an ASSERT_EXPR with an equality comparison against a
14187 2017-05-06 Jeff Law <law@redhat.com>
14189 PR tree-optimization/78496
14190 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14193 PR tree-optimization/78496
14194 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14195 (simplify_stmt_using_ranges): Call it.
14196 (vrp_dom_walker::before_dom_children): Extract equivalences
14197 from an ASSERT_EXPR with an equality comparison against a
14200 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14202 * lra-constraints.c (lra_copy_reg_equiv): New function.
14203 (split_reg): Use it to copy equivalence information from the
14204 original register to the spill register.
14206 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14208 PR rtl-optimization/75964
14209 * simplify-rtx.c (simplify_const_relational_operation): Remove
14210 invalid handling of comparisons of integer ABS.
14212 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14214 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14215 initialize to zero.
14216 (init_regs): Remove declaration.
14217 (function_arg_advance_32): Initialize error_p as boolean variable.
14219 2017-05-05 Nathan Sidwell <nathan@acm.org>
14221 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14222 lines. Use for (;;).
14224 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14226 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14227 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14228 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14229 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14230 VF=2 that require versioning.
14232 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14234 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14237 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14239 * diagnostic.h (diagnostic_override_option_index): Convert from
14240 macro to inline function.
14242 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14244 * diagnostic.c (last_module_changed_p): New function.
14245 (set_last_module): New function.
14246 (diagnostic_report_current_module): Convert macro usage to
14247 the above functions.
14248 * diagnostic.h (diagnostic_context::last_module): Strengthen
14249 from const line_map * to const line_map_ordinary *.
14250 (diagnostic_last_module_changed): Delete macro.
14251 (diagnostic_set_last_module): Delete macro.
14253 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14255 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14256 with diagnostic_report_diagnostic.
14257 (diagnostic_n_impl_richloc): Likewise.
14258 * diagnostic.h (report_diagnostic): Delete macro.
14259 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14260 with diagnostic_report_diagnostic.
14261 * substring-locations.c (format_warning_va): Likewise.
14263 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14265 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14266 save/restor of format_spec. Move option-printing code to...
14267 (print_option_information): ...this new function, and
14268 reimplement by simply printing to the pretty_printer,
14269 rather than appending to the format string.
14271 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14273 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14274 handling logic into...
14275 (update_effective_level_from_pragmas): ...this new function.
14277 2017-05-04 Andrew Waterman <andrew@sifive.com>
14279 * config/riscv/riscv.opt (mstrict-align): New option.
14280 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14281 (SLOW_UNALIGNED_ACCESS): Define.
14282 (riscv_slow_unaligned_access): Declare.
14283 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14285 (riscv_slow_unaligned_access): New variable.
14286 (rocket_tune_info): Set slow_unaligned_access to true.
14287 (optimize_size_tune_info): Set slow_unaligned_access to false.
14288 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14289 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14290 (riscv_option_override): Set riscv_slow_unaligned_access.
14291 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14293 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14295 * config/riscv/riscv.md: Unify indentation.
14297 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14302 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14304 (extendsi<mode>2): Add support for doing sign extension via
14305 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14306 don't have ISA 3.0 instructions.
14307 (extendsi<mode>2 splitter): Likewise.
14308 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14309 generate the normal insns since SImode can now go in vector
14310 registers. Disallow the special UNSPECs needed for previous
14311 machines to hide SImode being used. Add new insns
14312 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14313 (fix_trunc<mode>si2_stfiwx): Likewise.
14314 (fix_trunc<mode>si2_internal): Likewise.
14315 (fixuns_trunc<mode>si2): Likewise.
14316 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14317 (fctiw<u>z_<mode>_smallint): Likewise.
14318 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14319 of floating point to 32-bit integer from doing a direct move to
14320 the GPR registers to do a store.
14321 (fctiwz_<mode>): Break long line.
14323 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14325 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14326 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14327 (addr_list, addr_offset_valid_p): New.
14328 (split_address_groups): Check offset validity with above function.
14329 (gt-tree-ssa-loop-ivopts.h): Include header file.
14331 2017-05-05 Nathan Sidwell <nathan@acm.org>
14333 * config.gcc (arm*-*-*): Add missing 'fi'.
14335 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14337 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14338 included in -fopt-info does not matter.
14339 * doc/optinfo.texi (-fopt-info): Fix description of default
14340 behavour. Explicitly say order of options included in -fopt-info
14343 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14345 * config.gcc: Allow combinations of aprofile and rmprofile values for
14346 --with-multilib-list.
14347 * config/arm/t-multilib: New file.
14348 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14349 variables. Remove setting of ISA and floating-point ABI in
14350 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14351 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14352 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14353 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14355 * config/arm/t-rmprofile: Likewise except for the matches changes.
14356 * doc/install.texi (--with-multilib-list): Document the combination of
14357 aprofile and rmprofile values and warn about pitfalls in doing that.
14359 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14361 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14362 (movdi_aarch64): Likewise.
14364 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14366 PR tree-optimization/80632
14367 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14369 (build_arrays): Initialize it for virtual phis.
14370 (fix_phi_nodes): Use it for virtual phis.
14372 PR tree-optimization/80558
14373 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14374 [x, y] op z into [x op, y op z] for op & or | if conditions
14377 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14378 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14381 * config/arm/arm.md (use_literal_pool): Remove.
14382 (64-bit immediate split): No longer takes cost into consideration
14383 if arm_disable_literal_pool is enabled.
14384 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14385 used when arm_disable_literal_pool is enabled.
14386 (arm_max_const_double_inline_cost): Remove use of
14387 arm_disable_literal_pool.
14388 (push_minipool_fix): Add assert.
14389 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14390 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14391 (no_literal_pool_sf_immediate): New.
14393 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14395 PR tree-optimization/80613
14396 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14397 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14399 2017-05-05 Richard Biener <rguenther@suse.de>
14401 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14403 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14405 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14406 of this flag from insn conditions due to removal from r247495.
14408 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14410 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14412 (arm_early_store_addr_dep_ptr): Likewise.
14413 * config/arm/aarch-common-protos.h
14414 (arm_early_load_addr_dep_ptr): Add prototype.
14415 (arm_early_store_addr_dep_ptr): Likewise.
14416 * config/arm/cortex-a53.md: Add new bypasses.
14418 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14420 * tree.c (next_type_uid): Change type to unsigned.
14421 (type_hash_canon): Decrement back next_type_uid if
14422 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14423 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14426 2017-05-04 Martin Sebor <msebor@redhat.com>
14428 * builtins.c: Fix a trivial typo in a comment.
14430 PR middle-end/79234
14431 * builtins.c (check_sizes): Adjust to handle reading past the end.
14432 Avoid printing excessive upper bound of ranges. Use %E to print
14433 tree nodes instead of converting them to %wu.
14434 (expand_builtin_memchr): New function.
14435 (compute_dest_size): Rename...
14436 (compute_objsize): ...to this.
14437 (expand_builtin_memcpy): Adjust.
14438 (expand_builtin_mempcpy): Adjust.
14439 (expand_builtin_strcat): Adjust.
14440 (expand_builtin_strcpy): Adjust.
14441 (check_strncat_sizes): Adjust.
14442 (expand_builtin_strncat): Adjust.
14443 (expand_builtin_strncpy): Adjust and simplify.
14444 (expand_builtin_memset): Adjust.
14445 (expand_builtin_bzero): Adjust.
14446 (expand_builtin_memcmp): Adjust.
14447 (expand_builtin): Handle memcmp.
14448 (maybe_emit_chk_warning): Check strncat just once.
14450 2017-05-04 Martin Sebor <msebor@redhat.com>
14452 PR preprocessor/79214
14453 PR middle-end/79222
14454 PR middle-end/79223
14455 * builtins.c (check_sizes): Add inlining context and issue
14456 warnings even when -Wno-system-headers is set.
14457 (check_strncat_sizes): Same.
14458 (expand_builtin_strncat): Same.
14459 (expand_builtin_memmove): New function.
14460 (expand_builtin_stpncpy): Same.
14461 (expand_builtin): Handle memmove and stpncpy.
14463 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14465 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14466 which is not used any more.
14468 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14470 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14472 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14474 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14475 (cortexa53_tunings): Likewise.
14476 (cortexa57_tunings): Likewise.
14477 (cortexa72_tunings): Likewise.
14478 (cortexa73_tunings): Likewise.
14480 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14482 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14483 Set loop alignment to 8.
14485 2017-05-04 Martin Sebor <msebor@redhat.com>
14487 PR translation/80280
14488 * builtins.c (expand_builtin_object_size): Add missing quoting to
14489 %D and like directives.
14490 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14491 (hsa_type_for_tree_type): Same.
14492 (verify_function_arguments): Same.
14493 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14494 * varasm.c (get_section): Same.
14497 2017-05-04 Martin Sebor <msebor@redhat.com>
14499 PR translation/80280
14500 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14502 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14504 * config/aarch64/aarch64.c (generic_addrcost_table):
14505 Change HI/TI mode setting.
14507 2017-05-04 Martin Jambor <mjambor@suse.cz>
14509 PR tree-optimization/80622
14510 * tree-sra.c (comes_initialized_p): New function.
14511 (build_accesses_from_assign): Only set write lazily when
14512 comes_initialized_p is false.
14513 (analyze_access_subtree): Use comes_initialized_p.
14514 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14515 instead of testing for PARM_DECL.
14517 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14519 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14520 constraint on operand 0 to allow more general addressing modes.
14521 Adjust output template.
14522 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14524 * config/aarch64/aarch64-protos.h
14525 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14526 * config/aarch64/constraints.md (Dp): New address constraint.
14527 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14530 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14532 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14533 update use of estimate_ipcp_clone_size_and_time.
14534 (estimate_local_effects): Update use of
14535 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14536 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14537 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14538 Return nonspecialized time.
14540 2017-05-04 Richard Biener <rguenther@suse.de>
14542 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14543 for the last VUSE which def dominates the PHI. Directly call
14545 (get_continuation_for_phi_1): Remove.
14547 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14549 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14550 to explain the use of truncating division. Cap the number of
14551 iterations to the maximum given by nb_iterations_upper_bound,
14554 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14556 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14557 * configure: Regenerate.
14558 * config.in: Regenerate.
14559 * config/i386/driver-mingw32.c: new file.
14560 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14561 * config.host: Link driver-mingw32.o on MinGW host.
14562 * doc/install.texi: Document new --enable-mingw-wildcard configure
14565 2017-05-04 Marek Polacek <polacek@redhat.com>
14567 PR tree-optimization/80612
14568 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14570 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14571 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14573 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14574 (movt splitter): Likewise.
14575 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14576 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14577 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14578 block for Thumb-1 with MOVT.
14579 (thumb2_legitimate_address_p): Move code block ...
14580 (can_avoid_literal_pool_for_label_p): ... into this new function.
14581 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14583 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14584 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14585 "M-profile targets with the MOVT instruction".
14587 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14589 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14590 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14591 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14593 2017-05-04 Martin Liska <mliska@suse.cz>
14595 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14596 variable cond_code.
14598 2017-05-04 Richard Biener <rguenther@suse.de>
14600 * tree.c (array_at_struct_end_p): Handle arrays at struct
14601 end with flexarrays more conservatively. Refactor and treat
14602 arrays of arrays or aggregates more strict. Fix
14603 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14604 * tree.c (array_at_struct_end_p): Adjust prototype.
14605 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14606 * gimple-fold.c (get_range_strlen): Likewise.
14607 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14609 2017-05-04 Richard Biener <rguenther@suse.de>
14611 PR tree-optimization/31130
14612 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14614 (supports_overflow_infinity): Likewise.
14615 (is_negative_overflow_infinity): Likewise.
14616 (is_positive_overflow_infinity): Likewise.
14617 (is_overflow_infinity): Likewise.
14618 (stmt_overflow_infinity): Likewise.
14619 (overflow_infinity_range_p): Likewise.
14620 (usable_range_p): Remove as always returning true.
14621 (make_overflow_infinity): Remove.
14622 (negative_overflow_infinity): Likewise.
14623 (positive_overflow_infinity): Likewise.
14624 (avoid_overflow_infinity): Likewise.
14625 (set_value_range): Adjust accordingly.
14626 (set_value_range_to_nonnegative): Likewise, remove now unused
14627 overflow_infinity arg.
14628 (vrp_operand_equal_p): Adjust.
14629 (update_value_range): Likewise.
14630 (range_int_cst_singleton_p): Likewise.
14631 (operand_less_p): Likewise.
14632 (compare_values_warnv): Likewise.
14633 (extract_range_for_var_from_comparison_expr): Likewise.
14634 (vrp_int_const_binop): Likewise.
14635 (zero_nonzero_bits_from_vr): Likewise.
14636 (extract_range_from_multiplicative_op_1): Likewise.
14637 (extract_range_from_binary_expr_1): Likewise.
14638 (extract_range_from_unary_expr): Likewise.
14639 (extract_range_from_comparison): Likewise.
14640 (extract_range_basic): Likewise.
14641 (adjust_range_with_scev): Likewise.
14642 (compare_ranges): Likewise.
14643 (compare_range_with_value): Likewise.
14644 (dump_value_range): Likewise.
14645 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14647 (simplify_cond_using_ranges): Adjust.
14649 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14651 * brig-builtins.def: Added a builtin for class_f64.
14652 * builtin-types.def: Added a builtin type needed by class_f64.
14654 2017-05-03 Jason Merrill <jason@redhat.com>
14656 * timevar.def: Add TV_CONSTEXPR.
14658 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14660 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14662 2017-05-03 Martin Jambor <mjambor@suse.cz>
14664 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14665 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14666 * ipa-cp.c (ipcp_propagate_stage): Do not call
14667 ipa_update_after_lto_read.
14668 * ipa-inline.c (ipa_inline): Likewise.
14670 2017-05-03 Martin Jambor <mjambor@suse.cz>
14672 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14673 tag. Added a default constructor and a destructor.
14674 (ipa_edge_args_sum_t): New class;
14675 (ipa_edge_args_sum): Declare.
14676 (ipa_edge_args_vector): Remove declaration.
14677 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14678 (ipa_free_edge_args_substructures): Remove declaration.
14679 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14680 (ipa_edge_args_info_available_for_edge_p): Likewise.
14681 * ipa-prop.c (ipa_edge_args_vector): Removed.
14682 (edge_removal_hook_holder): Likewise.
14683 (edge_duplication_hook_holder): Likewise.
14684 (ipa_edge_args_sum): New variable.
14685 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14686 ipa_edge_args_vector.
14687 (ipa_free_edge_args_substructures): Likewise.
14688 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14689 ipa_edge_args_vector.
14690 (ipa_edge_removal_hook): Turned into method
14691 ipa_edge_args_sum_t::remove.
14692 (ipa_edge_duplication_hook): Turned into method
14693 ipa_edge_args_sum_t::duplicate.
14694 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14695 registering edge hooks.
14696 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14697 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14698 ipa_edge_args_sum instead of ipa_edge_args_vector.
14699 * ipa-profile.c (ipa_profile): Likewise.
14701 2017-05-03 Martin Jambor <mjambor@suse.cz>
14703 * symbol-summary.h (function_summary): New method exists.
14704 (function_summary::symtab_removal): Deallocate through release.
14705 (call_summary): New class.
14706 (gt_ggc_mx): New overload.
14707 (gt_pch_nx): Likewise.
14708 (gt_pch_nx): Likewise.
14710 2017-05-03 Jeff Law <law@redhat.com>
14712 PR tree-optimization/78496
14713 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14714 from simplify_cond_using_ranges. Split off code to walk
14715 backwards through casts into ...
14716 (simplify_cond_using_ranges_2): New function.
14717 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14718 (execute_vrp): After identifying jump threads, call
14719 simplify_cond_using_ranges_2.
14721 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14724 * ipa-inline.h (inline_summary): Add ctor.
14725 (create_ggc): Do not use ggc_cleared_alloc.
14727 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14728 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14730 * gcc.c (handle_braces): Support escaping in switch matching
14732 * doc/invoke.texi (Spec Files): Document it.
14733 Remove superfluous @code markup in items.
14735 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14737 * diagnostic-show-locus.c (struct column_range): New struct.
14738 (get_affected_columns): New function.
14739 (get_printed_columns): New function.
14740 (struct correction): New struct.
14741 (correction::ensure_capacity): New function.
14742 (correction::ensure_terminated): New function.
14743 (struct line_corrections): New struct.
14744 (line_corrections::~line_corrections): New dtor.
14745 (line_corrections::add_hint): New function.
14746 (layout::print_trailing_fixits): Reimplement in terms of the new
14748 (selftest::test_overlapped_fixit_printing): New function.
14749 (selftest::diagnostic_show_locus_c_tests): Call it.
14751 2017-05-03 Nathan Sidwell <nathan@acm.org>
14753 Canonicalize canonical type hashing
14754 * tree.h (type_hash_canon_hash): Declare.
14755 * tree.c (type_hash_list, attribute_hash_list): Move into
14756 type_hash_canon_hash.
14757 (build_type_attribute_qual_variant): Break out hash code calc into
14758 type_hash_canon_hash.
14759 (type_hash_canon_hash): New. Generic type hash computation.
14760 (build_range_type_1, build_array_type_1, build_function_type,
14761 build_method_type_directly, build_offset_type, build_complex_type,
14762 make_vector_type): Call it.
14764 2017-05-03 Richard Biener <rguenther@suse.de>
14766 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14767 When all DRs have unknown misaligned do not always peel
14768 when there is a store but apply the same costing model as if
14769 there were only loads.
14771 2017-05-03 Richard Biener <rguenther@suse.de>
14774 PR tree-optimization/80492
14775 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14776 compare_base_decls returning dont-know properly.
14778 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14780 * config/arm/iterators.md (CCSI): New mode iterator.
14781 (arch): New mode attribute.
14782 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14783 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14784 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14785 code iterator for success result mode.
14786 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14787 the corresponding new insn generators.
14789 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14792 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14793 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14795 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14797 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14798 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14799 (DDR_A): Wrap DDR argument in brackets.
14800 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14801 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14802 (DDR_REVERSED_P): Likewise.
14804 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14806 PR tree-optimization/79472
14807 * tree-switch-conversion.c (struct switch_conv_info): Add
14808 contiguous_range and default_case_nonstandard fields.
14809 (collect_switch_conv_info): Compute contiguous_range and
14810 default_case_nonstandard fields, don't clear final_bb if
14811 contiguous_range and only the default case doesn't have the required
14813 (check_all_empty_except_final): Set default_case_nonstandard instead
14814 of failing if contiguous_range and the default case doesn't have empty
14816 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14817 and only the default case doesn't have the required constants. Skip
14819 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14820 if default_case_nonstandard.
14821 (build_constructors): Build constant 1 just once. Assert that default
14822 values aren't inserted in between cases if contiguous_range. Skip
14824 (build_arrays): Skip virtual phis.
14825 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14826 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14827 Handle virtual phis.
14828 (gen_inbound_check): Handle default_case_nonstandard case.
14829 (process_switch): Adjust check_final_bb caller. Call
14830 gather_default_values with the first non-default case instead of
14831 default case if default_case_nonstandard.
14833 2017-05-02 Nathan Sidwell <nathan@acm.org>
14835 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14836 check. Fix formatting.
14838 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14840 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14841 errors when comparing specialized and unspecialized times.
14843 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14845 * diagnostic-show-locus.c
14846 (layout::should_print_annotation_line_p): Make private.
14847 (layout::print_annotation_line): Make private.
14848 (layout::annotation_line_showed_range_p): Make private.
14849 (layout::show_ruler): Make private.
14850 (layout::print_source_line): Make private. Pass in line and
14851 line_width, rather than calling location_get_source_line. Drop
14853 (layout::print_leading_fixits): New method.
14854 (layout::print_any_fixits): Rename to...
14855 (layout::print_trailing_fixits): ...this, and make private.
14856 Don't print newline fixits.
14857 (diagnostic_show_locus): Move logic for printing one row into...
14858 (layout::print_line): ...this new function. Move the
14859 location_get_source_line call and error-handling from
14860 print_source_line to here. Call print_leading_fixits, and rename
14861 print_any_fixits to print_trailing_fixits.
14862 (selftest::test_fixit_insert_containing_newline): Update now that
14863 newlines are partially supported.
14864 (selftest::test_fixit_insert_containing_newline_2): New test.
14865 (selftest::test_fixit_replace_containing_newline): Update comments.
14866 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14867 * edit-context.c (class added_line): New class.
14868 (class edited_line): Describe newline handling in comment.
14869 (edited_line::actually_edited_p): New method.
14870 (edited_line::print_content): Delete redundant decl.
14871 (edited_line::m_predecessors): New field.
14872 (edited_file::print_content): Call edited_line::print_content.
14873 (edited_file::print_diff): Update to support newlines.
14874 (edited_file::print_diff_hunk): Likewise.
14875 (edited_file::print_run_of_changed_lines): New function.
14876 (edited_file::print_diff_line): Convert to...
14877 (print_diff_line): ...this.
14878 (edited_file::get_effective_line_count): New function.
14879 (edited_line::edited_line): Initialize new field m_predecessors.
14880 (edited_line::~edited_line): Clean up m_predecessors.
14881 (edited_line::apply_fixit): Handle newlines.
14882 (edited_line::get_effective_line_count): New function.
14883 (edited_line::print_content): New function.
14884 (edited_line::print_diff_lines): New function.
14885 (selftest::test_applying_fixits_insert_containing_newline): New
14887 (selftest::test_applying_fixits_replace_containing_newline): New
14889 (selftest::insert_line): New function.
14890 (selftest::test_applying_fixits_multiple_lines): Add example of
14892 (selftest::edit_context_c_tests): Call the new tests.
14894 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14896 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14897 parameter cand. Update dump information.
14898 (get_computation_cost): Update uses.
14900 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14902 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14903 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14904 (get_computation_at, rewrite_use_address): Update use of
14905 get_computation_aff.
14907 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14909 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14910 (get_computation): Delete.
14911 (get_computation_cost): Implement like get_computation_cost_at.
14912 Use get_computation_at.
14913 (get_computation_cost_at): Delete.
14914 (rewrite_use_nonlinear_expr): Use get_computation_at.
14915 (rewrite_use_compare, remove_unused_ivs): Ditto.
14917 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14919 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14921 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14923 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14924 (ivopts_global_cost_for_size): Rename parameter and update uses.
14925 (iv_ca_recount_cost): Update uses.
14926 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14927 candidates seperately in n_invs and n_cands.
14928 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14930 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14932 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14933 (find_inv_vars_cb): New.
14934 (find_depends): Renamed to ...
14935 (find_inv_vars): ... this.
14936 (add_candidate_1, force_var_cost): Call find_inv_vars.
14937 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14939 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14941 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14942 inv_vars. Add inv_exprs.
14943 (struct iv_cand): Rename depends_on to inv_vars.
14944 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14945 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14946 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14947 (dump_cand): Dump inv_vars.
14948 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14949 (record_invariant, find_depends, add_candidate_1): Ditto.
14950 (set_group_iv_cost, force_var_cost): Ditto.
14951 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14952 (get_computation_cost_at, get_computation_cost): Ditto.
14953 (determine_group_iv_cost_generic): Ditto.
14954 (determine_group_iv_cost_address): Ditto.
14955 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14956 (determine_group_iv_costs): Ditto.
14957 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14958 (iv_ca_set_remove_invariants): Renamed to ...
14959 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14960 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14961 (iv_ca_set_add_invariants): Renamed to ...
14962 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14963 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14964 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14965 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14966 (create_new_ivs): Remove useless dump.
14968 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14970 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14972 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14973 (iv_ca_set_no_cp, create_new_iv): Ditto.
14975 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14977 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14979 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14981 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14982 function tree_check2.
14984 2017-05-02 Martin Liska <mliska@suse.cz>
14986 * doc/gcov.texi: Add missing preposition.
14987 * gcov.c (function_info::function_info): Properly fill up
14988 all member variables.
14990 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14992 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14994 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14996 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14998 2017-05-02 Martin Liska <mliska@suse.cz>
15001 * lto-streamer-in.c (lto_read_tree_1): Remove
15002 LTO_STREAMER_DEBUG.
15003 * lto-streamer.c (struct tree_hash_entry): Likewise.
15004 (struct tree_entry_hasher): Likewise.
15005 (tree_entry_hasher::hash): Likewise.
15006 (tree_entry_hasher::equal): Likewise.
15007 (lto_streamer_init): Likewise.
15008 (lto_orig_address_map): Likewise.
15009 (lto_orig_address_get): Likewise.
15010 (lto_orig_address_remove): Likewise.
15011 * lto-streamer.h: Likewise.
15012 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15013 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15015 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15017 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15018 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15019 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15020 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15021 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15022 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15023 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15024 (mm_maskz_sub_ss): New intrinsics.
15025 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15026 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15027 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15028 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15029 (__builtin_ia32_subss_mask_round): New builtins.
15030 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15031 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15032 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15034 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15035 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15037 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15040 2017-05-02 Martin Jambor <mjambor@suse.cz>
15042 PR tree-optimization/78687
15043 * tree-sra.c (access): New field parent.
15044 (process_subtree_disqualification): New function.
15045 (disqualify_candidate): Call it.
15046 (build_accesses_from_assign): Reset write flag if creating an
15048 (build_access_subtree): Fill in parent field and also prpagate
15049 down grp_write flag.
15050 (create_artificial_child_access): New parameter set_grp_write, set
15051 grp_write to its value.
15052 (propagate_subaccesses_across_link): Also propagate grp_write flag
15054 (propagate_all_subaccesses): Push the closest parent back to work
15055 queue if add_access_to_work_queue returned true.
15057 2017-05-02 Richard Biener <rguenther@suse.de>
15059 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15060 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15061 -fstrict-overflow documentation.
15062 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15063 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15064 flag_strict_overflow.
15065 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15066 * lto-opts.c (lto_write_options): Do not stream it.
15067 * lto-wrapper.c (merge_and_complain): Do not handle it.
15068 * opts.c (default_options_table): Do not set -fstrict-overflow.
15069 (finish_options): Likewise do not clear it when sanitizing.
15070 * simplify-rtx.c (simplify_const_relational_operation): Do not
15071 test flag_strict_overflow.
15073 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15075 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15076 using enabled attribute.
15077 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15078 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15079 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15080 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15081 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15082 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15083 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15084 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15085 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15086 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15088 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15090 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15092 2017-05-02 Richard Biener <rguenther@suse.de>
15094 PR tree-optimization/80591
15096 2017-04-10 Richard Biener <rguenther@suse.de>
15098 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15101 2017-05-02 Richard Biener <rguenther@suse.de>
15103 PR tree-optimization/80549
15104 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15105 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15106 headers if they do not have a preheader.
15108 2017-05-02 Martin Liska <mliska@suse.cz>
15111 * common.opt: Fix typo.
15112 * doc/invoke.texi: Likewise.
15114 2017-05-01 Jan Beulich <jbeulich@suse.com>
15116 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15117 swapping, add (x,x,m,x,n) alternative.
15119 2017-05-01 Nathan Sidwell <nathan@acm.org>
15121 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15122 unnecessary unadjusted_alignment check.
15124 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15127 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15129 * gimplify.c (gimplify_cilk_detach): New function.
15130 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15131 * tree-core.h: Document EXPR_CILK_SPAWN.
15132 * tree.h (EXPR_CILK_SPAWN): Define.
15134 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15136 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15137 to use new fixit_hint representation, using the "replace" logic.
15138 (get_line_span_for_fixit_hint): Likewise.
15139 (layout::print_any_fixits): Likewise.
15140 (selftest::test_one_liner_many_fixits): Rename to...
15141 (selftest::test_one_liner_many_fixits_1): ...this, and update
15142 comment and expected output to reflect that the multiple fix-it
15143 hints are now consolidated into one insertion.
15144 (selftest::test_one_liner_many_fixits_2): New test.
15145 (selftest::test_diagnostic_show_locus_one_liner): Update for
15147 (selftest::test_fixit_consolidation): Update for fix-it API
15149 * diagnostic.c (print_parseable_fixits): Likewise.
15150 * edit-context.c (edited_line::m_line_events): Convert from
15151 auto_vec <line_event *> to auto_vec <line_event>.
15152 (class line_event): Convert from abstract base class to a concrete
15153 class, taking over the role of replace_event.
15154 (class insert_event): Delete.
15155 (class replace_event): Rename to class line_event. Convert to
15157 (edit_context::add_fixits): Reimplement.
15158 (edit_context::apply_insert): Delete.
15159 (edit_context::apply_replace): Rename to...
15160 (edit_context::apply_fixit): ...this. Convert to half-open range.
15161 (edited_file::apply_insert): Delete.
15162 (edited_file::apply_replace): Rename to...
15163 (edited_file::apply_fixit): ...this.
15164 (edited_line::~edited_line): Drop deletion of events.
15165 (edited_line::apply_insert): Delete.
15166 (edited_line::apply_replace): Rename to...
15167 (edited_line::apply_fixit): ...this. Convert to half-open range.
15168 Update for change to type of m_line_events.
15169 * edit-context.h (edit_context::apply_insert): Delete.
15170 (edit_context::apply_replace): Rename to...
15171 (edit_context::apply_fixit): ...this.
15173 2017-05-01 Martin Sebor <msebor@redhat.com>
15175 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15178 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15181 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15182 __get_cpuid_max returns 0.
15183 (__get_cpuid_count): Ditto.
15185 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15187 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15188 replacement expression is another instance of one of its arguments.
15190 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15193 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15194 check for stack push/pop autoinc.
15195 * config/i386/i386.c (ix86_agi_dependent): Return false
15196 if the only reason why modified_in_p returned true is that
15197 addr is SP based and set_insn is a push or pop.
15199 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15201 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15204 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15207 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15208 (account_size_time): Use two predicates - exec_pred and
15210 (evaluate_conditions_for_known_args): Compute both clause and
15212 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15213 (inline_summary_t::duplicate): Update.
15214 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15216 (compute_inline_parameters): Likewise.
15217 (estimate_edge_size_and_time): Update caluclation of time.
15218 (estimate_node_size_and_time): Compute both time and nonspecialized
15220 (estimate_ipcp_clone_size_and_time): Update.
15221 (inline_merge_summary): Update.
15222 (do_estimate_edge_time): Update.
15223 (do_estimate_edge_size): Update.
15224 (do_estimate_edge_hints): Update.
15225 (inline_read_section, inline_write_summary): Stream both new predicates.
15226 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15228 (compute_inlined_call_time): Cleanup.
15229 (big_speedup_p): Update.
15230 (edge_badness): Update.
15231 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15232 (size_time_entry): Replace predicate by exec_predicate and
15233 nonconst_predicate.
15234 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15235 (estimate_edge_time): Return also nonspec_time.
15236 (reset_edge_growth_cache): Update.
15238 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15240 PR rtl-optimization/80491
15241 * ifcvt.c (noce_process_if_block): When looking for x setter
15242 with missing else_bb, don't check only the insn right before
15243 cond_earliest, but look for the last insn that x is modified in
15244 within the same bb.
15246 PR rtl-optimization/80491
15247 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15249 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15251 PR tree-optimization/80487
15252 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15254 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15256 PR tree-optimization/79697
15257 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15258 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15259 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15261 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15262 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15264 2017-04-28 Martin Sebor <msebor@redhat.com>
15266 PR tree-optimization/80523
15267 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15268 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15270 (maybe_warn, format_directive, parse_directive): Use new functions.
15271 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15273 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15275 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15277 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15279 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15280 target_header_dir): Set correctly.
