1 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6 (V_INT_EQUIV): ...this.
7 (v_cmp_result): Rename to...
8 (v_int_equiv): ...this.
9 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11 (copysign<mode>3): Likewise.
12 (aarch64_simd_bsl<mode>_internal): Likewise.
13 (aarch64_simd_bsl<mode>): Likewise.
14 (vec_cmp<mode><mode>): Likewise.
15 (vcond<mode><mode>): Likewise.
16 (vcond<v_cmp_mixed><mode>): Likewise.
17 (vcondu<mode><v_cmp_mixed>): Likewise.
18 (aarch64_cm<optab><mode>): Likewise.
19 (aarch64_cmtst<mode>): Likewise.
20 (aarch64_fac<optab><mode>): Likewise.
21 (vec_perm_const<mode>): Likewise.
22 (vcond_mask_<mode><v_cmp_result>): Rename to...
23 (vcond_mask_<mode><v_int_equiv>): ...this.
24 (vec_cmp<mode><v_cmp_result>): Rename to...
25 (vec_cmp<mode><v_int_equiv>): ...this.
27 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
28 Alan Hayward <alan.hayward@arm.com>
29 David Sherwood <david.sherwood@arm.com>
31 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
33 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
34 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
35 (UNSPEC_LD4_DREG): New unspecs.
36 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
37 (aarch64_ld2<mode>_dreg_be): Replace with...
38 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
40 (aarch64_ld3<mode>_dreg_le)
41 (aarch64_ld3<mode>_dreg_be): Replace with...
42 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
44 (aarch64_ld4<mode>_dreg_le)
45 (aarch64_ld4<mode>_dreg_be): Replace with...
46 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
49 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
51 PR tree-optimization/81987
52 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
53 insert an initializer in a location not dominated by the stride
56 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
58 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
59 on the entire header of the finally block in the fallthru case.
61 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
63 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
65 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
67 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
68 rs6000_emit_move_from_cr and call renamed function.
69 (rs6000_emit_prologue): Call renamed functions.
70 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
71 movesi_from_cr, remove volatile CRs.
73 2017-08-30 Jon Beniston <jon@beniston.com>
74 Richard Biener <rguenther@suse.de>
76 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
77 of VECTOR_MODE_P check.
78 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
81 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
83 * df.h (df_read_modify_subreg_p): Remove in favor of...
84 * rtl.h (read_modify_subreg_p): ...this new function. Take a
85 const_rtx instead of an rtx.
86 * cprop.c (local_cprop_find_used_regs): Update accordingly.
87 * df-problems.c (df_word_lr_mark_ref): Likewise.
88 * ira-lives.c (mark_pseudo_reg_live): Likewise.
89 (mark_pseudo_reg_dead): Likewise.
90 (mark_ref_dead): Likewise.
91 * reginfo.c (init_subregs_of_mode): Likewise.
92 * sched-deps.c (sched_analyze_1): Likewise.
93 * df-scan.c (df_def_record_1): Likewise.
94 (df_uses_record): Likewise.
95 (df_read_modify_subreg_p): Remove in favor of...
96 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
97 const_rtx instead of an rtx.
99 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
100 Alan Hayward <alan.hayward@arm.com>
101 David Sherwood <david.sherwood@arm.com>
103 * rtl.h (partial_subreg_p): New function.
104 * caller-save.c (save_call_clobbered_regs): Use it.
105 * calls.c (expand_call): Likewise.
106 * combine.c (combinable_i3pat): Likewise.
107 (simplify_set): Likewise.
108 (make_extraction): Likewise.
109 (make_compound_operation_int): Likewise.
110 (gen_lowpart_or_truncate): Likewise.
111 (force_to_mode): Likewise.
112 (make_field_assignment): Likewise.
113 (reg_truncated_to_mode): Likewise.
114 (record_truncated_value): Likewise.
115 (move_deaths): Likewise.
116 * cse.c (record_jump_cond): Likewise.
117 (cse_insn): Likewise.
118 * cselib.c (cselib_lookup_1): Likewise.
119 * expmed.c (extract_bit_field_using_extv): Likewise.
120 * function.c (assign_parm_setup_reg): Likewise.
121 * ifcvt.c (noce_convert_multiple_sets): Likewise.
122 * ira-build.c (create_insn_allocnos): Likewise.
123 * lra-coalesce.c (merge_pseudos): Likewise.
124 * lra-constraints.c (match_reload): Likewise.
125 (simplify_operand_subreg): Likewise.
126 (curr_insn_transform): Likewise.
127 * lra-lives.c (process_bb_lives): Likewise.
128 * lra.c (new_insn_reg): Likewise.
129 (lra_substitute_pseudo): Likewise.
130 * regcprop.c (mode_change_ok): Likewise.
131 (maybe_mode_change): Likewise.
132 (copyprop_hardreg_forward_1): Likewise.
133 * reload.c (push_reload): Likewise.
134 (find_reloads): Likewise.
135 (find_reloads_subreg_address): Likewise.
136 * reload1.c (alter_reg): Likewise.
137 (eliminate_regs_1): Likewise.
138 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
140 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
142 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
143 back to if statements, including unpack.
145 2017-08-30 Martin Liska <mliska@suse.cz>
148 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
150 (func_checker::compare_asm_inputs_outputs): ... this function.
151 (func_checker::compare_gimple_asm): Use the function to compare
153 * ipa-icf-gimple.h: Rename the function.
155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
156 Alan Hayward <alan.hayward@arm.com>
157 David Sherwood <david.sherwood@arm.com>
159 * coretypes.h (complex_mode): New type.
160 * gdbhooks.py (build_pretty_printer): Handle it.
161 * machmode.h (complex_mode): New class.
162 (complex_mode::includes_p): New function.
163 (is_complex_int_mode): Likewise.
164 (is_complex_float_mode): Likewise.
165 * genmodes.c (get_mode_class): Handle complex mode classes.
166 * function.c (expand_function_end): Use is_complex_int_mode.
168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
169 Alan Hayward <alan.hayward@arm.com>
170 David Sherwood <david.sherwood@arm.com>
172 * coretypes.h (scalar_mode_pod): New typedef.
173 * gdbhooks.py (build_pretty_printer): Handle it.
174 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
175 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
176 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
177 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * machmode.h (mode_for_vector): Take a scalar_mode instead
186 * stor-layout.c (mode_for_vector): Likewise.
187 * explow.c (promote_mode): Use as_a <scalar_mode>.
188 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
191 Alan Hayward <alan.hayward@arm.com>
192 David Sherwood <david.sherwood@arm.com>
194 * target.def (preferred_simd_mode): Take a scalar_mode
195 instead of a machine_mode.
196 * targhooks.h (default_preferred_simd_mode): Likewise.
197 * targhooks.c (default_preferred_simd_mode): Likewise.
198 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
199 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
200 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
201 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
202 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
203 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
204 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
205 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
207 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
208 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
209 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
210 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
211 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
212 * doc/tm.texi: Regenerate.
213 * optabs-query.c (can_vec_mask_load_store_p): Return false for
216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
217 Alan Hayward <alan.hayward@arm.com>
218 David Sherwood <david.sherwood@arm.com>
220 * target.def (scalar_mode_supported_p): Take a scalar_mode
221 instead of a machine_mode.
222 * targhooks.h (default_scalar_mode_supported_p): Likewise.
223 * targhooks.c (default_scalar_mode_supported_p): Likewise.
224 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
225 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
226 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
227 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
228 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
229 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
230 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
231 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
232 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
233 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
234 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
235 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
237 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
238 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
239 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
240 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
241 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
243 * doc/tm.texi: Regenerate.
245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
246 Alan Hayward <alan.hayward@arm.com>
247 David Sherwood <david.sherwood@arm.com>
249 * coretypes.h (opt_scalar_mode): New typedef.
250 * gdbhooks.py (build_pretty_printers): Handle it.
251 * machmode.h (mode_iterator::get_2xwider): Add overload for
253 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
255 * expr.c (convert_mode_scalar): Likewise.
256 * omp-low.c (omp_clause_aligned_alignment): Likewise.
257 * optabs.c (expand_float): Likewise.
258 (expand_fix): Likewise.
259 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
262 Alan Hayward <alan.hayward@arm.com>
263 David Sherwood <david.sherwood@arm.com>
265 * optabs.c (expand_float): Explicitly check for scalars before
266 using a branching expansion.
267 (expand_fix): Likewise.
269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
270 Alan Hayward <alan.hayward@arm.com>
271 David Sherwood <david.sherwood@arm.com>
273 * expr.c (convert_mode): Split scalar handling out into...
274 (convert_mode_scalar): ...this new function. Treat the modes
277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
278 Alan Hayward <alan.hayward@arm.com>
279 David Sherwood <david.sherwood@arm.com>
281 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
283 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
286 Alan Hayward <alan.hayward@arm.com>
287 David Sherwood <david.sherwood@arm.com>
289 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
290 rather than a machine_mode.
291 (fixed_from_string): Likewise.
292 (fixed_convert): Likewise.
293 (fixed_convert_from_int): Likewise.
294 (fixed_convert_from_real): Likewise.
295 (real_convert_from_fixed): Likewise.
296 * fixed-value.c (fixed_from_double_int): Likewise.
297 (fixed_from_string): Likewise.
298 (fixed_convert): Likewise.
299 (fixed_convert_from_int): Likewise.
300 (fixed_convert_from_real): Likewise.
301 (real_convert_from_fixed): Likewise.
302 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
305 Alan Hayward <alan.hayward@arm.com>
306 David Sherwood <david.sherwood@arm.com>
308 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
309 of separate mode class checks. Do not allow vector modes here.
310 (immed_wide_int_const): Use as_a <scalar_mode>.
311 * explow.c (trunc_int_for_mode): Likewise.
312 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
313 (wi::shwi): Likewise.
314 (wi::min_value): Likewise.
315 (wi::max_value): Likewise.
316 * dwarf2out.c (loc_descriptor): Likewise.
317 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
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 * tree.h (SCALAR_TYPE_MODE): New macro.
325 * expr.c (expand_expr_addr_expr_1): Use it.
326 (expand_expr_real_2): Likewise.
327 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
328 (fold_convert_const_fixed_from_int): Likewise.
329 (fold_convert_const_fixed_from_real): Likewise.
330 (native_encode_fixed): Likewise
331 (native_encode_complex): Likewise
332 (native_encode_vector): Likewise.
333 (native_interpret_fixed): Likewise.
334 (native_interpret_real): Likewise.
335 (native_interpret_complex): Likewise.
336 (native_interpret_vector): Likewise.
337 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
338 (simd_clone_adjust_argument_types): Likewise.
339 (simd_clone_init_simd_arrays): Likewise.
340 (simd_clone_adjust): Likewise.
341 * stor-layout.c (layout_type): Likewise.
342 * tree.c (build_minus_one_cst): Likewise.
343 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
344 * tree-inline.c (estimate_move_cost): Likewise.
345 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
346 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
347 (vectorizable_reduction): Likewise.
348 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
349 (vect_recog_mixed_size_cond_pattern): Likewise.
350 (check_bool_pattern): Likewise.
351 (adjust_bool_pattern): Likewise.
352 (search_type_for_mask_1): Likewise.
353 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
354 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
355 (vectorizable_load): Likewise.
356 (vectorizable_store): Likewise.
357 * ubsan.c (ubsan_encode_value): Likewise.
358 * varasm.c (output_constant): Likewise.
360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
361 Alan Hayward <alan.hayward@arm.com>
362 David Sherwood <david.sherwood@arm.com>
364 * coretypes.h (scalar_mode): New class.
365 * machmode.h (scalar_mode): Likewise.
366 (scalar_mode::includes_p): New function.
367 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
368 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
369 * genmodes.c (get_mode_class): Handle remaining scalar modes.
370 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
371 * expmed.c (store_bit_field_1): Likewise.
372 (extract_bit_field_1): Likewise.
373 * expr.c (write_complex_part): Likewise.
374 (read_complex_part): Likewise.
375 (emit_move_complex_push): Likewise.
376 (expand_expr_real_2): Likewise.
377 * function.c (assign_parm_setup_reg): Likewise.
378 (assign_parms_unsplit_complex): Likewise.
379 * optabs.c (expand_binop): Likewise.
380 * rtlanal.c (subreg_get_info): Likewise.
381 * simplify-rtx.c (simplify_immed_subreg): Likewise.
382 * varasm.c (output_constant_pool_2): Likewise.
384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
385 Alan Hayward <alan.hayward@arm.com>
386 David Sherwood <david.sherwood@arm.com>
388 * expmed.c (extract_high_half): Use scalar_int_mode and remove
390 (expmed_mult_highpart_optab): Likewise.
391 (expmed_mult_highpart): Likewise.
393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
394 Alan Hayward <alan.hayward@arm.com>
395 David Sherwood <david.sherwood@arm.com>
397 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
398 instead of a machine_mode.
399 (builtin_memset_read_str): Likewise.
400 * builtins.c (c_readstr): Likewise.
401 (builtin_memcpy_read_str): Likewise.
402 (builtin_strncpy_read_str): Likewise.
403 (builtin_memset_read_str): Likewise.
404 (builtin_memset_gen_str): Likewise.
405 (expand_builtin_signbit): Use scalar_int_mode for local variables.
406 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
407 instead of a machine_mode.
408 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
410 (make_extraction): Likewise.
411 (try_widen_shift_mode): Take and return scalar_int_modes instead
413 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
414 a scalar_int_mode instead of a machine_mode.
415 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
416 (avr_addr_space_pointer_mode): Likewise.
417 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
418 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
419 (msp430_unwind_word_mode): Likewise.
420 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
421 (spu_addr_space_pointer_mode): Likewise.
422 (spu_addr_space_address_mode): Likewise.
423 (spu_libgcc_cmp_return_mode): Likewise.