15281 * configure: Regenerated.
15282 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15283 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15284 instead of SYSTEM_HEADER_DIR.
15286 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15288 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15289 (estimate_local_effects): Likewise.
15290 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15291 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15292 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15293 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15294 do_estimate_edge_time, estimate_edge_time): Likewise.
15295 * ipa-inline-analysis.c (estimate_node_size_and_time,
15296 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15297 (estimate_time_after_inlining): Remove.
15299 2017-04-28 Martin Liska <mliska@suse.cz>
15301 * doc/gcov.texi: Enhance documentation of gcov.
15303 2017-04-28 Martin Liska <mliska@suse.cz>
15305 * doc/gcov.texi: Sort options in alphabetic order.
15306 * doc/gcov-dump.texi: Likewise.
15307 * doc/gcov-tool.texi: Likewise.
15308 * gcov.c (print_usage): Likewise.
15309 * gcov-dump.c (print_usage): Likewise.
15310 * gcov-tool.c (print_merge_usage_message): Likewise.
15311 (print_rewrite_usage_message): Likewise.
15312 (print_overlap_usage_message): Likewise.
15314 2017-04-28 Martin Liska <mliska@suse.cz>
15316 PR gcov-profile/53915
15317 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15319 2017-04-28 Martin Liska <mliska@suse.cz>
15321 PR gcov-profile/79891
15322 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15323 is marked by compiler as living on a line.
15324 (get_cycles_count): Remove usage of the union.
15325 (output_intermediate_file): Likewise.
15326 (find_source): Fix GNU coding style.
15327 (accumulate_line_counts): Remove old non-all block mode.
15328 (output_lines): Remove usage of the union.
15329 * profile.c (output_location): Include all BBs, even if
15330 belonging to a same line (and file) as a previous BB.
15332 2017-04-28 Martin Liska <mliska@suse.cz>
15334 * gcov.c (process_args): Handle new argument 'w'.
15335 (read_graph_file): Assign ID to BBs.
15336 (output_branch_count): Display BB # if verbose flag is set.
15337 (output_lines): Likewise for arcs.
15338 (print_usage): Add '--verbose' option help.
15339 * doc/gcov.texi: Document --verbose (-w) option.
15341 2017-04-28 Martin Liska <mliska@suse.cz>
15343 * gcov.c (struct block_location_info): New struct.
15344 (process_file): Fill up the new structure.
15345 (read_graph_file): Replace usage of encoding by the newly added
15347 (add_line_counts): Likewise.
15348 (accumulate_line_counts): Remove usage of the union.
15349 (function_info::function_info): New function.
15350 (function_info::~function_info): Likewise.
15351 (process_file): Call delete instead of release_function.
15352 (release_function): Release the function.
15353 (release_structures): Call delete instead of release_function.
15354 (solve_flow_graph): Replace usage of num_blocks.
15355 (find_exception_blocks): Likewise.
15356 (output_lines): Fix GNU coding style.
15358 2017-04-28 Martin Liska <mliska@suse.cz>
15361 * coverage.c (coverage_remove_note_file): New function.
15362 * coverage.h: Declare the function.
15363 * toplev.c (finalize): Clean if an error has been seen.
15365 2017-04-28 Martin Liska <mliska@suse.cz>
15367 PR gcov-profile/80031
15368 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15369 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15370 * gcov.c (read_graph_file): Read just number of blocks.
15371 * profile.c (branch_prob): Do not stream 0 flags per a basic
15374 2017-04-28 Martin Liska <mliska@suse.cz>
15376 * gcov-dump.c (tag_*): Add new argument to declarations.
15377 (dump_gcov_file): Likewise.
15378 (tag_blocks): Add and use new argument depth.
15379 (tag_arcs): Likewise.
15380 (tag_lines): Likewise.
15381 (tag_counters): Likewise.
15382 (tag_summary): Likewise.
15383 (dump_working_sets): Use depth to do a proper indentation.
15385 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15388 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15389 * symtab.c (symtab_node::debug_symtab): Move definition here.
15391 2017-04-28 Richard Biener <rguenther@suse.de>
15393 * lto-streamer.h (LTO_major_version): Bump to 7.
15395 2017-04-28 Richard Biener <rguenther@suse.de>
15397 * tree-vrp.c (assert_info): New struct.
15398 (add_assert_info): New helper.
15399 (register_edge_assert_for_2): Refactor to add asserts to a vector
15401 (register_edge_assert_for_1): Likewise.
15402 (register_edge_assert_for): Likewise.
15403 (finish_register_edge_assert_for): New helper actually registering
15404 asserts where live on edge.
15405 (find_conditional_asserts): Adjust.
15406 (find_switch_asserts): Likewise.
15407 (evrp_dom_walker::try_find_new_range): Generalize.
15408 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15410 2017-04-27 Marek Polacek <polacek@redhat.com>
15413 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15414 arg10 and arg11 to itype.
15416 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15418 * doc/extend.texi (Object Size Checking): Improve grammar.
15420 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15423 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15424 that the logic for permitting reciprocal estimates matches that
15427 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15430 * tree.c (type_cache_hasher::equal): Only compare
15431 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15432 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15433 non-aggregate element types.
15434 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15435 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15437 2017-04-27 Richard Biener <rguenther@suse.de>
15439 PR middle-end/80533
15440 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15441 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15442 keeping a reference to a trailing array.
15444 2017-04-27 Richard Biener <rguenther@suse.de>
15446 PR middle-end/80539
15447 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15448 being in loop-closed SSA form conservatively.
15449 (chrec_fold_multiply_poly_poly): Likewise.
15451 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15453 PR middle-end/79665
15454 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15455 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15457 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15460 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15461 (aarch64_function_arg_alignment): Return unsigned int again, but still
15462 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15463 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15464 Don't emit -Wpsabi note.
15465 (aarch64_function_arg_boundary): Likewise.
15466 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15469 2017-04-26 Nathan Sidwell <nathan@acm.org>
15471 * tree.h (crc32_unsigned_n): Declare.
15472 (crc32_unsigned, crc32_unsigned): Make inline.
15473 * tree.c (crc32_unsigned_bits): Replace with ...
15474 (crc32_unsigned_n): ... this.
15475 (crc32_unsigned, crc32_byte): Remove.
15476 (crc32_string): Remove unnecessary braces.
15478 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15480 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15481 * ipa-inline-analysis.c (MAX_TIME): Remove.
15482 (account_size_time): Use sreal for time.
15483 (dump_inline_summary): Update.
15484 (estimate_function_body_sizes): Update.
15485 (estimate_edge_size_and_time): Update.
15486 (estimate_calls_size_and_time): Update.
15487 (estimate_node_size_and_time): Update.
15488 (inline_merge_summary): Update.
15489 (inline_update_overall_summary): Update.
15490 (estimate_time_after_inlining): Update.
15491 (inline_read_section): Update.
15492 (inline_write_summary): Update.
15493 * ipa-inline.c (compute_uninlined_call_time): Update.
15494 (compute_inlined_call_time): Update.
15495 (recursive_inlining): Update.
15496 (inline_small_functions): Update.
15497 (dump_overall_stats): Update.
15498 * ipa-inline.h: Include sreal.h.
15499 (size_time_entry): Turn time to sreal.
15500 (inline_summary): Turn self_time nad time to sreal.
15502 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15504 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15506 (sreal::stream_out, sreal::stream_in): New.
15507 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15509 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15511 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15514 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15517 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15519 (dimode_scalar_chain::compute_convert_gain): Ditto.
15520 (dimode_scalar_chain::make_vector_copies): Ditto.
15521 (dimode_scalar_chain::convert_reg): Ditto.
15522 (dimode_scalar_chain::convert_insn): Ditto.
15523 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15524 (VI248_AVX512BW_1): New mode iterator.
15525 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15526 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15529 2017-04-25 Martin Sebor <msebor@redhat.com>
15531 PR tree-optimization/80497
15532 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15533 constants are representable in HOST_WIDE_INT.
15534 (parse_directive): Ditto.
15536 2017-04-25 Martin Sebor <msebor@redhat.com>
15539 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15540 (new_zero_array): Adjust signature.
15541 (dom_info::dom_init): Used unsigned rather that size_t.
15542 (dom_info::dom_info): Same.
15544 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15545 Jakub Jelinek <jakub@redhat.com>
15548 * config/arm/arm.c: Include gimple.h.
15549 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15550 returns negative, increment ncrn only if it returned positive.
15551 (arm_needs_doubleword_align): Return int instead of bool,
15552 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15553 members, but if there is any such non-FIELD_DECL
15554 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15555 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15556 returns negative, increment nregs only if it returned positive.
15557 (arm_setup_incoming_varargs): Likewise.
15558 (arm_function_arg_boundary): Emit -Wpsabi note if
15559 arm_needs_doubleword_align returns negative, return
15560 DOUBLEWORD_ALIGNMENT only if it returned positive.
15562 2017-04-25 Marek Polacek <polacek@redhat.com>
15565 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15566 first argument to type.
15568 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15571 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15572 type checks to test for compatibility instead of equality.
15574 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15575 Jakub Jelinek <jakub@redhat.com>
15578 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15580 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15581 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15582 the alignment computation, but return their maximum in warn_alignment.
15583 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15584 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15586 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15587 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15590 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15592 * config/arc/simdext.md (dmpyh): Fix typo.
15594 2017-04-25 Richard Biener <rguenther@suse.de>
15596 PR tree-optimization/80492
15597 * alias.c (compare_base_decls): Handle registers with asm
15598 specification conservatively.
15599 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15600 compare_base_decls returning dont-know properly.
15602 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15604 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15605 (legitimate_offset_address_p): New function.
15606 (arc_legitimate_address_p): Use above function.
15608 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15610 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15612 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15614 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15615 ACCH registers whenever they are available.
15617 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15619 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15620 double regs fix when not used.
15622 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15624 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15626 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15627 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15629 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15631 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15632 when emitting short ADD instructions.
15634 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15636 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15638 (cmpsi_cc_c_insn): Likewise.
15639 (cbranchsi4_scratch): Compute proper instruction length using
15640 compact_hreg_operand.
15641 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15643 2017-04-25 Richard Biener <rguenther@suse.de>
15645 PR middle-end/80509
15646 * passes.c (pass_manager::pass_manager): Initialize
15647 m_name_to_pass_map.
15649 2017-04-25 Richard Biener <rguenther@suse.de>
15651 PR tree-optimization/79201
15652 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15654 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15657 * config/s390/vector.md: Split MEM->GPR vector moves for
15658 non-s_operand addresses.
15660 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15663 * config/s390/predicates.md (reload_const_wide_int_operand): New
15665 * config/s390/s390.md ("movti"): Remove d/P alternative.
15666 ("movti_bigconst"): New pattern definition.
15668 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15671 * s390-protos.h (s390_expand_cs_hqi): Removed.
15672 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15673 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15674 modes as well as CCZ1mode and CCZmode.
15675 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15676 signature of s390_emit_compare_and_swap.
15677 (s390_expand_cs_hqi): Likewise, make static.
15678 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15679 compare-and-swap, in some cases.
15680 (s390_expand_cs): Wrapper function.
15681 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15683 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15684 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15685 patterns for small and large integers. Forbid symref memory operands.
15686 Move expander to s390.c. Require cc register.
15687 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15688 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15689 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15690 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15691 symref memory operands. Remove CC mode and call s390_match_ccmode
15693 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15695 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15697 * config/s390/s390.md (define_peephole2): New peephole to help
15698 combining the load-and-test pattern with volatile memory.
15700 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15702 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15703 with CCZmode for TARGET_Z196.
15705 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15707 PR rtl-optimization/80501
15708 * combine.c (make_compound_operation_int): Set subreg_code to SET
15709 even for AND with mask of the sign bit of mode.
15711 PR rtl-optimization/80500
15712 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15713 sum's initial value.
15715 2017-04-25 Julian Brown <julian@codesourcery.com>
15716 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15718 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15720 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15722 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15724 2017-04-25 Julian Brown <julian@codesourcery.com>
15725 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15727 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15728 (thunderx2t99_sha): New Reservation.
15730 2017-04-25 Julian Brown <julian@codesourcery.com>
15731 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15733 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15734 type for 1-element load.
15736 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15738 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15740 2017-04-24 Martin Jambor <mjambor@suse.cz>
15742 PR tree-optimization/80293
15743 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15744 char arrays not totally scalarizable if it is false.
15745 (analyze_all_variable_accesses): Pass correct value in the new
15746 parameter. Add a statistics counter.
15748 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15750 PR middle-end/79931
15751 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15753 2017-04-24 Richard Biener <rguenther@suse.de>
15755 PR tree-optimization/80494
15756 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15757 out for complex types.
15759 2017-04-24 Richard Biener <rguenther@suse.de>
15761 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15762 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15763 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15764 (DFS): Adjust and never fail.
15765 (sccvn_dom_walker::fail): Remove.
15766 (sccvn_dom_walker::before_dom_children): Adjust.
15767 (run_scc_vn): Likewise and never fail.
15768 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15769 (pass_fre::execute): Likewise.
15771 2017-04-24 Richard Biener <rguenther@suse.de>
15773 PR tree-optimization/79725
15774 * tree-ssa-sink.c (statement_sink_location): Return whether
15775 failure reason was zero uses. Move that check later.
15776 (sink_code_in_bb): Deal with zero uses by removing the stmt
15779 2017-04-24 Richard Biener <rguenther@suse.de>
15782 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15783 pointer-based references.
15785 2017-04-24 Richard Biener <rguenther@suse.de>
15788 * pass_manager.h (pass_manager::operator new): Remove.
15789 (pass_manager::operator delete): Likewise.
15790 * passes.c (pass_manager::operator new): Remove.
15791 (pass_manager::operator delete): Likewise.
15792 (pass_manager::pass_manager): Zero individual pass members.
15794 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15797 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15798 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15799 Check "XEXP (src, 1)" operand here.
15800 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15801 Check "XEXP (src, 1)" operand here.
15802 (dimode_scalar_chain::make_vector_copies): Detect count register
15803 of a shift instruction. Zero extend count register from QImode
15804 to DImode to satisfy vector shift pattern count operand predicate.
15805 Substitute vector shift count operand with a DImode copy.
15806 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15809 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15811 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15812 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15813 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15814 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15815 (UNSPEC_NOREX_MEM): Remove definition.
15817 2017-04-21 Richard Biener <rguenther@suse.de>
15819 PR tree-optimization/79547
15820 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15821 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15822 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15823 without any constraints.
15825 2017-04-21 Richard Biener <rguenther@suse.de>
15827 PR tree-optimization/78847
15828 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15830 2017-04-21 Richard Biener <rguenther@suse.de>
15832 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15833 (build_distinct_type_copy): Likewise.
15834 (build_variant_type_copy): Likewise.
15835 * tree.c (build_qualified_type): Pass down mem-stat info.
15836 (build_distinct_type_copy): Likewise.
15837 (build_variant_type_copy): Likewise.
15839 2017-04-21 Richard Biener <rguenther@suse.de>
15841 PR tree-optimization/80237
15842 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15844 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15845 for a simplified result.
15847 2016-04-21 Richard Biener <rguenther@suse.de>
15849 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15850 sth as strict as a simple_iv but a chrec without symbols and an
15851 operand defined in the loop we are peeling (and not some subloop).
15852 (propagate_constants_for_unrolling): Propagate all constants.
15854 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15857 * config/i386/i386.c (print_reg): Remove assert for disalowed
15858 regno values, call output_operand_lossage instead.
15860 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15863 * config/i386/constraints.md (Yc): New register constraint.
15864 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15865 Use Yc constraint for alternative 2 of operand 0. Remove
15866 preferred_for_speed attribute.
15868 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15870 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15871 lastprivate clauses in SIMT case.
15873 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15875 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15877 2017-04-20 Richard Biener <rguenther@suse.de>
15879 PR tree-optimization/57796
15880 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15881 as N scalar stores.
15882 (vect_model_load_cost): Cost gathers as N scalar loads.
15884 2017-04-20 Richard Biener <rguenther@suse.de>
15886 * ggc-page.c (ggc_allocated_p): Rename to ...
15887 (safe_lookup_page_table_entry): ... this and return the lookup
15889 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15891 2017-04-20 Richard Biener <rguenther@suse.de>
15893 PR tree-optimization/80453
15894 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15895 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15896 from the conditions.
15897 (vn_phi_eq): Pass them down.
15898 (vn_phi_lookup): Record them.
15899 (vn_phi_insert): Likewise.
15901 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15903 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15904 uninitialized variable warning to avoid buffer overrun.
15906 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15909 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15910 is suppressed for '{ 0 }' in C.
15912 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15914 * BASE-VER: Set to 8.0.0.
15916 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15918 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15919 priority .init_array and .fini_array section with SECTION_NOTYPE
15922 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15924 PR middle-end/80423
15925 * tree.h (build_array_type): Add typeless_storage default argument.
15926 * tree.c (type_cache_hasher::equal): Also compare
15927 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15928 (build_array_type): Add typeless_storage argument, set
15929 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15931 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15932 (build_array_type): Likewise. Add typeless_storage argument.
15934 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15935 Jakub Jelinek <jakub@redhat.com>
15937 PR tree-optimization/80426
15938 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15939 operation on symbolic operands, also compute the overflow for the
15940 invariant part when the operation degenerates into a negation.
15942 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15945 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15946 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15949 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15951 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15954 * config/avr/avr.c (tree.h): Include it.
15955 (cgraph.h): Include it.
15956 (avr_encode_section_info): Don't warn for uninitialized progmem
15957 variable if it's just an alias.
15959 2017-04-19 Richard Biener <rguenther@suse.de>
15962 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15963 when needed by AutoPGO.
15965 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15968 * doc/lto.texi: Remove an extra 'that'.
15970 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15972 PR rtl-optimization/80429
15973 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15974 are only used in debug insns.
15976 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15977 Vladimir Makarov <vmakarov@redhat.com>
15979 * config/sparc/predicates.md (input_operand): Add comment. Return
15980 true for any memory operand when LRA is in progress.
15981 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15983 2017-04-18 Jeff Law <law@redhat.com>
15986 * mips.md ({return,simple_return}_internal): Do not overwrite
15989 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15991 PR tree-optimization/80443
15992 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15993 instead of adding 1, subtract -1 and similarly instead of subtracting
15996 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15998 PR rtl-optimization/80357
15999 * haifa-sched.c (tmp_bitmap): New variable.
16000 (model_recompute): Handle duplicate use records.
16001 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16002 (free_global_sched_pressure_data): Free it.
16004 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16007 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16008 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16009 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16010 instead of SYSTEM_HEADER_DIR.
16012 2017-04-18 Jeff Law <law@redhat.com>
16014 PR middle-end/80422
16015 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16016 predecessors after walking up the insn chain.
16018 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16021 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16022 sizetype type into debug info.
16024 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16027 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16028 unneeded test for TARGET_UPPER_REGS_SF.
16029 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16031 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16034 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16035 instead of gsi_after_labels.
16037 2017-04-18 Jeff Law <law@redhat.com>
16039 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16043 2017-04-13 Jeff Law <law@redhat.com>
16044 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16045 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16047 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16050 * config/avr/avr.c (intl.h): Include it.
16051 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16053 2017-04-18 Martin Liska <mliska@suse.cz>
16055 PR gcov-profile/78783
16056 * gcov-tool.c (gcov_output_files): Validate that destination
16057 file is either removed by the tool or by a user.
16059 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16060 Guy Benyei <guybe@mellanox.com>
16062 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16063 block, and do not negate it, the stored id is already negative.
16065 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16067 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16069 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16072 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16073 masks of options that should be turned off if the VSX vector
16074 options are turned off.
16075 (OTHER_P8_VECTOR_MASKS): Likewise.
16076 (OTHER_VSX_VECTOR_MASKS): Likewise.
16077 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16078 rs6000_disable_incompatible_switches to validate no type switches
16080 (rs6000_incompatible_switch): New function to disallow turning on
16081 other vector options if -mno-vsx, -mno-power8-vector, or
16082 -mno-power9-vector are specified.
16084 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16086 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16088 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16090 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16091 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16092 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16093 (ARG_POINTER_CFA_OFFSET): Likewise.
16095 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16097 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16098 conditions to take advantage of various optimizations.
16100 2017-04-13 Jeff Law <law@redhat.com>
16102 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16103 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16104 (zero_extendsidi2_dext): Likewise.
16106 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16109 * fold-const.c (fold_ternary_loc): Revert
16110 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16113 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16115 PR rtl-optimization/80343
16116 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16117 new scratch pseudo.
16119 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16122 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16123 to ubsan_encode_value.
16125 2017-04-13 Jeff Law <law@redhat.com>
16127 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16128 appearing in DEBUG_INSNs.
16130 2017-04-13 Martin Liska <mliska@suse.cz>
16132 PR gcov-profile/80413
16133 * gcov-io.c (gcov_write_string): Copy to buffer just when
16134 allocated size is greater than zero.
16136 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16139 * dwarf2out.c (decls_for_scope): Ignore declarations of
16140 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16142 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16145 * ipa-visibility.c (non_local_p): Fix typos.
16146 (localize_node): When localizing symbol in same comdat group,
16147 dissolve the group only when we know external symbols are going
16149 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16151 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16153 PR tree-optimization/79390
16154 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16155 order does not result in usable sequence, retry with reversed operand
16161 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16162 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16163 op0 instead of fold_convert_loc (loc, type, arg0).
16165 2017-04-12 Jeff Law <law@redhat.com>
16167 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16168 has a delay slot in the generated code.
16170 * config/cris/cris.md (cris_preferred_reload_class): Return
16171 GENNONACR_REGS rather than GENERAL_REGS.
16173 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16176 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16177 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16178 signedness of the result type.
16180 2017-04-12 Richard Biener <rguenther@suse.de>
16181 Jeff Law <law@redhat.com>
16183 PR tree-optimization/80359
16184 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16185 trim stores to TARGET_MEM_REFs.
16187 2017-04-12 Richard Biener <rguenther@suse.de>
16189 PR tree-optimization/79390
16190 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16191 threading case even more.
16193 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16196 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16197 for quad_address_p for TImode, instead of just not indexed_address.
16199 2017-04-12 Richard Biener <rguenther@suse.de>
16200 Bernd Edlinger <bernd.edlinger@hotmail.de>
16202 PR middle-end/79671
16203 * alias.c (component_uses_parent_alias_set_from): Handle
16204 TYPE_TYPELESS_STORAGE.
16205 (get_alias_set): Likewise.
16206 * tree-core.h (tree_type_common): Add typeless_storage flag.
16207 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16208 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16209 for types containing members with TYPE_TYPELESS_STORAGE.
16210 (place_field): Likewise.
16211 (layout_type): Likewise for ARRAY_TYPE.
16212 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16213 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16214 TYPE_TYPELESS_STORAGE.
16215 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16217 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16220 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16221 first argument to type.
16223 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16227 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16228 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16229 (rs6000_expand_binop_builtin): Likewise.
16230 (rs6000_expand_ternop_builtin): Likewise; also add missing
16231 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16232 vshasigma built-ins.
16233 * doc/extend.texi: Document that vec_xxpermdi's third argument
16234 must be a constant.
16236 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16238 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16239 Use shift_const cost parameter when calculating gain of STV shifts.
16241 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16243 PR rtl-optimization/70478
16244 * lra-constraints.c (process_alt_operands): Check memory for
16245 disfavoring memory insn operand.
16247 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16249 PR middle-end/80100
16250 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16251 left shift in unsigned HOST_WIDE_INT type.
16253 PR rtl-optimization/80385
16254 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16255 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16258 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16259 if they have any depend clauses.
16261 2017-04-11 Martin Liska <mliska@suse.cz>
16264 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16265 * ipa-split.c (split_function): Create a local comdat symbol
16266 if caller is in a comdat group.
16268 2017-04-11 Martin Liska <mliska@suse.cz>
16271 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16274 2017-04-11 Martin Sebor <msebor@redhat.com>
16276 PR middle-end/80364
16277 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16278 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16280 (directive::set_width, directive::set_precision, format_character):
16282 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16285 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16288 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16289 conflict, set target->arch_name instead of target->cpu_name.
16291 2017-04-11 Richard Biener <rguenther@suse.de>
16293 PR tree-optimization/80374
16294 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16295 build_zero_cst, remove fold_convertible_p check again.
16297 2017-04-11 Martin Liska <mliska@suse.cz>
16300 * ubsan.c (instrument_object_size): Do not instrument register
16303 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16306 * config/i386/i386-builtin-types.def
16307 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16308 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16309 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16310 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16311 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16312 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16313 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16314 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16315 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16316 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16317 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16318 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16319 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16320 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16321 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16322 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16323 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16324 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16325 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16326 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16327 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16328 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16329 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16330 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16331 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16332 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16333 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16334 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16335 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16336 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16337 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16338 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16339 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16340 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16341 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16342 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16343 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16344 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16345 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16346 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16347 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16348 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16349 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16350 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16351 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16352 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16354 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16355 flag to second_arg_count, handle 4 argument function type _COUNT
16356 aliases, handle second_arg_count on second argument rather than last.
16358 2017-04-10 Jeff Law <law@redhat.com>
16360 PR tree-optimization/80374
16361 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16362 record anything if we can not convert integer_zero_node to the
16365 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16368 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16369 Enhance special handling given to the TARGET_P9_MINMAX option in
16370 relation to certain other options.
16372 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16374 PR tree-optimization/80153
16375 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16376 remove POINTER_PLUS_EXPR's base part directly, rather than through
16379 2017-04-10 Richard Biener <rguenther@suse.de>
16380 Bin Cheng <bin.cheng@arm.com>
16382 PR tree-optimization/80153
16383 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16384 the first element of pointer type aff_tree. Build result expr in
16386 (add_elt_to_tree): Convert to type unconditionally. Remove other
16387 fold_convert calls.
16388 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16389 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16391 2017-04-10 Richard Biener <rguenther@suse.de>
16393 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16396 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16398 PR rtl-optimization/70478
16399 * lra-constraints.c (curr_small_class_check): New.
16400 (update_and_check_small_class_inputs): New.
16401 (process_alt_operands): Update curr_small_class_check. Disfavor
16402 alternative insn memory operands. Check available regs for small
16405 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16408 * config/mips/mips.opt (-mvirt): Update description.
16409 * doc/invoke.texi (-mvirt): Likewise.
16411 2017-04-10 Richard Biener <rguenther@suse.de>
16413 PR middle-end/80362
16414 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16415 looking for NEGATE_EXPR in -A / -B to A / B folding.
16417 2017-04-10 Martin Liska <mliska@suse.cz>
16419 PR gcov-profile/80224
16420 * gcov.c (print_usage): Fix usage string.
16421 (get_gcov_intermediate_filename): Remove.
16422 (output_gcov_file): Use both for normal and intermediate format.
16423 (generate_results): Do not initialize special file for
16424 intermediate format.
16426 2017-04-10 Richard Biener <rguenther@suse.de>
16428 PR tree-optimization/80304
16429 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16432 2017-04-10 Nathan Sidwell <nathan@acm.org>
16435 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16436 (rs6000_init_builtins): Use it.
16438 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16440 * config/arm/arm.md (<mrc>): Add mode to SET source.
16441 (<mrrc>): Likewise.
16443 2017-04-10 Richard Biener <rguenther@suse.de>
16445 PR middle-end/80344
16446 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16448 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16451 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16452 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16453 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16454 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16455 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16456 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16457 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16458 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16459 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16460 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16461 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16462 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16463 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16464 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16465 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16466 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16467 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16468 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16469 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16470 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16471 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16472 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16473 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16475 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16477 PR rtl-optimization/70478
16478 * lra-constraints.c: Reverse the last patch.
16480 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16482 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16483 Add comment for WCHAR_T.
16485 2017-04-08 Martin Liska <mliska@suse.cz>
16488 2017-04-07 Martin Liska <mliska@suse.cz>
16491 * ipa-split.c (split_function): Add function part to a same comdat
16494 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16497 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16499 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16501 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16502 (vec_pairing): ...to this.
16503 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16504 (rs6000_sched_init): Adjust for name change.
16505 (struct rs6000_sched_context): Likewise.
16506 (rs6000_init_sched_context): Likewise.
16507 (rs6000_set_sched_context): Likewise.
16509 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16515 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16517 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16518 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16519 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16521 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16523 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16525 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16527 PR rtl-optimization/70703
16528 * ira-color.c (update_conflict_hard_regno_costs): Use
16529 int64_t instead of HOST_WIDE_INT.
16531 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16533 PR rtl-optimization/70478
16534 * lra-constraints.c (process_alt_operands): Disfavor alternative
16535 insn memory operands.
16537 2017-04-07 Jeff Law <law@redhat.com>
16539 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16540 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16542 2017-04-07 Martin Liska <mliska@suse.cz>
16545 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16546 Show error message instead of an ICE.
16548 2017-04-07 Martin Liska <mliska@suse.cz>
16551 * ipa-split.c (split_function): Add function part to a same comdat
16554 2017-04-07 Richard Biener <rguenther@suse.de>
16556 PR middle-end/80341
16557 * tree.c (get_unwidened): Also handle ! for_type case for
16559 * convert.c (do_narrow): Split out from ...
16560 (convert_to_integer_1): ... here. Do not pass final truncation
16561 type to get_unwidened for TRUNC_DIV_EXPR.
16563 2017-04-07 Richard Biener <rguenther@suse.de>
16565 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16567 (aff_combination_const): Adjust.
16568 (aff_combination_scale): Likewise.
16569 (aff_combination_add_elt): Likewise.
16570 (aff_combination_add_cst): Likewise.
16571 (aff_combination_convert): Likewise.
16572 (add_elt_to_tree): Likewise. Remove unused argument.
16573 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16575 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16577 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16579 * config/arm/arm.c (arm_default_short_enums): Use
16580 ARM_DEFAULT_SHORT_ENUMS.
16581 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16583 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16586 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16587 members with redundant out-of-class redeclaration.
16589 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16592 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16593 * config/i386/i386.md (*zero_extendsidi2):
16594 Add (?*x,*x) and (?*v,*v) alternatives.
16596 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16599 * config/i386/i386.c (ix86_expand_builtin)
16600 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16601 mode from insn data. Convert operands to insn operand mode.
16602 Copy operands that don't satisfy insn predicate to a register.
16604 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16606 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16609 2017-04-06 Richard Biener <rguenther@suse.de>
16611 PR tree-optimization/80334
16612 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16613 preserve alignment of accesses.
16615 2017-04-06 Richard Biener <rguenther@suse.de>
16617 PR tree-optimization/80262
16618 * tree-sra.c (build_ref_for_offset): Preserve address-space
16620 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16621 Drop useless address-space information on MEM_REF offsets.
16623 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16625 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16627 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16629 PR rtl-optimization/70703
16630 * ira-color.c (update_conflict_hard_regno_costs): Use
16631 HOST_WIDE_INT instead of long.
16633 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16636 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16637 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16639 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16640 for x86_64 target. Handle -m3dnowa option.
16642 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16644 PR rtl-optimization/70703
16645 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16646 (update_conflict_hard_regno_costs): Use long instead of unsigned
16647 arithmetic for cost calculation.
16649 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16650 Bernd Edlinger <bernd.edlinger@hotmail.de>
16653 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16656 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16659 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16660 ptr_mode with Pmode throughout.
16661 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16662 into probe_stack_range and use DImode.
16664 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16667 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16668 call_eh_return is true.
16670 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16672 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16673 Initialize last_match_fntype_index.
16675 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16678 * tree-nvr.c: Include internal-fn.h.
16679 (pass_return_slot::execute): Ignore internal calls without
16682 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16683 Richard Biener <rguenther@suse.de>
16686 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16687 captures used multiple times, except for the last use.
16688 * generic-match-head.c: Include gimplify.h.
16690 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16692 PR tree-optimization/79390
16693 * target.h (struct noce_if_info): Declare.
16694 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16695 * target.def (noce_conversion_profitable_p): New target hook.
16696 * ifcvt.h (struct noce_if_info): New type, moved from ...
16697 * ifcvt.c (struct noce_if_info): ... here.
16698 (noce_conversion_profitable_p): Renamed to ...
16699 (default_noce_conversion_profitable_p): ... this. No longer
16701 (noce_try_store_flag_constants, noce_try_addcc,
16702 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16703 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16704 instead of noce_conversion_profitable_p.
16705 * config/i386/i386.c: Include ifcvt.h.
16706 (ix86_option_override_internal): Don't override
16707 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16708 (ix86_noce_conversion_profitable_p): New function.
16709 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16710 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16711 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16712 * doc/tm.texi: Regenerated.
16714 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16716 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16719 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16722 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16723 instructions for small multiply cores.
16725 2017-04-04 Jeff Law <law@redhat.com>
16727 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16729 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16730 that are not set by the loop over the elements.
16732 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16735 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16736 int mode, convert_modes it to mode as unsigned, otherwise use
16737 lowpart_subreg to mode rather than SImode.
16738 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16739 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16740 Use DImode instead of SImode for the shift count operand.
16741 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16744 2017-04-04 Richard Biener <rguenther@suse.de>
16746 PR middle-end/80281
16747 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16748 arithmetic done for the negate or the plus. Simplify.
16749 (A - (-B) -> A + B): Likewise.
16750 * fold-const.c (split_tree): Make sure to not negate pointers.
16752 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16754 PR rtl-optimization/60818
16755 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16756 a compare of comparisons with the thing compared if this results
16757 in a different machine mode.
16759 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16761 * alias.c (base_alias_check): Fix typo in comment.
16762 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16763 * cgraphunit.c (symbol_table::compile): Likewise.
16764 * collect2.c (maybe_run_lto_and_relink): Likewise.
16765 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16766 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16767 * config/avr/avr.c (avr_map_op_t): Likewise.
16768 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16769 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16770 * config/epiphany/epiphany.md (movcc): Likewise.
16771 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16772 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16774 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16775 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16776 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16777 * config/sh/sh.c (sh_rtx_costs): Likewise.
16778 * fold-const.c (fold_truth_andor): Likewise.
16779 * genautomata.c (collapse_flag): Likewise.
16780 * gengtype.h (struct type::u::s): Likewise.
16781 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16782 * input.c (FORMAT_AMOUNT): Likewise.
16783 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16784 (known_aggs_to_agg_replacement_list): Likewise.
16785 * ipa-inline-analysis.c: Likewise.
16786 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16787 * ipa-polymorphic-call.c
16788 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16789 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16790 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16792 * modulo-sched.c (apply_reg_moves): Likewise.
16793 * omp-expand.c (build_omp_regions_1): Likewise.
16794 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16795 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16796 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16797 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16798 * value-prof.c: Likewise.
16799 * var-tracking.c (val_reset): Likewise.
16801 2017-04-03 Richard Biener <rguenther@suse.de>
16803 PR tree-optimization/80275
16804 * fold-const.c (split_address_to_core_and_offset): Handle
16807 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16809 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16810 descriptors is at least equal to that of functions.
16812 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16814 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16816 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16819 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16820 (mov<IMOD4:mode>): New expander.
16821 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16823 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16825 PR rtl-optimization/79405
16826 * fwprop.c (propagations_left): New variable.
16827 (forward_propagate_into): Decrement it.
16828 (fwprop_init): Initialize it.
16829 (fw_prop): If the variable has reached zero, stop propagating.
16830 (fwprop_addr): Ditto.
16832 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16835 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16836 a FUNCTION_DECL, pass it as decl instead of origin to
16839 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16841 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16844 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16847 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16848 TARGET_VSX_SMALL_INTEGER.
16850 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16852 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16853 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16855 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16857 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16858 extraction from odd-numbered MSA register.
16860 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16862 PR middle-end/80173
16863 * expmed.c (store_bit_field_1): Don't attempt to create
16864 a word subreg out of hard registers wider than word if they
16865 have HARD_REGNO_NREGS of 1 for their mode.
16867 PR middle-end/80163
16868 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16869 conversions to integer types wider than word and pointer.
16872 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16873 (rtx_equal_for_cselib_p): Pass 0 to it.
16874 * cselib.c (cselib_hasher::equal): Likewise.
16875 (rtx_equal_for_cselib_1): Add depth argument. If depth
16876 is 128, don't look up VALUE locs and punt. Increment
16877 depth in recursive calls when walking VALUE locs.
16879 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16881 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16882 (make_gcov_file_name): Use the canonical path name for generating
16884 (read_line): Fix handling of files with ascii null bytes.
16886 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16888 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16889 to initialise a vector register instead
16890 of using a const_int.
16892 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16894 PR translation/80189
16895 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16896 diagnostic messages.
16898 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16901 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16902 (dfp_diex_<mode>): Update mode of operand 1.
16903 * doc/extend.texi (dxex, dxexq): Document change to return type.
16904 (diex, diexq): Document change to argument type.
16906 2017-03-30 Martin Jambor <mjambor@suse.cz>
16909 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16910 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16911 it reflects the signature changes performed at the callee side.
16912 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16913 to cgraph_build_function_type_skip_args.
16914 (build_function_decl_skip_args): Adjust call to the above function.
16916 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16919 * config/i386/sse.md
16920 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16921 register as dest whenever it is a MEM not rtx_equal_p to the
16922 corresponding dup operand, and when forcing into reg move the
16923 reg into the memory afterwards.
16924 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16925 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16926 for the force_reg mode.
16927 (avx512vl_vextractf128<mode>): Use register as dest either
16928 always when a MEM, or when it is a MEM not rtx_equal_p to the
16929 corresponding dup operand, or even not when it is a CONST_VECTOR
16930 depending on the mode and lo vs. hi.
16931 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16933 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16934 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16935 Likewise. Require that operands[2] is even.
16936 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16937 Remove extraneous parens. Require that operands[2] is a multiple
16939 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16940 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16941 disallow memory then.
16943 2017-03-30 Richard Biener <rguenther@suse.de>
16945 PR tree-optimization/77498
16946 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16947 to non-constants over backedges.
16949 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16951 PR rtl-optimization/80233
16952 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16953 as last_combined_insn. Do not test for BARRIER_P separately.
16955 2017-03-29 Andreas Schwab <schwab@suse.de>
16958 * calls.c (prepare_call_address): Convert funexp to Pmode before
16959 copying to temp reg.
16961 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16963 PR tree-optimization/80158
16964 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16965 Handle possible future case of more than one alternate
16967 (replace_rhs_if_not_dup): Likewise.
16968 (replace_one_candidate): Likewise.
16970 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16972 PR rtl-optimization/80193
16973 * ira.c (ira): Do not check allocation for LRA.
16975 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16977 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16978 (nvptx_output_simt_exit): Declare.
16979 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16980 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16981 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16982 (nvptx_declare_function_name): ...here. Emit declaration of local
16983 memory space buffer for omp_simt_enter insn.
16984 (nvptx_output_unisimt_switch): New.
16985 (nvptx_output_softstack_switch): New.
16986 (nvptx_output_simt_enter): New.
16987 (nvptx_output_simt_exit): New.
16988 * config/nvptx/nvptx.h (struct machine_function): New fields
16989 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16990 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16991 (UNSPECV_SIMT_EXIT): Ditto.
16992 (omp_simt_enter_insn): New insn.
16993 (omp_simt_enter): New expansion.
16994 (omp_simt_exit): New insn.
16995 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16997 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16998 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16999 (expand_GOMP_SIMT_EXIT): New.
17000 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17001 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17002 (GOMP_SIMT_EXIT): Ditto.
17003 * target-insns.def (omp_simt_enter): New insn.
17004 (omp_simt_exit): Ditto.
17005 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17007 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17008 (lower_rec_input_clauses): Likewise.
17009 (lower_lastprivate_clauses): Handle SIMT privatization.
17011 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17012 (ompdevlow_adjust_simt_enter): New.
17013 (find_simtpriv_var_op): New.
17014 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17015 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17017 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17018 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17019 (copy_decl_for_dup_finish): Ditto.
17021 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17023 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17026 * config/i386/i386.c (ix86_option_override_internal): Always
17027 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17029 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17031 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17033 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17035 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17036 mark new edge's irreducible flag accordign to it.
17037 (vect_do_peeling): Check loop preheader edge's irreducible flag
17038 and pass it to function slpeel_add_loop_guard.
17040 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17042 PR tree-optimization/80218
17043 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17044 Update block frequencies and counts.
17046 2017-03-28 Richard Biener <rguenther@suse.de>
17048 PR tree-optimization/78644
17049 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17050 of a simplification result we may not use it at all.
17052 2017-03-28 Richard Biener <rguenther@suse.de>
17055 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17056 without arguments, generate default definition of a SSA name.
17058 2017-03-28 Richard Biener <rguenther@suse.de>
17060 PR middle-end/80222
17061 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17062 TYPE_REF_CAN_ALIAS_ALL references.
17063 * fold-const.c (fold_indirect_ref_1): Likewise.
17065 2017-03-28 Martin Liska <mliska@suse.cz>
17068 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17069 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17071 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17072 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17074 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17075 (EXTRA_SPECS): Define.
17076 (SUBTARGET_EXTRA_SPECS): Likewise.
17077 (SUBTARGET_CPP_SPEC): Likewise.
17078 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17079 SUBTARGET_EXTRA_SPECS.
17080 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17082 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17084 * config/arc/simdext.md (vst64_insn): Update pattern.
17085 (vld32wh_insn): Likewise.
17086 (vld32wl_insn): Likewise.
17087 (vld64_insn): Likewise.
17088 (vld32_insn): Likewise.
17090 2017-03-28 Marek Polacek <polacek@redhat.com>
17093 * fold-const.c (fold_comparison): Use protected_set_expr_location
17094 instead of SET_EXPR_LOCATION.
17096 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17098 * tree.c (add_expr): Avoid name lookup warning.
17100 2017-03-27 Jeff Law <law@redhat.com>
17102 PR tree-optimization/80216
17103 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17104 function name. Limit recursion depth.
17105 (record_temporary_equivalences): Corresponding changes.
17107 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17109 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17112 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17115 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17117 * cfgcleanup.c (reg_note_cfa_p): New array.
17118 (insns_have_identical_cfa_notes): New function.
17119 (old_insns_match_p): Don't cross-jump in between /f
17120 and non-/f instructions. If both i1 and i2 are frame related,
17121 verify all CFA notes, their order and content.
17123 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17126 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17127 HImode and SImode with zero extend to DImode to one insn.
17128 (bswap<mode>2_extenddi): Likewise.
17129 (bswapsi2_extenddi): Likewise.
17130 (bswaphi2_extendsi): Likewise.
17131 (bswaphi2): Combine bswap HImode and SImode into one insn.
17132 Separate memory insns from swapping register.
17133 (bswapsi2): Likewise.
17134 (bswap<mode>2): Likewise.
17135 (bswaphi2_internal): Delete, no longer used.
17136 (bswapsi2_internal): Likewise.
17137 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17138 store, and gpr<-gpr swap insns.
17139 (bswap<mode>2_store): Likewise.
17140 (bswaphi2_reg): Register only splitter, combine with the splitter.
17141 (bswaphi2 splitter): Likewise.
17142 (bswapsi2_reg): Likewise.
17143 (bswapsi2 splitter): Likewise.
17144 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17145 the insns into load, store, and register/register insns.
17146 (bswapdi2_ldbrx): Likewise.
17147 (bswapdi2_load): Likewise.
17148 (bswapdi2_store): Likewise.
17149 (bswapdi2_reg): Likewise.
17151 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17153 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17154 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17156 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17159 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17161 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17162 special handling for target option conflicts between dform
17163 options (-mpower9-dform, -mpower9-dform-vector,
17164 -mpower9-dform-scalar) and -mno-direct-move.
17166 2017-03-27 Richard Biener <rguenther@suse.de>
17168 PR tree-optimization/80181
17169 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17171 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17173 * config/arc/predicates.md (move_double_src_operand): Replace the
17174 call to move_double_src_operand with a call to address_operand.
17176 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17178 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17179 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17180 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17182 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17184 * config/arc/predicates.md (long_immediate_loadstore_operand):
17185 Consider scaled addresses cases.
17187 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17189 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17190 restored when in interrupt.
17191 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17192 doesn't have delay slot.
17194 2017-03-27 Richard Biener <rguenther@suse.de>
17197 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17198 inlined thunk clones.
17200 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17203 * asan.c (instrument_derefs): Copy over last operand from
17204 original COMPONENT_REF to the new COMPONENT_REF with
17205 DECL_BIT_FIELD_REPRESENTATIVE.
17206 * ubsan.c (instrument_object_size): Likewise.
17208 2017-03-27 Richard Biener <rguenther@suse.de>
17210 PR tree-optimization/80170
17211 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17212 sure DR/SCEV didnt fold in constants we do not see when looking
17213 at the reference base alignment.
17215 2017-03-27 Richard Biener <rguenther@suse.de>
17217 PR middle-end/80171
17218 * gimple-fold.c (fold_ctor_reference): Properly guard against
17219 NULL return value from canonicalize_constructor_val.
17221 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17224 * config/i386/i386.c (ix86_expand_builtin)
17225 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17226 flags reg setting and flags reg using instructions.
17227 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17228 clobbering instructions to zero extend op2.
17230 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17232 * doc/install.texi (Configuration) <--with-aix-soname>:
17233 Update link to AIX ld.
17235 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17237 PR rtl-optimization/80160
17238 PR rtl-optimization/80159
17239 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17240 reg_alternate_class into account.
17242 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17245 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17246 to consider in curr_insn_transform.
17248 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17250 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17251 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17252 and emit_mode_inner.
17254 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17256 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17257 argument to the overloaded builtin variants. Use the new flag to
17258 deprecate certain builtin variants.
17259 * config/s390/s390-builtin-types.def: Add new builtin types.
17260 * config/s390/s390-builtins.h: Support new flags field for
17261 overloaded builtins.
17262 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17263 (s390_macro_to_expand): Enable vector float data type.
17264 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17265 builtins by incrementing the __VEC__ version number.
17266 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17268 (s390_resolve_overloaded_builtin): Emit error messages depending
17269 on the builtin flags.
17270 * config/s390/s390.c (s390_expand_builtin): Support additional
17271 flags argument. Change error message to match the messages
17272 emitted in s390-c.c.
17273 * config/s390/s390.md: New UNSPEC_* constants.
17274 (op_type): Add new instruction types.
17275 * config/s390/vecintrin.h: Add new builtins and test data class
17277 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17278 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17279 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17280 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17281 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17282 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17284 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17285 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17286 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17287 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17289 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17290 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17291 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17292 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17294 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17295 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17296 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17297 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17298 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17299 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17300 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17302 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17303 ("vec_scatter_element<V_HW_4:mode>_DI")
17304 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17305 ("vec_fpint<mode>", "vflls")
17306 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17307 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17308 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17309 ("*vec_cmphe<mode>_cc"): ... these.
17311 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17312 mode constant instead of magic value.
17314 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17316 * config/s390/s390.c (s390_expand_vec_compare): Support other
17317 vector floating point modes than just V2DF.
17318 (s390_expand_vcond): Likewise.
17319 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17320 (s390_cannot_change_mode_class): Prevent mode changes between TF
17321 and V1TF in vector registers.
17322 * config/s390/s390.md (DF, SF): New mode attributes.
17323 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17324 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17325 SFmode support for VRs.
17326 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17328 (VFT, VF_HW): New mode iterators.
17329 (vw, sdx): New mode attributes.
17330 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17331 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17332 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17333 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17334 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17335 also the new vector floating point modes. Renaming to ...
17337 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17338 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17339 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17340 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17341 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17342 ("vec_unordered<mode>"): ... these.
17344 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17345 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17346 ("*vec_extendv2df"): New insn definitions.
17348 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17350 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17351 ("mulditi3_2", "*muldi3_sign"): New patterns.
17352 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17353 rename the pattern definition.
17355 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17357 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17359 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17361 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17363 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17364 instruction if possible.
17365 * config/s390/vector.md (vec_halfnumelts): New mode
17367 ("*vec_vllezlf<mode>"): New pattern.
17369 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17371 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17372 ("popcountv4si2", "popcountv2di2"): Rename to ...
17373 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17374 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17376 ("popcount<mode>2_vxe"): New pattern.
17378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17380 * common/config/s390/s390-common.c (processor_flags_table): Add
17382 * config.gcc: Add arch12.
17383 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17384 Default to arch12 for unknown CPU model numbers.
17385 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17386 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17387 PROCESSOR_max sanity check.
17388 * config/s390/s390-opts.h (enum processor_type): Add
17390 * config/s390/s390.c (processor_table): Add arch12.
17391 (s390_expand_builtin): Add check for B_VXE flag.
17392 (s390_issue_rate): Add PROCESSOR_ARCH12.
17393 (s390_get_sched_attrmask): Likewise.
17394 (s390_get_unit_mask): Likewise.
17395 (s390_sched_score): Enable z13 scheduling for arch12.
17396 (s390_sched_reorder): Likewise.
17397 (s390_sched_variable_issue): Likewise.
17398 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17400 (s390_tune_attr): Use z13 scheduling also for arch12.
17401 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17402 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17403 (TARGET_VXE_P): New macros.
17404 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17405 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17406 * config/s390/s390.opt: Add arch12 as processor_type.
17408 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17410 * config/s390/s390.md
17411 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17412 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17413 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17415 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17416 Rename expanders to ...
17418 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17419 ("fixuns_truncdddi2_emu"): ... these.
17421 ("fixuns_trunc<mode>si2_emu"): New expander.
17423 ("*fixuns_truncdfdi2_z13"): Rename to ...
17424 ("*fixuns_truncdfdi2_vx"): ... this.
17426 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17428 * config/s390/2964.md: Remove the single element vector compare
17429 instructions which are no longer used.
17430 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17432 (s390_canonicalize_comparison): Remove handling of DFmode
17434 (s390_expand_vec_compare_scalar): Remove function.
17435 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17436 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17438 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17440 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17442 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17444 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17445 will anyway by matched by mov<mode>_64dfp.
17447 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17449 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17450 vlef/vstef. Add missing operand to vleif.
17452 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17454 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17455 pair for all vector types with 64 bit elements.
17456 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17457 * config/s390/vector.md (V_HW_64): ... here.
17458 (V_128_NOSINGLE): New mode iterator.
17459 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17460 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17461 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17462 ("*vec_load_pairv2di"): Change to ...
17463 ("*vec_load_pair<mode>"): ... this one.
17465 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17467 * config/s390/constraints.md: Add comments.
17468 (jKK): Reject element sizes > 8 bytes.
17469 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17471 * config/s390/s390.md: Add the s_operand checks formerly in
17472 s390_split_ok_p to various splitters where they are still
17474 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17475 for 128 bit vectors. Plus two splitters.
17477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17479 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17482 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17485 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17486 error if the boundary argument is not constant.
17488 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17490 PR rtl-optimization/80112
17491 * loop-doloop.c (doloop_condition_get): Don't check condition
17492 if cmp isn't SET with IF_THEN_ELSE src.
17494 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17496 PR tree-optimization/80158
17497 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17498 replacing a candidate statement, also replace it for the
17499 candidate's alternate interpretation.
17500 (replace_rhs_if_not_dup): Likewise.
17501 (replace_one_candidate): Likewise.
17503 2017-03-24 Richard Biener <rguenther@suse.de>
17505 PR tree-optimization/80167
17506 * graphite-isl-ast-to-gimple.c
17507 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17509 (translate_isl_ast_to_gimple::get_rename): Likewise.
17511 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17513 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17514 handling of certain combinations of target options, including the
17515 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17516 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17518 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17521 * config/arm/arm.md (*load_multiple): Add reload_completed to
17522 matching condition.
17524 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17525 Richard Biener <rguenth@suse.de>
17527 PR tree-optimization/79908
17528 PR tree-optimization/80136
17529 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17530 been cast away, gimplify_and_add suffices.
17532 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17534 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17536 2017-03-23 Richard Biener <rguenther@suse.de>
17538 PR tree-optimization/80032
17539 * gimplify.c (gimple_push_cleanup): Forced unconditional
17540 cleanups still have to go to the conditional_cleanups
17543 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17545 PR tree-optimization/80072
17546 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17548 (next_operand_entry_id): Change type to unsigned int.
17549 (sort_by_operand_rank): Make sure to return the right return value
17550 even if unsigned fields are bigger than INT_MAX.
17551 (struct oecount): Change cnt and id type to unsigned int.
17552 (oecount_hasher::equal): Formatting fix.
17553 (oecount_cmp): Make sure to return the right return value
17554 even if unsigned fields are bigger than INT_MAX.
17555 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17558 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17559 TREE_READONLY on result if writing it more than once.
17562 * doc/invoke.texi (-fsanitize=thread): Document that with
17563 -fnon-call-exceptions atomics are not able to throw
17567 * tsan.c: Include tree-eh.h.
17568 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17569 maybe_clean_or_replace_eh_stmt where needed.
17570 (instrument_memory_accesses): Add cfg_changed argument.
17571 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17572 if it returned true.
17573 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17575 PR rtl-optimization/63191
17576 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17577 wrapper function, moved the whole old content into ...
17578 (ix86_delegitimize_address_1): ... this. New inline function.
17579 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17580 true as last argument instead of ix86_delegitimize_address.
17582 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17584 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17585 cortexa57_branch_cost.
17587 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17589 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17591 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17594 * doc/md.texi (Constraints): Document wA constraint.
17595 * config/rs6000/constraints.md (wA): New.
17596 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17597 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17598 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17599 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17601 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17604 * gimplify.c (is_oacc_declared): New function.
17605 (oacc_default_clause): Use it to set default flags for acc declared
17606 variables inside parallel regions.
17607 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17608 declared variables.
17609 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17610 declare attribute to any decl as necessary.
17612 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17615 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17616 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17617 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17618 (arm_arch_lpae): This.
17619 * config/arm/arm.c (arm_arch7ve): Rename into ...
17620 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17621 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17624 2017-03-22 Martin Liska <mliska@suse.cz>
17627 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17628 error message instead of an ICE.
17630 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17632 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17634 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17636 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17638 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17641 2017-03-21 Martin Sebor <msebor@redhat.com>
17643 * doc/extend.texi: Use "cannot" instead of "can't."
17644 * doc/hostconfig.texi: Same.
17645 * doc/install.texi: Same.
17646 * doc/invoke.texi: Same.
17647 * doc/loop.texi: Same.
17648 * doc/md.texi: Same.
17649 * doc/objc.texi: Same.
17650 * doc/rtl.texi: Same.
17651 * doc/tm.texi: Same.
17652 * doc/tm.texi.in: Same.
17653 * doc/trouble.texi: Same.
17655 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17658 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17659 (collect_checksum_attributes): Set it.
17660 (die_checksum_ordered): Use it.
17662 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17664 PR tree-optimization/79908
17665 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17666 change: For a VA_ARG whose LHS has been cast away, use
17667 force_gimple_operand to construct the side effects.
17669 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17671 PR translation/80001
17672 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17673 more amenable to translation.
17674 (oacc_loop_auto_partitions): Likewise.
17676 2017-03-21 Marek Polacek <polacek@redhat.com>
17677 Martin Sebor <msebor@redhat.com>
17679 PR tree-optimization/80109
17680 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17681 on INTEGRAL_TYPE_P.
17683 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17684 Segher Boessenkool <segher@kernel.crashing.org>
17687 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17688 check reg_used_between_p between insn and one of succ or succ2
17689 depending on if succ is artificial insn not inserted into insn
17692 2017-03-21 Martin Liska <mliska@suse.cz>
17694 PR gcov-profile/80081
17695 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17696 * doc/gcc.texi: Include gcov-dump stuff.
17697 * doc/gcov-dump.texi: New file.
17699 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17701 PR rtl-optimization/79150
17702 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17703 conditional jump, if the jump is the last insn of the loop.
17705 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17706 Richard Biener <rguenth@suse.de>
17708 PR tree-optimization/79908
17709 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17710 been cast away, use force_gimple_operand to construct the side
17713 2017-03-21 Martin Liska <mliska@suse.cz>
17715 PR libfortran/79956
17716 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17719 2017-03-21 Brad Spengler <spender@grsecurity.net>
17722 * plugin.c (htab_hash_plugin): New function.
17723 (add_new_plugin): Use it and adjust.
17724 (parse_plugin_arg_opt): Adjust.
17725 (init_one_plugin): Likewise.
17727 2017-03-21 Richard Biener <rguenther@suse.de>
17729 PR tree-optimization/80032
17730 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17731 if set force the cleanup to happen unconditionally.
17732 (gimplify_target_expr): Push inserted clobbers with force_uncond
17733 to avoid them being removed by control-dependent DCE.
17735 2017-03-21 Richard Biener <rguenther@suse.de>
17737 PR tree-optimization/80122
17738 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17739 va_arg_pack_len when the inlined call stmt requires pack
17741 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17743 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17746 * tsan.c (instrument_builtin_call): If the memory model argument
17747 is not a constant, assume it is valid.
17750 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17753 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17755 PR rtl-optimization/79910
17756 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17757 if its dest is used by an insn before I2 (other than the combined
17758 insns themselves, which are properly handled already).
17760 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17763 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17765 * combine.c (record_used_regs): New static function.
17766 (try_combine): Handle situations where there is an additional
17767 instruction between I2 and I3 which needs to have a LOG_LINK
17771 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17773 * combine.c (try_combine): Delete redundant i1 test. Call
17774 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17776 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17779 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17780 alternatives 13/14.
17782 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17784 PR tree-optimization/80054
17785 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17786 the optimization if a PHI or any of its arguments is not dominated
17787 by the candidate's basis. Use gphi* rather than gimple* as
17789 (replace_profitable_candidates): Clean up a gimple* variable that
17790 should be a gphi* variable.
17792 2017-03-20 Martin Sebor <msebor@redhat.com>
17795 * doc/extend.texi (attribute constructor): Document present limitation.
17797 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17800 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17801 __POWER9_VECTOR__ #ifdef control, change template definition to
17802 use Power9-specific built-in function.
17803 (vec_any_eq): Likewise.
17804 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17805 to control outcomes from this test.
17806 (vector_ae_<mode>p): For VEC_F modes, likewise.
17808 2017-03-20 Ian Lance Taylor <iant@google.com>
17810 * config/i386/i386.c (ix86_function_regparm): Save an extra
17811 register for -fsplit-stack with DECL_STATIC_CHAIN.
17813 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17816 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17817 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17819 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17821 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17823 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17826 2017-03-20 Marek Polacek <polacek@redhat.com>
17829 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17831 2017-03-20 Richard Biener <rguenther@suse.de>
17833 PR tree-optimization/80113
17834 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17835 allocate extra SSA name for PHI def.
17836 (add_close_phis_to_outer_loops): Likewise.
17837 (add_close_phis_to_merge_points): Likewise.
17838 (copy_loop_close_phi_args): Likewise.
17839 (copy_cond_phi_nodes): Likewise.
17841 2017-03-20 Martin Liska <mliska@suse.cz>
17843 PR middle-end/79753
17844 * tree-chkp.c (chkp_build_returned_bound): Do not build
17845 returned bounds for a LHS that's not a BOUNDED_P type.
17847 2017-03-20 Martin Liska <mliska@suse.cz>
17851 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17852 COMPLEX_CST and VECTOR_CST.
17854 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17857 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17858 target operand. A new splitter adds the clobber statement in case
17859 the target operand is dead anyway.
17861 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17863 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17864 to age-old versions of binutils and glibc.
17866 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17868 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17870 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17872 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17874 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17876 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17877 requirement for binutils 2.13.
17879 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17881 * combine.c (try_combine): Delete redundant i1 test. Call
17882 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17884 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17886 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17887 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17889 <riscv64-*-elf>: Re-arrange section
17890 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17891 <riscv32-*-linux>: Likewise.
17892 <riscv64-*-elf>: Likewise
17893 <riscv64-*-linux>: Likewise.
17895 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17898 * aarch64.opt(verbose-cost-dump): Fix typo.
17900 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17903 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17904 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17906 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17908 * reload.c (find_reloads): When reloading a nonoffsettable address,
17909 use RELOAD_OTHER for it and its address reloads.
17911 PR rtl-optimization/79910
17912 * combine.c (record_used_regs): New static function.
17913 (try_combine): Handle situations where there is an additional
17914 instruction between I2 and I3 which needs to have a LOG_LINK
17917 2017-03-17 Jeff Law <law@redhat.com>
17919 PR tree-optimization/71437
17920 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17921 conditional in the hash table first.
17922 (vrp_dom_walker::before_dom_children): Extract condition from
17923 ASSERT_EXPR. Record condition, its inverion and any implied
17924 conditions as well.
17926 2017-03-17 Marek Polacek <polacek@redhat.com>
17927 Markus Trippelsdorf <markus@trippelsdorf.de>
17929 PR tree-optimization/80079
17930 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17933 2017-03-17 Richard Biener <rguenther@suse.de>
17935 PR middle-end/80075
17936 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17937 Properly verify the LHS before the RHS possibly claims to be
17939 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17942 2017-03-17 Martin Jambor <mjambor@suse.cz>
17944 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17945 (List of -O2 options): Likewise.
17946 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17949 2017-03-17 Tom de Vries <tom@codesourcery.com>
17951 * gcov-dump.c (print_usage): Print bug_report_url.
17953 2017-03-17 Richard Biener <rguenther@suse.de>
17955 PR middle-end/80050
17956 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17957 (parser::peek): Likewise.
17959 2017-03-17 Richard Biener <rguenther@suse.de>
17961 PR tree-optimization/80048
17962 * sese.c (free_sese_info): Properly release rename_map and
17963 copied_bb_map elements.
17965 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17967 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17968 Add linked-list forward and backlinks. Insert on
17969 construction, remove on destruction.
17970 (class pass_store_merging): Add m_stores_head field.
17971 (pass_store_merging::terminate_and_process_all_chains):
17972 Iterate over m_stores_head list.
17973 (pass_store_merging::terminate_all_aliasing_chains):
17975 (pass_store_merging::execute): Check for debug stmts first.
17976 Push new chains onto the m_stores_head stack.
17978 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17981 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17982 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17983 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17985 2017-03-16 Jeff Law <law@redhat.com>
17987 PR tree-optimization/71437
17988 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17989 member function. Implementation moved into after_dom_children
17990 member function and into the threader's thread_outgoing_edges
17992 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17993 some code into new thread_outgoing_edges.
17994 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17995 definition. Simplify marker handling (do it here). Assume we always
17996 have the available expression and the const/copies tables.
17997 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17999 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18000 * tree-vrp.c (equiv_stack): No longer file scoped.
18001 (vrp_dom_walker): New class.
18002 (vrp_dom_walker::before_dom_children): New member function.
18003 (vrp_dom_walker::after_dom_children): Likewise.
18004 (identify_jump_threads): Setup domwalker. Use it rather than
18005 walking edges in a random order by hand. Simplify setup/finalization.
18006 (finalize_jump_threads): Remove.
18007 (vrp_finalize): Do not call identify_jump_threads here.