424 (spu_libgcc_shift_count_mode): Likewise.
425 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
426 (rl78_addr_space_pointer_mode): Likewise.
427 (fl78_unwind_word_mode): Likewise.
428 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
430 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
431 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
432 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
433 (mips_valid_pointer_mode): Likewise.
434 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
435 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
436 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
438 (ft32_addr_space_address_mode): Likewise.
439 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
440 scalar_int_mode instead of a machine_mode.
441 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
443 (m32c_addr_space_address_mode): Likewise.
444 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
445 (rs6000_eh_return_filter_mode): Likewise.
446 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
447 (rs6000_eh_return_filter_mode): Likewise.
448 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
449 (s390_libgcc_shift_count_mode): Likewise.
450 (s390_unwind_word_mode): Likewise.
451 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
453 * target.def (mode_rep_extended): Likewise.
454 (valid_pointer_mode): Likewise.
455 (addr_space.valid_pointer_mode): Likewise.
456 (eh_return_filter_mode): Return a scalar_int_mode rather than
458 (libgcc_cmp_return_mode): Likewise.
459 (libgcc_shift_count_mode): Likewise.
460 (unwind_word_mode): Likewise.
461 (addr_space.pointer_mode): Likewise.
462 (addr_space.address_mode): Likewise.
463 * doc/tm.texi: Regenerate.
464 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
466 (do_jump): Use scalar_int_mode for local variables.
467 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
468 rather than a machine_mode.
469 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
470 (scompare_loc_descriptor_wide): Likewise.
471 (scompare_loc_descriptor_narrow): Likewise.
472 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
474 * except.c (sjlj_emit_dispatch_table): Likewise.
475 (expand_builtin_eh_copy_values): Likewise.
476 * explow.c (convert_memory_address_addr_space_1): Likewise.
477 Take a scalar_int_mode rather than a machine_mode.
478 (convert_memory_address_addr_space): Take a scalar_int_mode rather
480 (memory_address_addr_space): Use scalar_int_mode for local variables.
481 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
482 rather than a machine_mode.
483 * expmed.c (mask_rtx): Likewise.
484 (init_expmed_one_conv): Likewise.
485 (expand_mult_highpart_adjust): Likewise.
486 (extract_high_half): Likewise.
487 (expmed_mult_highpart_optab): Likewise.
488 (expmed_mult_highpart): Likewise.
489 (expand_smod_pow2): Likewise.
490 (expand_sdiv_pow2): Likewise.
491 (emit_store_flag_int): Likewise.
492 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
494 (extract_low_bits): Likewise.
495 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
497 * expr.c (pieces_addr::adjust): Likewise.
498 (can_store_by_pieces): Likewise.
499 (store_by_pieces): Likewise.
500 (clear_by_pieces_1): Likewise.
501 (expand_expr_addr_expr_1): Likewise.
502 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
503 (expand_expr_real_1): Likewise.
504 (try_casesi): Likewise.
505 * final.c (shorten_branches): Likewise.
506 * fold-const.c (fold_convert_const_int_from_fixed): Change the
507 type of "mode" to machine_mode.
508 * internal-fn.c (expand_arith_overflow_result_store): Take a
509 scalar_int_mode rather than a machine_mode.
510 (expand_mul_overflow): Use scalar_int_mode for local variables.
511 * loop-doloop.c (doloop_modify): Likewise.
512 (doloop_optimize): Likewise.
513 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
515 (expand_doubleword_shift_condmove): Likewise.
516 (expand_doubleword_shift): Likewise.
517 (expand_doubleword_clz): Likewise.
518 (expand_doubleword_popcount): Likewise.
519 (expand_doubleword_parity): Likewise.
520 (expand_absneg_bit): Use scalar_int_mode for local variables.
521 (prepare_float_lib_cmp): Likewise.
522 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
523 rather than a machine_mode.
524 (convert_memory_address_addr_space): Likewise.
525 (get_mode_bounds): Likewise.
526 (get_address_mode): Return a scalar_int_mode rather than a
528 * rtlanal.c (get_address_mode): Likewise.
529 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
531 * targhooks.c (default_mode_rep_extended): Likewise.
532 (default_valid_pointer_mode): Likewise.
533 (default_addr_space_valid_pointer_mode): Likewise.
534 (default_eh_return_filter_mode): Return a scalar_int_mode rather
536 (default_libgcc_cmp_return_mode): Likewise.
537 (default_libgcc_shift_count_mode): Likewise.
538 (default_unwind_word_mode): Likewise.
539 (default_addr_space_pointer_mode): Likewise.
540 (default_addr_space_address_mode): Likewise.
541 * targhooks.h (default_eh_return_filter_mode): Likewise.
542 (default_libgcc_cmp_return_mode): Likewise.
543 (default_libgcc_shift_count_mode): Likewise.
544 (default_unwind_word_mode): Likewise.
545 (default_addr_space_pointer_mode): Likewise.
546 (default_addr_space_address_mode): Likewise.
547 (default_mode_rep_extended): Take a scalar_int_mode rather than
549 (default_valid_pointer_mode): Likewise.
550 (default_addr_space_valid_pointer_mode): Likewise.
551 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
553 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
554 rather than a machine_mode.
555 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
557 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
558 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
566 the mode argument to scalar_int_mode.
567 (do_jump_by_parts_zero_rtx): Likewise.
568 (do_jump_by_parts_equality_rtx): Likewise.
569 (do_jump_by_parts_greater): Take a mode argument.
570 (do_jump_by_parts_equality): Likewise.
571 (do_jump_1): Update calls accordingly.
573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
574 Alan Hayward <alan.hayward@arm.com>
575 David Sherwood <david.sherwood@arm.com>
577 * is-a.h (safe_dyn_cast): New function.
578 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
579 (jump_table_for_label): Likewise.
580 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
581 instead of an rtx_insn *.
582 (shorten_branches): Use dyn_cast instead of LABEL_P and
583 JUMP_TABLE_DATA_P. Use jump_table_for_label and
584 rtx_jump_table_data::get_data_mode.
585 (final_scan_insn): Likewise.
587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
588 Alan Hayward <alan.hayward@arm.com>
589 David Sherwood <david.sherwood@arm.com>
591 * combine.c (try_combine): Use is_a <scalar_int_mode> when
592 trying to combine a full-register integer set with a subreg
595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
596 Alan Hayward <alan.hayward@arm.com>
597 David Sherwood <david.sherwood@arm.com>
599 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
600 that is always either address_mode or pointer_mode.
602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
603 Alan Hayward <alan.hayward@arm.com>
604 David Sherwood <david.sherwood@arm.com>
606 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
607 when the two are known to be equal.
609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
610 Alan Hayward <alan.hayward@arm.com>
611 David Sherwood <david.sherwood@arm.com>
613 * simplify-rtx.c (simplify_const_unary_operation): Use
614 is_a <scalar_int_mode> instead of checking for a nonzero
615 precision. Forcibly convert op_mode to a scalar_int_mode
616 in that case. More clearly differentiate the operand and
617 result modes and use the former when deciding what the value
618 of a count-bits operation should be. Use is_int_mode instead
619 of checking for a MODE_INT. Remove redundant check for whether
620 this mode has a zero precision.
622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
623 Alan Hayward <alan.hayward@arm.com>
624 David Sherwood <david.sherwood@arm.com>
626 * optabs.c (widen_leading): Change the type of the mode argument
627 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
628 (widen_bswap): Likewise.
629 (expand_parity): Likewise.
630 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
631 (expand_ffs): Likewise.
632 (epand_unop): Check for scalar integer modes before calling the
635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
636 Alan Hayward <alan.hayward@arm.com>
637 David Sherwood <david.sherwood@arm.com>
639 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
641 (expand_expr_real_1): Update call accordingly.
643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
644 Alan Hayward <alan.hayward@arm.com>
645 David Sherwood <david.sherwood@arm.com>
647 * expmed.c (store_bit_field_using_insv): Add op0_mode and
648 value_mode arguments. Use scalar_int_mode internally.
649 (store_bit_field_1): Rename the new integer mode from imode
650 to op0_mode and use it instead of GET_MODE (op0). Update calls
651 to store_split_bit_field, store_bit_field_using_insv and
652 store_fixed_bit_field.
653 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
654 Use scalar_int_mode internally. Use a bit count rather than a mode
655 when calculating the largest bit size for get_best_mode.
656 Update calls to store_split_bit_field and store_fixed_bit_field_1.
657 (store_fixed_bit_field_1): Add mode and value_mode arguments.
658 Remove assertion that OP0 has a scalar integer mode.
659 (store_split_bit_field): Add op0_mode and value_mode arguments.
660 Update calls to extract_fixed_bit_field.
661 (extract_bit_field_using_extv): Add an op0_mode argument.
662 Use scalar_int_mode internally.
663 (extract_bit_field_1): Rename the new integer mode from imode to
664 op0_mode and use it instead of GET_MODE (op0). Update calls to
665 extract_split_bit_field, extract_bit_field_using_extv and
666 extract_fixed_bit_field.
667 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
668 to extract_split_bit_field and extract_fixed_bit_field_1.
669 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
670 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
672 (extract_split_bit_field): Add an op0_mode argument. Update call
673 to extract_fixed_bit_field.
675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
676 Alan Hayward <alan.hayward@arm.com>
677 David Sherwood <david.sherwood@arm.com>
679 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
680 * explow.c (hard_function_value): Likewise.
681 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
682 convert_to_mode call outside the loop.
683 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
684 for the mode iterator. Require the mode specified by max_pieces
686 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
688 (copy_blkmode_to_reg): Likewise.
689 (set_storage_via_setmem): Likewise.
690 * optabs.c (prepare_cmp_insn): Likewise.
691 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
692 * stor-layout.c (finish_bitfield_representative): 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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
699 * expr.c (convert_move): Use them.
700 (convert_modes): Likewise.
701 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
708 parameter for the mode of "x". Remove the "known_x", "known_mode"
709 and "known_ret" arguments. Change the type of the mode argument
711 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
712 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
713 (reg_num_sign_bit_copies_for_combine): Likewise.
714 * rtlanal.c (nonzero_bits1): Likewise.
715 (num_sign_bit_copies1): Likewise.
716 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
717 (reg_num_sign_bit_copies_general): Likewise.
718 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
719 (reg_nonzero_bits_general): Likewise.
721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
722 Alan Hayward <alan.hayward@arm.com>
723 David Sherwood <david.sherwood@arm.com>
725 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
726 than in subroutines. Return 1 for non-integer modes.
727 (cached_num_sign_bit_copies): Change the type of the mode parameter
729 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
730 classes. Handle CONST_INT_P first and then check whether X also
731 has a scalar integer mode. Check the same thing for inner registers
732 of a SUBREG and for values that are being extended or truncated.
734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
735 Alan Hayward <alan.hayward@arm.com>
736 David Sherwood <david.sherwood@arm.com>
738 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
739 in subroutines. Return the mode mask for non-integer modes.
740 (cached_nonzero_bits): Change the type of the mode parameter
742 (nonzero_bits1): Likewise. Remove early exit for other mode
743 classes. Handle CONST_INT_P first and then check whether X
744 also has a scalar integer mode.
746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
747 Alan Hayward <alan.hayward@arm.com>
748 David Sherwood <david.sherwood@arm.com>
750 * expr.c (widest_int_mode_for_size): Make the comment match the code.
751 Return a scalar_int_mode and assert that the size is greater than
753 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
754 (op_by_pieces_d::op_by_pieces_d): Likewise.
755 (op_by_pieces_d::run): Likewise.
756 (can_store_by_pieces): Likewise.
758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
759 Alan Hayward <alan.hayward@arm.com>
760 David Sherwood <david.sherwood@arm.com>
762 * combine.c (extract_left_shift): Add a mode argument and update
764 (make_compound_operation_int): Change the type of the mode parameter
765 to scalar_int_mode and update the call to extract_left_shift.
767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
768 Alan Hayward <alan.hayward@arm.com>
769 David Sherwood <david.sherwood@arm.com>
771 * combine.c (simplify_and_const_int): Change the type of the mode
772 parameter to scalar_int_mode.
773 (simplify_and_const_int_1): Likewise. Update recursive call.
775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
776 Alan Hayward <alan.hayward@arm.com>
777 David Sherwood <david.sherwood@arm.com>
779 * combine.c (simplify_compare_const): Check that the mode is a
780 scalar_int_mode (rather than VOIDmode) before testing its
782 (simplify_comparison): Move COMPARISON_P handling out of the
783 loop and restrict the latter part of the loop to scalar_int_modes.
784 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
785 and when considering SUBREG_REGs. Use is_int_mode instead of
786 checking GET_MODE_CLASS against MODE_INT.
788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
789 Alan Hayward <alan.hayward@arm.com>
790 David Sherwood <david.sherwood@arm.com>
792 * combine.c (try_widen_shift_mode): Move check for equal modes to...
793 (simplify_shift_const_1): ...here. Use scalar_int_mode for
794 shift_unit_mode and for modes involved in scalar shifts.
796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
797 Alan Hayward <alan.hayward@arm.com>
798 David Sherwood <david.sherwood@arm.com>
800 * combine.c (force_int_to_mode): New function, split out from...
801 (force_to_mode): ...here. Keep xmode up-to-date and use it
802 instead of GET_MODE (x).
804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
805 Alan Hayward <alan.hayward@arm.com>
806 David Sherwood <david.sherwood@arm.com>
808 * optabs-query.h (extraction_insn::struct_mode): Change type to
809 opt_scalar_int_mode and update comment.
810 (extraction_insn::field_mode): Change type to scalar_int_mode.
811 (extraction_insn::pos_mode): Likewise.
812 * combine.c (make_extraction): Update accordingly.
813 * optabs-query.c (get_traditional_extraction_insn): Likewise.
814 (get_optab_extraction_insn): Likewise.