18008 (execute_vrp): Do it here instead and call thread_through_all_blocks
18011 PR tree-optimization/71437
18012 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18014 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18016 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18017 (dom_opt_dom_walker::thread_across_edge): Remove
18018 handle_dominating_asserts argument. All callers changed.
18019 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18020 changes. Remove calls to lhs_of_dominating_assert. Other
18021 uses of handle_dominating_asserts turn into unconditional code
18022 (simplify_control_stmt_condition_1): Likewise.
18023 (simplify_control_stmt_condition): Likewise.
18024 (thread_through_normal_block, thread_across_edge): Likewise.
18025 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18026 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18027 object if it is not an SSA_NAME.
18028 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18029 before calling into the VRP specific simplifiers.
18030 (identify_jump_threads): Remove handle_dominating_asserts
18033 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18036 * tree-diagnostic.c (default_tree_printer): No longer static.
18037 * tree-diagnostic.h (default_tree_printer): New prototype.
18039 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18041 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18042 Change ins into fmov.
18044 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18046 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18047 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18048 Use h_con constraint for operand 1.
18049 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18050 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18052 2017-03-15 Jeff Law <law@redhat.com>
18054 PR tree-optimization/71437
18055 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18056 (record_temporary_equivalences): Use it.
18058 PR tree-optimization/71437
18059 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18060 tree-ssa-scopedtables.
18061 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18062 (record_conditions, record_cond, vuse_eq): Likewise.
18063 (record_edge_info): Adjust to API tweak of record_conditions.
18064 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18065 (record_temporary_equivalences, optimize_stmt): Likewise.
18066 (eliminate_redundant_computations): Likewise.
18067 (record_equivalences_from_stmt): Likewise.
18068 * tree-ssa-scopedtables.c: Include options.h and params.h.
18069 (vuse_eq): New function, moved from tree-ssa-dom.c
18070 (build_and_record_new_cond): Likewise.
18071 (record_conditions): Likewise. Accept vector of conditions rather
18072 than edge_equivalence structure for first argument.
18073 for the first argument.
18074 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18075 from tree-ssa-dom.c.
18076 (avail_exprs_stack::record_cond): Likewise.
18077 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18078 from tree-ssa-dom.c.
18079 (avail_exprs_stack): Add new member functions lookup_avail_expr
18081 (record_conditions): Declare.
18083 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18086 * lra-constraints.c (process_alt_operands): Increase reject for
18087 reloading an input/output operand.
18089 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18092 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18093 insns to convert from signed/unsigned char/short to IEEE 128-bit
18095 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18097 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18100 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18101 subreg of inner mode for values already in registers.
18103 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18105 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18106 iteration reg is used after the loop.
18108 2017-03-14 Martin Sebor <msebor@redhat.com>
18110 PR tree-optimization/79800
18111 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18112 precision in negative-positive range.
18113 (format_floating): Call non-const overload with adjusted precision.
18115 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18118 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18121 2017-03-14 Martin Sebor <msebor@redhat.com>
18123 PR middle-end/80020
18124 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18125 * builtins.def (aligned_alloc): Use it.
18128 * Makefile.in (GTFILES): Add calls.c.
18129 * calls.c: Include "gt-calls.h".
18131 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18133 PR rtl-optimization/79728
18134 * regs.h (struct target_regs): New field
18135 x_contains_allocatable_regs_of_mode.
18136 (contains_allocatable_regs_of_mode): New macro.
18137 * reginfo.c (init_reg_sets_1): Initialize it, and change
18138 contains_reg_of_mode so it includes global regs as well.
18139 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18140 rather than contains_regs_of_mode.
18142 2017-03-14 Martin Liska <mliska@suse.cz>
18144 * doc/invoke.texi: Document options that can't be combined with
18145 -fcheck-pointer-bounds.
18147 2017-03-14 Martin Liska <mliska@suse.cz>
18149 PR middle-end/79831
18150 * doc/invoke.texi (-Wchkp): Document the option.
18152 2017-03-14 Martin Liska <mliska@suse.cz>
18154 * Makefile.in: Install gcov-dump.
18156 2017-03-14 Martin Liska <mliska@suse.cz>
18158 * multiple_target.c (expand_target_clones): Bail out for
18159 an invalid attribute.
18161 2017-03-14 Richard Biener <rguenther@suse.de>
18163 * alias.c (struct alias_set_entry): Pack properly.
18164 * cfgloop.h (struct loop): Likewise.
18165 * cse.c (struct set): Likewise.
18166 * ipa-utils.c (struct searchc_env): Likewise.
18167 * loop-invariant.c (struct invariant): Likewise.
18168 * lra-remat.c (struct cand): Likewise.
18169 * recog.c (struct change_t): Likewise.
18170 * rtl.h (struct address_info): Likewise.
18171 * symbol-summary.h (function_summary): Likewise.
18172 * tree-loop-distribution.c (struct partition): Likewise.
18173 * tree-object-size.c (struct object_size_info): Likewise.
18174 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18175 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18176 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18177 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18178 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18179 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18180 (struct _stmt_vec_info): Likewise.
18182 2017-03-14 Martin Liska <mliska@suse.cz>
18185 * multiple_target.c (create_dispatcher_calls): Check that
18186 a target can create a function dispatcher.
18188 2017-03-14 Martin Liska <mliska@suse.cz>
18191 * multiple_target.c (expand_target_clones): Drop local.local
18192 flag for default implementation.
18194 2017-03-14 Richard Biener <rguenther@suse.de>
18196 PR tree-optimization/80030
18197 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18199 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18201 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18202 gcc_fallthrough() instead of __attribute__((fallthrough));
18204 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18206 * doc/gcc.texi: Remove "up" link to (DIR).
18207 * doc/gccint.texi: Ditto.
18209 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18211 * doc/install.texi (Specific) <avr>: Remove reference to
18214 2017-03-13 Jeff Law <law@redhat.com>
18216 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18217 attribute rather than comments.
18219 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18220 match_scratch operand is highest.
18222 2017-03-13 Martin Liska <mliska@suse.cz>
18224 PR middle-end/78339
18225 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18226 is a CHKP clone, use original declaration.
18228 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18230 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18231 (arc_conditional_register_usage): Use a different allocation order
18232 when optimizing for size.
18233 * common/config/arc/arc-common.c (arc_option_optimization_table):
18234 Section anchors default on when optimizing for size.
18236 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18238 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18240 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18242 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18243 * config/arc/arc.md (cpu_facility): Add cd variant.
18244 (*movqi_insn): Add code density variant.
18245 (*movhi_insn): Likewise.
18246 (*movqi_insn): Likewise.
18247 (*addsi3_mixed): Likewise.
18248 (subsi3_insn): Likewise.
18250 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18252 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18254 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18256 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18257 expressions with MINUS and UNARY ops.
18259 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18262 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18264 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18265 between vec_select and vector argument.
18266 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18267 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18268 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18269 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18270 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18271 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18273 2017-03-13 Richard Biener <rguenther@suse.de>
18276 * params.def (vect-max-peeling-for-alignment): Fix typo.
18278 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18280 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18281 issue that only occurred with binutils below 2.18.
18283 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18285 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18286 refer to binutils 2.11/2.12 minimum.
18288 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18290 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18291 ftp.kernel.org and simplify binutils requirement.
18293 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18295 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18297 (Optimize Options): Ditto. Also remove redundancy.
18299 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18301 PR translation/79848
18302 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18304 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18305 to G_ to avoid double translation.
18307 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18309 PR translation/79923
18310 * auto-profile.c (get_combined_location): Convert leading
18311 character of diagnostics to lower case and remove trailing period.
18312 (read_profile): Likewise for various diagnostics.
18313 * config/arm/arm.c (arm_option_override): Remove trailing period
18314 from various diagnostics.
18315 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18316 (msp430_expand_delay_cycles): Likewise.
18318 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18321 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18322 full command-line argument, rather than just "str".
18323 (aarch64_validate_march): Likewise.
18324 (aarch64_validate_mtune): Likewise.
18326 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18328 PR rtl-optimization/78911
18329 * lra-assigns.c (must_not_spill_p): New function.
18330 (spill_for): Use it.
18332 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18334 PR tree-optimization/79981
18335 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18336 ATOMIC_COMPARE_EXCHANGE ifn result.
18337 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18338 IFN_ATOMIC_COMPARE_EXCHANGE.
18340 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18343 * opts.c (parse_sanitizer_options): Add missing question mark to
18344 "did you mean" message.
18346 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18348 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18350 (VMULEUH_UNS): Likewise.
18351 (VMULOUB_UNS): Likewise.
18352 (VMULOUH_UNS): Likewise.
18353 * config/rs6000/rs6000.c (builtin_function_type): Remove
18354 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18356 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18359 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18360 x with result of extra_parsing_for_operand_code_0.
18361 (function_reader::extra_parsing_for_operand_code_0): Convert
18362 return type from void to rtx, returning x. When reading
18363 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18364 larger size containing struct block_symbol.
18366 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18369 -mfloat128-hardware without -m64.
18371 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18374 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18375 entries to the case statement that marks unsigned arguments to
18376 overloaded functions.
18378 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18380 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18381 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18383 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18386 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18387 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18389 2017-03-10 Martin Liska <mliska@suse.cz>
18393 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18394 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18397 2017-03-10 Olivier Hainque <hainque@adacore.com>
18399 * tree-switch-conversion (array_value_type): Start by resetting
18400 candidate type to it's main variant.
18402 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18404 PR rtl-optimization/79909
18405 * combine.c (try_combine): Use simplify_replace_rtx on individual
18406 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18407 of the whole CALL_INSN_FUNCTION_USAGE.
18409 PR tree-optimization/79972
18410 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18411 get_range_info on SSA_NAMEs. Formatting fixes.
18413 2017-03-10 Richard Biener <rguenther@suse.de>
18414 Jakub Jelinek <jakub@redhat.com>
18416 PR tree-optimization/77975
18417 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18418 edge to be constant.
18419 (get_val_for): For constant x return it. Formatting fix.
18420 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18421 would use the same bases as the current one.
18423 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18425 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18426 instead of vec_select for V1TImode.
18427 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18429 (VSX_LE_128): Add V1TI to this mode iterator.
18430 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18431 (*vsx_le_perm_store_<mode>): Likewise.
18432 (pre-reload splitter for VSX stores): Likewise.
18433 (post-reload splitter for VSX stores): Likewise.
18434 (*vsx_xxpermdi2_le_<mode>): Likewise.
18435 (*vsx_lxvd2x2_le_<mode>): Likewise.
18436 (*vsx_stxvd2x2_le_<mode>): Likewise.
18438 2017-03-09 Michael Eager <eager@eagercon.com>
18440 Correct failures with --enable-checking=yes,rtl.
18442 * config/microblaze/microblaze.c (microblaze_expand_shift):
18443 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18444 test for const0_rtx.
18445 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18446 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18448 2017-03-09 Richard Biener <rguenther@suse.de>
18450 PR tree-optimization/79977
18451 * graphite-scop-detection.c (scop_detection::merge_sese):
18452 Handle the case of extra exits to blocks dominating the entry.
18454 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18456 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18459 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18461 PR rtl-optimization/79949
18462 * lra-constraints.c (process_alt_operands): Check memory when
18463 trying to predict a cycle. Print about the overall increase.
18465 2017-03-09 Richard Biener <rguenther@suse.de>
18467 PR middle-end/79971
18468 * gimple-expr.c (useless_type_conversion_p): Preserve
18469 TYPE_SATURATING for fixed-point types.
18471 2017-03-09 Richard Biener <rguenther@suse.de>
18474 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18475 alignment of BLKmode params.
18477 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18480 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18481 (VALL_NO_V2Q): Likewise.
18483 * config/aarch64/aarch64-simd.md
18484 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18486 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18487 VALL_NO_V2Q mode iterator.
18488 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18490 2017-03-09 Martin Liska <mliska@suse.cz>
18492 PR tree-optimization/79631
18493 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18494 tree_int_cst_sign_bit just for INTEGER constants.
18496 2017-03-09 Martin Liska <mliska@suse.cz>
18500 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18503 2017-03-09 Marek Polacek <polacek@redhat.com>
18506 * tree.c (inchash::add_expr): Handle TREE_VEC.
18508 2017-03-09 Martin Liska <mliska@suse.cz>
18511 (chkp_narrow_size_and_offset): New function.
18512 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18513 (void chkp_parse_bit_field_ref): New function.
18514 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18515 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18517 2017-03-09 Martin Liska <mliska@suse.cz>
18520 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18521 (chkp_find_bounds_1): Remove gcc_unreachable.
18523 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18526 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18527 BUILT_IN_SYNC*, determine the access type from the size suffix and
18528 always build a MEM_REF with that type. Handle forgotten
18529 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18532 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18533 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18534 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18535 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18536 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18537 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18538 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18539 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18540 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18541 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18542 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18543 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18544 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18545 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18546 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18547 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18548 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18549 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18550 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18551 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18552 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18553 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18554 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18555 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18556 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18557 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18558 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18559 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18560 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18561 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18562 definitions outside of __OPTIMIZE__ guarded section.
18565 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18566 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18567 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18568 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18571 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18573 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18574 ("vfenez<mode>"): Add missing constraints.
18576 2017-03-08 Martin Sebor <msebor@redhat.com>
18579 * config/nds32/nds32.c (nds32_option_override):
18580 Fix misspelled diagnostic.
18582 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18585 * gimplify.c (gimplify_omp_for): Replace index var in outer
18586 taskloop statement with an artificial variable and add
18587 OMP_CLAUSE_PRIVATE clause for it.
18589 2017-03-08 Richard Biener <rguenther@suse.de>
18591 PR tree-optimization/79955
18592 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18593 for accesses that are completely outside of the variable.
18595 2017-03-08 Andrew Haley <aph@redhat.com>
18597 PR tree-optimization/79943
18598 * tree-ssa-loop-split.c (compute_new_first_bound): When
18599 calculating the new upper bound, (END-BEG) should be added, not
18602 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18604 * config/avr/avr.md (setmemhi): Make sure match_dup
18605 operand number comes before match_scratch.
18607 2017-03-08 Richard Biener <rguenther@suse.de>
18609 PR tree-optimization/79920
18610 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18611 with ncopies == 1 to ...
18612 (vect_transform_slp_perm_load): ... here. Properly compute
18613 all element loads by iterating VF times over the group. Do
18614 not handle ncopies (computed in a broken way) in
18615 vect_create_mask_and_perm.
18617 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18620 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18621 is a uniform vector, use uniform_vector_p return value instead of
18622 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18624 2017-03-07 Marek Polacek <polacek@redhat.com>
18626 PR middle-end/79809
18627 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18628 (alloca_call_type): Likewise.
18630 2017-03-07 Martin Liska <mliska@suse.cz>
18632 * gcov.c (process_args): Put comment to correct location.
18634 2017-03-07 Martin Liska <mliska@suse.cz>
18636 PR middle-end/68270
18637 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18638 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18639 (chkp_narrow_bounds_for_field): Likewise.
18640 (chkp_parse_array_and_component_ref): Pass one more argument to
18643 2017-03-07 Richard Biener <rguenther@suse.de>
18645 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18648 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18650 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18651 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18653 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18656 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18657 to end of description.
18658 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18660 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18662 PR rtl-optimization/79901
18663 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18665 (*avx512f_<code><mode>3<mask_name>): ... this.
18666 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18667 iterator instead of VI8_AVX2_AVX512BW.
18669 PR rtl-optimization/79901
18670 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18671 min/max expander, expand it using expand_vec_cond_expr.
18674 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18677 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18680 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18681 to void * for PCH reasons.
18682 * dwarf2out.c (output_loc_operands, output_die): Cast
18683 v.val_vec.array to unsigned char *.
18685 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18688 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18691 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18693 PR rtl-optimization/79571
18694 * lra-constraints.c (process_alt_operands): Calculate static
18695 reject and subtract it from overall when only addresses will be
18698 2017-03-06 Julia Koval <julia.koval@intel.com>
18701 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18702 incoming stack boundary to 128 for 64-bit targets.
18704 2017-03-06 Richard Biener <rguenther@suse.de>
18706 PR tree-optimization/79894
18707 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18708 to NULL after folding it.
18710 2017-03-06 Richard Biener <rguenther@suse.de>
18712 PR tree-optimization/79824
18713 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18714 check disabling peeling for gaps.
18716 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18718 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18719 attributes): Document gettimeofday.
18721 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18723 * config/s390/s390.c (s390_option_override_internal): Set
18724 PARAM_MIN_VECT_LOOP_BOUND
18726 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18728 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18729 * config/s390/s390.md: Likewise.
18731 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18734 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18735 (<avx2_avx512>_perm<mode>): Rename to ...
18736 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18738 (<avx512>_perm<mode>_mask): Rename to ...
18739 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18741 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18742 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18743 instead of VI8F_256_512.
18744 (avx512f_perm<mode>): New define_expand.
18745 (avx512f_perm<mode>_mask): Likewise.
18746 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18747 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18749 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18751 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18752 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18754 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18756 2017-03-06 Martin Liska <mliska@suse.cz>
18759 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18760 when having a SSA NAME w/o VAR_DECL assigned to it.
18762 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18764 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18765 msa_dpsub_<su>_d): Fix MODE for vec_select.
18767 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18769 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18771 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18773 2017-03-06 Richard Biener <rguenther@suse.de>
18775 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18776 * plugin.c (register_plugin_info): Likewise.
18777 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18779 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18781 * config/i386/sse.md (sse_storehps, sse_storelps,
18782 avx_<castmode><avxsizesuffix>_<castmode>,
18783 avx512f_<castmode><avxsizesuffix>_<castmode>,
18784 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18785 in condition that at least one operand is not a MEM.
18787 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18789 PR middle-end/79805
18790 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18791 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18793 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18794 gimple_call_nothrow_p flag based on whether original builtin can throw.
18795 If it can, emit following stmts on the fallthrough edge.
18796 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18797 don't create new bb if inserting just debug stmts on the edge, try to
18798 insert them on the fallthru bb or just reset debug stmts.
18800 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18803 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18804 restore recog_data (including the operand rtxes inside it) around
18805 the call to get_insn_template.
18807 2017-03-03 Martin Sebor <msebor@redhat.com>
18809 PR tree-optimization/79699
18810 * context.c (context::~context): Free MPFR caches to avoid
18811 a memory leak on program exit.
18813 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18815 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18816 Use wide_int::ulow () instead of .elt (0).
18818 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18820 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18821 (*pushxf): Limit oF constraint to 32bit targets and add oC
18822 constraint for 64bit targets.
18823 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18824 (*pushdf): Change rmF constraint to rmC.
18826 2017-03-03 Martin Liska <mliska@suse.cz>
18828 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18829 Remove unused variable.
18831 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18834 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18835 is a memory operand, increase num_memory.
18836 (ix86_expand_args_builtin): Likewise.
18838 2017-03-03 Jan Hubicka <jh@suse.cz>
18841 * ipa-devirt.c (maybe_record_node): Properly handle
18842 __cxa_pure_virtual visibility.
18844 2017-03-03 Martin Liska <mliska@suse.cz>
18846 PR tree-optimization/79803
18847 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18849 (pass_loop_prefetch::execute): Disabled optimization if an
18850 assumption about L1 cache size is not met.
18852 2017-03-03 Martin Liska <mliska@suse.cz>
18854 PR rtl-optimization/79574
18855 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18856 (hash_scan_set): Likewise.
18857 (dump_hash_table): Likewise.
18858 (hoist_code): Likewise.
18860 2017-03-03 Richard Biener <rguenther@suse.de>
18862 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18863 in place of uhigh ().
18864 (fixed_convert_from_real): Likewise.
18866 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18869 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18871 2017-03-03 Richard Biener <rguenther@suse.de>
18873 PR middle-end/79818
18874 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18875 TYPE_OVERFLOW_UNDEFINED check.
18877 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18879 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18881 (vector_ae_<mode>_p): Likewise.
18882 (vector_nez_<mode>_p): Likewise.
18883 (vector_ne_v2di_p): Likewise.
18884 (vector_ae_v2di_p): Likewise.
18885 (vector_ne_<mode>_p): Likewise.
18886 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18888 (vsx_tsqrt<mode>2_fe): Likewise.
18890 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18893 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18895 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18897 PR rtl-optimization/79780
18898 * cprop.c (one_cprop_pass): When second and further conditional trap
18899 in a single basic block is turned into an unconditional trap, turn it
18900 into a deleted note to avoid RTL verification failures.
18902 2017-03-02 Richard Biener <rguenther@suse.de>
18904 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18906 2017-03-02 Richard Biener <rguenther@suse.de>
18908 PR tree-optimization/79345
18910 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18911 param and abort the walk, returning -1 if it is hit.
18912 (walk_aliased_vdefs): Take a limit param and pass it on.
18913 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18914 defaulting to 0 and return a signed int.
18915 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18916 (check_defs): New helper.
18917 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18918 about uninitialized memory.
18919 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18920 bogus uninitialized warning.
18921 (fixed_convert_from_real): Likewise.
18923 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18925 PR tree-optimization/66768
18926 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18927 iv_use if base object can't be determined.
18929 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18931 PR tree-optimization/79345
18932 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18933 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18934 (get_pattern_stats): Initialize it.
18935 * genemit.c (gen_expand): Verify match_scratch numbers come after
18936 match_operand/match_dup numbers.
18937 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18938 match_scratch numbers.
18939 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18941 * config/s390/s390.md (trunctdsd2): Likewise.
18943 2017-03-02 Richard Biener <rguenther@suse.de>
18945 * wide-int.h (wide_int_storage::operator=): Implement in terms
18948 2017-03-02 Richard Biener <rguenther@suse.de>
18950 PR tree-optimization/79777
18951 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18952 the to insert expression to sth existing.
18954 2017-03-01 Martin Sebor <msebor@redhat.com>
18956 PR middle-end/79692
18957 * gimple-ssa-sprintf.c
18958 (directive::known_width_and_precision): New function.
18959 (format_integer): Use it.
18960 (get_mpfr_format_length): Consider the full range of precision
18961 when computing %g output with the # flag. Set the likely byte
18962 count to 3 rather than 1 when precision is indeterminate.
18963 (format_floating): Correct the lower bound of precision.
18965 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18967 * doc/invoke.texi: Document default code model for 64-bit Linux.
18969 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18972 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18973 udiv rather than div since input pattern is unsigned.
18975 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18977 * config/i386/i386.c (print_reg): Warn for values of
18978 unsupported size in integer register.
18980 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18983 * config/rs6000/predicates.md (current_file_function_operand): Do
18984 not allow self calls to be local if the function is replaceable.
18986 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18989 * config/rs6000/altivec.h (vec_ctz and others): Change the
18990 preprocessor macro that controls conditional compilation from
18991 _ARCH_PWR9 to __POWER9_VECTOR__.
18992 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18993 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18994 control (instead of _ARCH_PWR9 control) so that template
18995 definition uses power9-specific function.
18996 (vec_any_eq): Likewise.
18997 (vec_all_ne): Change macro definition to use a power9-specific
18998 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18999 _ARCH_PWR9 control).
19000 (vec_any_eq) Likewise.
19001 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19002 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19003 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19004 support for xvcmpnedp instruction.
19005 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19006 macro expansion so that Power9 implementation of vec_all_ne does
19007 not use the AltiVec predicate framework.
19008 (VCMPNEH_P): Likewise.
19009 (VCMPNEW_P): Likewise.
19010 (VCMPNED_P): Likewise.
19011 (VCMPNEFP_P): Likewise.
19012 (VCMPNEDP_P): Likewise.
19013 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19014 implementation of vec_any_eq to not use AltiVec predicate
19016 (VCMPAEH_P): Likewise.
19017 (VCMPAEW_P): Likewise.
19018 (VCMPAED_P): Likewise.
19019 (VCMPAEFP_P): Likewise.
19020 (VCMPAEDP_P): Likewise.
19021 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19022 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19023 not use the AltiVec predicate framework.
19024 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19025 of vec_any_eq to not use AltiVec predicate framework.
19026 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19027 support for predefined __POWER9_VECTOR__ macro to indicate that
19028 Power9 instruction selection is enabled.
19029 (altivec_overloaded_builtins): Remove extraneous
19030 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19031 function argument types RS6000_BTI_bool_V16QI and
19032 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19033 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19034 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19035 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19036 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19037 Power9 for implementations of vec_cmpne. Change the signature for
19038 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19039 (representing vec_all_ne) to remove the previously described first
19040 argument of type RS6000_BTI_INTSI, as this was an artifact of
19041 reliance on the AltiVec predicate framework, which is no longer
19042 used in the implementation of these functions. Add
19043 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19044 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19045 since, unlike the AltiVec predicate framework implementation, we
19046 do not share function descriptors between vec_alle and vec_anyeq.
19047 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19048 set of modes that receive special treatment even when
19049 TARGET_P9_VECTOR is true. The special treatment emits code that
19050 does not depend on Power9 instructions.
19051 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19052 define_expand to not rely on AltiVec predicate framework.
19053 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19055 (vector_ne_v2di_p): Change this define_expand to not rely on
19056 AltiVec predicate framework.
19057 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19059 (vector_ne_<mode>_p): Change this define_expand to not rely on
19060 AltiVec predicate framework.
19061 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19063 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19064 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19065 define_insn pattern.
19066 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19067 define_insn pattern because the xvcmpne<VSs>. instruction is not
19069 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19070 instruction is not supported.
19072 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19074 * config/nvptx/nvptx.c: Include intl.h.
19076 2017-03-01 Martin Jambor <mjambor@suse.cz>
19079 * ipa-prop.h (ipa_bits): Removed field known.
19080 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19081 to pointers. Adjusted their comments to warn about their sharing.
19082 (ipcp_transformation_summary): Change bits to a vector of pointers.
19083 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19084 (ipa_get_ipa_bits_for_value): Declare.
19085 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19086 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19087 (ipa_bits_hash_table): Likewise.
19088 (ipa_vr_ggc_hash_traits): Likewise.
19089 (ipa_vr_hash_table): Likewise.
19090 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19091 being pointers and vr_known being removed.
19092 (ipa_set_jf_unknown): Likewise.
19093 (ipa_get_ipa_bits_for_value): New function.
19094 (ipa_set_jfunc_bits): Likewise.
19095 (ipa_get_value_range): New overloaded functions.
19096 (ipa_set_jfunc_vr): Likewise.
19097 (ipa_compute_jump_functions_for_edge): Use the above functions to
19098 construct bits and vr parts of jump functions.
19099 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19100 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19102 (ipcp_grow_transformations_if_necessary): Also allocate
19103 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19105 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19106 them. Fix too long lines.
19107 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19108 vr_known being removed.
19109 (ipa_read_jump_function): Use new setter functions to construct bits
19110 and vr parts of jump functions or set them to NULL.
19111 (write_ipcp_transformation_info): Adjust for bits being pointers.
19112 (read_ipcp_transformation_info): Likewise.
19113 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19115 Include gt-ipa-prop.h.
19116 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19118 (ipcp_store_bits_results): Likewise.
19119 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19120 Do not write to existing jump functions but use a temporary instead.
19122 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19125 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19126 attempt to use its first operand as BIT_FIELD_REF base.
19128 2017-03-01 Richard Biener <rguenther@suse.de>
19130 PR middle-end/79721
19131 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19132 interpolating formula in wrapping arithmetic.
19133 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19135 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19137 PR tree-optimization/79734
19138 * tree-vect-generic.c (expand_vector_condition): Optimize
19139 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19140 Handle VEC_COND_EXPR where comparison has different inner width from
19141 type's inner width.
19143 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19145 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19146 markup, and similar issues. Remove @opindex entries for things
19147 that aren't options. Add missing -mmpy-option entries.
19149 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19151 PR tree-optimization/79737
19152 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19153 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19154 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19155 instead of byte_size. Formatting fix.
19156 (shift_bytes_in_array_right): Formatting fix.
19158 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19161 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19162 condition on optimize for the leaf function test.
19164 2017-02-28 Martin Liska <mliska@suse.cz>
19167 * read-rtl-function.c (function_reader::handle_unknown_directive):
19168 Bail out when one uses -flto.
19170 2017-02-28 Martin Liska <mliska@suse.cz>
19172 * common.opt: Replace space with tabular for options of <number>
19174 * config/i386/i386.opt: Show <number> value for
19175 -mlarge-data-threshold.
19176 * opts.c (print_filtered_help): Do not display number in hexadecimal
19179 2017-02-28 Martin Liska <mliska@suse.cz>
19181 * common.opt: Fix --help=option -Q for options which are of
19184 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19186 * config/i386/i386.c (print_reg): Error out for values
19187 of 8-bit size in invalid integer register.
19189 2017-02-28 Martin Sebor <msebor@redhat.com>
19191 PR tree-optimization/79691
19192 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19194 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19197 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19198 gcc_unreachable with output_operand_lossage.
19200 2017-02-28 Richard Biener <rguenther@suse.de>
19202 PR tree-optimization/79740
19203 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19205 (visit_nary_op): Insert the nary into the hashtable if we
19206 pattern-matched sth.
19207 * tree-ssa-pre.c (eliminate_insert): Robustify.
19209 2017-02-28 Richard Biener <rguenther@suse.de>
19211 PR middle-end/79731
19212 * fold-const.c (decode_field_reference): Reject out-of-bound
19215 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19217 * config/i386/i386.c: Include intl.h.
19218 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19219 instead of just cond ? "..." : "...".
19220 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19221 * coverage.c (read_counts_file): Likewise.
19222 * omp-offload.c: Include intl.h.
19223 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19224 of just cond ? "..." : "...".
19225 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19226 of just cond ? "..." : "...".
19228 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19231 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19233 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19234 'tune for' CPU name.
19235 * config/arm/arm-cpu-data.h: Regenerated.
19237 2017-02-28 Richard Biener <rguenther@suse.de>
19239 PR tree-optimization/79732
19240 * tree-inline.c (expand_call_inline): Do not shadow var.
19242 2017-02-28 Richard Biener <rguenther@suse.de>
19244 PR tree-optimization/79723
19245 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19246 address-space properly.
19248 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19250 * doc/optinfo.texi (Optimization groups): Fix option used for
19252 * doc/invoke.texi (-fopt-info): Document "omp".
19253 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19254 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19255 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19256 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19257 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19259 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19261 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19262 OPTGROUP_OMP with "omp".
19264 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19267 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19268 for arithmetic shift of unsigned V2DI.
19270 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19272 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19273 arc/linux.h headers.
19274 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19275 (LINK_SPEC): Likewise.
19276 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19277 (EXTRA_SPECS): Likewise.
19278 (STARTFILE_SPEC): Likewise.
19279 (ENDFILE_SPEC): Likewise.
19280 (LIB_SPEC): Likewise.
19281 (TARGET_SDATA_DEFAULT): Likewise.
19282 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19283 (MULTILIB_DEFAULTS): Likewise.
19284 (DWARF2_UNWIND_INFO): Likewise.
19285 * config/arc/big.h: New file.
19286 * config/arc/elf.h: Likewise.
19287 * config/arc/linux.h: Likewise.
19288 * config/arc/t-uClibc: Remove.
19290 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19292 PR tree-optimization/77536
19293 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19294 (tree_transform_and_unroll_loop): Use above function to compute the
19295 estimated niter of unrolled loop and use it when scaling profile.
19296 Also use count info rather than frequency if it's non-zero.
19297 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19298 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19299 (vect_transform_loop): Call above function.
19301 2017-02-27 Richard Biener <rguenther@suse.de>
19303 PR tree-optimization/45397
19304 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19305 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19306 (visit_nary_op): Add pattern matching for CSEing sign-changed
19307 or truncated operations with wider ones.
19309 2017-02-27 Richard Biener <rguenther@suse.de>
19311 PR tree-optimization/79690
19312 * tree-vect-stmts.c (vectorizable_store): Use vector type
19313 built from the DR with address-space.
19315 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19317 * doc/invoke.texi (Optimize Options): Refine the description
19318 of asan-use-after-return.
19320 2017-02-25 Alan Modra <amodra@gmail.com>
19322 PR rtl-optimization/79584
19323 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19324 base, not ad->base_term, the reg within base. Remove assertion
19325 that ad->base == ad->base_term. Replace gen_int_mode using
19326 bogus mode with const0_rtx.
19328 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19330 PR middle-end/79396
19331 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19332 FMA_EXPR like tcc_binary or tcc_unary.
19334 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19337 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19339 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19340 (output_loc_operands): Handle DW_OP_call_ref and
19341 DW_OP_GNU_variable_value.
19342 (struct variable_value_struct): New type.
19343 (struct variable_value_hasher): Likewise.
19344 (variable_value_hash): New variable.
19345 (string_types): Remove.
19346 (copy_loc_descr): New function.
19347 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19348 (prepend_loc_descr_to_each): New function.
19349 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19350 instead of add_loc_descr_to_each if the first argument is single
19351 location list and the second has multiple.
19352 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19353 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19354 when looking for variable value which doesn't have other location info.
19355 (loc_list_from_tree): Formatting fix.
19356 (gen_array_type_die): Simplify DW_AT_string_length handling.
19357 (adjust_string_types): Remove.
19358 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19359 string_types. Call resolve_variable_values.
19360 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19361 (resolve_addr_in_expr): Likewise. Add A argument.
19362 (copy_deref_exprloc): Remove deref argument. Adjust for the
19363 original expression being DW_OP_GNU_variable_value with optionally
19364 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19365 optionally after it.
19366 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19367 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19368 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19369 (variable_value_hasher::hash, variable_value_hasher::equal): New
19371 (resolve_variable_value_in_expr, resolve_variable_value,
19372 resolve_variable_values, note_variable_value_in_expr,
19373 note_variable_value): New functions.
19374 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19377 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19380 * opts.h (handle_generated_option): Add GENERATED_P argument.
19381 * opts-common.c (handle_option): Adjust function comment.
19382 (handle_generated_option): Add GENERATED_P argument, pass it to
19384 (control_warning_option): Pass false to handle_generated_option
19386 * opts.c (maybe_default_option): Pass true to handle_generated_option
19388 * optc-gen.awk: Likewise.
19390 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19392 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19393 a REG, look at the REG it is a SUBREG of.
19394 (splitter for cmpeqsi_t): Ditto.
19396 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19398 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19399 the special USEs with the pattern of the insn, not the insn itself.
19401 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19404 * doc/invoke.texi: Document -mload-store-pairs.
19406 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19407 Sandra Loosemore <sandra@codesourcery.com>
19409 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19410 argument isn't a CONST_INT.
19411 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19412 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19413 (nios2_validate_compare): Bypass alternate compare logic if *op2
19414 is not a CONST_INT.
19415 (ldstwm_operation_p): Return false if first_base is not a REG or
19416 if first_offset is not a CONST_INT.
19418 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19420 * config/cris/cris.md: Use correct operand in a define_peephole2.
19422 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19424 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19426 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19428 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19429 this_insn if it is an INSN or JUMP_INSN.
19430 (force_offsettable): Look at base, not at addr.
19431 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19432 on things that aren't necessarily CONST_INTs.
19434 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19436 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19437 -mfpmath=sse is the default also for x86-32 targets with SSE2
19438 instruction set when @option{-ffast-math} is enabled
19440 2017-02-24 Jeff Law <law@redhat.com>
19442 PR rtl-optimizatoin/79286
19443 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19446 2017-02-24 Richard Biener <rguenther@suse.de>
19448 PR tree-optimization/79389
19449 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19452 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19454 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19455 function comment to reflect reality.
19456 (loop_exits_before_overflow): Fix typo in function description.
19458 2017-02-24 Richard Biener <rguenther@suse.de>
19460 PR tree-optimization/79389
19461 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19462 properly that a threading opportunity exists. Detect conditional
19463 copy/constant propagation opportunities.
19465 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19467 * config/visium/visium.md (type): Add trap.
19468 (b): New mode attribute.
19469 (*btst): Rename into...
19470 (*btst<mode>): ...this and adjust.
19471 (*cbranchsi4_btst_insn): Rename into...
19472 (*cbranch<mode>4_btst_insn): ...this and adjust.
19473 (trap): New define_insn.
19475 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19477 PR tree-optimization/79389
19478 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19479 (noce_reversed_cond_code): New function.
19480 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19481 reversed_comparison_code. Formatting fix.
19482 (noce_try_store_flag): Test rev_cond != NULL in addition to
19483 reversed_comparison_code.
19484 (noce_try_store_flag_constants): Likewise.
19485 (noce_try_store_flag_mask): Likewise.
19486 (noce_try_addcc): Use rev_cond if non-NULL instead of
19487 reversed_comparison_code.
19488 (noce_try_cmove_arith): Likewise. Formatting fixes.
19489 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19490 (noce_find_if_block): Initialize rev_cond.
19491 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19492 instead of false as last argument never attempt to reverse it
19495 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19497 PR tree-optimization/79663
19498 * tree-predcom.c (combine_chains): Process refs in reverse order
19499 only for ZERO length chains, and add explaining comment.
19501 2017-02-23 Jeff Law <law@redhat.com>
19503 PR tree-optimization/79578
19504 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19505 in call to operand_equal_p.
19507 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19510 * config/i386/cpuid.h: Fix another undefined behavior.
19512 2017-02-23 Richard Biener <rguenther@suse.de>
19514 PR tree-optimization/79683
19515 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19516 vector types for data-refs.
19518 2017-02-23 Martin Liska <mliska@suse.cz>
19520 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19522 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19524 PR middle-end/79665
19525 * internal-fn.c (get_range_pos_neg): Moved to ...
19526 * tree.c (get_range_pos_neg): ... here. No longer static.
19527 * tree.h (get_range_pos_neg): New prototype.
19528 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19529 are known to be in between 0 and signed maximum inclusive, try to
19530 expand both unsigned and signed divmod and use the cheaper one from
19533 2017-02-22 Jeff Law <law@redhat.com>
19535 PR tree-optimization/79578
19536 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19537 to compare base operands.
19539 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19542 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19543 gpc_reg_operand instead of fpr_reg_operand.
19545 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19547 * config/mips/mips.c (mips_return_in_memory): Force FP
19548 vector types to be returned in memory for o32 ABI.
19550 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19552 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19553 instead of DW_TAG_member for static data member declarations and don't
19554 set no_linkage_name for static inline data members.
19555 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19558 2017-02-22 Martin Liska <mliska@suse.cz>
19560 * doc/invoke.texi: Replace inequality signs with square brackets
19563 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19565 PR tree-optimization/68644
19566 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19568 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19571 * lra-constraints.c (simplify_operand_subreg): Handle
19572 WORD_REGISTER_OPERATIONS targets.
19574 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19577 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19578 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19579 elimination by swapping fld*.
19581 2017-02-22 Richard Biener <rguenther@suse.de>
19583 PR tree-optimization/79673
19584 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19585 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19586 irrelevant address-space qualifiers and avoiding a
19587 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19589 2017-02-22 Richard Biener <rguenther@suse.de>
19591 PR tree-optimization/79666
19592 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19593 to not symbolically negate if that may introduce undefined
19596 2017-02-22 Martin Liska <mliska@suse.cz>
19599 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19600 * data-streamer-out.c (streamer_write_gcov_count_stream):
19602 * value-prof.c (stream_out_histogram_value): Make assert more
19603 precise based on type of counter.
19605 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19608 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19609 Use nonimmediate_operand instead of memory_operand for operand 1.
19610 (float-extend standard_x87sse_constant_load splitter): Ditto.
19612 2017-02-21 Jeff Law <law@redhat.com>
19614 PR tree-optimization/79621
19615 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19616 blocks with edges to themselves.
19618 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19621 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19622 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19623 Use gimple_call_builtin_p.
19626 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19627 on temporarily removed DEBUG_INSNs.
19629 PR tree-optimization/79649
19630 * tree-loop-distribution.c (classify_partition): Give up on
19631 non-generic address space loads/stores.
19633 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19635 * doc/loop.texi (Loop manipulation): Remove nonexistent
19636 tree_ssa_loop_version from the documentation.
19637 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19639 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19642 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19643 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19644 * config/rs6000/rs6000.c: Include except.h.
19645 (rs6000_expand_split_stack_prologue): Call
19646 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19648 2017-02-21 Martin Jambor <mjambor@suse.cz>
19651 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19652 have been analyzed.
19654 2017-02-21 Martin Jambor <mjambor@suse.cz>
19656 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19657 for backward compatibility only.
19658 * doc/invoke.texi (Option Summary): Remove all references to
19659 -fipa-cp-alignment.
19661 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19665 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19667 * lra-constraints.c (curr_insn_transform): Handle
19668 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19670 2017-02-21 Martin Liska <mliska@suse.cz>
19672 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19674 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19676 PR translation/79638
19677 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19679 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19682 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19683 (arm_function_ok_for_sibcall): Return false for an indirect call by
19684 descriptor if all the argument registers are used.
19685 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19686 alignment of the function.
19688 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19690 PR tree-optimization/61441
19691 * simplify-rtx.c (simplify_const_unary_operation): For
19692 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19693 the sNaN unmodified.
19695 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19697 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19698 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19699 instead of SYSTEM_HEADER_DIR.
19701 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19702 Martin Liška <mliska@suse.cz>
19704 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19705 Fix typos and grammar, use active voice, and clarify.
19707 2017-02-20 Marek Polacek <polacek@redhat.com>
19709 PR middle-end/79537
19710 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19713 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19715 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19718 * config/i386/i386.c (ix86_expand_builtin): Handle
19719 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19720 ix86_builtins_isa[fcode].isa as a requirement of those
19721 flags and any other flag in the bitmask.
19722 (ix86_init_mmx_sse_builtins): Use 0 instead of
19723 ~OPTION_MASK_ISA_64BIT as mask.
19724 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19725 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19726 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19727 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19729 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19732 * lra-constraints.c (split_reg): Check requested split mode
19733 is supported by the register.
19735 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19737 * lra-constraints.c (simplify_operand_subreg): Remove early
19740 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19743 * lra-constraints.c (curr_insn_transform): Tighten condition
19744 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19746 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19749 * lra-constraints.c (curr_insn_transform): Handle
19750 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19752 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19755 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19757 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19759 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19762 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19765 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19767 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19770 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19772 * final.c (last_columnnum, override_columnnum): New variables.
19773 (final_start_function): Set last_columnnum, pass it to begin_prologue
19774 hook and pass 0 to dwarf2out_begin_prologue.
19775 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19776 to source_line debug hook.
19777 (notice_source_line): Compute last_columnnum and for debug_column_info
19778 return true on column changes.
19779 * debug.h (struct gcc_debug_hooks): Add column argument to
19780 source_line and begin_prologue hooks.
19781 (debug_nothing_int_charstar_int_bool): Remove prototype.
19782 (debug_nothing_int_int_charstar,
19783 debug_nothing_int_int_charstar_int_bool): New prototypes.
19784 (dwarf2out_begin_prologue): Add column argument.
19785 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19786 begin_prologue hooks.
19787 (debug_nothing_int_charstar_int_bool): Remove.
19788 (debug_nothing_int_int_charstar,
19789 debug_nothing_int_int_charstar_int_bool): New functions.
19790 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19791 through to dwarf2out_source_line.
19792 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19793 (dwarf2out_source_line): Add column argument, emit it if requested.
19794 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19796 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19797 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19798 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19799 through to dwarf2out_begin_prologue.
19800 (vmsdbgout_source_line): Add column argument, pass it through to
19801 dwarf2out_source_line.
19802 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19803 dbxout_source_line caller.
19804 (dbxout_source_line): Add column argument.
19806 * common.opt (gno-column-info, gcolumn-info): New options.
19807 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19808 (check_die): Also test for multiple DW_AT_decl_column attributes.
19809 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19810 DW_AT_decl_column if requested.
19811 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19813 (gen_variable_die): Likewise.
19814 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19815 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19818 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19819 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19820 (ix86_handle_option): Handle OPT_m3dnowa.
19821 * doc/invoke.texi (-m3dnowa): Document.
19822 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19823 -m3dnowa instead of -m3dnow -march=athlon.
19826 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19827 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19829 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19832 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19833 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19834 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19835 generator for vsx_xxpermdi_<mode>_be.
19836 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19837 force big-endian semantics.
19838 (vsx_xxpermdi_<mode>_be): New define_expand with same
19839 implementation as previous version of vsx_xxpermdi_<mode>.
19841 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19843 PR tree-optimization/79327
19844 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19845 variable, its initialization and use.
19847 2017-02-17 Julia Koval <julia.koval@intel.com>
19849 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19850 (OPTION_MASK_ISA_PKU_UNSET): New.
19851 (ix86_handle_option): Handle -mrdpid.
19852 * config/i386/cpuid.h (bit_RDPID): New.
19853 * config/i386/driver-i386.c (host_detect_local_cpu):
19854 Detect RDPID feature.
19855 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19856 * config/i386/i386-c.c (ix86_target_macros_internal):
19858 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19859 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19860 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19861 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19862 * config/i386/i386.md (define_insn "rdpid"): New.
19863 * config/i386/i386.opt Add -mrdpid.
19864 * config/i386/immintrin.h (_rdpid_u32): New.
19866 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19868 PR rtl-optimization/79541
19869 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19870 instead of transforming it into USE.
19872 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19874 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19875 If HONOR_SNANS (SFmode) force the input to a register.
19876 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19877 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19878 an frsp or similar insn.
19880 2017-02-17 Martin Liska <mliska@suse.cz>
19882 PR rtl-optimization/79577
19883 * params.def (selsched-max-sched-times): Increase minimum to 1.
19885 2017-02-17 Martin Liska <mliska@suse.cz>
19887 PR rtl-optimization/79574
19888 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19890 2017-02-17 Martin Liska <mliska@suse.cz>
19892 PR tree-optimization/79529
19893 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19894 ssa_defined_default_def_p to handle cases which are implicitly
19896 * tree-ssa.c (ssa_defined_default_def_p): New function.
19897 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19898 which are implicitly defined.
19899 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19901 2017-02-17 Richard Biener <rguenther@suse.de>
19903 PR middle-end/79576
19904 * params.def (max-ssa-name-query-depth): Limit to 10.
19906 2017-02-17 Richard Biener <rguenther@suse.de>
19908 PR tree-optimization/79552
19909 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19912 2017-02-17 Richard Biener <rguenther@suse.de>
19915 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19917 2017-02-17 Marek Polacek <polacek@redhat.com>
19919 PR middle-end/79536
19920 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19921 (fold_negate_expr): New wrapper.
19923 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19925 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19926 Correct terminology and de-emphasize pre-standard behavior.
19928 2017-02-16 Alan Modra <amodra@gmail.com>
19930 PR rtl-optimization/79286
19931 * ira.c (def_dominates_uses): New function.
19932 (update_equiv_regs): Don't create an equivalence for insns that
19933 may trap where the register def does not dominate the use.
19935 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19937 PR rtl-optimization/78127
19938 * lra.c (lra): Call lra_eliminate before finish the loop after
19941 2017-02-16 Richard Biener <rguenther@suse.de>
19943 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19945 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19946 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19947 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19948 (isl_val_int_from_wi): New function.
19949 (extract_affine_gmp): Rename to ...
19950 (extract_affine_wi): ... this, take a widest_int.
19951 (extract_affine_int): Just wrap extract_affine_wi.
19952 (add_param_constraints): Use isl_val_int_from_wi.
19953 (add_loop_constraints): Likewise, and extract_affine_wi.
19955 2017-02-15 Jeff Law <law@redhat.com>
19957 PR middle-end/79521
19958 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19959 ira_init_register_move_cost_if_necessary.
19961 2017-02-15 Martin Sebor <msebor@redhat.com>
19963 PR middle-end/32003
19964 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19965 removed in a prior commit.
19967 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19969 PR tree-optimization/79347
19970 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19971 counters during peeling.
19973 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19975 * Makefile.in (site.exp): Remove "set ISLVER".
19977 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19980 * real.c (real_from_integer): Call real_convert even for decimal.
19982 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19985 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19987 2017-02-14 Andrew Pinski <apinski@cavium.com>
19989 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19990 cores and change the partno/implementer to be correct.
19991 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19992 the 'B" as the implementer.
19993 * config/aarch64/aarch64-tune.md: Regenerate.
19995 2017-02-14 Carl Love <cel@us.ibm.com>
19997 * config/rs6000/rs6000.c: Add case statement entry to make the
19998 xvcvuxdsp built-in argument unsigned.
19999 * config/rs6000/vsx.md: Fix the source and return operand types so they
20000 match the instruction definitions from the ISA document. Fix typo
20001 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20004 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20007 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20008 member early_clobber_alts.
20009 * lra-lives.c (reg_early_clobber_p): New.
20010 (process_bb_lives): Use it.
20011 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20012 (debug_operand_data): Initialize early_clobber_alts.
20013 (setup_operand_alternative): Set up early_clobber_alts.
20014 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20015 alternatives to new_insn_reg.
20016 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20018 (lra_update_insn_regno_info): Pass the new arg.
20020 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20022 PR middle-end/79505
20023 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20024 (new_oacc_loop_raw): Don't clear already cleared fields.
20027 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20028 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20029 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20031 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20034 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20036 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20039 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20040 the extra instruction to the right place to store 128-bit constant
20043 2017-02-14 Martin Sebor <msebor@redhat.com>
20045 PR middle-end/79448
20046 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20047 warning for strings of unknown length.
20049 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20051 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20053 2017-02-14 Jeff Law <law@redhat.com>
20056 * ira-costs.c (scan_one_insn): Initialize register move costs
20057 for pseudos seen in USE/CLOBBER insns.
20059 PR tree-optimization/79095
20060 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20061 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20062 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20063 if the operands are known to be not equal, then the resulting range
20065 (intersect_ranges): If the new range is ~[0,0] and the old range is
20066 wide, then prefer ~[0,0].
20067 * tree-vrp.c (overflow_comparison_p_1): New function.
20068 (overflow_comparison_p): New function.
20069 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20070 if NAME is used in an overflow test.
20071 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20072 overflow check that can be expressed as an equality test, then adjust
20073 ops to be that equality test.
20075 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20077 * config/s390/s390-builtin-types.def: Remove flags argument.
20078 * config/s390/s390.c (s390_init_builtins): Likewise.
20080 2017-02-14 Martin Liska <mliska@suse.cz>
20082 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20083 vector. Fix trailing white spaces.
20085 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20087 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20090 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20092 PR rtl-optimization/68664
20093 * config/arm/arm.c (arm_sched_can_speculate_insn):
20094 New function. Declare prototype.
20095 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20097 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20099 PR rtl-optimization/68664
20100 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20102 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20104 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20106 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20107 max skip bytes for function, loop and jump.
20109 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20111 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20112 ABS_EXPR for gimple dump.
20114 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20117 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20119 PR tree-optimization/79408
20120 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20121 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20122 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20123 also if rhs1 is INTEGER_CST.
20125 2017-02-14 Richard Biener <rguenther@suse.de>
20127 PR middle-end/79432
20128 * tree-into-ssa.c (insert_phi_nodes): When the function can
20129 have abnormal edges rewrite SSA names with broken use-def
20130 dominance out of SSA and register them for PHI insertion.
20132 2017-02-13 Martin Sebor <msebor@redhat.com>
20134 PR middle-end/79496
20135 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20136 clearing info.nowrite flag when snprintf size argument is a range.
20138 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20140 * cprop.c (cprop_jump): Add missing space in string literal.
20141 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20142 (get_constraint_for_component_ref): Likewise.
20143 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20144 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20145 * lra-constraints.c (process_alt_operands): Likewise.
20146 * ipa-inline.c (inline_small_functions): Likewise.
20147 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20148 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20149 * trans-mem.c (diagnose_tm_1_op): Likewise.
20150 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20151 (grid_parallel_clauses_gridifiable): Likewise.
20153 * config/nvptx/mkoffload.c (process): Add space in between
20156 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20157 "MOD4_SSE_REGS" and "ALL_REGS".
20159 * spellcheck.c (test_data): Add , in between "foo" and "food".
20161 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20164 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20165 boundary crossing check and subsequent code generation agree.
20167 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20169 * config/aarch64/aarch64.c (has_memory_op): Delete.
20170 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20173 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20175 PR rtl-optimization/79388
20176 PR rtl-optimization/79450
20177 * combine.c (distribute_notes): When removing TEM_INSN for which
20178 corresponding dest has last value recorded, invalidate that last
20181 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20183 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20184 of explicit '@'. Add missing assembly comment marker on branch costs
20187 2017-02-13 Nathan Sidwell <nathan@acm.org>
20189 * gengtype-lex.l (<in_struct>): Add '/'.
20191 2017-02-13 Martin Liska <mliska@suse.cz>
20194 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20196 2017-02-13 Richard Biener <rguenther@suse.de>
20198 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20200 * configure: Re-generate.
20201 * config.in: Likewise.
20202 * graphite-dependences.c: Simplify as if
20203 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20204 * graphite-isl-ast-to-gimple.c: Likewise.
20205 * graphite-optimize-isl.c: Likewise.
20206 * graphite-poly.c: Likewise.
20207 * graphite-sese-to-poly.c: Likewise.
20208 * graphite.h: Likewise.
20209 * toplev.c: Include isl/version.h and use isl_version () for
20210 printing the ISL version.
20211 * doc/install.texi: Update ISL requirement.
20213 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20215 * doc/standards.texi (Standards): Update reference to
20218 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20220 * doc/extend.texi (Named Address Spaces): sourceware.org now
20222 * doc/install.texi (Binaries): Ditto.
20225 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20227 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20228 standard terminology "stringize"/"stringizing" throughout.
20229 * doc/cppinternals.texi: Likewise.
20231 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20233 * doc/extend.texi: Fix some spelling mistakes and typos.
20234 * doc/invoke.texi: Likewise.
20236 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20239 * params.def (inline-min-speedup) Change from 10 to 8.
20241 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20243 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20246 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20249 * ipa-inline-analysis.c (get_minimal_bb): New function.
20250 (record_modified): Use it.
20251 (remap_edge_change_prob): Handle also ancestor functions.
20253 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20255 * doc/contrib.texi (Contributors): Remove broken link into
20256 the Mauve CVS repository.
20258 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20260 PR middle-end/79454
20261 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20262 result computation whenever lhs doesn't have vector mode, not
20263 just when it has BLKmode.
20265 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20267 * doc/makefile.texi (profiledbootstrap): Refer to the
20268 installation instructions only in textual form.
20270 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20273 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20275 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20277 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20278 (Specific): Update mingw-w64 reference.
20280 (Specific): Remove broken link to Renesas RX processor.
20282 2017-02-10 Richard Biener <rguenther@suse.de>
20284 * toplev.c (process_options): Do not mention obsolete graphite
20285 options when printing sorry message about missing graphite support.
20286 Mention -floop-nest-optimize.
20288 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20290 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20291 (vtst_p16): Likewise.
20292 (vtstq_p8): Likewise.
20293 (vtstq_p16): Likewise.
20295 (vtstq_p64): Likewise.
20296 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20297 (vset_lane_p64): New.
20298 (vsetq_lane_p64): New.
20300 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20302 PR tree-optimization/79411
20303 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20304 stmt operands are SSA_NAMEs used in abnormal phis.
20305 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20308 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20311 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20314 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20316 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20318 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20320 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20324 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20325 not arbitrary TREE_CONSTANT.
20328 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20329 "omp declare target link" attribute unless is_global_var.
20330 * omp-offload.c (find_link_var_op): Likewise.
20332 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20333 Chung-Lin Tang <cltang@codesourcery.com>
20335 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20337 (gimplify_adjust_omp_clauses): Don't delete TILE.
20338 (gimplify_omp_for): Deal with TILE.
20339 * internal-fn.c (expand_GOACC_TILE): New function.
20340 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20342 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20343 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20345 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20346 avoid DIV for outermost collapse var.
20347 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20348 Remove out of date comments, fix whitespace.
20349 * omp-general.c (omp_extract_for_data): Deal with tiling.
20350 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20351 adjust OLF_DIM_BASE value.
20352 (struct omp_for_data): Add tiling field.
20353 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20354 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20355 for auto loops. Remove default auto determining, moved to
20356 oacc_loop_fixed_partitions.
20357 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20358 stmts, add e_mask field.
20359 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20360 (oacc_thread_numbers): Use oacc_dim_call.
20361 (oacc_xform_tile): New.
20362 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20363 (finish_oacc_loop): Adjust for ifns vector.
20364 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20365 add case for GOACC_TILE fns.
20366 (oacc_loop_xform_loop): Delete.
20367 (oacc_loop_process): Iterate over call list directly, and add
20368 handling for GOACC_TILE fns.
20369 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20371 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20372 vector partitioning to outer loops. Assign 2 partitions to loops
20373 when available. Add TILE handling.
20374 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20375 (execite_oacc_device_lower): Process GOACC_TILE fns,
20376 ignore unknown specs.
20377 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20378 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20379 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20381 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20383 * configure.ac (ACX_BUGURL): Update.
20384 * configure: Regenerate.
20386 2017-02-09 Richard Biener <rguenther@suse.de>
20388 PR tree-optimization/69823
20389 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20390 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20392 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20394 * config/arc/arc-c.def: Add __NPS400__ definition.
20395 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20396 (TARGET_NPS400): Define.
20398 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20400 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20402 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20403 pointer, arch_info.
20404 (arc_cpu_types): Fill the arch_info field with a pointer into the
20405 arc_arch_types table.
20406 (arc_selected_cpu): Declare.
20407 * config/arc/arc.c (arc_selected_cpu): Make global.
20408 (arc_selected_arch): Delete.
20409 (arc_base_cpu): Delete.
20410 (arc_override_options): Remove references to deleted variables,
20411 update access to arch information.
20412 (ARC_OPT): Update access to arch information.
20413 (ARC_OPTX): Likewise.
20414 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20415 (TARGET_ARC600): Update access to arch information.
20416 (TARGET_ARC601): Likewise.
20417 (TARGET_ARC700): Likewise.
20418 (TARGET_EM): Likewise.
20419 (TARGET_HS): Likewise.
20420 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20423 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20426 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20427 condition/operands for integer GE/LE/GEU/LEU operations.
20429 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20431 PR translation/79397
20432 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20435 2017-02-08 Martin Jambor <mjambor@suse.cz>
20438 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20439 whether allocation happened.
20440 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20441 nothing was allocated.
20443 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20445 PR tree-optimization/79408
20446 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20447 constant, but SSA_NAME with a known integer range, use the minimum
20448 of that range instead of op1 to determine if modulo can be replaced
20449 with its first operand.
20451 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20453 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20455 2017-02-08 Richard Biener <rguenther@suse.de>
20457 PR tree-optimization/71824
20458 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20459 Check all loops contained in the merged region.
20461 2017-02-07 Andrew Pinski <apinski@cavium.com>
20463 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20465 2017-02-07 Andrew Pinski <apinski@cavium.com>
20467 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20468 (thunderxt88): Likewise.
20469 (thunderxt81): Disable LSE and change v8.1 to v8.
20470 (thunderxt83): Likewise.
20472 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20473 Richard Biener <rguenther@suse.de>
20475 PR middle-end/79399
20476 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20477 type from int to size_t.
20478 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20480 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20482 PR rtl-optimization/79386
20483 * cprop.c (bypass_conditional_jumps): Initialize
20484 bypass_last_basic_block already before splitting bbs after
20485 unconditional traps...
20486 (bypass_conditional_jumps): ... rather than here.
20489 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20490 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20491 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20492 fix -masm=intel patterns.
20494 2017-02-07 Richard Biener <rguenther@suse.de>
20496 PR tree-optimization/79256
20497 PR middle-end/79278
20498 * builtins.c (get_object_alignment_2): Use min_align_of_type
20499 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20500 and ADJUST_FIELD_ALIGN.
20502 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20504 * doc/tm.texi: Regenerate.
20505 * stor-layout.c (layout_decl): Adjust.
20506 (update_alignment_for_field): Likewise.
20507 (place_field): Likewise.
20508 (min_align_of_type): Likewise.
20509 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20510 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20511 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20512 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20513 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20514 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20515 * config/i386/i386.c (x86_field_alignment): Likewise.
20516 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20517 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20518 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20519 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20520 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20521 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20525 2017-01-30 Richard Biener <rguenther@suse.de>
20527 PR tree-optimization/79256
20528 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20529 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20532 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20534 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20535 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20536 builtins to SImode and emit a zero-extend, if necessary.
20538 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20540 * docs/invoke.texi (RISC-V Options): Alphabetize.
20542 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20544 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20547 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20549 * config/riscv/riscv.c: New file.
20550 * common/config/riscv/riscv-common.c: Likewise.
20551 * config.gcc: Likewise.
20552 * config/riscv/constraints.md: Likewise.
20553 * config/riscv/elf.h: Likewise.
20554 * config/riscv/generic.md: Likewise.
20555 * config/riscv/linux.h: Likewise.
20556 * config/riscv/multilib-generator: Likewise.
20557 * config/riscv/peephole.md: Likewise.
20558 * config/riscv/pic.md: Likewise.
20559 * config/riscv/predicates.md: Likewise.
20560 * config/riscv/riscv-builtins.c: Likewise.
20561 * config/riscv/riscv-c.c: Likewise.
20562 * config/riscv/riscv-ftypes.def: Likewise.
20563 * config/riscv/riscv-modes.def: Likewise.
20564 * config/riscv/riscv-opts.h: Likewise.
20565 * config/riscv/riscv-protos.h: Likewise.
20566 * config/riscv/riscv.h: Likewise.
20567 * config/riscv/riscv.md: Likewise.
20568 * config/riscv/riscv.opt: Likewise.
20569 * config/riscv/sync.md: Likewise.
20570 * config/riscv/t-elf-multilib: Likewise.
20571 * config/riscv/t-linux: Likewise.
20572 * config/riscv/t-linux-multilib: Likewise.
20573 * config/riscv/t-riscv: Likewise.
20574 * configure.ac: Likewise.
20575 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20576 Waterman as RISC-V maintainers.
20577 * doc/install.texi: Add RISC-V entries.
20578 * doc/invoke.texi: Add RISC-V options section.
20579 * doc/md.texi: Add RISC-V constraints section.
20580 * configure: Regenerated.
20582 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20585 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20586 false values to be constant vectors with all 0 or all 1 bits set.
20587 (vcondu<mode><mode>): Likewise.
20588 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20590 (fpmask_comparison_operator): Update comment.
20591 (vecint_comparison_operator): New predicate.
20592 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20593 vector conditionals when the true and false values are constant
20594 vectors with all 0 bits or all 1 bits set.
20596 2017-02-06 Martin Sebor <msebor@redhat.com>
20598 PR tree-optimization/79376
20599 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20601 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20603 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20604 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20605 to simplify split condition.
20607 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20609 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20610 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20613 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20615 PR rtl-optimization/68664
20616 * target.def (can_speculate_insn): New hook.
20617 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20618 * doc/tm.texi: Regenerate.