815 * recog.c (simplify_while_replacing): Likewise.
816 * expmed.c (narrow_bit_field_mem): Change the type of the mode
817 parameter to opt_scalar_int_mode.
819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
820 Alan Hayward <alan.hayward@arm.com>
821 David Sherwood <david.sherwood@arm.com>
823 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
824 to a scalar_int_mode instead of a machine_mode.
825 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
826 (get_best_mode): Return a boolean and use a pointer argument to store
827 the selected mode. Replace the limit mode parameter with a bit limit.
828 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
829 for the values returned by bit_field_mode_iterator::next_mode.
830 (store_bit_field): Update call to get_best_mode.
831 (store_fixed_bit_field): Likewise.
832 (extract_fixed_bit_field): Likewise.
833 * expr.c (optimize_bitfield_assignment_op): Likewise.
834 * fold-const.c (optimize_bit_field_compare): Likewise.
835 (fold_truth_andor_1): Likewise.
836 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
837 Update for new type of m_mode.
838 (get_best_mode): As above.
840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
841 Alan Hayward <alan.hayward@arm.com>
842 David Sherwood <david.sherwood@arm.com>
844 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
845 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
846 (store_bit_field): Check is_a <scalar_int_mode> before calling
847 strict_volatile_bitfield_p.
848 (extract_bit_field): Likewise.
850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
851 Alan Hayward <alan.hayward@arm.com>
852 David Sherwood <david.sherwood@arm.com>
854 * target.def (cstore_mode): Return a scalar_int_mode.
855 * doc/tm.texi: Regenerate.
856 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
857 * targhooks.h (default_cstore_mode): Likewise.
858 * targhooks.c (default_cstore_mode): Likewise, using a forced
860 * expmed.c (emit_cstore): Expect the target of the cstore to be
863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
864 Alan Hayward <alan.hayward@arm.com>
865 David Sherwood <david.sherwood@arm.com>
867 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
869 (niter_desc): Likewise mode.
870 (iv_analyze): Add a mode parameter.
872 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
873 and change its type to scalar_int_mode.
874 * loop-iv.c: Update commentary at head of file.
875 (iv_constant): Pass the mode paraeter before the rtx it describes
876 and change its type to scalar_int_mode. Remove VOIDmode handling.
877 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
878 (iv_extend): Likewise.
879 (shorten_into_mode): Likewise.
880 (iv_add): Use scalar_int_mode.
882 (iv_shift): Likewise.
883 (canonicalize_iv_subregs): Likewise.
884 (get_biv_step_1): Pass the outer_mode parameter before the rtx
885 it describes and change its mode to scalar_int_mode. Also change
886 the type of the returned inner_mode to scalar_int_mode.
887 (get_biv_step): Likewise, turning outer_mode from a pointer
888 into a direct parameter. Update call to get_biv_step_1.
889 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
890 iv_constant and get_biv_step.
891 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
892 and change its type to scalar_int_mode. Don't initialise iv->mode
893 to VOIDmode and remove later checks for its still being VOIDmode.
894 Update calls to iv_analyze_op and iv_analyze_expr. Check
895 is_a <scalar_int_mode> when changing the mode under consideration.
896 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
897 Update call to iv_analyze_expr.
898 (iv_analyze_op): Add a mode parameter. Reject subregs whose
899 inner register is not also a scalar_int_mode. Update call to
901 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
902 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
903 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
904 separate mode class checks. Update calls to iv_analyze. Remove
905 fix-up of VOIDmodes after iv_analyze_biv.
906 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
907 don't have a scalar_int_mode. Update call to biv_p.
909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
910 Alan Hayward <alan.hayward@arm.com>
911 David Sherwood <david.sherwood@arm.com>
913 * cfgexpand.c (convert_debug_memory_address): Use
914 as_a <scalar_int_mode>.
915 * combine.c (expand_compound_operation): Likewise.
916 (make_extraction): Likewise.
917 (change_zero_ext): Likewise.
918 (simplify_comparison): Likewise.
919 * cse.c (cse_insn): Likewise.
920 * dwarf2out.c (minmax_loc_descriptor): Likewise.
921 (mem_loc_descriptor): Likewise.
922 (loc_descriptor): Likewise.
923 * expmed.c (init_expmed_one_mode): Likewise.
924 (synth_mult): Likewise.
925 (emit_store_flag_1): Likewise.
926 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
927 of a comparison with size.
928 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
929 (reduce_to_bit_field_precision): Likewise.
930 * function.c (expand_function_end): Likewise.
931 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
932 * loop-doloop.c (doloop_modify): Likewise.
933 * optabs.c (expand_binop): Likewise.
934 (expand_unop): Likewise.
935 (expand_copysign_absneg): Likewise.
936 (prepare_cmp_insn): Likewise.
937 (maybe_legitimize_operand): Likewise.
938 * recog.c (const_scalar_int_operand): Likewise.
939 * rtlanal.c (get_address_mode): Likewise.
940 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
941 (simplify_cond_clz_ctz): Likewise.
942 * tree-nested.c (get_nl_goto_field): Likewise.
943 * tree.c (build_vector_type_for_mode): Likewise.
944 * var-tracking.c (use_narrower_mode): Likewise.
946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
947 Alan Hayward <alan.hayward@arm.com>
948 David Sherwood <david.sherwood@arm.com>
950 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
951 * builtins.c (expand_builtin_signbit): Use it.
952 * cfgexpand.c (expand_debug_expr): Likewise.
953 * dojump.c (do_jump): Likewise.
954 (do_compare_and_jump): Likewise.
955 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
956 * expmed.c (make_tree): Likewise.
957 * expr.c (expand_expr_real_2): Likewise.
958 (expand_expr_real_1): Likewise.
959 (try_casesi): Likewise.
960 * fold-const-call.c (fold_const_call_ss): Likewise.
961 * fold-const.c (unextend): Likewise.
962 (extract_muldiv_1): Likewise.
963 (fold_single_bit_test): Likewise.
964 (native_encode_int): Likewise.
965 (native_encode_string): Likewise.
966 (native_interpret_int): Likewise.
967 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
968 * internal-fn.c (expand_addsub_overflow): Likewise.
969 (expand_neg_overflow): Likewise.
970 (expand_mul_overflow): Likewise.
971 (expand_arith_overflow): Likewise.
972 * match.pd: Likewise.
973 * stor-layout.c (layout_type): Likewise.
974 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
975 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
976 * tree-ssanames.c (get_range_info): Likewise.
977 * tree-switch-conversion.c (array_value_type) Likewise.
978 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
979 (vect_recog_divmod_pattern): Likewise.
980 (vect_recog_mixed_size_cond_pattern): Likewise.
981 * tree-vrp.c (extract_range_basic): Likewise.
982 (simplify_float_conversion_using_ranges): Likewise.
983 * tree.c (int_fits_type_p): Likewise.
984 * ubsan.c (instrument_bool_enum_load): Likewise.
985 * varasm.c (mergeable_string_section): Likewise.
986 (narrowing_initializer_constant_valid_p): Likewise.
987 (output_constant): Likewise.
989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
990 Alan Hayward <alan.hayward@arm.com>
991 David Sherwood <david.sherwood@arm.com>
993 * machmode.h (NARROWEST_INT_MODE): New macro.
994 * expr.c (alignment_for_piecewise_move): Use it instead of
995 GET_CLASS_NARROWEST_MODE (MODE_INT).
996 (push_block): Likewise.
997 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
999 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1002 Alan Hayward <alan.hayward@arm.com>
1003 David Sherwood <david.sherwood@arm.com>
1005 * postreload.c (move2add_valid_value_p): Change the type of the
1006 mode parameter to scalar_int_mode.
1007 (move2add_use_add2_insn): Add a mode parameter and use it instead
1009 (move2add_use_add3_insn): Likewise.
1010 (reload_cse_move2add): Update accordingly.
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 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
1018 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
1019 * optabs.c (expand_unop): Likewise.
1021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1022 Alan Hayward <alan.hayward@arm.com>
1023 David Sherwood <david.sherwood@arm.com>
1025 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
1026 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
1027 (popcount_loc_descriptor): Likewise.
1028 (bswap_loc_descriptor): Likewise.
1029 (rotate_loc_descriptor): Likewise.
1030 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
1031 calling the functions above.
1033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1034 Alan Hayward <alan.hayward@arm.com>
1035 David Sherwood <david.sherwood@arm.com>
1037 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
1039 (try_combine): Likewise.
1040 (simplify_if_then_else): Likewise.
1041 * cse.c (cse_insn): Likewise.
1042 * dwarf2out.c (mem_loc_descriptor): Likewise.
1043 * emit-rtl.c (gen_lowpart_common): Likewise.
1044 * simplify-rtx.c (simplify_truncation): Likewise.
1045 (simplify_binary_operation_1): Likewise.
1046 (simplify_const_relational_operation): Likewise.
1047 (simplify_ternary_operation): Likewise.
1048 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
1050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1051 Alan Hayward <alan.hayward@arm.com>
1052 David Sherwood <david.sherwood@arm.com>
1054 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
1055 * reload.c (push_reload): Likewise.
1056 (find_reloads): Likewise.
1058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1059 Alan Hayward <alan.hayward@arm.com>
1060 David Sherwood <david.sherwood@arm.com>
1062 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
1063 (make_compound_operation_int): Likewise.
1064 (change_zero_ext): Likewise.
1065 * expr.c (convert_move): Likewise.
1066 (convert_modes): Likewise.
1067 * fwprop.c (forward_propagate_subreg): Likewise.
1068 * loop-iv.c (get_biv_step_1): Likewise.
1069 * optabs.c (widen_operand): Likewise.
1070 * postreload.c (move2add_valid_value_p): Likewise.
1071 * recog.c (simplify_while_replacing): Likewise.
1072 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1073 (simplify_binary_operation_1): Likewise. Remove redundant
1074 mode equality check.
1076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1077 Alan Hayward <alan.hayward@arm.com>
1078 David Sherwood <david.sherwood@arm.com>
1080 * combine.c (combine_simplify_rtx): Add checks for
1081 is_a <scalar_int_mode>.
1082 (simplify_if_then_else): Likewise.
1083 (make_field_assignment): Likewise.
1084 (simplify_comparison): Likewise.
1085 * ifcvt.c (noce_try_bitop): Likewise.
1086 * loop-invariant.c (canonicalize_address_mult): Likewise.
1087 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1090 Alan Hayward <alan.hayward@arm.com>
1091 David Sherwood <david.sherwood@arm.com>
1093 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1094 is_a <scalar_int_mode> instead of != BLKmode.
1096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1097 Alan Hayward <alan.hayward@arm.com>
1098 David Sherwood <david.sherwood@arm.com>
1100 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1101 instead of != VOIDmode.
1102 * combine.c (if_then_else_cond): Likewise.
1103 (change_zero_ext): Likewise.
1104 * dwarf2out.c (mem_loc_descriptor): Likewise.
1105 (loc_descriptor): Likewise.
1106 * rtlanal.c (canonicalize_condition): Likewise.
1107 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1110 Alan Hayward <alan.hayward@arm.com>
1111 David Sherwood <david.sherwood@arm.com>
1113 * simplify-rtx.c (simplify_binary_operation_1): Use
1114 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
1116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1117 Alan Hayward <alan.hayward@arm.com>
1118 David Sherwood <david.sherwood@arm.com>
1120 * wide-int.h (int_traits<unsigned char>) New class.
1121 (int_traits<unsigned short>) Likewise.
1122 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1123 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1125 * combine.c (set_nonzero_bits_and_sign_copies): Use
1126 is_a <scalar_int_mode>.
1127 (find_split_point): Likewise.
1128 (combine_simplify_rtx): Likewise.
1129 (simplify_logical): Likewise.
1130 (expand_compound_operation): Likewise.
1131 (expand_field_assignment): Likewise.
1132 (make_compound_operation): Likewise.
1133 (extended_count): Likewise.
1134 (change_zero_ext): Likewise.
1135 (simplify_comparison): Likewise.
1136 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1137 (ucompare_loc_descriptor): Likewise.
1138 (minmax_loc_descriptor): Likewise.
1139 (mem_loc_descriptor): Likewise.
1140 (loc_descriptor): Likewise.
1141 * expmed.c (init_expmed_one_mode): Likewise.
1142 * lra-constraints.c (lra_constraint_offset): Likewise.
1143 * optabs.c (prepare_libcall_arg): Likewise.
1144 * postreload.c (move2add_note_store): Likewise.
1145 * reload.c (operands_match_p): Likewise.
1146 * rtl.h (load_extend_op): Likewise.
1147 * rtlhooks.c (gen_lowpart_general): Likewise.
1148 * simplify-rtx.c (simplify_truncation): Likewise.
1149 (simplify_unary_operation_1): Likewise.
1150 (simplify_binary_operation_1): Likewise.
1151 (simplify_const_binary_operation): Likewise.
1152 (simplify_const_relational_operation): Likewise.
1153 (simplify_subreg): Likewise.
1154 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1155 * var-tracking.c (adjust_mems): Likewise.
1156 (prepare_call_arguments): Likewise.
1158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * machmode.h (is_int_mode): New fuction.
1163 * combine.c (find_split_point): Use it.
1164 (combine_simplify_rtx): Likewise.
1165 (simplify_if_then_else): Likewise.
1166 (simplify_set): Likewise.
1167 (simplify_shift_const_1): Likewise.
1168 (simplify_comparison): Likewise.
1169 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1170 * cse.c (notreg_cost): Likewise.
1171 (cse_insn): Likewise.
1172 * cselib.c (cselib_lookup_1): Likewise.
1173 * dojump.c (do_jump_1): Likewise.
1174 (do_compare_rtx_and_jump): Likewise.
1175 * dse.c (get_call_args): Likewise.
1176 * dwarf2out.c (rtl_for_decl_init): Likewise.