20619 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20620 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20621 (rs6000_sched_can_speculate_insn): New function.
20623 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20625 PR tree-optimization/79284
20626 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20627 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20628 vectorizable_mask_load_store, vectorizable_operation,
20629 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20630 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20631 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20632 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20633 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20634 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20635 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20636 is_gimple_assign (stmt). Replace another such test with
20637 is_gimple_assign (stmt).
20639 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20642 * config/avr/avr.c (rtl-iter.h): Include it.
20643 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20644 (avr_legitimate_combined_insn): ...and implementation.
20646 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20648 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20649 * config/s390/s390.c (s390_const_operand_ok)
20650 (s390_canonicalize_comparison, s390_extract_part)
20651 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20652 (s390_contiguous_bitmask_p, s390_rtx_costs)
20653 (legitimize_pic_address): Likewise.
20654 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20655 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20656 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20657 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20658 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20660 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20662 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20663 REGNO($0) == REGNO($1).
20665 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20667 * config/s390/linux.h(SIZE_TYPE): Add comment.
20669 2017-02-06 Julian Brown <julian@codesourcery.com>
20670 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20671 Virendra Pathak <virendra.pathak@broadcom.com>
20673 * config/aarch64/aarch64-cores.def: Change the scheduler
20675 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20676 * config/aarch64/thunderx2t99.md: New file.
20678 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20680 * doc/standards.texi (Go Language): Update link to language
20683 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20685 * tree-eh.c (lower_resx): Sanitize profile.
20686 (cleanup_empty_eh_move_lp): Likewise.
20688 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20691 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20693 * cfgloopmanip.h (loop_version): Update prototype.
20694 * modulo-sched.c (sms_schedule): Update call of loop_version.
20695 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20696 * tree-parloops.c (gen_parallel_loop): Likewise.
20697 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20698 * tree-ssa-loop-split.c (split_loop): Likewise.
20699 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20700 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20702 2017-02-05 Martin Liska <mliska@suse.cz>
20705 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20707 (print_operand_address): Initialize a struct to zero.
20709 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20711 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20712 garbage collector only in textual form.
20714 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20716 * doc/extend.texi (x86 specific memory model extensions for
20717 transactional memory): Simplify a phrase.
20719 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20722 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20723 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20724 (atomic_storedi_1): Likewise.
20726 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20728 PR tree-optimization/79338
20729 * tree-parloops.c (gather_scalar_reductions): Don't call
20730 vect_analyze_loop_form for loop->inner before destroying loop's
20733 2017-02-03 Martin Sebor <msebor@redhat.com>
20735 PR tree-optimization/79327
20736 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20737 when precision has resulted in leading zeros.
20738 (format_integer): Adjust the likely counter to assume an unknown
20739 argument that may be zero is non-zero.
20741 2017-02-03 Jason Merrill <jason@redhat.com>
20744 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20745 avoid copying non-taken branch.
20747 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20749 PR tree-optimization/79340
20750 * tree-vect-loop.c (vectorizable_reduction): Release
20751 vec_defs elements after safe_splicing them into other vectors.
20754 PR tree-optimization/79327
20755 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20756 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20758 (format_integer): Use wide_int_to_tree instead of build_int_cst
20759 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20760 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20761 of shortest and longest sequence.
20763 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20765 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20766 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20768 2017-02-03 Walter Lee <walt@tilera.com>
20771 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20772 after initial stackframe link reg save.
20773 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20775 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20778 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20779 wu for stxssp alternative.
20781 2017-02-03 Martin Sebor <msebor@redhat.com>
20783 PR tree-optimization/79352
20784 * gimple-fold.c (get_range_strlen): Add argument.
20785 (get_range_strlen): Change return type to bool.
20786 (get_maxval_strlen): Pass in a dummy argument.
20787 * gimple-fold.h (get_range_strlen): Change return type to bool.
20788 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20789 * tree.h (array_at_struct_end_p): Add argument.
20790 * tree.c (array_at_struct_end_p): Handle it.
20792 2017-02-03 Martin Liska <mliska@suse.cz>
20795 * multiple_target.c (create_dispatcher_calls): Redirect edge
20796 from a caller of a dispatcher.
20797 (expand_target_clones): Make the clones local.
20798 (ipa_target_clone): Do both target clones and resolvers.
20799 (ipa_dispatcher_calls): Remove the pass.
20800 (pass_dispatcher_calls::gate): Likewise.
20801 (make_pass_dispatcher_calls): Likewise.
20802 * passes.def (pass_target_clone): Put as very first IPA early
20805 2017-02-03 Martin Liska <mliska@suse.cz>
20807 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20808 in case of a function with ifunc attribute.
20810 2017-02-03 Martin Liska <mliska@suse.cz>
20812 * cgraph.c (cgraph_node::dump): Dump function version info.
20813 * symtab.c (symtab_node::dump_base): Add missing new line.
20815 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20817 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20818 (ifcombine_ifandif): Use it.
20820 2017-02-03 Martin Liska <mliska@suse.cz>
20822 * doc/invoke.texi: Document default value for
20823 use-after-scope-direct-emission-threshold.
20825 2017-02-03 Martin Liska <mliska@suse.cz>
20827 PR tree-optimization/79339
20828 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20829 (format_floating): Likewise.
20831 2017-02-03 Martin Liska <mliska@suse.cz>
20834 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20836 (ipa_node_params_t::remove): Likewise.
20837 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20838 initialization from removed ipa_node_params_t::insert.
20839 (ipa_node_params::~ipa_node_params): Move from removed
20840 ipa_node_params_t::release.
20841 * symbol-summary.h (symbol_summary::m_released): New member.
20842 Do not release a summary twice. Do not allow to call finalizer
20843 for types of a summary that live in GGC memory.
20845 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20847 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20850 2017-02-02 Martin Sebor <msebor@redhat.com>
20852 PR middle-end/79275
20853 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20854 (format_string): Tighten up the range of output for non-constant
20855 strings and correct the expected range for wide non-constant strings.
20857 2017-02-02 Martin Sebor <msebor@redhat.com>
20859 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20861 PR middle-end/32003
20862 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20864 (-fdump-tree-@var): Add to index and document how to come up
20865 with pass-specific option and dump file names.
20866 (-fdump-passes): Clarify where to look for output.
20868 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20870 PR middle-end/77445
20871 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20872 statistics of the analyzed path; allow threading for speed when
20873 any of BBs along the path are optimized for speed.
20875 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20877 PR middle-end/78468
20878 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20879 settings of the virtual registers.
20882 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20884 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20885 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20888 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20890 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20891 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20893 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20895 * config/s390/s390.md: Add missing comments with the expanded
20897 * config/s390/vector.md: Likewise.
20898 * config/s390/vx-builtins.md: Likewise.
20900 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20903 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20904 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20905 conditions on a single line.
20907 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20909 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20910 __S390_VX__ to __VX__.
20912 2017-02-01 Andrew Pinski <apinski@cavium.com>
20914 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20915 stmt_info to record_stmt_cost.
20916 (vect_get_known_peeling_cost): Pass stmt_info if known to
20918 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20919 cpu_vector_cost field into
20920 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20921 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20922 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20923 splitting of scalar_stmt_cost and vec_stmt_cost.
20924 (thunderx_vector_cost): Likewise.
20925 (cortexa57_vector_cost): LIkewise.
20926 (exynosm1_vector_cost): Likewise.
20927 (xgene1_vector_cost): Likewise.
20928 (thunderx2t99_vector_cost): Improve after the splitting of the two
20930 (aarch64_builtin_vectorization_cost): Update for the splitting of
20931 scalar_stmt_cost and vec_stmt_cost.
20933 2017-02-01 Torvald Riegel <triegel@redhat.com>
20934 Richard Henderson <rth@redhat.com>
20936 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20937 conditional on existance of a fast atomic load.
20938 * optabs-query.c (can_atomic_load_p): New function.
20939 * optabs-query.h (can_atomic_load_p): Declare it.
20940 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20941 no fast atomic load is available for the particular size of access.
20942 (expand_atomic_compare_and_swap): Likewise.
20943 (expand_atomic_load): Likewise.
20944 (expand_atomic_store): Likewise.
20945 (expand_atomic_fetch_op): Likewise.
20946 * testsuite/lib/target-supports.exp
20947 (check_effective_target_sync_int_128): Remove x86 because it provides
20948 no fast atomic load.
20949 (check_effective_target_sync_int_128_runtime): Likewise.
20951 2017-02-01 Richard Biener <rguenther@suse.de>
20953 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20954 (graphite_transform_loops): Provide opt-info for optimized nests.
20955 * tree-parloop.c (parallelize_loops): Provide opt-info for
20956 parallelized loops.
20958 2017-02-01 Richard Biener <rguenther@suse.de>
20960 PR middle-end/79315
20961 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20962 was not set before.
20964 2017-02-01 Richard Biener <rguenther@suse.de>
20966 PR tree-optimization/71824
20967 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20968 Verify the loops are valid in the merged SESE region.
20969 (scop_detection::can_represent_loop_1): Check analyzing the
20970 evolution of the number of iterations in the region succeeds.
20972 2017-01-31 Ian Lance Taylor <iant@golang.org>
20974 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20975 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20977 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20979 PR preprocessor/79210
20980 * input.c (get_substring_ranges_for_loc): Replace line_width
20981 assertion with error-handling.
20983 2017-01-31 Richard Biener <rguenther@suse.de>
20985 PR tree-optimization/77318
20986 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20987 (create_pw_aff_from_tree): Take loop parameter.
20988 (add_condition_to_pbb): Pass loop of the condition to
20989 create_pw_aff_from_tree.
20991 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20993 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20994 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20996 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21000 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21002 (convert_int_to_float128): Likewise.
21003 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21004 (convert_int_to_float128): Likewise.
21005 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21006 (UNSPEC_IEEE128_CONVERT): Likewise.
21007 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21008 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21009 Use local variables for IBM extended format.
21010 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21011 (fix_trunc<mode>si2_fprs): Likewise.
21012 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21013 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21014 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21015 to know that we can now have integers of all sizes in vector
21017 (fix<uns>_<mode>di2_hw): Likewise.
21018 (float<uns>_<mode>si2_hw): Likewise.
21019 (fix_<mode>si2_hw): Likewise.
21020 (fixuns_<mode>si2_hw): Likewise.
21021 (float<uns>_<mode>di2_hw): Likewise.
21022 (float_<mode>di2_hw): Likewise.
21023 (float_<mode>si2_hw): Likewise.
21024 (floatuns_<mode>di2_hw): Likewise.
21025 (floatuns_<mode>si2_hw): Likewise.
21026 (xscvqp<su>wz_<mode>): Delete, no longer used.
21027 (xscvqp<su>dz_<mode>): Likewise.
21028 (xscv<su>dqp_<mode>): Likewise.
21029 (ieee128_mfvsrd_64bit): Likewise.
21030 (ieee128_mfvsrd_32bit): Likewise.
21031 (ieee128_mfvsrwz): Likewise.
21032 (ieee128_mtvsrw): Likewise.
21033 (ieee128_mtvsrd_64bit): Likewise.
21034 (ieee128_mtvsrd_32bit): Likewise.
21036 2017-01-31 Martin Liska <mliska@suse.cz>
21039 * ipa-prop.c (ipa_free_all_node_params): Call release method
21040 instead of ~sumbol_summary to not to trigger double times
21043 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21045 PR tree-optimization/71691
21046 * bitmap.h (class auto_bitmap): New.
21047 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21048 is_maybe_undefined instead of ssa_undefined_value_p.
21050 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21052 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21053 __S390_ARCH_LEVEL__ to __ARCH__.
21055 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21057 PR tree-optimization/79267
21058 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21059 if should_remove_lhs_p is true.
21061 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21064 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21065 (add_alignment_attribute): New.
21066 (base_type_die): Add alignment attribute.
21067 (subrange_type_die): Likewise.
21068 (modified_type_die): Likewise.
21069 (gen_array_type_die): Likewise.
21070 (gen_descr_array_type_die: Likewise.
21071 (gen_enumeration_type_die): Likewise.
21072 (gen_subprogram_die): Likewise.
21073 (gen_variable_die): Likewise.
21074 (gen_field_die): Likewise.
21075 (gen_ptr_to_mbr_type_die): Likewise.
21076 (gen_struct_or_union_type_die): Likewise.
21077 (gen_subroutine_type_die): Likewise.
21078 (gen_typedef_die): Likewise.
21079 (base_type_cmp): Compare alignment attribute.
21081 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21084 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21085 (setb_unsigned) New pattern for setb with CCUNS.
21086 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21087 subfc./subfe sequence to avoid overflow problems. Generate a
21088 shorter sequence with cmpld/setb for power9.
21089 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21090 for generating subfc. instruction.
21091 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21092 now uses this instruction.
21094 2017-01-30 Ian Lance Taylor <iant@google.com>
21097 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21098 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21100 2017-01-30 Martin Sebor <msebor@redhat.com>
21102 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21103 Move constant to the right of a relational operator.
21104 (get_mpfr_format_length, format_character, format_string): Ditto.
21105 (should_warn_p, maybe_warn): Same.
21107 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21109 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21112 * asan.c (get_translation_unit_decl): Remove function.
21113 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21115 2017-01-30 Martin Liska <mliska@suse.cz>
21117 PR gcov-profile/79259
21118 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21119 -fprofile-generate.
21121 2017-01-30 Martin Liska <mliska@suse.cz>
21124 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21125 Initialize variables with NULL value.
21127 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21130 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21132 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21134 2017-01-30 Richard Biener <rguenther@suse.de>
21136 PR tree-optimization/79276
21137 * tree-vrp.c (process_assert_insertions): Properly adjust common
21138 when removing a duplicate.
21140 * gcc.dg/torture/pr79276.c: New testcase.
21142 2017-01-30 Richard Biener <rguenther@suse.de>
21144 PR tree-optimization/79256
21145 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21146 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21148 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21150 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21153 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21154 ("*r<noxa>sbg_<mode>_sll_bitmask")
21155 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21156 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21157 Use contiguous_bitmask_nowrap_operand.
21159 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21162 * config/rs6000/altivec.h (vec_xl): Revise #define.
21163 (vec_xst): Likewise.
21165 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21167 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21169 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21171 PR rtl-optimization/79194
21172 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21173 traps before call to bypass_conditional_jumps.
21175 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21177 PR tree-optimization/71374
21178 * lra-constraints.c (check_conflict_input_operands): New.
21179 (match_reload): Use it.
21181 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21184 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21185 account to calculate conflict_set.
21187 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21189 PR rtl-optimization/78559
21190 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21191 other_insn in combine.
21193 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21195 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21196 uint16_type_node for BT_UINT16.
21198 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21200 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21201 "RTL Tests" to menu.
21202 (GIMPLE Tests): New node.
21203 (RTL Tests): New node.
21205 2017-01-27 Richard Biener <rguenther@suse.de>
21207 PR tree-optimization/79245
21208 * tree-loop-distribution.c (distribute_loop): Apply cost
21209 modeling also to detected patterns.
21211 2017-01-27 Richard Biener <rguenther@suse.de>
21213 PR tree-optimization/71433
21214 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21215 (compare_assert_loc): New function.
21216 (process_assert_insertions): Sort and optimize assert locations
21217 to remove duplicates and push down identical assertions on
21218 edges to their destination block.
21220 2017-01-27 Richard Biener <rguenther@suse.de>
21222 PR tree-optimization/79244
21223 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21224 out SSA names even if abnormal.
21226 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21228 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21229 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21230 instead of MPFR_RNDN.
21232 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21235 * arm.c (arm_option_override): Don't call build_target_option_node
21236 until after doing all option overrides.
21237 (arm_valid_target_attribute_tree): Likewise.
21239 2017-01-27 Martin Liska <mliska@suse.cz>
21241 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21242 for {cd}tors and C++ {cd}tors.
21244 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21246 * config/s390/s390.md ("*setmem_long_and")
21247 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21249 2017-01-26 Martin Sebor <msebor@redhat.com>
21251 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21254 2017-01-26 Martin Sebor <msebor@redhat.com>
21256 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21257 HAVE_DFmode before using XFmode or DFmode.
21258 (parse_directive): Avoid using the z length modifier to avoid
21259 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21261 PR middle-end/78703
21262 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21263 to accept adjustment as an array.
21264 (get_int_range): New function.
21265 (struct directive): Make width and prec arrays.
21266 (directive::set_width, directive::set_precision): Call get_int_range.
21267 (format_integer, format_floating): Handle width and precision ranges.
21268 (format_string, parse_directive): Same.
21270 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21273 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21274 comdat_type_p set, just clone them, but keep the children in the
21278 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21279 which have direct callers with -fvar-tracking-assignments enabled
21281 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21282 inside of type units.
21284 2017-01-26 Martin Sebor <msebor@redhat.com>
21286 PR middle-end/78703
21287 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21289 (struct format_result): Replace number_chars, number_chars_min,
21290 and number_chars_max with a single member of struct result_range.
21292 (format_result::operator+=): Adjust.
21293 (struct fmtresult): Remove bounded. Handle likely and unlikely
21295 (fmtresult::adjust_for_width_or_precision): New function.
21296 (fmtresult:type_max_digits): New function.
21297 (bytes_remaining): Handle likely and unlikely counters.
21298 (min_bytes_remaining): Remove.
21299 (format_percent): Simplify.
21300 (format_integer, format_floating): Set likely and unlikely counters.
21301 (get_string_length, format_character, format_string): Same.
21302 (format_plain, should_warn_p): New function.
21303 (maybe_warn): Call should_warn_p. Update diagnostic messages
21304 and handle those for all directives, including plain strings.
21305 (format_directive): Handle likely and unlikely counters.
21306 Remove unnecessary quoting from diagnostics. Add an informational
21308 (add_bytes): Remove.
21309 (pass_sprintf_length::compute_format_length): Simplify.
21310 (try_substitute_return_value): Handle likely and unlikely counters.
21312 2017-01-26 Carl Love <cel@us.ibm.com>
21314 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21315 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21317 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21320 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21321 endianess for subregs into account.
21322 * lra-constraints.c (lra_constraints): Do risky transformations
21323 always on the first iteration.
21324 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21325 last_call_used_reg_set.
21326 (process_bb_lives): Define and use last_call_used_reg_set.
21327 * lra.c (lra): Always continue after lra_constraints on the first
21330 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21332 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21334 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21336 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21338 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21339 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21340 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21341 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21342 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21343 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21344 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21345 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21346 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21348 2017-01-26 Marek Polacek <polacek@redhat.com>
21351 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21352 for the third operand.
21354 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21356 PR middle-end/79236
21357 * omp-low.c (struct omp_context): Add simt_stmt field.
21358 (scan_omp_for): Return omp_context *.
21359 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21360 context to the _simt_ SIMD stmt.
21361 (lower_omp_for): For combined SIMD with sibling _simt_
21362 SIMD, make sure to use the same decls in _looptemp_
21363 clauses as in the sibling.
21365 2017-01-26 David Sherwood <david.sherwood@arm.com>
21367 PR middle-end/79212
21368 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21371 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21374 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21375 emit fld b; fld a; if possible.
21377 * brig-builtins.def: Update copyright years.
21378 * config/arm/arm_acle_builtins.def: Update copyright years.
21380 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21383 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21384 constraint instead of o for the stxsd instruction.
21386 2017-01-25 Carl Love <cel@us.ibm.com>
21388 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21389 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21391 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21393 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21395 2017-01-25 Richard Biener <rguenther@suse.de>
21397 PR tree-optimization/69264
21398 * target.def (vector_alignment_reachable): Improve documentation.
21399 * doc/tm.texi: Regenerate.
21400 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21402 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21403 earlier changes with respect to TYPE_USER_ALIGN.
21404 (vector_alignment_reachable_p): Likewise. Improve dumping.
21406 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21409 * config/arm/arm.md (xordi3): Force constant operand into a register
21412 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21414 * doc/invoke.texi (-fstore-merging): Correct default optimization
21415 levels at which it is enabled.
21416 (-O): Move -fstore-merging from list to...
21419 2017-01-25 Richard Biener <rguenther@suse.de>
21422 * omp-expand.c: Include debug.h.
21423 (expand_omp_taskreg): Make sure to generate early debug before
21424 outlining anything from a function.
21425 (expand_omp_target): Likewise.
21426 (grid_expand_target_grid_body): Likewise.
21428 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21431 * asan.c (get_translation_unit_decl): New function.
21432 (asan_add_global): Extract modules file name from globals
21433 TRANSLATION_UNIT_DECL name.
21435 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21438 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21439 for long calls with APCS frame and VFP.
21441 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21443 * cfg.c (original_copy_tables_initialized_p): New function.
21444 * cfg.h (original_copy_tables_initialized_p): New decl.
21445 * cfgrtl.c (relink_block_chain): Guard the call to
21446 free_original_copy_tables with a call to
21447 original_copy_tables_initialized_p.
21448 * cgraph.h (symtab_node::native_rtl_p): New decl.
21449 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21450 (symtab_node::needed_p): Don't assert for early assembly output
21451 for __RTL functions.
21452 (cgraph_node::finalize_function): Set "force_output" for __RTL
21454 (cgraph_node::analyze): Bail out early for __RTL functions.
21455 (analyze_functions): Update assertion to support __RTL functions.
21456 (cgraph_node::expand): Bail out early for __RTL functions.
21457 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21459 * function.h (struct function): Update comment for field
21461 * gimple-expr.c: Include "tree-pass.h".
21462 (gimple_has_body_p): Return false for __RTL functions.
21463 * Makefile.in (OBJS): Add run-rtl-passes.o.
21464 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21466 (gcc::pass_manager::get_clean_slate): New accessor.
21467 * passes.c: Include "insn-addr.h".
21468 (should_skip_pass_p): Add logging. Update logic for running
21469 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21470 property-provider override so it is only done for gimple passes.
21472 (skip_pass): New function.
21473 (execute_one_pass): Call skip_pass when skipping passes.
21474 * read-md.c (md_reader::read_char): Support filtering
21475 the input to a subset of line numbers.
21476 (md_reader::md_reader): Initialize fields
21477 m_first_line and m_last_line.
21478 (md_reader::read_file_fragment): New function.
21479 * read-md.h (md_reader::read_file_fragment): New decl.
21480 (md_reader::m_first_line): New field.
21481 (md_reader::m_last_line): New field.
21482 * read-rtl-function.c (function_reader::create_function): Only
21483 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21484 curr_properties. Set DECL_INITIAL to a dummy block.
21485 (read_rtl_function_body_from_file_range): New function.
21486 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21488 * run-rtl-passes.c: New file.
21489 * run-rtl-passes.h: New file.
21491 2017-01-24 Jeff Law <law@redhat.com>
21493 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21496 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21498 PR tree-optimization/79159
21499 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21500 (record_nonwrapping_iv): Improve boundary using above function if no
21501 value range information.
21503 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21504 Martin Jambor <mjambor@suse.cz>
21506 * brig-builtins.def: New file.
21507 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21508 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21509 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21510 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21511 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21512 * builtin-types.def (BT_INT8): New.
21513 (BT_INT16): Likewise.
21514 (BT_UINT8): Likewise.
21515 (BT_UINT16): Likewise.
21516 (BT_FN_ULONG): Likewise.
21517 (BT_FN_UINT_INT): Likewise.
21518 (BT_FN_UINT_ULONG): Likewise.
21519 (BT_FN_UINT_LONG): Likewise.
21520 (BT_FN_UINT_PTR): Likewise.
21521 (BT_FN_ULONG_PTR): Likewise.
21522 (BT_FN_INT8_FLOAT): Likewise.
21523 (BT_FN_INT16_FLOAT): Likewise.
21524 (BT_FN_UINT32_FLOAT): Likewise.
21525 (BT_FN_UINT16_FLOAT): Likewise.
21526 (BT_FN_UINT8_FLOAT): Likewise.
21527 (BT_FN_UINT64_FLOAT): Likewise.
21528 (BT_FN_UINT16_UINT32): Likewise.
21529 (BT_FN_UINT32_UINT16): Likewise.
21530 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21531 (BT_FN_INT_PTR_INT): Likewise.
21532 (BT_FN_UINT_PTR_UINT): Likewise.
21533 (BT_FN_LONG_PTR_LONG): Likewise.
21534 (BT_FN_ULONG_PTR_ULONG): Likewise.
21535 (BT_FN_VOID_UINT64_UINT64): Likewise.
21536 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21537 (BT_FN_INT8_INT8_INT8): Likewise.
21538 (BT_FN_INT16_INT16_INT16): Likewise.
21539 (BT_FN_INT_INT_INT): Likewise.
21540 (BT_FN_UINT_FLOAT_UINT): Likewise.
21541 (BT_FN_FLOAT_UINT_UINT): Likewise.
21542 (BT_FN_ULONG_UINT_UINT): Likewise.
21543 (BT_FN_ULONG_UINT_PTR): Likewise.
21544 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21545 (BT_FN_UINT_UINT_UINT): Likewise.
21546 (BT_FN_VOID_UINT_PTR): Likewise.
21547 (BT_FN_UINT_UINT_PTR: Likewise.
21548 (BT_FN_UINT32_UINT64_PTR): Likewise.
21549 (BT_FN_INT_INT_UINT_UINT): Likewise.
21550 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21551 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21552 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21553 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21554 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21555 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21556 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21557 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21558 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21559 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21560 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21561 * doc/frontends.texi: List BRIG FE.
21562 * doc/install.texi (Testing): Add BRIG tesring requirements.
21563 * doc/invoke.texi (Overall Options): Mention BRIG.
21564 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21566 2017-01-24 Richard Biener <rguenther@suse.de>
21568 PR translation/79208
21569 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21571 2017-01-24 Martin Jambor <mjambor@suse.cz>
21574 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21575 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21576 and known_contexts.
21578 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21580 PR middle-end/79123
21581 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21582 casts from signed to unsigned really don't have a range.
21584 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21586 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21587 GMP_RNDx for compatiblity.
21589 2017-01-24 Martin Liska <mliska@suse.cz>
21592 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21593 that would prevent us to call alloca with -1 as argument.
21595 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21597 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21598 Avoid -Wformat-security warning.
21600 2017-01-23 Andrew Pinski <apinski@cavium.com>
21602 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21605 2017-01-23 Martin Sebor <msebor@redhat.com>
21607 PR middle-end/78703
21608 * gimple-ssa-sprintf.c (warn_level): New global.
21609 (format_integer): Use it here and throughout the rest of the file.
21610 Use the same switch to compute sign as base.
21611 (maybe_warn): New function.
21612 (format_directive): Factor out warnings into maybe_warn.
21613 Add debugging output. Use warn_level.
21614 (add_bytes): Use warn_level.
21615 (pass_sprintf_length::compute_format_length): Add debugging output.
21616 (try_substitute_return_value): Same.
21617 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21619 PR middle-end/78703
21620 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21621 (struct fmtresult, format_integer, format_floating): Adjust.
21622 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21623 (get_string_length, format_string,format_directive): Same.
21624 (pass_sprintf_length::compute_format_length): Same.
21625 (try_substitute_return_value): Simplify slightly.
21627 PR middle-end/78703
21628 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21629 (fmtresult::operator+=): Outlined.
21630 (struct fmtresult): Add ctors.
21631 (struct conversion_spec): Rename...
21632 (struct directive): ...to this. Add and remove data members.
21633 (directive::set_width, directive::set_precision): New functions.
21634 (format_percent): Use fmtresult ctor.
21635 (get_width_and_precision): Remove.
21636 (format_integer): Make naming changes. Avoid computing width and
21638 (format_floating): Same. Adjust indentation.
21639 (format_character, format_none): New functions.
21640 (format_string): Moved character handling to format_character.
21641 (format_directive): Remove arguments, change return type.
21642 (parse_directive): New function.
21643 (pass_sprintf_length::compute_format_length): Move directive
21644 parsing to parse_directive.
21646 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21648 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21649 (assign_assembler_name_if_needed): ... this.
21650 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21651 (assign_assembler_name_if_needed): ... this.
21652 (free_lang_data_in_cgraph): Adjust callers.
21653 * cgraphunit.c (cgraph_node::analyze): Likewise.
21654 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21657 2017-01-23 Richard Biener <rguenther@suse.de>
21659 PR tree-optimization/79088
21660 PR tree-optimization/79188
21661 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21662 resetting loop bounds after last path deletion. Reset loop
21663 bounds of the target loop, make code match the comments.
21664 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21665 Make sure loops need no fixups.
21667 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21669 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21670 exponent support with double type for first argument.
21671 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21672 type returned by __builtin_vec_extract_sig,
21673 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21674 functions from "vector int" to "vector unsigned int" or from
21675 "vector long long int" to "vector unsigned long long int".
21676 Changed type returned by __builtin_vec_extract_exp,
21677 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21678 functions from "vector int" to "vector unsigned int" or from
21679 "vector long long int" to "vector unsigned long long int".
21680 Changed return type of __builtin_vec_test_data_class,
21681 __builtin_vec_test_data_class_sp, and
21682 __builtin_vec_test_data_class_dp from "vector int" to
21683 "vector bool int" or from "vector long long int" to "vector bool
21684 long long int" and changed second argument type from "unsigned
21685 int" to "int". Added new overloaded function forms "vector float
21686 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21687 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21688 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21689 double, vector unsigned long long int)" and "vector double
21690 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21691 long int)". Changed return type of
21692 __builtin_scalar_test_data_class and
21693 __builtin_scalar_test_data_class_sp and
21694 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21695 int" and changed second argument from "unsigned int" to "int".
21696 Changed type returned by __builtin_scalar_test_neg,
21697 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21698 from "int" to "bool int". Added new overloaded function form
21699 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21700 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21701 exponent double-precision with floating point first argument.
21702 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21703 documentation of scalar_test_data_class, scalar_test_neg,
21704 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21705 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21706 vec_test_data_class built-in functions to reflect refinements in
21707 their type signatures.
21709 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21711 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21713 (aarch64_elf_asm_destructor): Likewise.
21715 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21717 PR rtl-optimization/78634
21718 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21719 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21720 * ifcvt.c (noce_try_cmove): Add missing cost check.
21722 PR rtl-optimization/71724
21723 * combine.c (if_then_else_cond): Look for situations where it is
21724 beneficial to undo the work of one of the recursive calls.
21726 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21728 PR tree-optimization/70754
21729 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21730 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21731 combined stmt before it if not NULL.
21732 (combine_chains): Process refs reversely and compute dominance point
21735 2017-01-23 Martin Liska <mliska@suse.cz>
21737 PR tree-optimization/79196
21738 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21739 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21741 (strlen_optimize_stmt): Call the renamed function.
21743 2017-01-23 Michael Matz <matz@suse.de>
21745 PR tree-optimization/78384
21746 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21748 2017-01-23 Richard Biener <rguenther@suse.de>
21750 PR tree-optimization/79186
21751 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21752 both incoming edges before moving an assert.
21754 2017-01-23 Martin Jambor <mjambor@suse.cz>
21756 * ipa-prop.c (load_from_param_1): Removed.
21757 (load_from_unmodified_param): Bits from load_from_param_1 put back
21759 (load_from_param): Removed.
21760 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21761 with stmt. Reverted back to use of load_from_unmodified_param.
21763 2017-01-23 Martin Jambor <mjambor@suse.cz>
21766 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21767 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21768 field a pointer to garbage collected vector, mark lattices and
21769 ipcp_orig_node with GTY((skip)).
21770 (ipa_get_param_count): Adjust to descriptors being a pointer.
21771 (ipa_get_param): Likewise.
21772 (ipa_get_type): Likewise.
21773 (ipa_get_param_move_cost): Likewise.
21774 (ipa_set_param_used): Likewise.
21775 (ipa_get_controlled_uses): Likewise.
21776 (ipa_set_controlled_uses): Likewise.
21777 (ipa_is_param_used): Likewise.
21778 (ipa_node_params_t): Move into garbage collector. New methods insert
21780 (ipa_node_params_sum): Annotate wth GTY(()).
21781 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21783 (ipa_load_from_parm_agg): Adjust declaration.
21784 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21785 * ipa-profile.c (ipa_profile): Likewise.
21786 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21787 (ipa_populate_param_decls): Make descriptors parameter garbage
21789 (ipa_dump_param): Adjust to descriptors being a pointer.
21790 (ipa_alloc_node_params): Likewise.
21791 (ipa_initialize_node_params): Likewise.
21792 (load_from_param_1): Make descriptors parameter garbage collected.
21793 (load_from_unmodified_param): Likewise.
21794 (load_from_param): Likewise.
21795 (ipa_load_from_parm_agg): Likewise.
21796 (ipa_node_params::~ipa_node_params): Removed.
21797 (ipa_free_all_node_params): Remove call to delete operator.
21798 (ipa_node_params_t::insert): New.
21799 (ipa_node_params_t::remove): Likewise.
21800 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21801 copy known_csts and known_contexts vectors.
21802 (ipa_read_node_info): Adjust to descriptors being a pointer.
21803 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21805 (ipcp_transform_function): Make descriptors variable garbage
21808 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21810 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21811 * config/i386/avx512dqintrin.h: Ditto.
21812 * config/i386/avx512fintrin.h: Ditto.
21813 * config/i386/i386.c: Handle new builtins.
21814 * config/i386/i386-builtin.def: Add new builtins.
21815 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21816 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21818 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21819 Martin Liska <mliska@suse.cz>
21821 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21822 * asan.c (asan_expand_poison_ifn): Support stores and use
21823 appropriate ASAN report function.
21824 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21825 * internal-fn.def (ASAN_POISON_USE): Declare.
21826 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21827 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21828 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21829 ASAN_POISON calls w/o LHS.
21830 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21831 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21832 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21833 * gimplify.c (asan_poison_variables): Add attribute
21834 use_after_scope_memory to variables that really needs to live
21836 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21837 having the attribute.
21839 2017-01-23 Martin Liska <mliska@suse.cz>
21841 * asan.c (create_asan_shadow_var): New function.
21842 (asan_expand_poison_ifn): Likewise.
21843 * asan.h (asan_expand_poison_ifn): New declaration.
21844 * internal-fn.c (expand_ASAN_POISON): Likewise.
21845 * internal-fn.def (ASAN_POISON): New builtin.
21846 * sanopt.c (pass_sanopt::execute): Expand
21847 asan_expand_poison_ifn.
21848 * tree-inline.c (copy_decl_for_dup_finish): Make function
21850 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21851 * tree-ssa.c (is_asan_mark_p): New function.
21852 (execute_update_addresses_taken): Rewrite local variables
21853 (identified just by use-after-scope as addressable) into SSA.
21855 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21857 * doc/install.texi (Specific): opensource.apple.com uses https
21858 now. Remove trailing slash.
21860 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21862 * README.Portability: Remove note on an Irix compatibility issue.
21864 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21866 * gcov.c (INCLUDE_ALGORITHM): Define.
21867 (INCLUDE_VECTOR): Define.
21868 No longer include <vector> and <algorithm> directly.
21870 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21872 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21874 * doc/invoke.texi (Code Gen Options): Ditto.
21876 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21879 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21881 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21883 rtl-optimization/79125
21884 * cprop.c (local_cprop_pass): Handle cases where we make an
21885 unconditional trap.
21887 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21891 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21892 read from, for big endian.
21894 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21896 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21897 register pauth builtins for LP64 only.
21899 2017-01-20 Marek Polacek <polacek@redhat.com>
21902 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21905 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21907 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21909 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21910 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21911 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21913 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21916 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21917 in big-endian mode when they are not a single duplicated value.
21919 2017-01-20 Richard Biener <rguenther@suse.de>
21921 * BASE-VER: Bump to 7.0.1.
21923 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21925 * omp-low.c (omplow_simd_context): New struct. Use it...
21926 (lower_rec_simd_input_clauses): ...here and...
21927 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21928 references to idx, lane, max_vf, is_simt.
21930 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21932 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21935 2017-01-20 Martin Jambor <mjambor@suse.cz>
21937 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21938 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21940 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21941 (GTFILES): Rename hsa.c to hsa-common.c.
21942 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21943 * hsa-dump.c: Likewise.
21944 * hsa-gen.c: Likewise.
21945 * hsa-regalloc.c: Likewise.
21946 * ipa-hsa.c: Likewise.
21947 * omp-expand.c: Likewise.
21948 * omp-low.c: Likewise.
21949 * toplev.c: Likewise.
21951 2017-01-20 Marek Polacek <polacek@redhat.com>
21954 * doc/invoke.texi: Document -Wduplicated-branches.
21955 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21956 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21957 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21958 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21959 return 0 only when not OEP_LEXICOGRAPHIC.
21960 (fold_build_cleanup_point_expr): Use the expression
21961 location when building CLEANUP_POINT_EXPR.
21962 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21963 * tree.c (add_expr): Handle error_mark_node.
21965 2017-01-20 Martin Liska <mliska@suse.cz>
21968 * tree-profile.c (init_ic_make_global_vars): Do not call
21970 (gimple_init_gcov_profiler): Likewise.
21972 2017-01-20 Martin Liska <mliska@suse.cz>
21975 * cgraph.h (maybe_create_reference): Remove argument and
21977 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21979 * ipa-cp.c (create_specialized_node): Likewise.
21980 * symtab.c (symtab_node::maybe_create_reference): Handle
21981 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21983 2017-01-20 Martin Liska <mliska@suse.cz>
21985 * read-rtl-function.c (function_reader::create_function): Use
21986 build_decl instread of build_decl_stat.
21988 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21990 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21991 * config/i386/avx512dqintrin.h: Ditto.
21992 * config/i386/avx512fintrin.h: Ditto.
21993 * config/i386/i386-builtin-types.def: Add new types.
21994 * config/i386/i386.c: Handle new types.
21995 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21996 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21997 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21998 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21999 (__builtin_ia32_kshiftridi): New.
22000 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22002 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22006 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22007 define to rs6000_init_stack_protect_guard.
22008 (rs6000_init_stack_protect_guard): New function.
22010 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22011 Yunqiang Su <yunqiang.su@imgtec.com>
22013 * config.gcc (supported_defaults): Add madd4.
22014 (with_madd4): Add validation.
22015 (all_defaults): Add madd4.
22016 * config/mips/mips.opt (mmadd4): New option.
22017 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22019 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22021 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22022 (ISA_HAS_FUSED_MADD4): Likewise.
22023 * doc/invoke.texi (-mmadd4): Document the new option.
22024 * doc/install.texi (--with-madd4): Document the new option.
22026 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22028 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22029 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22030 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22031 (aarch64_init_pauth_hint_builtins): New.
22032 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22033 (aarch64_expand_builtin): Expand new builtins.
22035 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22037 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22038 * combine-stack-adj.c (no_unhandled_cfa): Handle
22039 REG_CFA_TOGGLE_RA_MANGLE.
22040 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22041 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22042 info for return address signing.
22043 (aarch64_expand_epilogue): Likewise.
22045 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22047 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22048 * config/aarch64/aarch64-protos.h
22049 (aarch64_return_address_signing_enabled): New declaration.
22050 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22052 (aarch64_expand_prologue): Sign return address before it's pushed onto
22054 (aarch64_expand_epilogue): Authenticate return address fetched from
22056 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22057 (aarch64_attributes): New function attributes for "sign-return-address".
22058 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22059 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22060 ("*do_return"): Generate combined instructions according to key index.
22061 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22062 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22064 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22065 * config/aarch64/aarch64.opt (msign-return-address=): New.
22066 * doc/extend.texi (AArch64 Function Attributes): Documents
22067 "sign-return-address=".
22068 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22070 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22072 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22073 overall option summary.
22075 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22077 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22078 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22079 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22080 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22082 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22084 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22085 -mpower9-minmax by default for -mcpu=power9.
22086 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22087 128-bit floating point.
22089 2017-01-20 Alan Modra <amodra@gmail.com>
22091 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22092 optimizing for size.
22094 2017-01-20 Alan Modra <amodra@gmail.com>
22097 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22098 for strcmp and strncmp from corresponding builtin decl.
22100 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22102 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22103 instead of i386/rtems-64.h.
22104 * config/i386/rtems-64.h: Remove.
22106 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22110 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22112 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22114 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22116 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22117 Change int to HOST_WIDE_INT.
22118 * config/aarch64/aarch64-protos.h
22119 (aarch64_simd_gen_const_vector_dup): Likewise.
22120 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22122 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22124 * langhooks-def.h (lhd_type_for_size): New decl.
22125 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22126 * langhooks.c (lhd_type_for_size): New function, taken from
22129 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22131 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22132 define_bypass for CR latency.
22133 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22134 (power9-alu2): Add define_bypass for CR latency.
22136 (power9-mul): Update insn latency.
22137 (power9-mul-compare): Update insn latency, bypass latency and remove
22140 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22142 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22144 * config/aarch64/aarch64.md
22145 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22146 aarch64_nopcrelative_literal_loads.
22147 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22149 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22151 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22152 TARGET_LOONGSON_3A.
22153 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22155 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22158 * config.gcc (supported_defaults): Add lxc1-sxc1.
22159 (with_lxc1_sxc1): Add validation.
22160 (all_defaults): Add lxc1-sxc1.
22161 * config/mips/mips.opt (mlxc1-sxc1): New option.
22162 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22164 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22165 __mips_no_lxc1_sxc1.
22166 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22167 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22168 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22170 2017-01-19 Richard Biener <rguenther@suse.de>
22172 PR tree-optimization/72488
22173 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22174 sure to restore SSA info.
22175 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22177 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22179 PR rtl-optimization/79121
22180 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22181 of the inner type when shifting an extended value.
22183 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22186 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22187 interposable aliases.
22189 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22192 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22193 Use the evmergelohi instruction.
22194 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22195 (mov_sitf_e500_subreg8_2_be): Likewise.
22196 (mov_sitf_e500_subreg12_2_le): Likewise.
22197 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22198 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22199 (mov_sitf_e500_subreg8_2_le): Likewise.
22200 (mov_sitf_e500_subreg12_2_be): Likewise.
22202 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22204 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22205 attribute from vecsimple to vecperm.
22206 (altivec_vbpermq2): Likewise.
22208 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22211 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22213 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22214 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22215 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22216 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22217 case where N arg is SIZE_MAX.
22218 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22219 (cmpstrsi): Add pattern.
22221 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22223 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22224 __builtin_vec_revb builtins.
22225 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22226 built-in functions to support generation of the ISA 3.0 XXBR<x>
22227 vector byte reverse instructions.
22228 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22229 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22230 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22231 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22232 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22233 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22234 (P9V_BUILTIN_VEC_REVB): Likewise.
22235 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22236 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22237 (p9_xxbrq_v16qi): Likewise.
22238 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22239 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22240 (p9_xxbrh_v8hi): Likewise.
22241 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22242 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22243 vec_revb built-in functions.
22245 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22247 PR rtl-optimization/78952
22248 * config/i386/i386.md (any_extract): New code iterator.
22249 (*insvqi_2): Use any_extract for source operand.
22250 (*insvqi_3): Use any_shiftrt for source operand.
22252 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22254 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22256 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22258 2017-01-18 Matthias Klose <doko@ubuntu.com>
22260 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22262 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22264 * config/rs6000/altivec.h (vec_bperm): Change #define.
22265 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22266 (altivec_vbpermq2): New define_insn.
22267 (altivec_vbpermd): Likewise.
22268 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22269 function interface.
22270 (VBPERMD): Likewise.
22271 (VBPERM): New polymorphic function interface.
22272 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22273 Add entries for P9V_BUILTIN_VEC_VBPERM.
22274 * doc/extend.texi: Add interfaces for vec_bperm.
22276 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22278 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22279 first letter of error messages.
22280 (s390_resolve_overloaded_builtin): Likewise.
22281 * config/s390/s390.c (s390_expand_builtin): Likewise.
22282 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22283 (s390_valid_target_attribute_inner_p): Likewise.
22284 * config/s390/s390.md ("tabort"): Likewise.
22286 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22288 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22289 (ISA_AVOID_DIV_HILO): New macro.
22290 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22291 (ISA_HAS_DDIV): Likewise.
22293 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22295 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22297 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22299 * doc/invoke.texi (fabi-version): Spelling fix.
22301 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22304 * doc/invoke.texi (fabi-version): Mention mangling fix for
22307 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22310 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22312 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22315 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22316 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22318 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22320 (stack_protect_test): Ditto.
22321 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22322 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22324 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22325 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22326 -mstack-protector-guard-offset=.
22327 (RS/6000 and PowerPC Options): Ditto.
22329 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22331 * config/i386/i386.h (MASK_CLASS_P): New define.
22332 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22333 there are no registers from different register sets also when
22334 mask registers are used. Update function comment.
22335 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22336 to (*k/*r) and (*k/*km) alternatives.
22338 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22340 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22341 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22342 (EH_RETURN_HANDLER_RTX): New define.
22343 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22344 Force frame pointer in EH return functions.
22345 (aarch64_expand_epilogue): Add barrier for eh_return.
22346 (aarch64_final_eh_return_addr): Remove.
22347 (aarch64_eh_return_handler_rtx): New function.
22348 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22350 (aarch64_eh_return_handler_rtx): New prototype.
22352 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22354 * config/rs6000/altivec.h (vec_rlmi): New #define.
22355 (vec_vrlnm): Likewise.
22356 (vec_rlnm): Likewise.
22357 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22358 (UNSPEC_VRLNM): Likewise.
22359 (VIlong): New mode iterator.
22360 (altivec_vrl<VI_char>mi): New define_insn.
22361 (altivec_vrl<VI_char>nm): Likewise.
22362 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22364 (VRLDNM): Likewise.
22365 (RLNM): New polymorphic function entry.
22366 (VRLWMI): New monomorphic function entry.
22367 (VRLDMI): Likewise.
22368 (RLMI): New polymorphic function entry.
22369 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22370 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22371 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22374 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22377 * dwarf2out.c (field_byte_offset): Restore the
22378 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22379 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22380 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22383 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22386 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22388 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22391 * dwarf2out.c (add_data_member_location_attribute): For constant
22392 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22393 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22394 DW_AT_byte_size attributes.
22396 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22398 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22399 after forcing to constant memory when the code model is medium.
22401 2017-01-17 Julia Koval <julia.koval@intel.com>
22404 * config/i386/avx512fintrin.h
22405 (_mm512_i32gather_ps): Change __addr type to void const*.
22406 (_mm512_mask_i32gather_ps): Ditto.
22407 (_mm512_i32gather_pd): Ditto.
22408 (_mm512_mask_i32gather_pd): Ditto.
22409 (_mm512_i64gather_ps): Ditto.
22410 (_mm512_mask_i64gather_ps): Ditto.
22411 (_mm512_i64gather_pd): Ditto.
22412 (_mm512_mask_i64gather_pd): Ditto.
22413 (_mm512_i32gather_epi32): Ditto.
22414 (_mm512_mask_i32gather_epi32): Ditto.
22415 (_mm512_i32gather_epi64): Ditto.
22416 (_mm512_mask_i32gather_epi64): Ditto.
22417 (_mm512_i64gather_epi32): Ditto.
22418 (_mm512_mask_i64gather_epi32): Ditto.
22419 (_mm512_i64gather_epi64): Ditto.
22420 (_mm512_mask_i64gather_epi64): Ditto.
22421 (_mm512_i32scatter_ps): Change __addr type to void*.
22422 (_mm512_mask_i32scatter_ps): Ditto.
22423 (_mm512_i32scatter_pd): Ditto.
22424 (_mm512_mask_i32scatter_pd): Ditto.
22425 (_mm512_i64scatter_ps): Ditto.
22426 (_mm512_mask_i64scatter_ps): Ditto.
22427 (_mm512_i64scatter_pd): Ditto.
22428 (_mm512_mask_i64scatter_pd): Ditto.
22429 (_mm512_i32scatter_epi32): Ditto.
22430 (_mm512_mask_i32scatter_epi32): Ditto.
22431 (_mm512_i32scatter_epi64): Ditto.
22432 (_mm512_mask_i32scatter_epi64): Ditto.
22433 (_mm512_i64scatter_epi32): Ditto.
22434 (_mm512_mask_i64scatter_epi32): Ditto.
22435 (_mm512_i64scatter_epi64): Ditto.
22436 (_mm512_mask_i64scatter_epi64): Ditto.
22437 * config/i386/avx512pfintrin.h
22438 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22439 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22440 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22441 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22442 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22443 (_mm512_prefetch_i32scatter_ps): Ditto.
22444 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22445 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22446 (_mm512_prefetch_i64scatter_pd): Ditto.
22447 (_mm512_prefetch_i64scatter_ps): Ditto.
22448 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22449 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22450 * config/i386/avx512vlintrin.h
22451 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22452 (_mm_mmask_i32gather_ps): Ditto.
22453 (_mm256_mmask_i32gather_pd): Ditto.
22454 (_mm_mmask_i32gather_pd): Ditto.
22455 (_mm256_mmask_i64gather_ps): Ditto.
22456 (_mm_mmask_i64gather_ps): Ditto.
22457 (_mm256_mmask_i64gather_pd): Ditto.
22458 (_mm_mmask_i64gather_pd): Ditto.
22459 (_mm256_mmask_i32gather_epi32): Ditto.
22460 (_mm_mmask_i32gather_epi32): Ditto.
22461 (_mm256_mmask_i32gather_epi64): Ditto.
22462 (_mm_mmask_i32gather_epi64): Ditto.
22463 (_mm256_mmask_i64gather_epi32): Ditto.
22464 (_mm_mmask_i64gather_epi32): Ditto.
22465 (_mm256_mmask_i64gather_epi64): Ditto.
22466 (_mm_mmask_i64gather_epi64): Ditto.
22467 (_mm256_i32scatter_ps): Change __addr type to void*.
22468 (_mm256_mask_i32scatter_ps): Ditto.
22469 (_mm_i32scatter_ps): Ditto.
22470 (_mm_mask_i32scatter_ps): Ditto.
22471 (_mm256_i32scatter_pd): Ditto.
22472 (_mm256_mask_i32scatter_pd): Ditto.
22473 (_mm_i32scatter_pd): Ditto.
22474 (_mm_mask_i32scatter_pd): Ditto.
22475 (_mm256_i64scatter_ps): Ditto.
22476 (_mm256_mask_i64scatter_ps): Ditto.
22477 (_mm_i64scatter_ps): Ditto.
22478 (_mm_mask_i64scatter_ps): Ditto.
22479 (_mm256_i64scatter_pd): Ditto.
22480 (_mm256_mask_i64scatter_pd): Ditto.
22481 (_mm_i64scatter_pd): Ditto.
22482 (_mm_mask_i64scatter_pd): Ditto.
22483 (_mm256_i32scatter_epi32): Ditto.
22484 (_mm256_mask_i32scatter_epi32): Ditto.
22485 (_mm_i32scatter_epi32): Ditto.
22486 (_mm_mask_i32scatter_epi32): Ditto.
22487 (_mm256_i32scatter_epi64): Ditto.
22488 (_mm256_mask_i32scatter_epi64): Ditto.
22489 (_mm_i32scatter_epi64): Ditto.
22490 (_mm_mask_i32scatter_epi64): Ditto.
22491 (_mm256_i64scatter_epi32): Ditto.
22492 (_mm256_mask_i64scatter_epi32): Ditto.
22493 (_mm_i64scatter_epi32): Ditto.
22494 (_mm_mask_i64scatter_epi32): Ditto.
22495 (_mm256_i64scatter_epi64): Ditto.
22496 (_mm256_mask_i64scatter_epi64): Ditto.
22497 (_mm_i64scatter_epi64): Ditto.
22498 (_mm_mask_i64scatter_epi64): Ditto.
22499 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22500 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22501 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22502 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22503 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22504 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22505 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22506 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22507 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22508 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22509 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22510 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22511 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22512 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22513 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22514 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22515 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22516 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22517 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22518 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22519 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22520 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22521 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22522 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22523 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22524 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22525 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22526 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22527 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22528 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22529 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22530 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22531 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22532 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22533 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22534 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22535 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22536 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22537 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22538 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22539 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22540 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22541 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22542 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22543 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22544 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22545 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22546 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22547 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22548 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22549 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22550 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22551 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22552 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22553 definitions accordingly.
22555 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22556 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22559 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22562 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22565 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22566 hard regs for inner regno.
22568 2017-01-17 Martin Liska <mliska@suse.cz>
22571 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22572 assumption and add comment.
22574 2017-01-17 Nathan Sidwell <nathan@acm.org>
22576 * ipa-visibility.c (localize_node): New function, broken out of ...
22577 (function_and_variable_visibility): ... here. Call it.
22579 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22581 PR middle-end/77445
22582 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22583 correctly set frequency of oudgoing edge.
22584 (duplicate_thread_path): Fix profile updating.
22586 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22589 * configure.ac: Add GCC_BASE_VER.
22590 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22591 version from BASE-VER file.
22592 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22593 (gcc.o): Depend on $(BASEVER).
22594 * common.opt (dumpfullversion): New option.
22595 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22596 * doc/invoke.texi: Document -dumpfullversion.
22597 * doc/install.texi: Document --with-gcc-major-version-only.
22598 * configure: Regenerated.
22600 2017-01-17 Richard Biener <rguenther@suse.de>
22602 PR tree-optimization/71433
22603 * tree-vrp.c (register_new_assert_for): Merge same asserts
22604 on all incoming edges.
22605 (process_assert_insertions_for): Handle insertions at the
22608 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22610 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22611 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22613 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22616 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22619 2017-01-17 Alan Modra <amodra@gmail.com>
22622 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22623 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22624 symbolic stack limit when pic.
22626 2017-01-16 Martin Sebor <msebor@redhat.com>
22628 PR tree-optimization/78608
22629 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22631 2017-01-16 Jeff Law <law@redhat.com>
22634 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22635 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22636 for several include directories that may be relative to sysroot.
22637 * config/i386/x-mingw32 (gplus_includedir): Define.
22638 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22639 (native_system_includedir): Likewise.
22640 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22641 override if TARGET_SYSTEM_ROOT is defined.
22642 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22644 PR tree-optimization/79090
22645 PR tree-optimization/33562
22646 PR tree-optimization/61912
22647 PR tree-optimization/77485
22648 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22649 and computed trims into the dump file.
22651 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22653 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22655 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22658 * gimplify.c (gimplify_init_constructor): If want_value and
22659 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22663 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22664 sequence. Formatting fixes.
22665 (doloop_optimize): Formatting fixes.
22668 * gcc.c (debug_level_greater_than_spec_func): New function.
22669 (static_spec_functions): Add debug-level-gt spec function.
22670 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22672 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22673 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22674 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22675 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22676 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22677 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22679 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22681 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22682 QImode fixups to general and mask registers only.
22684 2017-01-16 Carl Love <cel@us.ibm.com>
22686 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22687 for built-in functions
22688 vector signed char vec_nabs (vector signed char)
22689 vector signed short vec_nabs (vector signed short)
22690 vector signed int vec_nabs (vector signed int)
22691 vector signed long long vec_nabs (vector signed long long)
22692 vector float vec_nabs (vector float)
22693 vector double vec_nabs (vector double)
22694 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22696 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22697 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22698 * doc/extend.texi: Update the documentation file for the new built-in
22701 2017-01-16 Martin Sebor <msebor@redhat.com>
22703 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22706 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22708 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22709 UNSPEC_VSX__XXSPLTD to require special splat handling.
22711 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22714 * system.h: Poison strndup.
22716 2017-01-16 Alan Modra <amodra@gmail.com>
22719 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22722 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22724 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22726 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22728 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22729 call recog here. Assert that INSN_CODE (insn) is non-negative.
22731 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22734 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22736 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22737 in the currently scheduled RTL fragment.
22739 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22741 PR rtl-optimization/78751
22742 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22745 2017-01-14 Jeff Law <law@redhat.com>
22747 PR tree-optimization/79090
22748 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22749 variable length stores.
22750 (compute_trims): Delete dead assignment to *trim_tail.
22751 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22754 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22756 PR rtl-optimization/78626
22757 PR rtl-optimization/78727
22758 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22759 of a block, and split such blocks after everything else is finished.
22761 2017-01-14 Alan Modra <amodra@gmail.com>
22764 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22765 target legitimate_combined_insn.
22766 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22767 (rs6000_legitimate_combined_insn): New function.
22768 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22770 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22771 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22772 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22774 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22776 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22778 2017-01-13 Jeff Law <law@redhat.com>
22780 PR tree-optimization/33562
22781 PR tree-optimization/61912
22782 PR tree-optimization/77485
22783 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22785 (delete_dead_assignment): Likewise.
22786 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22787 statement to delete_dead_call and delete_dead_assignment.
22789 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22792 * substring-locations.c (format_warning_va): Strengthen case 1 so
22793 that both endpoints of the substring must be within the format
22794 range for just the substring to be printed.
22796 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22798 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22799 * config/i386/i386.c (ix86_target_string): Add missing options
22800 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22801 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22802 flags_other and ix86_target_other to flags2_other. Display unknown
22804 (ix86_valid_target_attribute_inner_p): Add missing options and
22805 reorder options by implied ISAs, as in ix86_target_string.
22807 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22809 * hash-table.h (hash_table::too_empty_p): New function.
22810 (hash_table::expand): Use it.
22811 (hash_table::traverse): Likewise.
22812 (hash_table::empty_slot): Use sizeof (value_type) instead of
22813 sizeof (PTR) to convert bytes to elements. Shrink the table
22814 if the current size is excessive for the current number of
22817 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22819 * ira-costs.c (record_reg_classes): Break from the inner loop
22820 early once alt_fail is known to be true. Update outer loop
22821 handling accordingly.
22823 2017-01-13 Jeff Law <law@redhat.com>
22825 * tree-ssa-dse.c (decrement_count): New function.
22826 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22827 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22828 when we know the partially dead statement is a mem* function.
22830 PR tree-optimization/61912
22831 PR tree-optimization/77485
22832 * tree-ssa-dse.c: Include expr.h.
22833 (maybe_trim_constructor_store): New function.
22834 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22836 PR tree-optimization/33562
22837 PR tree-optimization/61912
22838 PR tree-optimization/77485
22839 * doc/invoke.texi: Document new dse-max-object-size param.
22840 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22841 * tree-ssa-dse.c: Include params.h.
22842 (dse_store_status): New enum.
22843 (initialize_ao_ref_for_dse): New, partially extracted from
22845 (valid_ao_ref_for_dse, normalize_ref): New.
22846 (setup_live_bytes_from_ref, compute_trims): Likewise.
22847 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22848 (maybe_trim_partially_dead_store): Likewise.
22849 (maybe_trim_complex_store): Likewise.
22850 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22851 Track what bytes live from the original store. Return tri-state
22852 for dead, partially dead or live.
22853 (dse_dom_walker): Add constructor, destructor and new private members.
22854 (delete_dead_call, delete_dead_assignment): New extracted from
22856 (dse_optimize_stmt): Make a member of dse_dom_walker.
22857 Use initialize_ao_ref_for_dse.
22859 PR tree-optimization/33562
22860 PR tree-optimization/61912
22861 PR tree-optimization/77485
22862 * sbitmap.h (bitmap_count_bits): Prototype.
22863 (bitmap_clear_range, bitmap_set_range): Likewise.
22864 * sbitmap.c (bitmap_clear_range): New function.
22865 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22867 2017-01-13 Martin Liska <mliska@suse.cz>
22870 * function.c (set_cfun): Add new argument force.
22871 * function.h (set_cfun): Likewise.
22872 * ipa-inline-transform.c (inline_call): Use the function when
22873 strict alising from is dropped for function we inline to.
22875 2017-01-13 Richard Biener <rguenther@suse.de>
22877 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22878 for dumping GIMPLE INTEGER_CSTs.
22880 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22882 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22883 to 201112L since C++17.
22885 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22888 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22889 if -fsanitize=kernel-address is present.
22891 2017-01-13 Richard Biener <rguenther@suse.de>
22893 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22894 as _Literal ( type ) number in case usual suffixes do not
22895 preserve all information.
22897 2017-01-13 Richard Biener <rguenther@suse.de>
22899 PR tree-optimization/77283
22900 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22901 and ssa-iterators.h.
22902 (is_feasible_trace): Implement a cost model based on joiner
22905 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22908 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22909 char or short to __float128/_Float128 directly.
22911 2017-01-12 Martin Sebor <msebor@redhat.com>
22913 to -Wformat-overflow.
22914 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22915 (min_bytes_remaining): Same.
22916 (get_string_length): Same.
22917 (format_string): Same.
22918 (format_directive): Same.
22920 (pass_sprintf_length::handle_gimple_call): Same.
22922 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22924 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22925 info.nowrite calls with no lhs that can't throw. Return bool
22926 whether gsi_remove has been called or not.
22927 (pass_sprintf_length::handle_gimple_call): Return bool whether
22928 try_substitute_return_value called gsi_remove. Formatting fix.
22929 (pass_sprintf_length::execute): Don't use gsi_remove if
22930 handle_gimple_call returned true.
22933 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22934 be removed due to side-effects, don't remove following barrier nor
22935 turn the successor edge into fallthru edge.
22937 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22940 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22941 element-reversing loads and stores as not swappable.
22943 2017-01-12 Nathan Sidwell <nathan@acm.org>
22944 Nicolai Stange <nicstange@gmail.com>
22946 * combine.c (try_combine): Don't ignore result of overlap checking
22947 loop. Combine overlap & asm check into single loop.
22949 2017-01-12 Richard Biener <rguenther@suse.de>
22951 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22952 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22954 2017-01-12 Richard Biener <rguenther@suse.de>
22956 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22957 and TS_TARGET_OPTION directly derive from TS_BASE.
22958 * tree-core.h (tree_optimization_option): Derive from tree_base.
22959 (tree_target_option): Likewise.
22961 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22963 * config/i386/i386.c (memory_address_length): Increase len
22964 only when rip_relative_addr_p returns false.
22966 2017-01-11 Julia Koval <julia.koval@intel.com>
22968 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22969 (OPTION_MASK_ISA_SGX_SET): New.
22970 (ix86_handle_option): Handle OPT_msgx.
22971 * config.gcc: Added sgxintrin.h.
22972 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22973 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22974 * config/i386/i386.c (ix86_target_string): Add -msgx.
22976 (ix86_option_override_internal): Handle new options.
22977 (ix86_valid_target_attribute_inner_p): Add sgx.
22978 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22979 * config/i386/i386.opt: Add msgx.
22980 * config/i386/sgxintrin.h: New file.
22981 * config/i386/x86intrin.h: Add sgxintrin.h.
22983 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22986 * fold-const.c (maybe_nonzero_address): Return 1 for function
22988 (tree_single_nonzero_warnv_p): Don't handle function local objects
22992 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22995 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22998 * opts.c: Include "spellcheck.h"
22999 (struct string_fragment): New struct.
23000 (struct edit_distance_traits<const string_fragment &>): New
23002 (get_closest_sanitizer_option): New function.