1177 (native_encode_initializer): Likewise.
1178 * expmed.c (emit_store_flag_1): Likewise.
1179 (emit_store_flag): Likewise.
1180 * expr.c (convert_modes): Likewise.
1181 (store_field): Likewise.
1182 (expand_expr_real_1): Likewise.
1183 * fold-const.c (fold_read_from_constant_string): Likewise.
1184 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1185 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1186 * optabs.c (expand_binop): Likewise.
1187 (expand_unop): Likewise.
1188 (expand_abs_nojump): Likewise.
1189 (expand_one_cmpl_abs_nojump): Likewise.
1190 * simplify-rtx.c (mode_signbit_p): Likewise.
1191 (val_signbit_p): Likewise.
1192 (val_signbit_known_set_p): Likewise.
1193 (val_signbit_known_clear_p): Likewise.
1194 (simplify_relational_operation_1): Likewise.
1195 * tree.c (vector_type_mode): Likewise.
1197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1198 Alan Hayward <alan.hayward@arm.com>
1199 David Sherwood <david.sherwood@arm.com>
1201 * machmode.h (smallest_mode_for_size): Fix formatting.
1202 (smallest_int_mode_for_size): New function.
1203 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1204 instead of smallest_mode_for_size.
1205 * combine.c (make_extraction): Likewise.
1206 * config/arc/arc.c (arc_expand_movmem): Likewise.
1207 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1208 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1209 * config/s390/s390.c (s390_expand_insv): Likewise.
1210 * config/sparc/sparc.c (assign_int_registers): Likewise.
1211 * config/spu/spu.c (spu_function_value): Likewise.
1212 (spu_function_arg): Likewise.
1213 * coverage.c (get_gcov_type): Likewise.
1214 (get_gcov_unsigned_t): Likewise.
1215 * dse.c (find_shift_sequence): Likewise.
1216 * expmed.c (store_bit_field_1): Likewise.
1217 * expr.c (convert_move): Likewise.
1218 (store_field): Likewise.
1219 * internal-fn.c (expand_arith_overflow): Likewise.
1220 * optabs-query.c (get_best_extraction_insn): Likewise.
1221 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1222 * stor-layout.c (layout_type): Likewise.
1223 (initialize_sizetypes): Likewise.
1224 * targhooks.c (default_get_mask_mode): Likewise.
1225 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1228 Alan Hayward <alan.hayward@arm.com>
1229 David Sherwood <david.sherwood@arm.com>
1231 * machmode.h (opt_mode::else_blk): New function.
1232 (int_mode_for_mode): Declare.
1233 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1234 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1236 * cfgexpand.c (expand_debug_expr): Likewise.
1237 * combine.c (gen_lowpart_or_truncate): Likewise.
1238 (gen_lowpart_for_combine): Likewise.
1239 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1240 * config/avr/avr.c (avr_to_int_mode): Likewise.
1241 (avr_out_plus_1): Likewise.
1242 (avr_out_plus): Likewise.
1243 (avr_out_round): Likewise.
1244 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1245 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1246 (s390_expand_vcond): Likewise.
1247 * config/spu/spu.c (spu_split_immediate): Likewise.
1248 (spu_expand_mov): Likewise.
1249 * dse.c (get_stored_val): Likewise.
1250 * expmed.c (store_bit_field_1): Likewise.
1251 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1253 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1254 (extract_low_bits): Likewise.
1255 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1256 handling rather than repeating the check.
1257 (emit_group_store): Likewise.
1258 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1259 * optabs.c (expand_absneg_bit): Likewise.
1260 (expand_copysign_absneg): Likewise.
1261 (expand_copysign_bit): Likewise.
1262 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1263 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1264 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1265 * var-tracking.c (prepare_call_arguments): Likewise.
1266 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1267 int_mode_for_mode instead of mode_for_size.
1268 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1271 Alan Hayward <alan.hayward@arm.com>
1272 David Sherwood <david.sherwood@arm.com>
1274 * machmode.h (int_mode_for_size): New function.
1275 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1276 instead of mode_for_size.
1277 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1279 * combine.c (expand_field_assignment): Use int_mode_for_size
1280 instead of mode_for_size.
1281 (make_extraction): Likewise.
1282 (simplify_shift_const_1): Likewise.
1283 (simplify_comparison): Likewise.
1284 * dojump.c (do_jump): Likewise.
1285 * dwarf2out.c (mem_loc_descriptor): Likewise.
1286 * emit-rtl.c (init_derived_machine_modes): Likewise.
1287 * expmed.c (flip_storage_order): Likewise.
1288 (convert_extracted_bit_field): Likewise.
1289 * expr.c (copy_blkmode_from_reg): Likewise.
1290 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1291 * internal-fn.c (expand_mul_overflow): Likewise.
1292 * lower-subreg.c (simple_move): Likewise.
1293 * optabs-libfuncs.c (init_optabs): Likewise.
1294 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1295 * tree.c (vector_type_mode): Likewise.
1296 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1297 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1298 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1299 * tree-vect-stmts.c (vectorizable_load): Likewise.
1300 (vectorizable_store): Likewise.
1302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1303 Alan Hayward <alan.hayward@arm.com>
1304 David Sherwood <david.sherwood@arm.com>
1306 * coretypes.h (pod_mode): New type.
1307 (scalar_int_mode_pod): New typedef.
1308 * machmode.h (pod_mode): New class.
1309 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1310 * genmodes.c (emit_mode_int_n): Update accordingly.
1311 * lower-subreg.h (target_lower_subreg): Change type to
1312 scalar_int_mode_pod.
1313 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1314 scalar_int_mode_pod.
1316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1317 Alan Hayward <alan.hayward@arm.com>
1318 David Sherwood <david.sherwood@arm.com>
1320 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1321 machine_mode to scalar_int_mode.
1322 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1323 (rs6000_option_override_internal): Remove cast to int.
1324 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1325 machine_mode to scalar_int_mode.
1326 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1327 (rs6000_option_override_internal): Remove cast to int.
1328 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1329 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1331 * config/s390/s390.c (s390_expand_builtin): Likewise.
1332 * coretypes.h (scalar_int_mode): New type.
1333 (opt_scalar_int_mode): New typedef.
1334 * machmode.h (scalar_int_mode): New class.
1335 (scalar_int_mode::includes_p): New function.
1336 (byte_mode): Change type to scalar_int_mode.
1337 (word_mode): Likewise.
1338 (ptr_mode): Likewise.
1339 * emit-rtl.c (byte_mode): Likewise.
1340 (word_mode): Likewise.
1341 (ptr_mode): Likewise.
1342 (init_derived_machine_modes): Update accordingly.
1343 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1344 and MODE_PARTIAL_INT.
1345 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1346 opt_scalar_int_mode.
1348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1349 Alan Hayward <alan.hayward@arm.com>
1350 David Sherwood <david.sherwood@arm.com>
1352 * target.def (libgcc_floating_mode_supported_p): Take a
1354 * doc/tm.texi: Regenerate.
1355 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1357 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1358 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1362 Alan Hayward <alan.hayward@arm.com>
1363 David Sherwood <david.sherwood@arm.com>
1365 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1366 * doc/tm.texi: Regenerate.
1367 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1368 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1369 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1370 * targhooks.h (default_floatn_mode): Likewise.
1371 * targhooks.c (default_floatn_mode): Likewise.
1372 * tree.c (build_common_tree_nodes): Update accordingly.
1374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1379 (mode_iterator::iterate_p): Likewise.
1380 (mode_iterator::get_wider): Likewise.
1381 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1384 Alan Hayward <alan.hayward@arm.com>
1385 David Sherwood <david.sherwood@arm.com>
1387 * coretypes.h (opt_scalar_float_mode): New typedef.
1388 * machmode.h (float_mode_for_size): New function.
1389 * emit-rtl.c (double_mode): Delete.
1390 (init_emit_once): Use float_mode_for_size.
1391 * stor-layout.c (layout_type): Likewise.
1392 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1395 Alan Hayward <alan.hayward@arm.com>
1396 David Sherwood <david.sherwood@arm.com>
1398 * output.h (assemble_real): Take a scalar_float_mode.
1399 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1400 * config/arm/arm.md (consttable_4): Likewise.
1401 (consttable_8): Likewise.
1402 (consttable_16): Likewise.
1403 * config/mips/mips.md (consttable_float): Likewise.
1404 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1405 * varasm.c (assemble_real): Take a scalar_float_mode.
1406 (output_constant_pool_2): Update accordingly.
1407 (output_constant): Likewise.
1409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1410 Alan Hayward <alan.hayward@arm.com>
1411 David Sherwood <david.sherwood@arm.com>
1413 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1414 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1415 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1416 (native_encode_real): Likewise.
1417 (native_interpret_real): Likewise.
1418 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1419 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1422 Alan Hayward <alan.hayward@arm.com>
1423 David Sherwood <david.sherwood@arm.com>
1425 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1426 <scalar_float_mode>. Simplify.
1427 (gen_extend_conv_libfunc): Likewise.
1429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1430 Alan Hayward <alan.hayward@arm.com>
1431 David Sherwood <david.sherwood@arm.com>
1433 * coretypes.h (scalar_float_mode): New type.
1434 * machmode.h (mode_traits::from_int): Use machine_mode if
1435 USE_ENUM_MODES is defined.
1436 (is_a): New function.
1438 (dyn_cast): Likewise.
1439 (scalar_float_mode): New class.
1440 (scalar_float_mode::includes_p): New function.
1441 (is_float_mode): Likewise.
1442 * gdbhooks.py (MachineModePrinter): New class.
1443 (build_pretty_printer): Use it for scalar_float_mode.
1444 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1445 (format_helper::format_helper): Turn into a template.
1446 * genmodes.c (get_mode_class): New function.
1447 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1448 or machine_mode if none.
1449 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1450 as_a <scalar_float_mode>.
1451 * dwarf2out.c (mem_loc_descriptor): Likewise.
1452 (insert_float): Likewise.
1453 (add_const_value_attribute): Likewise.
1454 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1455 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1456 (expand_unop): Update accordingly.
1457 (expand_abs_nojump): Likewise.
1458 (expand_copysign_absneg): Take a scalar_float_mode.
1459 (expand_copysign_bit): Likewise.
1460 (expand_copysign): Update accordingly.
1462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * coretypes.h (opt_mode): New class.
1467 * machmode.h (opt_mode): Likewise.
1468 (opt_mode::else_void): New function.
1469 (opt_mode::require): Likewise.
1470 (opt_mode::exists): Likewise.
1471 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1472 (GET_MODE_2XWIDER_MODE): Likewise.
1473 (mode_iterator::get_wider): Update accordingly.
1474 (mode_iterator::get_2xwider): Likewise.
1475 (mode_iterator::get_known_wider): Likewise, turning into a template.
1476 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1477 forcing a wider mode to exist.
1478 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1479 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1480 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1481 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1482 * lower-subreg.c (init_lower_subreg): Likewise.
1483 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1484 on the final iteration.
1485 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1486 a wider mode exists before asking for a move pattern.
1487 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1488 forcing a wider mode to exist.
1489 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1490 returning false if no such mode exists.
1491 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1492 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1493 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1494 Avoid checking for a MODE_INT if we already know the mode is not a
1496 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1497 forcing a wider mode to exist.
1498 (expmed_mult_highpart_optab): Likewise.
1499 (expmed_mult_highpart): Likewise.
1500 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1502 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1503 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1504 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1505 * internal-fn.c (expand_mul_overflow): Update use of
1506 GET_MODE_2XWIDER_MODE.
1507 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1508 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1509 GET_MODE_WIDER_MODE.
1510 (convert_plusminus_to_widen): Likewise.
1511 * tree-switch-conversion.c (array_value_type): Likewise.
1512 * var-tracking.c (emit_note_insn_var_location): Likewise.
1513 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1514 Return false inside rather than outside the loop if no wider mode
1516 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1517 and GET_MODE_2XWIDER_MODE
1518 (can_compare_p): Use else_void.
1519 * gdbhooks.py (OptMachineModePrinter): New class.
1520 (build_pretty_printer): Use it for opt_mode.
1522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1523 Alan Hayward <alan.hayward@arm.com>
1524 David Sherwood <david.sherwood@arm.com>
1526 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1527 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
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 * machmode.h (mode_traits): New structure.
1534 (get_narrowest_mode): New function.
1535 (mode_iterator::start): Likewise.
1536 (mode_iterator::iterate_p): Likewise.
1537 (mode_iterator::get_wider): Likewise.
1538 (mode_iterator::get_known_wider): Likewise.
1539 (mode_iterator::get_2xwider): Likewise.
1540 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1541 (FOR_EACH_MODE): Likewise.
1542 (FOR_EACH_MODE_FROM): Likewise.
1543 (FOR_EACH_MODE_UNTIL): Likewise.
1544 (FOR_EACH_WIDER_MODE): Likewise.
1545 (FOR_EACH_2XWIDER_MODE): Likewise.
1546 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1547 * combine.c (simplify_comparison): Likewise
1548 * config/i386/i386.c (type_natural_mode): Likewise.
1549 * cse.c (cse_insn): Likewise.
1550 * dse.c (find_shift_sequence): Likewise.
1551 * emit-rtl.c (init_derived_machine_modes): Likewise.
1552 (init_emit_once): Likewise.
1553 * explow.c (hard_function_value): Likewise.
1554 * expmed.c (extract_fixed_bit_field_1): Likewise.
1555 (extract_bit_field_1): Likewise.
1556 (expand_divmod): Likewise.
1557 (emit_store_flag_1): Likewise.
1558 * expr.c (init_expr_target): Likewise.
1559 (convert_move): Likewise.
1560 (alignment_for_piecewise_move): Likewise.
1561 (widest_int_mode_for_size): Likewise.
1562 (emit_block_move_via_movmem): Likewise.
1563 (copy_blkmode_to_reg): Likewise.
1564 (set_storage_via_setmem): Likewise.
1565 (compress_float_constant): Likewise.
1566 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1567 * optabs-query.c (get_best_extraction_insn): Likewise.
1568 * optabs.c (expand_binop): Likewise.
1569 (expand_twoval_unop): Likewise.
1570 (expand_twoval_binop): Likewise.
1571 (widen_leading): Likewise.
1572 (widen_bswap): Likewise.
1573 (expand_parity): Likewise.
1574 (expand_unop): Likewise.
1575 (prepare_cmp_insn): Likewise.
1576 (prepare_float_lib_cmp): Likewise.
1577 (expand_float): Likewise.
1578 (expand_fix): Likewise.
1579 (expand_sfix_optab): Likewise.
1580 * postreload.c (move2add_use_add2_insn): Likewise.
1581 * reg-stack.c (reg_to_stack): Likewise.
1582 * reginfo.c (choose_hard_reg_mode): Likewise.
1583 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1584 * stor-layout.c (mode_for_size): Likewise.
1585 (smallest_mode_for_size): Likewise.
1586 (mode_for_vector): Likewise.
1587 (finish_bitfield_representative): Likewise.
1588 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1589 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1590 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1591 * var-tracking.c (prepare_call_arguments): Likewise.
1593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1594 Alan Hayward <alan.hayward@arm.com>
1595 David Sherwood <david.sherwood@arm.com>
1597 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1598 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1599 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1600 * machmode.h (mode_size): Move earlier in file.
1601 (mode_precision): Likewise.
1602 (mode_inner): Likewise.
1603 (mode_nunits): Likewise.
1604 (mode_unit_size): Likewise.
1605 (unit_unit_precision): Likewise.
1606 (mode_wider): Likewise.
1607 (mode_2xwider): Likewise.
1608 (machine_mode): New class.
1609 (mode_to_bytes): New function.
1610 (mode_to_bits): Likewise.
1611 (mode_to_precision): Likewise.
1612 (mode_to_inner): Likewise.
1613 (mode_to_unit_size): Likewise.
1614 (mode_to_unit_precision): Likewise.
1615 (mode_to_nunits): Likewise.
1616 (GET_MODE_SIZE): Use mode_to_bytes.
1617 (GET_MODE_BITSIZE): Use mode_to_bits.
1618 (GET_MODE_PRECISION): Use mode_to_precision.
1619 (GET_MODE_INNER): Use mode_to_inner.
1620 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1621 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1622 (GET_MODE_NUNITS): Use mode_to_nunits.
1623 * system.h (ALWAYS_INLINE): New macro.
1624 * config/powerpcspe/powerpcspe-c.c
1625 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1626 int for arg1_mode and arg2_mode.
1628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1629 Alan Hayward <alan.hayward@arm.com>
1630 David Sherwood <david.sherwood@arm.com>
1632 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1633 Prefix mode names with E_ in case statements.
1634 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1635 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1636 (aarch64_split_simd_move): Likewise.
1637 (aarch64_gen_storewb_pair): Likewise.
1638 (aarch64_gen_loadwb_pair): Likewise.
1639 (aarch64_gen_store_pair): Likewise.
1640 (aarch64_gen_load_pair): Likewise.
1641 (aarch64_get_condition_code_1): Likewise.
1642 (aarch64_constant_pool_reload_icode): Likewise.
1643 (get_rsqrte_type): Likewise.
1644 (get_rsqrts_type): Likewise.
1645 (get_recpe_type): Likewise.
1646 (get_recps_type): Likewise.
1647 (aarch64_gimplify_va_arg_expr): Likewise.
1648 (aarch64_simd_container_mode): Likewise.
1649 (aarch64_emit_load_exclusive): Likewise.
1650 (aarch64_emit_store_exclusive): Likewise.
1651 (aarch64_expand_compare_and_swap): Likewise.
1652 (aarch64_gen_atomic_cas): Likewise.
1653 (aarch64_emit_bic): Likewise.
1654 (aarch64_emit_atomic_swap): Likewise.
1655 (aarch64_emit_atomic_load_op): Likewise.
1656 (aarch64_evpc_trn): Likewise.
1657 (aarch64_evpc_uzp): Likewise.
1658 (aarch64_evpc_zip): Likewise.
1659 (aarch64_evpc_ext): Likewise.
1660 (aarch64_evpc_rev): Likewise.
1661 (aarch64_evpc_dup): Likewise.
1662 (aarch64_gen_ccmp_first): Likewise.
1663 (aarch64_gen_ccmp_next): Likewise.
1664 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1665 (alpha_emit_xfloating_libcall): Likewise.
1666 (emit_insxl): Likewise.
1667 (alpha_arg_type): Likewise.
1668 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1669 (arc_preferred_simd_mode): Likewise.
1670 (arc_secondary_reload): Likewise.
1671 (get_arc_condition_code): Likewise.
1672 (arc_print_operand): Likewise.
1673 (arc_legitimate_constant_p): Likewise.
1674 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1675 * config/arc/arc.md (casesi_load): Likewise.
1676 (casesi_compact_jump): Likewise.
1677 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1678 (cc_use_register): Likewise.
1679 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1680 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1681 (arm_init_iwmmxt_builtins): Likewise.
1682 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1683 (neon_expand_vector_init): Likewise.
1684 (arm_attr_length_move_neon): Likewise.
1685 (maybe_get_arm_condition_code): Likewise.
1686 (arm_emit_vector_const): Likewise.
1687 (arm_preferred_simd_mode): Likewise.
1688 (arm_output_iwmmxt_tinsr): Likewise.
1689 (thumb1_output_casesi): Likewise.
1690 (thumb2_output_casesi): Likewise.
1691 (arm_emit_load_exclusive): Likewise.
1692 (arm_emit_store_exclusive): Likewise.
1693 (arm_expand_compare_and_swap): Likewise.
1694 (arm_evpc_neon_vuzp): Likewise.
1695 (arm_evpc_neon_vzip): Likewise.
1696 (arm_evpc_neon_vrev): Likewise.
1697 (arm_evpc_neon_vtrn): Likewise.
1698 (arm_evpc_neon_vext): Likewise.
1699 (arm_validize_comparison): Likewise.
1700 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1701 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1702 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1703 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1704 (c6x_preferred_simd_mode): Likewise.
1705 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1706 (epiphany_rtx_costs): Likewise.
1707 * config/epiphany/predicates.md (proper_comparison_operator):
1709 * config/frv/frv.c (condexec_memory_operand): Likewise.
1710 (frv_emit_move): Likewise.
1711 (output_move_single): Likewise.
1712 (output_condmove_single): Likewise.
1713 (frv_hard_regno_mode_ok): Likewise.
1714 (frv_matching_accg_mode): Likewise.
1715 * config/h8300/h8300.c (split_adds_subs): Likewise.
1716 (h8300_rtx_costs): Likewise.
1717 (h8300_print_operand): Likewise.
1718 (compute_mov_length): Likewise.
1719 (output_logical_op): Likewise.
1720 (compute_logical_op_length): Likewise.
1721 (compute_logical_op_cc): Likewise.
1722 (h8300_shift_needs_scratch_p): Likewise.
1723 (output_a_shift): Likewise.
1724 (compute_a_shift_length): Likewise.
1725 (compute_a_shift_cc): Likewise.
1726 (expand_a_rotate): Likewise.
1727 (output_a_rotate): Likewise.
1728 * config/i386/i386.c (classify_argument): Likewise.
1729 (function_arg_advance_32): Likewise.
1730 (function_arg_32): Likewise.
1731 (function_arg_64): Likewise.
1732 (function_value_64): Likewise.
1733 (ix86_gimplify_va_arg): Likewise.
1734 (ix86_legitimate_constant_p): Likewise.
1735 (put_condition_code): Likewise.
1736 (split_double_mode): Likewise.
1737 (ix86_avx256_split_vector_move_misalign): Likewise.
1738 (ix86_expand_vector_logical_operator): Likewise.
1739 (ix86_split_idivmod): Likewise.
1740 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1741 (ix86_build_const_vector): Likewise.
1742 (ix86_build_signbit_mask): Likewise.
1743 (ix86_match_ccmode): Likewise.
1744 (ix86_cc_modes_compatible): Likewise.
1745 (ix86_expand_branch): Likewise.
1746 (ix86_expand_sse_cmp): Likewise.
1747 (ix86_expand_sse_movcc): Likewise.
1748 (ix86_expand_int_sse_cmp): Likewise.
1749 (ix86_expand_vec_perm_vpermi2): Likewise.
1750 (ix86_expand_vec_perm): Likewise.
1751 (ix86_expand_sse_unpack): Likewise.
1752 (ix86_expand_int_addcc): Likewise.
1753 (ix86_split_to_parts): Likewise.
1754 (ix86_vectorize_builtin_gather): Likewise.
1755 (ix86_vectorize_builtin_scatter): Likewise.
1756 (avx_vpermilp_parallel): Likewise.
1757 (inline_memory_move_cost): Likewise.
1758 (ix86_tieable_integer_mode_p): Likewise.
1759 (x86_maybe_negate_const_int): Likewise.
1760 (ix86_expand_vector_init_duplicate): Likewise.
1761 (ix86_expand_vector_init_one_nonzero): Likewise.
1762 (ix86_expand_vector_init_one_var): Likewise.
1763 (ix86_expand_vector_init_concat): Likewise.
1764 (ix86_expand_vector_init_interleave): Likewise.
1765 (ix86_expand_vector_init_general): Likewise.
1766 (ix86_expand_vector_set): Likewise.
1767 (ix86_expand_vector_extract): Likewise.
1768 (emit_reduc_half): Likewise.
1769 (ix86_emit_i387_round): Likewise.
1770 (ix86_mangle_type): Likewise.
1771 (ix86_expand_round_sse4): Likewise.
1772 (expand_vec_perm_blend): Likewise.
1773 (canonicalize_vector_int_perm): Likewise.
1774 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1775 (expand_vec_perm_1): Likewise.
1776 (expand_vec_perm_interleave3): Likewise.
1777 (expand_vec_perm_even_odd_pack): Likewise.
1778 (expand_vec_perm_even_odd_1): Likewise.
1779 (expand_vec_perm_broadcast_1): Likewise.
1780 (ix86_vectorize_vec_perm_const_ok): Likewise.
1781 (ix86_expand_vecop_qihi): Likewise.
1782 (ix86_expand_mul_widen_hilo): Likewise.
1783 (ix86_expand_sse2_abs): Likewise.
1784 (ix86_expand_pextr): Likewise.
1785 (ix86_expand_pinsr): Likewise.
1786 (ix86_preferred_simd_mode): Likewise.
1787 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1788 * config/i386/sse.md (*andnot<mode>3): Likewise.
1789 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1790 (*<code><mode>3): Likewise.
1791 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1792 (ia64_expand_atomic_op): Likewise.
1793 (ia64_arg_type): Likewise.
1794 (ia64_mode_to_int): Likewise.
1795 (ia64_scalar_mode_supported_p): Likewise.
1796 (ia64_vector_mode_supported_p): Likewise.
1797 (expand_vec_perm_broadcast): Likewise.
1798 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1799 (iq2000_function_arg_advance): Likewise.
1800 (iq2000_function_arg): Likewise.
1801 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1802 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1803 (m68k_libcall_value): Likewise.
1804 (m68k_function_value): Likewise.
1805 (sched_attr_op_type): Likewise.
1806 * config/mcore/mcore.c (mcore_output_move): Likewise.
1807 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1809 (microblaze_function_arg): Likewise.
1810 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1811 (mips_print_operand): Likewise.
1812 (mips_mode_ok_for_mov_fmt_p): Likewise.
1813 (mips_vector_mode_supported_p): Likewise.
1814 (mips_preferred_simd_mode): Likewise.
1815 (mips_expand_vpc_loongson_even_odd): Likewise.
1816 (mips_expand_vec_unpack): Likewise.
1817 (mips_expand_vi_broadcast): Likewise.
1818 (mips_expand_vector_init): Likewise.
1819 (mips_expand_vec_reduc): Likewise.
1820 (mips_expand_msa_cmp): Likewise.
1821 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1822 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1823 (cc_flags_for_mode): Likewise.
1824 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1825 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1826 (nds32_output_casesi_pc_relative): Likewise.
1827 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1828 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1829 (nvptx_gen_unpack): Likewise.
1830 (nvptx_gen_pack): Likewise.
1831 (nvptx_gen_shuffle): Likewise.
1832 (nvptx_gen_wcast): Likewise.
1833 (nvptx_preferred_simd_mode): Likewise.
1834 * config/pa/pa.c (pa_secondary_reload): Likewise.
1835 * config/pa/predicates.md (base14_operand): Likewise.
1836 * config/powerpcspe/powerpcspe-c.c
1837 (altivec_resolve_overloaded_builtin): Likewise.
1838 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1840 (rs6000_preferred_simd_mode): Likewise.
1841 (output_vec_const_move): Likewise.
1842 (rs6000_expand_vector_extract): Likewise.
1843 (rs6000_split_vec_extract_var): Likewise.
1844 (reg_offset_addressing_ok_p): Likewise.
1845 (rs6000_legitimate_offset_address_p): Likewise.
1846 (rs6000_legitimize_address): Likewise.
1847 (rs6000_emit_set_const): Likewise.
1848 (rs6000_const_vec): Likewise.
1849 (rs6000_emit_move): Likewise.
1850 (spe_build_register_parallel): Likewise.
1851 (rs6000_darwin64_record_arg_recurse): Likewise.
1852 (swap_selector_for_mode): Likewise.
1853 (spe_init_builtins): Likewise.
1854 (paired_init_builtins): Likewise.
1855 (altivec_init_builtins): Likewise.