23003 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23005 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23007 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23009 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23010 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23011 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23012 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23013 for initial die_offset if dwarf_split_debug_info.
23014 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23015 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23017 (output_skeleton_debug_sections): Formatting fix. Use
23018 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23019 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23021 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23023 * config/arm/cortex-a53.md: Add bypasses for
23024 cortex_a53_r2f_cvt.
23025 (cortex_a53_r2f): Only use for transfers.
23026 (cortex_a53_f2r): Likewise.
23027 (cortex_a53_r2f_cvt): Add reservation for conversions.
23028 (cortex_a53_f2r_cvt): Likewise.
23030 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23032 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23033 to all inlined functions, change static to extern.
23035 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23038 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23040 (arm_assemble_integer): Likewise.
23042 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23044 * config.gcc: Use new awk script to check CPU, FPU and architecture
23045 parameters for --with-... options.
23046 * config/arm/parsecpu.awk: New file
23047 * config/arm/arm-cpus.in: New file.
23048 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23050 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23052 * config/arm/t-arm: Update dependency rules.
23053 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23054 of processing .def files.
23055 * config/arm/genopt.sh: Deleted.
23056 * config/arm/gentune.sh: Deleted.
23057 * config/arm/arm-cores.def: Deleted.
23058 * config/arm/arm-arches.def: Deleted.
23059 * config/arm/arm-fpus.def: Deleted.
23060 * config/arm/arm-tune.md: Regenerated.
23061 * config/arm/arm-tables.opt: Regenerated.
23062 * config/arm/arm-cpu.h: New generated file.
23063 * config/arm/arm-cpu-data.h: New generated file.
23064 * config/arm/arm-cpu-cdata.h: New generated file.
23066 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23069 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23071 (input_varpool_node): Unpack dynamically_initialized bit.
23073 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23075 PR rtl-optimization/79032
23076 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23077 the alignment of the adjusted memory reference against that of MODE,
23078 instead of the alignment of the original memory reference.
23080 2017-01-11 Martin Jambor <mjambor@suse.cz>
23082 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23084 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23085 decorated functions.
23087 2017-01-11 Richard Biener <rguenther@suse.de>
23089 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23090 set range/nonnull info for PHI results. Do not set it on
23091 stmts marked for removal.
23093 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23095 * expr.c (store_field): In the bitfield case, fetch the return value
23096 from the registers before applying a single big-endian adjustment.
23097 Always do a final load for a BLKmode value not larger than a word.
23099 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23102 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23103 that we correctly handle column numbers greater than
23104 LINE_MAP_MAX_COLUMN_NUMBER.
23106 2017-01-10 Martin Sebor <msebor@redhat.com>
23108 PR middle-end/78245
23109 * gimple-ssa-sprintf.c (get_destination_size): Call
23110 {init,fini}object_sizes.
23111 * tree-object-size.c (addr_object_size): Adjust.
23112 (pass_through_call): Adjust.
23113 (pass_object_sizes::execute): Adjust.
23114 * tree-object-size.h (fini_object_sizes): Declare.
23116 2017-01-10 Martin Sebor <msebor@redhat.com>
23118 PR tree-optimization/78775
23119 * builtins.c (get_size_range): Move...
23120 * calls.c: ...to here.
23121 (alloc_max_size): Accept zero argument.
23122 (operand_signed_p): Remove.
23123 (maybe_warn_alloc_args_overflow): Call get_size_range.
23124 * calls.h (get_size_range): Declare.
23126 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23128 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23129 from TI's devices.csv file as of September 2016.
23130 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23132 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23134 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23135 * doc/invoke.texi: Likewise.
23136 * doc/md.texi: Likewise.
23137 * doc/objc.texi: Likewise.
23139 2017-01-10 Joshua Conner <joshconner@google.com>
23141 * config/arm/fuchsia-elf.h: New file.
23142 * config/fuchsia.h: New file.
23143 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23144 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23146 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23148 2016-01-10 Richard Biener <rguenther@suse.de>
23150 PR tree-optimization/79034
23151 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23152 Propagate out degenerate PHIs in the joiner.
23154 2017-01-10 Martin Liska <mliska@suse.cz>
23156 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23157 (sort_congruence_classes_by_decl_uid): Likewise.
23158 (sort_congruence_class_groups_by_decl_uid): Likewise.
23159 (sem_item_optimizer::merge_classes): Sort class, groups in these
23160 classes and members in the groups by DECL_UID of declarations.
23161 This would make merge operations stable.
23163 2017-01-10 Martin Liska <mliska@suse.cz>
23165 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23166 usage of m_classes_vec.
23167 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23168 (sem_item_optimizer::get_group_by_hash): Likewise.
23169 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23170 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23171 (sem_item_optimizer::verify_classes): Likewise.
23172 (sem_item_optimizer::process_cong_reduction): Likewise.
23173 (sem_item_optimizer::dump_cong_classes): Likewise.
23174 (sem_item_optimizer::merge_classes): Likewise.
23175 * ipa-icf.h (congruence_class_hash): Rename from
23176 congruence_class_group_hash. Remove declaration of m_classes_vec.
23178 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23180 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23181 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23182 * config.gcc: Add avx512vpopcntdqintrin.h.
23183 * config/i386/avx512vpopcntdqintrin.h: New.
23184 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23185 * config/i386/i386-builtin-types.def: Add new types.
23186 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23187 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23188 __builtin_ia32_vpopcountq_v8di_mask): New.
23189 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23190 __AVX512VPOPCNTDQ__.
23191 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23192 (PTA_AVX512VPOPCNTDQ): Define.
23193 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23194 TARGET_AVX512VPOPCNTDQ_P): Define.
23195 * config/i386/i386.opt: Add mavx512vpopcntdq.
23196 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23197 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23199 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23201 PR middle-end/77484
23202 * predict.def (PRED_CALL): Set to 67.
23204 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23206 * expr.c (store_field): In the bitfield case, if the value comes from
23207 a function call and is of an aggregate type returned in registers, do
23208 not modify the field mode; extract the value in all cases if the mode
23209 is BLKmode and the size is not larger than a word.
23211 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23214 * config/i386/cpuid.h: Fix undefined behavior.
23216 2017-01-04 Jeff Law <law@redhat.com>
23218 PR tree-optimization/79007
23219 PR tree-optimization/67955
23220 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23221 conservative for pt.null when flag_non_call_exceptions is on.
23223 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23225 PR translation/79019
23226 PR translation/79020
23227 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23228 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23229 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23231 * config/avr/avr.opt (maccumulate-args): Likewise.
23232 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23233 * common.opt (freport-bug): Likewise.
23234 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23235 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23236 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23237 translatable string.
23238 * config/i386/i386.c (function_value_32): Likewise.
23239 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23240 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23242 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23243 * common/config/msp430/msp430-common.c (msp430_handle_option):
23245 * symtab.c (symtab_node::verify_base): Likewise.
23246 * opts.c (set_debug_level): Likewise.
23247 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23248 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23249 missing whitespace to translatable strings.
23250 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23251 * config/sh/superh.h: Likewise.
23252 * config/i386/xopintrin.h: Likewise.
23253 * config/i386/znver1.md: Likewise.
23254 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23255 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23256 * double-int.h (struct double_int): Likewise.
23257 * double-int.c (div_and_round_double): Likewise.
23258 * wide-int.cc: Likewise.
23259 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23260 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23261 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23262 (crossjumps_occurred): ... this.
23263 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23266 PR tree-optimization/78899
23267 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23268 returning bool return struct loop *, NULL for failure and the new
23270 (versionable_outer_loop_p): Don't version outer loop if it has
23271 dont_vectorized bit set.
23272 (tree_if_conversion): When versioning outer loop, ensure
23273 tree_if_conversion is performed also on the inner loop of the
23274 non-vectorizable outer loop copy.
23275 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23276 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23277 prevent vectorization of it.
23278 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23279 the outer loop vectorization of the non-scalar version is attempted
23280 before vectorization of the inner loop in scalar version. If
23281 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23282 vectorization of its inner loop.
23283 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23284 has 2 inner loops, rename also on edges from bb whose single pred
23285 is outer_loop->header. Fix typo in function comment.
23287 2017-01-09 Martin Sebor <msebor@redhat.com>
23290 * asan.c (asan_emit_stack_protection): Increase local buffer size
23291 to avoid snprintf truncation warning.
23293 2017-01-09 Andrew Pinski <apinski@cavium.com>
23295 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23296 to reference thunderx2t99 for the tuning structure
23297 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23299 (thunderx2t99_extra_costs): This.
23300 * config/aarch64/aarch64-tune.md: Regenerate.
23301 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23302 (vulcan_addrcost_table): This.
23303 (vulcan_regmove_cost): Rename to ...
23304 (thunderx2t99_regmove_cost): This.
23305 (vulcan_vector_cost): Rename to ...
23306 (thunderx2t99_vector_cost): this.
23307 (vulcan_branch_cost): Rename to ...
23308 (thunderx2t99_branch_cost): This.
23309 (vulcan_tunings): Rename to ...
23310 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23311 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23313 2017-01-09 Martin Jambor <mjambor@suse.cz>
23317 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23318 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23319 (propagate_vr_accross_jump_function): Use the above function for all
23320 value range computations for pass-through jump functions and type
23321 converasion from explicit value range values.
23322 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23323 parameters from TYPE_ARG_TYPES.
23324 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23325 (ipa_write_node_info): Stream type of the actual argument.
23326 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23328 2017-01-09 Martin Liska <mliska@suse.cz>
23331 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23332 (lookup_compiler): Do not show error message with have_E.
23334 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23336 PR tree-optimization/78938
23337 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23338 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23339 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23340 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23343 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23345 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23348 2017-01-09 Richard Biener <rguenther@suse.de>
23350 PR tree-optimization/78997
23351 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23352 name condition properly.
23354 2017-01-09 Richard Biener <rguenther@suse.de>
23357 * dwarf2out.c (is_cxx): New overload with context.
23358 (is_naming_typedef_decl): Use it.
23360 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23362 * invoke.texi (Option Summary): Correct spacing in option lists
23363 and add line breaks to fix over-long lines.
23365 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23367 PR middle-end/17660
23369 * extend.texi (Common Variable Attributes): Add xref to GCC
23370 Internals manual to explain mode attribute keywords.
23372 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23375 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23376 and Preprocessor Options.
23377 (Options for Linking): Document -pthread here....
23378 (RS/6000 and PowerPC Options): ...not here.
23379 (Solaris 2 Options): ...or here.
23380 * doc/cppopts.texi: Document -pthread.
23382 2017-01-08 Martin Sebor <msebor@redhat.com>
23384 PR middle-end/77708
23385 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23386 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23387 New member functions.
23388 (format_directive): Used them.
23390 (pass_sprintf_length::handle_gimple_call): Same.
23391 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23392 to avoid truncation for any argument.
23393 (extract_affine_mul): Same.
23394 * tree.c (get_file_function_name): Same.
23396 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23398 PR middle-end/77484
23399 * predict.def (PRED_INDIR_CALL): Set to 86.
23401 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23403 PR preprocessor/54124
23404 * doc/cppopts.texi: Reformat -d subtable to list the full name
23405 of the options. Add cross-reference to the docs for the general
23406 compiler -d options.
23407 * doc/invoke.texi (Developer Options): Add cross-reference to the
23408 preprocessor-specific -d option documentation.
23410 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23412 PR preprocessor/13498
23413 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23414 redudant material, and reflect new command-line options.
23415 (System Headers): Likewise.
23417 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23419 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23420 -isystem, and -idirafter. Copy-edit.
23421 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23422 default for -ftrack-macro-expansion. Delete obsolete and
23423 badly-formatted implementation details about -fdebug-cpp output.
23424 * doc/cppwarnopts.texi: Copy-edit.
23426 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23429 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23430 that the transition from a max line width >= 1<<10 to narrower
23431 lines works correctly.
23433 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23435 * doc/options.texi (PerFunction): New.
23436 * opt-functions.awk (switch_flags): Map both Optimization and
23437 PerFunction to CL_OPTIMIZATION.
23438 * opth-gen.awk: Test for PerFunction flag along with
23440 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23441 it only when the latter is present. Skip those that don't in
23442 the hash function generator.
23443 * common.opt (fvar-tracking): Mark as PerFunction instead of
23445 (fvar-tracking-assignments): Likewise.
23446 (fvar-tracking-assignments-toggle): Likewise.
23447 (fvar-tracking-uninit): Likewise.
23449 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23451 PR translation/79018
23452 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23455 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23458 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23459 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23460 TARGET_LONG_JUMP_TABLE_OFFSETS.
23461 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23462 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23463 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23464 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23465 * config/m68k/m68k.md (tablejump expander): Likewise.
23466 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23467 TARGET_LONG_JUMP_TABLE_OFFSETS.
23468 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23469 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23471 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23472 David Holsgrove <david.holsgrove@xilinx.com>
23474 * common/config/microblaze/microblaze-common.c
23475 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23476 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23478 * config/microblaze/microblaze.c (microblaze_must_save_register)
23479 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23481 (microblaze_eh_return): New function.
23482 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23483 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23484 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23485 * config/microblaze/microblaze.md (eh_return): New pattern.
23487 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23489 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23490 GCC_DIAGNOSTIC_STRINGIFY): Define.
23492 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23494 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23496 * config/arm/arm.md (<mcrr>): New.
23498 * config/arm/arm.c (arm_arch5te): New.
23499 (arm_option_override): Set arm_arch5te.
23500 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23502 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23503 (arm_mcrr_qualifiers): ... this. New.
23504 (MRRC_QUALIFIERS): Define to...
23505 (arm_mrrc_qualifiers): ... this. New.
23506 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23508 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23509 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23510 (MRRCI, mrrc, MRRC): New.
23511 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23512 VUNSPEC_MRRC2): New.
23514 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23516 * config/arm/arm.md (<mcr>): New.
23518 * config/arm/arm.c (arm_coproc_builtin_available): Add
23519 support for mcr, mrc, mcr2 and mrc2.
23520 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23521 (arm_mcr_qualifiers): ... this. New.
23522 (MRC_QUALIFIERS): Define to ...
23523 (arm_mrc_qualifiers): ... this. New.
23524 (MCR_QUALIFIERS): Define to ...
23525 (arm_mcr_qualifiers): ... this. New.
23526 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23528 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23529 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23530 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23531 VUNSPEC_MRC2): New.
23533 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23535 * config/arm/arm.md (*ldc): New.
23539 * config/arm/arm.c (arm_coproc_builtin_available): Add
23540 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23541 (arm_coproc_ldc_stc_legitimate_address): New.
23542 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23543 'qualifier_const_pointer'.
23544 (LDC_QUALIFIERS): Define to...
23545 (arm_ldc_qualifiers): ... this. New.
23546 (STC_QUALIFIERS): Define to...
23547 (arm_stc_qualifiers): ... this. New.
23548 * config/arm/arm-protos.h
23549 (arm_coproc_ldc_stc_legitimate_address): New.
23550 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23551 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23552 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23553 stc2, stcl, stc2l): New.
23554 * config/arm/constraints.md (Uz): New.
23555 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23556 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23557 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23558 VUNSPEC_STC2L): New.
23560 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23562 * config/arm/arm.md (<cdp>): New.
23563 * config/arm/arm.c (neon_const_bounds): Rename this ...
23564 (arm_const_bounds): ... this.
23565 (arm_coproc_builtin_available): New.
23566 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23567 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23568 (CDP_QUALIFIERS): Define to...
23569 (arm_cdp_qualifiers): ... this. New.
23571 (arm_expand_builtin_args): Add case for 6 arguments.
23572 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23573 (arm_const_bounds): ... this.
23574 (arm_coproc_builtin_available): New.
23575 * config/arm/arm_acle.h (__arm_cdp): New.
23577 * config/arm/arm_acle_builtins.def (cdp): New.
23579 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23580 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23581 'arm_const_bounds'.
23582 * config/arm/types.md (coproc): New.
23583 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23584 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23585 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23586 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23588 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23590 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23591 (UBINOP_QUALIFIERS): New.
23593 (acle_builtin_data): New. Change comment.
23594 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23595 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23596 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23597 arm_acle_builtins.def.
23598 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23599 (arm_init_acle_builtins): New.
23600 (CRC32_BUILTIN): Remove.
23601 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23602 crc32cb, crc32ch and crc32cw.
23603 (arm_init_crc32_builtins): Remove.
23604 (arm_init_builtins): Use arm_init_acle_builtins rather
23605 than arm_init_crc32_builtins.
23606 (arm_expand_acle_builtin): New.
23607 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23608 * config/arm/arm_acle_builtins.def: New.
23610 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23612 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23613 (arm_builtin_datum): ... this.
23614 (arm_init_neon_builtin): Rename to ...
23615 (arm_init_builtin): ... this. Add a new parameters PREFIX
23616 and USE_SIG_IN_NAME.
23617 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23618 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23619 'arm_builtin_datum'.
23620 (arm_init_vfp_builtins): Likewise.
23621 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23622 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23623 (arm_expand_neon_args): Rename to ...
23624 (arm_expand_builtin_args): ... this. Rename builtin_arg
23625 enum values and differentiate between ARG_BUILTIN_MEMORY
23626 and ARG_BUILTIN_NEON_MEMORY.
23627 (arm_expand_neon_builtin_1): Rename to ...
23628 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23629 values, arm_expand_builtin_args and add bool parameter NEON.
23630 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23631 (arm_expand_vfp_builtin): Likewise.
23632 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23634 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23636 PR middle-end/77484
23637 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23638 * predict.c (tree_estimate_probability_bb): Reverse direction of
23639 polymorphic call predictor.
23641 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23643 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23644 (determine_pass_name_match): ...this new function and...
23645 (should_skip_pass_p): ...this new function.
23647 2017-01-06 Nathan Sidwell <nathan@acm.org>
23649 * ipa-visibility.c (function_and_variable_visibility): Reformat
23650 comments and long lines. Remove extrneous if.
23651 * symtab.c (symtab_node::make_decl_local): Fix code format.
23652 (symtab_node::set_section_for_node): Fix comment typo.
23654 2017-01-06 Martin Liska <mliska@suse.cz>
23657 * lra-constraints.c: Rename invariant to lra_invariant.
23658 * predict.c (set_even_probabilities): Initialize e to NULL.
23660 2017-01-05 Martin Sebor <msebor@redhat.com>
23662 PR tree-optimization/78910
23663 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23664 (format_integer): Correct off-by-one error in the handling
23665 of precision with negative numbers in signed conversions..
23667 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23669 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23671 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23673 PR tree-optimization/71016
23674 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23675 factor_out_conditional_conversion. Formatting fix.
23676 (factor_out_conditional_conversion): Add cond_stmt argument.
23677 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23678 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23681 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23683 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23684 read-rtl-function.o, and selftest-rtl.o.
23685 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23686 (selftest::aarch64_test_loading_full_dump): New function.
23687 (selftest::aarch64_run_selftests): New function.
23688 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23689 selftest::aarch64_run_selftests.
23690 * config/i386/i386.c
23691 (selftest::ix86_test_loading_dump_fragment_1): New function.
23692 (selftest::ix86_test_loading_call_insn): New function.
23693 (selftest::ix86_test_loading_full_dump): New function.
23694 (selftest::ix86_test_loading_unspec): New function.
23695 (selftest::ix86_run_selftests): Call the new functions.
23696 * emit-rtl.c (maybe_set_max_label_num): New function.
23697 * emit-rtl.h (maybe_set_max_label_num): New decl.
23698 * function.c (instantiate_decls): Guard call to
23699 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23700 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23702 * gensupport.c (gen_reader::gen_reader): Pass "false"
23703 for new "compact" param of rtx_reader.
23704 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23705 rather than an empty string for NULL strings.
23706 * read-md.c: Potentially include config.h rather than bconfig.h.
23707 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23708 (have_error): New global, copied from errors.c.
23709 (md_reader::read_name): Rename to...
23710 (md_reader::read_name_1): ...this, adding "out_loc" param,
23711 and converting "missing name or number" to returning false, rather
23713 (md_reader::read_name): Reimplement in terms of read_name_1.
23714 (md_reader::read_name_or_nil): New function.
23715 (md_reader::read_string): Handle "(nil)" by returning NULL.
23716 (md_reader::md_reader): Add new param "compact".
23717 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23718 (md_reader::read_file): New method.
23719 * read-md.h (md_reader::md_reader): Add new param "compact".
23720 (md_reader::read_file): New method.
23721 (md_reader::is_compact): New accessor.
23722 (md_reader::read_name): Convert return type from void to file_location.
23723 (md_reader::read_name_or_nil): New decl.
23724 (md_reader::read_name_1): New decl.
23725 (md_reader::m_compact): New field.
23726 (noop_reader::noop_reader): Pass "false" for new "compact" param
23728 (rtx_reader::rtx_reader): Add new "compact" param.
23729 (rtx_reader::read_rtx_operand): Make virtual and convert return
23730 type from void to rtx.
23731 (rtx_reader::read_until): New decl.
23732 (rtx_reader::handle_any_trailing_information): New virtual function.
23733 (rtx_reader::postprocess): New virtual function.
23734 (rtx_reader::finalize_string): New virtual function.
23735 (rtx_reader::m_in_call_function_usage): New field.
23736 (rtx_reader::m_reuse_rtx_by_id): New field.
23737 * read-rtl-function.c: New file.
23738 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23739 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23740 (selftest::verify_three_block_rtl_cfg): New decl.
23741 * read-rtl-function.h: New file.
23742 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23743 For host, include function.h, memmodel.h, and emit-rtl.h.
23744 (one_time_initialization): New function.
23745 (struct compact_insn_name): New struct.
23746 (compact_insn_names): New array.
23747 (find_code): Handle insn codes in compact dumps.
23748 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23749 (bind_subst_iter_and_attr): Likewise.
23750 (add_condition_to_string): Likewise.
23751 (add_condition_to_rtx): Likewise.
23752 (apply_attribute_uses): Likewise.
23753 (add_current_iterators): Likewise.
23754 (apply_iterators): Likewise.
23755 (initialize_iterators): Guard usage of apply_subst_iterator with
23756 #ifdef GENERATOR_FILE.
23757 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23758 (md_reader::read_mapping): Likewise.
23759 (add_define_attr_for_define_subst): Likewise.
23760 (add_define_subst_attr): Likewise.
23761 (read_subst_mapping): Likewise.
23762 (check_code_iterator): Likewise.
23763 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23765 (one_time_initialization): New function.
23766 (rtx_reader::read_until): New method.
23767 (read_flags): New function.
23768 (parse_reg_note_name): New function.
23769 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23770 Handle reuse_rtx ids.
23771 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23772 Add parsing support for RTL dumps, mirroring the special-cases in
23773 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23774 values, and calling handle_any_trailing_information.
23775 (rtx_reader::read_rtx_operand): Convert return type from void
23776 to rtx, returning return_rtx. Handle case 'e'. Call
23777 finalize_string on XSTR and XTMPL fields.
23778 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23779 "(nil)" values were omitted. Call the postprocess vfunc on the
23781 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23782 class ctor. Initialize m_in_call_function_usage. Call
23783 one_time_initialization.
23784 * rtl-tests.c (selftest::test_uncond_jump): Call
23785 set_new_first_and_last_insn.
23786 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23787 * selftest-rtl.c: New file.
23788 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23789 (selftest::get_insn_by_uid): New decl.
23790 * selftest-run-tests.c (selftest::run_tests): Call
23791 read_rtl_function_c_tests.
23792 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23793 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23796 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23798 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23799 operands in a special way. Assert that pos+len <= mode precision.
23801 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23803 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23804 3 argument Alias with unlimited for the negative form.
23805 (fno-vect-cost-model): Removed.
23807 2017-01-05 Martin Liska <mliska@suse.cz>
23809 * hsa-gen.c (gen_hsa_divmod): New function.
23810 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23812 2017-01-05 Martin Liska <mliska@suse.cz>
23815 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23818 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23820 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23821 small constant length operands.
23823 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23825 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23826 between loop iterations.
23828 2017-01-05 Martin Liska <mliska@suse.cz>
23831 * gimplify.c (gimplify_decl_expr): Compare to
23832 asan_poisoned_variables instread of checking flags.
23833 (gimplify_target_expr): Likewise.
23834 (gimplify_expr): Likewise.
23835 (gimplify_function_tree): Conditionally initialize
23836 asan_poisoned_variables.
23838 2017-01-04 Jeff Law <law@redhat.com>
23840 PR tree-optimizatin/78812
23841 * rtl.h (contains_mem_rtx_p): Prototype.
23842 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23843 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23844 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23845 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23846 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23848 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23850 * input.c (assert_char_at_range): Default-initialize actual_range.
23852 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23854 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23855 to match the caller.
23857 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23859 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23860 insns after final jump in test to emit dummy move.
23862 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23864 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23865 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23867 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23869 * multiple_target.c (create_dispatcher_calls): Init e_next.
23870 * tree-ssa-loop-split.c (split_loop): Init border.
23871 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23874 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23879 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23880 (altivec_register_operand): Do not return true if the operand
23881 contains a SUBREG mixing SImode and SFmode.
23882 (vsx_register_operand): Likewise.
23883 (vsx_reg_sfsubreg_ok): New predicate.
23884 (vfloat_operand): Do not return true if the operand contains a
23885 SUBREG mixing SImode and SFmode.
23886 (vint_operand): Likewise.
23887 (vlogical_operand): Likewise.
23888 (gpc_reg_operand): Likewise.
23889 (int_reg_operand): Likewise.
23890 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23891 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23892 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23894 (rs6000_emit_move_si_sf_subreg): New helper function.
23895 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23896 fixup SUBREGs involving SImode and SFmode.
23897 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23898 numbers for the new peephole2 optimization.
23899 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23900 the GLIBC math library that do AND/IOR/XOR operations on single
23901 precision floating point.
23902 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23903 target macros to say whether we need to avoid SUBREGs mixing
23905 (TARGET_ALLOW_SF_SUBREG): Likewise.
23906 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23907 (UNSPEC_SI_FROM_SF): Likewise.
23908 (iorxor): Change spacing.
23909 (and_ior_xor): New iterator for AND, IOR, and XOR.
23910 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23911 (movdi_from_sf_zero_ext): Likewise.
23912 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23913 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23914 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23915 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23916 (fms<mode>4): Likewise.
23917 (fnma<mode>4): Likewise.
23918 (fnms<mode>4): Likewise.
23919 (nfma<mode>4): Likewise.
23920 (nfms<mode>4): Likewise.
23922 2017-01-04 Marek Polacek <polacek@redhat.com>
23925 * doc/invoke.texi: Document -Wpointer-compare.
23927 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23929 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23932 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23933 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23936 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23939 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23940 documentation of the powerpc_popcntb_ok attribute.
23941 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23942 code to issue warning messages if a requested CPU configuration is
23943 not supported by the binary (assembler and loader) toolchain.
23944 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23945 made to define a built-in function that has been disabled.
23946 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23947 made to define a built-in function that has been disabled.
23948 (altivec_init_builtins): Add comment explaining why definition
23949 of the DST built-in functions is not preceded by an assertion
23950 check. Add assertions to prevent ICE if attempts are made to
23951 define an altivec predicate or an abs* built-in function that has
23953 (htm_init_builtins): Add comment explaining why definition of the
23954 htm built-in functions is not preceded by an assertion check.
23956 2017-01-04 Jeff Law <law@redhat.com>
23958 PR tree-optimizatin/67955
23959 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23960 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23961 the points-to solution does not include pt_null. Use DECL_PT_UID
23964 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23966 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23967 Use gen_int_mode instead of gen_lopwart for const_int operands.
23969 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23971 PR tree-optimization/71563
23972 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23973 out of range value - has low bits known to be zero.
23975 2017-01-04 Alan Modra <amodra@gmail.com>
23977 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23978 * configure: Regenerate.
23979 * config.in: Regenerate.
23981 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23984 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23985 a substring of the message, but strcmp with the whole message. Ifdef
23986 ENABLE_NLS, translate the message first using dgettext.
23988 2017-01-03 Jeff Law <law@redhat.com>
23990 PR tree-optimizatin/78856
23991 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23992 (mark_threaded_blocks): Remove code to truncate thread paths that
23993 cross multiple loop headers. Instead invalidate the cached loop
23994 iteration information and handle case of a thread path walking
23995 into an irreducible region.
23997 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24000 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24001 assertions. Add support for doing the signbit if the IEEE 128-bit
24002 floating point value is in a GPR.
24003 * config/rs6000/rs6000.md (Fsignbit): Delete.
24004 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24005 Update the length attribute if the value is in a GPR.
24006 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24007 the sign or zero extension instruction, since the value is always 0/1.
24008 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24011 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24012 extracting SImode to a GPR register so that we can generate a
24013 store, limit the vector to be in a traditional Altivec register
24014 for the vextuwrx instruction.
24016 2017-01-03 Ian Lance Taylor <iant@google.com>
24018 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24020 2017-01-03 Martin Sebor <msebor@redhat.com>
24022 PR tree-optimization/78696
24023 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24024 precision. Use MPFR for %f for greater fidelity. Correct handling
24026 (pass_sprintf_length::compute_format_length): Set width and precision
24027 specified by asrerisk to void_node for vararg functions.
24028 (try_substitute_return_value): Adjust dump output.
24030 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24032 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24034 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24036 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24037 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24038 -mlra/-mno-lra was passed to the compiler.
24040 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24042 PR rtl-optimization/65618
24043 * emit-rtl.c (try_split): Move initialization of "before" and
24044 "after" to just before the call to emit_insn_after_setloc.
24046 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24048 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24050 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24052 * dwarf2out.c (gen_enumeration_type_die): When
24053 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24055 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24057 PR tree-optimization/78965
24058 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24059 Change first argument from const call_info & to call_info &. For %n
24060 set info.nowrite to false.
24062 PR middle-end/78901
24063 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24064 possibly throwing calls.
24066 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24067 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24068 and fns handling, rather than in a separate case SSA_NAME.
24070 2017-01-02 Jeff Law <law@redhat.com>
24072 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24073 fixes for first_period and second_period variables.
24075 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24078 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24079 (*insvqi_1): New insn pattern.
24080 (*insvqi_1_mem_rex64): Ditto.
24081 (*insvqi_2): Ditto.
24082 (*insvqi_3): Rename from *insvqi.
24084 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24086 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24088 * doc/cfg.texi (Edges): Remove reference to Java.
24089 (Maintaining the CFG): Ditto.
24091 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24093 PR middle-end/77674
24094 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24095 transparent aliases.
24097 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24099 PR middle-end/77484
24100 * predict.def (PRED_CALL): Update hitrate.
24101 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24102 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24103 into direct/indirect/polymorphic variants.
24105 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24107 Update copyright years.
24109 * gcc.c (process_command): Update copyright notice dates.
24110 * gcov-dump.c (print_version): Ditto.
24111 * gcov.c (print_version): Ditto.
24112 * gcov-tool.c (print_version): Ditto.
24113 * gengtype.c (create_file): Ditto.
24114 * doc/cpp.texi: Bump @copying's copyright year.
24115 * doc/cppinternals.texi: Ditto.
24116 * doc/gcc.texi: Ditto.
24117 * doc/gccint.texi: Ditto.
24118 * doc/gcov.texi: Ditto.
24119 * doc/install.texi: Ditto.
24120 * doc/invoke.texi: Ditto.
24122 Copyright (C) 2017 Free Software Foundation, Inc.
24124 Copying and distribution of this file, with or without modification,
24125 are permitted in any medium without royalty provided the copyright
24126 notice and this notice are preserved.