1856 (do_load_for_compare): Likewise.
1857 (rs6000_generate_compare): Likewise.
1858 (rs6000_expand_float128_convert): Likewise.
1859 (emit_load_locked): Likewise.
1860 (emit_store_conditional): Likewise.
1861 (rs6000_output_function_epilogue): Likewise.
1862 (rs6000_handle_altivec_attribute): Likewise.
1863 (rs6000_function_value): Likewise.
1864 (emit_fusion_gpr_load): Likewise.
1865 (emit_fusion_p9_load): Likewise.
1866 (emit_fusion_p9_store): Likewise.
1867 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1868 (fusion_gpr_mem_load): Likewise.
1869 (fusion_addis_mem_combo_load): Likewise.
1870 (fusion_addis_mem_combo_store): Likewise.
1871 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1872 (fusion_gpr_mem_load): Likewise.
1873 (fusion_addis_mem_combo_load): Likewise.
1874 (fusion_addis_mem_combo_store): Likewise.
1875 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1877 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1878 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1879 (rs6000_preferred_simd_mode): Likewise.
1880 (output_vec_const_move): Likewise.
1881 (rs6000_expand_vector_extract): Likewise.
1882 (rs6000_split_vec_extract_var): Likewise.
1883 (reg_offset_addressing_ok_p): Likewise.
1884 (rs6000_legitimate_offset_address_p): Likewise.
1885 (rs6000_legitimize_address): Likewise.
1886 (rs6000_emit_set_const): Likewise.
1887 (rs6000_const_vec): Likewise.
1888 (rs6000_emit_move): Likewise.
1889 (rs6000_darwin64_record_arg_recurse): Likewise.
1890 (swap_selector_for_mode): Likewise.
1891 (paired_init_builtins): Likewise.
1892 (altivec_init_builtins): Likewise.
1893 (rs6000_expand_float128_convert): Likewise.
1894 (emit_load_locked): Likewise.
1895 (emit_store_conditional): Likewise.
1896 (rs6000_output_function_epilogue): Likewise.
1897 (rs6000_handle_altivec_attribute): Likewise.
1898 (rs6000_function_value): Likewise.
1899 (emit_fusion_gpr_load): Likewise.
1900 (emit_fusion_p9_load): Likewise.
1901 (emit_fusion_p9_store): Likewise.
1902 * config/rx/rx.c (rx_gen_move_template): Likewise.
1903 (flags_from_mode): Likewise.
1904 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1905 (s390_slb_comparison): Likewise.
1906 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1907 (s390_vector_mode_supported_p): Likewise.
1908 (s390_cc_modes_compatible): Likewise.
1909 (s390_match_ccmode_set): Likewise.
1910 (s390_canonicalize_comparison): Likewise.
1911 (s390_emit_compare_and_swap): Likewise.
1912 (s390_branch_condition_mask): Likewise.
1913 (s390_rtx_costs): Likewise.
1914 (s390_secondary_reload): Likewise.
1915 (__SECONDARY_RELOAD_CASE): Likewise.
1916 (s390_expand_cs): Likewise.
1917 (s390_preferred_simd_mode): Likewise.
1918 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1919 * config/sh/sh.c (sh_print_operand): Likewise.
1920 (dump_table): Likewise.
1921 (sh_secondary_reload): Likewise.
1922 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1923 * config/sh/sh.md (casesi_worker_1): Likewise.
1924 (casesi_worker_2): Likewise.
1925 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1926 (fcc_comparison_operator): Likewise.
1927 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1928 (emit_soft_tfmode_cvt): Likewise.
1929 (sparc_preferred_simd_mode): Likewise.
1930 (output_cbranch): Likewise.
1931 (sparc_print_operand): Likewise.
1932 (sparc_expand_vec_perm_bmask): Likewise.
1933 (vector_init_bshuffle): Likewise.
1934 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1935 (spu_vector_mode_supported_p): Likewise.
1936 (spu_expand_insv): Likewise.
1937 (spu_emit_branch_or_set): Likewise.
1938 (spu_handle_vector_attribute): Likewise.
1939 (spu_builtin_splats): Likewise.
1940 (spu_builtin_extract): Likewise.
1941 (spu_builtin_promote): Likewise.
1942 (spu_expand_sign_extend): Likewise.
1943 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1944 (tilegx_simd_int): Likewise.
1945 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1946 (tilepro_simd_int): Likewise.
1947 * config/v850/v850.c (const_double_split): Likewise.
1948 (v850_print_operand): Likewise.
1949 (ep_memory_offset): Likewise.
1950 * config/vax/vax.c (vax_rtx_costs): Likewise.
1951 (vax_output_int_move): Likewise.
1952 (vax_output_int_add): Likewise.
1953 (vax_output_int_subtract): Likewise.
1954 * config/visium/predicates.md (visium_branch_operator): Likewise.
1955 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1956 (visium_print_operand_address): Likewise.
1957 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1958 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1959 (xtensa_expand_conditional_branch): Likewise.
1960 (xtensa_copy_incoming_a7): Likewise.
1961 (xtensa_output_literal): Likewise.
1962 * dfp.c (decimal_real_maxval): Likewise.
1963 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1966 Alan Hayward <alan.hayward@arm.com>
1967 David Sherwood <david.sherwood@arm.com>
1969 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1970 (mode_nunits_inline): Likewise.
1971 (mode_inner_inline): Likewise.
1972 (mode_unit_size_inline): Likewise.
1973 (mode_unit_precision_inline): Likewise.
1974 (emit_insn_modes_h): Likewise. Also emit a #define of the
1976 (emit_mode_wider): Add an E_ prefix to mode names.
1977 (emit_mode_complex): Likewise.
1978 (emit_mode_inner): Likewise.
1979 (emit_mode_adjustments): Likewise.
1980 (emit_mode_int_n): Likewise.
1981 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1982 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1983 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1984 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1985 (CRC32_BUILTIN, ENTRY): Likewise.
1986 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1987 (aarch64_pop_regs): Likewise.
1988 (aarch64_process_components): Likewise.
1989 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1990 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1991 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1992 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1993 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1994 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1996 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1997 (pcmpistr): Likewise.
1998 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1999 * config/mmix/mmix.c (mmix_output_condition): Likewise.
2000 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2002 * config/rl78/rl78.c (mduc_regs): Likewise.
2003 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2004 (htm_expand_builtin): Likewise.
2005 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
2006 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
2007 * config/xtensa/xtensa.c (print_operand): Likewise.
2008 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
2009 (NUM_MODE_VECTOR_INT): Likewise.
2010 * genoutput.c (null_operand): Likewise.
2011 (output_operand_data): Likewise.
2012 * genrecog.c (print_parameter_value): Likewise.
2013 * lra.c (debug_operand_data): Likewise.
2015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2017 * dwarf2out.c (scompare_loc_descriptor_wide)
2018 (scompare_loc_descriptor_narrow): New functions, split out from...
2019 (scompare_loc_descriptor): ...here.
2020 * expmed.c (emit_store_flag_int): New function, split out from...
2021 (emit_store_flag): ...here.
2023 2017-08-30 Richard Biener <rguenther@suse.de>
2025 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
2026 (dwarf2out_early_finish): Move setting of AT_pubnames from
2027 early debug output to early finish.
2029 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2031 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
2032 and -mdata-region to the assembler.
2034 2017-08-30 Richard Biener <rguenther@suse.de>
2036 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
2038 (gen_subprogram_die): Add DW_AT_object_pointer only early.
2039 (dwarf2out_early_global_decl): Only generate a DIE for the
2040 abstract origin if it doesn't already exist or is a declaration DIE.
2041 (resolve_addr): Do not add the linkage name twice when
2042 generating a stub DIE for the DW_TAG_GNU_call_site target.
2044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2046 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2047 Use machine_mode rather than int for arg1_mode.
2049 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
2052 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
2053 that the second argument of the built-in functions to unpack
2054 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
2055 switch statement instead a lot of if statements.
2056 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
2057 Allow 64-bit values to be in Altivec registers as well as
2058 traditional floating point registers.
2059 (pack<mode>, FMOVE128_VSX iterator): Likewise.
2061 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
2063 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2064 MAX_REGS_PER_ADDRESS == 1.
2066 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2068 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2069 * config/i386/i386.c (ix86_option_override_internal): Simplify
2070 setting of opts->x_flag_entry.
2072 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2073 Jakub Jelinek <jakub@redhat.com>
2074 Richard Biener <rguenther@suse.de>
2076 PR tree-optimization/81503
2077 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2078 folded constant fits in the target type; reorder tests for clarity.
2080 2017-08-29 Martin Liska <mliska@suse.cz>
2082 * passes.def: Include pass_lower_switch.
2083 * stmt.c (dump_case_nodes): Remove and move to
2084 tree-switch-conversion.
2085 (case_values_threshold): Likewise.
2086 (expand_switch_as_decision_tree_p): Likewise.
2087 (emit_case_decision_tree): Likewise.
2088 (expand_case): Likewise.
2089 (balance_case_nodes): Likewise.
2090 (node_has_low_bound): Likewise.
2091 (node_has_high_bound): Likewise.
2092 (node_is_bounded): Likewise.
2093 (emit_case_nodes): Likewise.
2094 (struct simple_case_node): New struct.
2095 (add_case_node): Remove.
2096 (emit_case_dispatch_table): Use vector instead of case_list.
2097 (reset_out_edges_aux): Remove.
2098 (compute_cases_per_edge): Likewise.
2099 (expand_case): Build list of simple_case_node.
2100 (expand_sjlj_dispatch_table): Use it.
2101 * tree-switch-conversion.c (struct case_node): Moved from
2102 stmt.c and adjusted.
2103 (emit_case_nodes): Likewise.
2104 (node_has_low_bound): Likewise.
2105 (node_has_high_bound): Likewise.
2106 (node_is_bounded): Likewise.
2107 (case_values_threshold): Likewise.
2108 (reset_out_edges_aux): Likewise.
2109 (compute_cases_per_edge): Likewise.
2110 (add_case_node): Likewise.
2111 (dump_case_nodes): Likewise.
2112 (balance_case_nodes): Likewise.
2113 (expand_switch_as_decision_tree_p): Likewise.
2114 (emit_jump): Likewise.
2115 (emit_case_decision_tree): Likewise.
2116 (try_switch_expansion): Likewise.
2117 (do_jump_if_equal): Likewise.
2118 (emit_cmp_and_jump_insns): Likewise.
2119 (fix_phi_operands_for_edge): New function.
2120 (record_phi_operand_mapping): Likewise.
2121 (class pass_lower_switch): New pass.
2122 (pass_lower_switch::execute): New function.
2123 (make_pass_lower_switch): Likewise.
2124 (conditional_probability):
2125 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2126 * tree-pass.h: Add make_pass_lower_switch.
2128 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2131 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2134 2017-08-29 Richard Biener <rguenther@suse.de>
2136 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2137 we do not add a DW_AT_inline attribute twice.
2138 (gen_subprogram_die): Remove code setting DW_AT_inline on
2139 DECL_ABSTRACT_P nodes.
2141 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2143 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2144 calls to internal functions.
2145 (gimplify_modify_expr): Likewise.
2146 * tree-call-cdce.c (use_internal_fn): Likewise.
2147 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2148 (convert_to_divmod): Set the nothrow flag.
2149 * tree-if-conv.c (predicate_mem_writes): Likewise.
2150 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2151 (vectorizable_call): Likewise.
2152 (vectorizable_store): Likewise.
2153 (vectorizable_load): Likewise.
2154 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2155 (vect_recog_mask_conversion_pattern): Likewise.
2157 2017-08-29 Martin Liska <mliska@suse.cz>
2160 * gcc.c (driver_handle_option): Add new argument.
2161 * opts-common.c (handle_option): Pass
2162 target_option_override_hook.
2163 * opts-global.c (lang_handle_option): Add new option.
2164 (set_default_handlers): Add new argument.
2165 (decode_options): Likewise.
2166 * opts.c (target_handle_option): Likewise.
2167 (common_handle_option): Call target_option_override_hook.
2168 * opts.h (struct cl_option_handler_func): Add hook for
2169 target option override.
2170 (struct cl_option_handlers): Likewise.
2171 (set_default_handlers): Add new argument.
2172 (decode_options): Likewise.
2173 (common_handle_option): Likewise.
2174 (target_handle_option): Likewise.
2175 * toplev.c (toplev::main): Pass targetm.target_option.override
2178 2017-08-29 Richard Biener <rguenther@suse.de>
2179 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2181 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2182 life to the active subtree.
2184 2017-08-28 Jeff Law <law@redhat.com>
2186 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2187 derive_equivalences.
2188 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2190 (edge_info::derive_equivalences): New private member function
2192 * tree-ssa-dom.c (class edge_info): Changed from a struct
2193 to a class. Add ctor/dtor, methods and data members.
2194 (edge_info::edge_info): Renamed from allocate_edge_info.
2195 Initialize additional members.
2196 (edge_info::~edge_info): New.
2197 (free_dom_edge_info): Delete the edge info.
2198 (record_edge_info): Use new class & associated member functions.
2199 Tighten forms for testing for edge equivalences.
2200 (record_temporary_equivalences): Iterate over the simple
2201 equivalences rather than assuming there's only one per edge.
2202 (cprop_into_successor_phis): Iterate over the simple
2203 equivalences rather than assuming there's only one per edge.
2204 (optimize_stmt): Use operand_equal_p rather than pointer
2205 equality for mini-DSE code.
2207 2017-08-28 Nathan Sidwell <nathan@acm.org>
2209 * gcc.c (execute): Fold SIGPIPE handling into switch
2210 statement. Adjust internal error message.
2212 2017-08-28 Richard Biener <rguenther@suse.de>
2215 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2216 Do nothing for removed DIEs.
2218 2017-08-28 Richard Biener <rguenther@suse.de>
2220 PR tree-optimization/81977
2221 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2224 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2227 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2228 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2229 using targetm.gen_mem_thread_fence.
2231 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2234 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2235 predicate to register_operand. Reorder operands.
2236 (*btr<mode>): Ditto.
2237 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2238 (*btr<mode>_mask): Ditto.
2240 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2242 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2243 * config/rs6000/xmmintrin.h: New file.
2244 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2246 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2249 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2250 parameter and_insn and return it.
2251 (recombine_lvx_pattern): Insert a copy to ensure availability of
2252 the base register of the copied masking operation at the point of
2253 the instruction replacement.
2254 (recombine_stvx_pattern): Likewise.
2256 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2258 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2259 undocumented switches.
2260 (-mpower9-dform-vector): Likewise.
2261 (-mpower9-dform): Likewise.
2262 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2263 comments to delete references to -mpower9-dform* switches.
2264 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2265 Delete reference to -mpower9-dform* switches, test for
2266 -mpower9-vector instead.
2267 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2268 (OTHER_P9_VECTOR_MASKS): Likewise.
2269 (POWERPC_MASKS): Likewise.
2270 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2271 tests against -mpower9-dform* to -mpower9-vector. Delete code
2272 that checked for -mpower9-dform* consistancy with other options.
2273 Add test for -mpower9-misc to enable other power9 switches.
2274 (rs6000_init_hard_regno_mode_ok): Likewise.
2275 (rs6000_option_override_internal): Likewise.
2276 (rs6000_emit_prologue): Likewise.
2277 (rs6000_emit_epilogue): Likewise.
2278 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2279 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2280 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2282 (emit_fusion_p9_store): Likewise.
2283 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2284 resetting these macros if the assembler does not support ISA 3.0
2286 (TARGET_P9_DFORM_VECTOR): Likewise.
2287 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2288 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2290 2017-08-25 Alan Modra <amodra@gmail.com>
2292 PR rtl-optimization/81747
2293 * cse.c (cse_extended_basic_block): Don't attempt to record
2294 equivalences for degenerate conditional branches that branch
2295 to their fall-through.
2297 2017-08-24 Martin Sebor <msebor@redhat.com>
2300 * gimple-fold.c (size_must_be_zero_p): New function.
2301 (gimple_fold_builtin_memory_op): Call it.
2303 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2305 * config/rs6000/mm_malloc.h: New file.
2307 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2309 PR tree-optimization/81913
2310 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2311 analysis when either IVs in condition can wrap.
2313 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2315 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2316 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2318 2017-08-24 Richard Biener <rguenther@suse.de>
2321 * targhooks.c (default_target_can_inline_p): Properly
2322 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2323 is present and always compare.
2324 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2325 infer -mfpmath=sse from TARGET_SSE_P.
2326 (ix86_can_inline_p): Properly use target_option_default_node when
2327 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2329 2017-08-24 Richard Biener <rguenther@suse.de>
2332 * dwarf2out.c (output_die): Handle flag_generate_offload like
2334 (output_comp_unit): Likewise.
2335 (gen_array_type_die): Likewise.
2336 (dwarf2out_early_finish): Likewise.
2337 (note_variable_value_in_expr): Likewise.
2338 (dwarf2out_finish): Likewise. Adjust assert.
2339 * cgraphunit.c (symbol_table::compile): Move setting of
2340 flag_generate_offload earlier ...
2341 (symbol_table::finalize_compilation_unit): ... here, before
2342 early debug finalization.
2344 2017-08-24 Richard Biener <rguenther@suse.de>
2346 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2347 and ipa-fnsummary.h.
2348 (ix86_can_inline_p): When ix86_fpmath flags do not match
2349 check whether the callee uses FP math at all.
2351 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2354 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2355 instead of TYPE_PRECISION.
2357 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2358 Alan Hayward <alan.hayward@arm.com>
2359 David Sherwood <david.sherwood@arm.com>
2361 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2363 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2365 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2367 (ix86_frame::stack_realign_allocate): New field.
2368 (struct machine_frame_state): Modify comments.
2369 (machine_frame_state::sp_realigned_fp_end): New field.
2370 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2372 (sp_valid_at): Add assertion to assure no attempt to access invalid
2373 offset of a realigned stack.
2374 (fp_valid_at): Likewise.
2375 (choose_baseaddr): Modify comments.
2376 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2377 ix86_expand_prologue.
2378 (ix86_expand_prologue): Modify stack realignment and allocation.
2379 (ix86_expand_epilogue): Modify comments.
2380 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2381 avx2_runtime, avx512f, and avx512f_runtime.
2383 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2385 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2386 (mstackrealign): Do not init to -1.
2387 * config/i386/i386.c (ix86_option_override_internal):
2388 Check opts_set, not opts when setting default value of
2389 opts->x_ix86_force_align_arg_pointer.
2391 2017-08-23 Richard Biener <rguenther@suse.de>
2393 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2394 lang_hooks.decl_printable_name.
2395 * print-rtl-function.c (print_rtx_function): Likewise.
2396 * tree-pretty-print.c (dump_function_header): Likewise.
2398 2017-08-23 Richard Biener <rguenther@suse.de>
2401 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2402 -g0 at compile-time.
2404 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2407 * doc/sourcebuild.texi (Other hardware attributes):
2410 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2413 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2414 Fix single-use check.
2416 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2418 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2420 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2422 * doc/install.texi: Modify to add more details on running selected
2425 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2427 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2428 is combined with -mabi=ms.
2429 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2432 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2434 PR tree-optimization/81488
2435 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2436 and cached_basis fields.
2437 (MAX_SPREAD): New constant.
2438 (alloc_cand_and_find_basis): Initialize new fields.
2439 (clear_visited): New function.
2440 (create_phi_basis_1): Rename from create_phi_basis, set visited
2441 and cached_basis fields.
2442 (create_phi_basis): New wrapper function.
2443 (phi_add_costs_1): Rename from phi_add_costs, add spread
2444 parameter, set visited field, short-circuit when limits reached.
2445 (phi_add_costs): New wrapper function.
2446 (record_phi_increments_1): Rename from record_phi_increments, set
2448 (record_phi_increments): New wrapper function.
2449 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2450 (phi_incr_cost): New wrapper function.
2451 (all_phi_incrs_profitable_1): Rename from
2452 all_phi_incrs_profitable, set visited field.
2453 (all_phi_incrs_profitable): New wrapper function.
2455 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2456 Alan Hayward <alan.hayward@arm.com>
2457 David Sherwood <david.sherwood@arm.com>
2459 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2460 that takes the outer and inner modes.
2461 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2462 comparison as the canonical test for a paradoxical subreg.
2463 * combine.c (simplify_set): Use paradoxical_subreg_p.
2464 (make_extraction): Likewise.
2465 (force_to_mode): Likewise.
2466 (rtx_equal_for_field_assignment_p): Likewise.
2467 (gen_lowpart_for_combine): Likewise.
2468 (simplify_comparison): Likewise.
2469 * cse.c (equiv_constant): Likewise.
2470 * expmed.c (store_bit_field_1): Likewise.
2471 * final.c (alter_subreg): Likewise.
2472 * fwprop.c (propagate_rtx): Likewise.
2473 (forward_propagate_subreg): Likewise.
2474 * ira-conflicts.c (ira_build_conflicts): Likewise.
2475 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2476 * lra-constraints.c (curr_insn_transform): Likewise.
2477 (split_reg): Likewise.
2478 * lra-eliminations.c (move_plus_up): Likewise.
2479 (lra_eliminate_regs_1): Likewise.
2480 * recog.c (general_operand): Likewise.
2481 * ree.c (combine_reaching_defs): Likewise.
2482 * reload.c (push_reload): Likewise.
2483 (find_reloads): Likewise.
2484 * reload1.c (elimination_effects): Likewise.
2485 (compute_reload_subreg_offset): Likewise.
2486 (choose_reload_regs): Likewise.
2487 * rtlanal.c (subreg_lsb_1): Likewise.
2488 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2489 (simplify_subreg): Likewise.
2490 * var-tracking.c (track_loc_p): Likewise.
2491 * emit-rtl.c (byte_lowpart_offset): Likewise.
2492 (paradoxical_subreg_p): Delete out-of-line definition.
2494 2017-08-22 Jeff Law <law@redhat.com>
2496 PR tree-optimization/81741
2497 PR tree-optimization/71947
2498 * tree-ssa-dom.c: Include tree-inline.h.
2499 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2500 equivalences if one is more expensive to compute than the other.
2501 * tree-ssa-scopedtables.h (class const_or_copies): Make
2502 record_const_or_copy_raw method private.
2503 (class avail_exprs_stack): New method simplify_binary_operation.
2504 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2505 avail_exprs_stack::simplify_binary_operation as needed.
2506 (avail_exprs_stack::simplify_binary_operation): New function.
2508 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2510 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2511 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2512 (DOT_SYMBOLS): Likewise.
2513 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2514 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2515 (RS6000_ABI_NAME): Likewise.
2516 (TARGET_CMODEL): Likewise.
2517 (TOC_SECTION_ASM_OP): Likewise.
2518 (SET_CMODEL): New macro.
2519 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2521 2017-08-22 Richard Biener <rguenther@suse.de>
2523 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2524 to allow for free-lang-data replacements similar to verify_type_variant.
2526 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2527 Alan Hayward <alan.hayward@arm.com>
2528 David Sherwood <david.sherwood@arm.com>
2530 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2531 VOIDmode for the LABEL_REF.
2533 2017-08-22 Richard Biener <rguenther@suse.de>
2535 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2537 2017-08-22 Richard Biener <rguenther@suse.de>
2539 * common.opt (feliminate-dwarf2-dups): Ignore.
2540 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2541 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2542 same_die_p_wrap, compute_section_prefix,
2543 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2544 (comdat_symbol_id, comdat_symbol_number): Likewise.
2545 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2547 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2548 (output_die): Mark unreachable path unreachable.
2549 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2550 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2551 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2552 (dwarf2out_early_finish): Likewise.
2554 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2556 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2558 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2561 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2562 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2563 (avr_attribute_table) <io, io_low, address>: Initialize
2564 .decl_required with true.
2566 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2568 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2569 undocumented debugging options.
2570 (-mvsx-scalar-double): Likewise.
2571 (-mallow-df-permute): Likewise.
2572 (-mvectorize-builtins): Likewise.
2573 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2574 (rs6000_builtin_vectorized_function): Likewise.
2575 (rs6000_builtin_md_vectorized_function): Likewise.
2576 (rs6000_opt_vars): Likewise.
2578 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2581 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2582 (*btrq_imm): Rename from *btrq.
2583 (*btcq_imm): Rename from *btcq.
2584 (btsc): New code attribute.
2585 (*<btsc><mode>): New insn pattern.
2586 (*btr<mode>): Ditto.
2587 (*<btsc><mode>_mask): New insn_and_split pattern.
2588 (*btr<mode>_mask): Ditto.
2590 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2591 Alan Hayward <alan.hayward@arm.com>
2592 David Sherwood <david.sherwood@arm.com>
2594 * function.c (pad_below): Simplify padding calculation.
2596 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2597 Alan Hayward <alan.hayward@arm.com>
2598 David Sherwood <david.sherwood@arm.com>
2600 * target.def (function_prologue): Remove frame size argument.
2601 (function_epilogue): Likewise.
2602 * doc/tm.texi: Regenerate.
2603 * final.c (final_start_function): Update call to function_prologue.
2604 (final_end_function): Update call to function_epilogue.
2605 (default_function_pro_epilogue): Remove frame size argument.
2606 * output.h (default_function_pro_epilogue): Likewise.
2607 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2608 (arm_output_function_prologue): Likewise.
2609 * config/frv/frv.c (frv_function_prologue): Likewise.
2610 (frv_function_epilogue): Likewise.
2611 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2612 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2613 (ia64_output_function_epilogue): Likewise.
2614 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2615 (m32r_output_function_epilogue): Likewise.
2616 * config/microblaze/microblaze.c (microblaze_function_prologue)
2617 (microblaze_function_epilogue): Likewise.
2618 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2619 (mips_output_function_epilogue): Likewise.
2620 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2621 (mmix_target_asm_function_epilogue): Likewise.
2622 * config/msp430/msp430.c (msp430_start_function): Likewise.
2623 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2624 (nds32_asm_function_epilogue): Likewise.
2625 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2626 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2627 (pa_output_function_epilogue): Likewise.
2628 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2629 (rs6000_output_function_epilogue): Likewise.
2630 * config/rl78/rl78.c (rl78_start_function): Likewise.
2631 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2632 (rs6000_output_function_epilogue): Likewise.
2633 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2634 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2635 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2636 (sparc_asm_function_epilogue): Likewise.
2638 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2640 * tree.h (type_has_mode_precision_p): New function.
2641 * convert.c (convert_to_integer_1): Use it.
2642 * expr.c (expand_expr_real_2): Likewise.
2643 (expand_expr_real_1): Likewise.
2644 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2645 * match.pd: Likewise.
2646 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2647 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2648 * tree-tailcall.c (process_assignment): Likewise.
2649 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2650 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2651 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2652 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2653 (vectorizable_assignment): Likewise.
2654 (vectorizable_shift): Likewise.
2655 (vectorizable_operation): Likewise.
2656 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2658 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2660 * match.pd: Add pow (C, x) simplification.
2662 2017-08-21 Richard Biener <rguenther@suse.de>
2664 PR tree-optimization/81900
2665 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2666 for blocks with abnormal predecessors.
2667 (compute_antic): Do not set visited flag prematurely.
2669 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2672 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2674 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2676 * stor-layout.h (vector_type_mode): Move to...
2677 * tree.h (vector_type_mode): ...here.
2678 * stor-layout.c (vector_type_mode): Move to...
2679 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2681 2017-08-21 Richard Biener <rguenther@suse.de>
2683 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2684 register_external_die hooks.
2685 (debug_false_tree_charstarstar_uhwistar): Declare.
2686 (debug_nothing_tree_charstar_uhwi): Likewise.
2687 * debug.c (do_nothing_debug_hooks): Adjust.
2688 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2689 (debug_nothing_tree_charstar_uhwi): Likewise.
2690 * dbxout.c (dbx_debug_hooks): Adjust.
2691 (xcoff_debug_hooks): Likewise.
2692 * sdbout.c (sdb_debug_hooks): Likewise.
2693 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2694 * dwarf2out.c (macinfo_label_base): New global.
2695 (dwarf2out_register_external_die): New function for the
2696 register_external_die hook.
2697 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2698 (dwarf2_debug_hooks): Use them.
2699 (dwarf2_lineno_debug_hooks): Adjust.
2700 (struct die_struct): Add with_offset flag.
2701 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2702 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2703 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2704 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2705 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2706 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2707 defining section names for the early LTO debug variants.
2708 (reset_indirect_string): New helper.
2709 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2710 (print_dw_val): Add support for offsetted symbol references.
2711 (get_ultimate_context): Split out from is_cxx.
2712 (is_cxx): Use get_ultimate_context.
2713 (is_fortran): Add decl overload.
2714 (compute_comp_unit_symbol): Split out worker from
2715 compute_section_prefix.
2716 (compute_section_prefix): Call compute_comp_unit_symbol and
2717 set comdat_type_p here.
2718 (output_die): Skip DIE symbol output for the LTO added one.
2719 Handle DIE symbol references with offset.
2720 (output_comp_unit): Guard section name mangling properly.
2721 For LTO debug sections emit a symbol at the section beginning
2722 which we use to refer to its DIEs.
2723 (add_abstract_origin_attribute): For DIEs registered via
2724 dwarf2out_register_external_die directly refer to the early
2725 DIE rather than indirectly through the shadow one we created.
2726 Remove obsolete call to dwarf2out_abstract_function for
2727 non-function/block origins.
2728 (gen_array_type_die): When generating early LTO debug do
2729 not emit DW_AT_string_length.
2730 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2731 late when in LTO. As suggested place a gcc_unreachable for
2732 the DECL_ABSTRACT_P case.
2733 (gen_subprogram_die): Avoid another specification DIE
2734 for early built declarations/definitions for the late LTO case.
2735 (gen_variable_die): Add type references for late duplicated VLA dies
2737 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2738 we have the abstract instance already.
2739 (process_scope_var): Adjust decl DIE contexts in LTO which
2740 first puts them in limbo.
2741 (gen_decl_die): Do not generate type DIEs late apart from
2742 types for VLAs or for decls we do not yet have a DIE. Do not
2743 call dwarf2out_abstract_function late.
2744 (dwarf2out_early_global_decl): Make sure to create DIEs
2745 for abstract instances of a decl first.
2746 (dwarf2out_late_global_decl): Adjust comment.
2747 (output_macinfo_op): With multiple macro sections use
2748 macinfo_label_base to distinguish labels.
2749 (output_macinfo): Likewise. Update macinfo_label_base.
2750 Pass in the line info label.
2751 (note_variable_value_in_expr): When generating LTO resolve
2752 all variable values here by generating DIEs as needed.
2753 (init_sections_and_labels): Add early LTO debug flag parameter
2754 and generate different sections and names if set. Add generation
2755 counter for the labels so we can have multiple of them.
2756 (reset_dies): Helper to allow DIEs to be output multiple times.
2757 (dwarf2out_finish): When outputting DIEs to the fat part of an
2758 LTO object first reset DIEs.
2759 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2760 (modified_type_die): Check for decl_ultimate_origin being self
2762 (gen_type_die_with_usage): Likewise.
2763 (gen_typedef_die): Allow decl_ultimate_origin being self.
2764 (set_decl_abstract_flags): Remove.
2765 (set_block_abstract_flags): Likewise.
2766 (dwarf2out_abstract_function): Treat the early generated DIEs
2767 as the abstract copy and only add DW_AT_inline and
2768 DW_AT_artificial here and call set_decl_origin_self.
2769 If the DIE has an abstract origin don't do anything.
2770 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2771 if we have none yet (Go fails to build one, PR78628).
2772 (variably_modified_type_p): Prevent endless recursion for Ada
2773 cyclic pointer types.
2774 * lto-streamer-in.c: Include debug.h.
2775 (dref_queue): New global.
2776 (lto_read_tree_1): Stream in DIE references.
2777 (lto_input_tree): Register DIE references.
2778 (input_function): Stream DECL_DEBUG_ARGS.
2779 * lto-streamer-out.c: Include debug.h.
2780 (lto_write_tree_1): Output DIE references.
2781 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2782 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2783 (output_function): Stream DECL_DEBUG_ARGS.
2784 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2785 Stream DECL_ABSTRACT_ORIGIN.
2786 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2787 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2788 DECL_CONTEXT for file-scope decls.
2789 * lto-streamer.h (struct dref_entry): Declare.
2790 (dref_queue): Likewise.
2791 * cfgexpand.c (pass_expand::execute): Do not call the
2792 outlining_inline_function hook here.
2793 * lto-wrapper.c (debug_obj): New global.
2794 (tool_cleanup): Unlink it if required.
2795 (debug_objcopy): New function.
2796 (run_gcc): Handle early debug sections in the IL files by
2797 extracting them to separate files, partially linkin them and
2798 feeding the result back as result to the linker.
2799 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2800 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2801 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2802 sections into a separate segment.
2803 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2805 (darwin_asm_dwarf_section): Likewise.
2806 (darwin_asm_output_dwarf_offset): Likewise.
2807 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2809 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2810 Alan Hayward <alan.hayward@arm.com>
2811 David Sherwood <david.sherwood@arm.com>
2813 * read-md.h (md_reader::record_potential_iterator_use): Replace
2814 pointer argument with an rtx and an index.
2815 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2816 (apply_mode_iterator): Likewise.
2817 (apply_code_iterator): Likewise.
2818 (apply_int_iterator): Likewise.
2819 (apply_subst_iterator): Likewise.
2820 (record_iterator_use): Likewise.
2821 (record_attribute_use): Likewise.
2822 (md_reader::record_potential_iterator_use): Likewise. Update calls
2823 to record_iterator_use and apply_iterator.
2824 (iterator_use): Replace ptr with x and index.
2825 (attribute_use): Likewise.
2826 (apply_attribute_uses): Update calls to apply_iterator.
2827 (apply_iterators): Likewise. Update initialization of iterator_use.
2828 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2829 and record_potential_iterator_use.
2830 (rtx_reader::read_rtx_operand): Likewise.
2832 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2833 Alan Hayward <alan.hayward@arm.com>
2834 David Sherwood <david.sherwood@arm.com>
2836 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2839 2017-08-21 Richard Biener <rguenther@suse.de>
2842 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2843 at struct end conservatively when comparing common bases.
2845 2017-08-21 Richard Biener <rguenther@suse.de>
2847 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2848 (mem_ref_in_stmt): Remove.
2849 (determine_max_movement): Use ref index to get at the reference.
2850 (invariantness_dom_walker::before_dom_children): Deal with
2851 lim data already initialized.
2852 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2854 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2856 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2857 (TARGET_ISA_ROUND): Ditto.
2858 (TARGET_ROUND): Ditto.
2859 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2860 * config/i386/i386.md: Ditto.
2861 * config/i386/sse.md: Ditto.
2862 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2863 with OPTION_MASK_ISA_SSE4_1.
2865 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2868 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2869 __builtin_ia32_lzcnt_u16.
2871 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2874 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2875 (rs6000_set_current_function): Rewrite function to use it.
2877 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2880 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2881 and TYPE_WARN_IF_NOT_ALIGN.
2882 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2883 (handle_warn_if_not_align): New.
2884 (place_union_field): Call handle_warn_if_not_align.
2885 (place_field): Call handle_warn_if_not_align.
2886 Copy TYPE_WARN_IF_NOT_ALIGN.
2887 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2888 (layout_type): Likewise.
2889 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2891 (tree_decl_common): Add warn_if_not_align.
2892 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2893 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2894 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2895 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2896 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2897 * doc/extend.texi: Document warn_if_not_aligned attribute.
2898 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2900 2017-08-17 Martin Liska <mliska@suse.cz>
2903 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2904 (get_data_dependence): Use it as pointer type.
2905 (distribute_loop): Likewise.
2907 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2909 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2910 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2911 (p8_vmrgow_<mode>_direct): New define_insn.
2912 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2913 handle endianness for vmrgew and vmrgow permute patterns.
2915 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2917 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2918 * config/rs6000/rs6000-cpus.def: Remove comment.
2919 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2920 (POWERPC_MASKS): Likewise.
2921 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2922 use of TARGET_VSX_TIMODE.
2923 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2924 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2925 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2926 (rs6000_option_override_internal): Remove dead code.
2927 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2928 (rs6000_legitimize_reload_address): Likewise.
2929 (rs6000_legitimate_address_p): Likewise.
2930 (rs6000_opt_masks): Delete "vsx-timode".
2931 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2932 from function comment.
2933 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2934 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2935 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2937 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2938 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2939 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2942 (define_peephole2): Likewise.
2944 2017-08-17 Martin Sebor <msebor@redhat.com>
2947 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2948 past the end of an array.
2949 (test_pp_format): Add test cases.
2951 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2953 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2954 missing ECF_NOTHROW flags.
2956 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2959 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2960 operands residing in integer registers.
2961 (*vsx_le_perm_load_<mode>): Likewise.
2962 (*vsx_le_perm_store_<mode>): Likewise.
2963 (define_peephole2): Add peepholes to optimize the above.
2965 2017-08-17 Marek Polacek <polacek@redhat.com>
2968 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2969 to mimic what shorten_compare did. Change the return type to bool.
2970 (fold_cond_expr_with_comparison): Update call to
2971 operand_equal_for_comparison_p.
2972 (fold_ternary_loc): Likewise.
2974 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2976 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2978 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2980 2017-08-17 Richard Biener <rguenther@suse.de>
2982 * tree-ssa-structalias.c (solve_graph): When propagating
2983 to successors update the graphs succ edges and avoid duplicate work.
2985 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2988 * config/i386/i386.c (ix86_option_override_internal): Save target
2989 specific options after ix86_stack_protector_guard_reg was changed.
2991 2017-08-17 Richard Biener <rguenther@suse.de>
2993 PR tree-optimization/81827
2994 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2995 (new_var_info): Initialize it conservatively.
2996 (get_call_vi): Mark register vars.
2997 (new_scalar_tmp_constraint_exp): Likewise.
2998 (handle_rhs_call): Likewise.
2999 (handle_const_call): Likewise.
3000 (create_function_info_for): Likewise.
3001 (solve_constraints): Sort varinfos to separate register from
3002 non-register vars to pack points-to solution bitmaps during
3005 2017-08-17 Marek Polacek <polacek@redhat.com>
3007 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
3009 2017-08-17 Richard Biener <rguenther@suse.de>
3011 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
3012 to true when overflow is undefined and we saturated the result.
3014 2017-08-17 Alan Modra <amodra@gmail.com>
3017 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
3018 Don't use store multiple if only one reg needs saving.
3019 (interesting_frame_related_regno): New function.
3020 (rs6000_frame_related): Don't emit frame info for regs that
3022 (rs6000_emit_epilogue): Likewise.
3024 2017-08-16 Nathan Sidwell <nathan@acm.org>
3026 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
3027 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
3028 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
3029 * tree.c (free_lang_data_in_type): Use else-if chain. Always
3030 clear TYPE_LANG_1. Remove obsolete member-function stripping.
3031 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
3032 (verify_type): Adjust for TYPE_BINFO move.
3033 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
3034 process TYPE_BINFO directly.
3035 (hash_tree): Likewise.
3036 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
3038 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
3041 2017-08-16 David Malcolm <dmalcolm@redhat.com>
3043 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
3045 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
3048 * config/i386/i386.md (*anddi_1_btr): Change predicates of
3049 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
3050 Add ix86_binary_operator_ok to insn constraint.
3051 (*iordi_1_bts): Ditto.
3052 (*xordi_1_btc): Ditto.
3053 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
3054 Update corresponding peephole2 pattern.
3058 2017-08-16 Bin Cheng <bin.cheng@arm.com>
3060 PR tree-optimization/81832
3061 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
3062 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3064 2017-08-16 Marek Polacek <polacek@redhat.com>
3067 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3068 regarding size_zero_node.
3070 2017-08-16 Martin Liska <mliska@suse.cz>
3073 * config.gcc: Respect previously set extra_objs in case
3076 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3078 PR tree-optimization/81835
3079 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3080 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3081 not depend on the phi.
3083 2017-08-16 Alan Modra <amodra@gmail.com>
3085 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3088 2017-08-16 Alan Modra <amodra@gmail.com>
3090 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3091 (save_reg_p): ..into this. Update all callers.
3092 (first_reg_to_save): Simplify.
3094 2017-08-16 Alan Modra <amodra@gmail.com>
3096 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3099 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3103 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3104 constant count if that count is less than 32.
3106 2017-08-15 Nathan Sidwell <nathan@acm.org>
3108 * gcc.c (execute): Emit friendlier message if inferior is killed
3109 by an external cause.
3111 2017-08-15 Richard Biener <rguenther@suse.de>
3113 PR tree-optimization/81790
3114 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3115 CONSTRUCTORs from simplifying and VN.
3117 2017-08-14 Martin Sebor <msebor@redhat.com>
3119 * builtin-attrs.def: Add comments.
3121 2017-08-14 Martin Sebor <msebor@redhat.com>
3124 * doc/extend.texi (attribute nonstring): Document new attribute.
3126 2017-08-14 Martin Sebor <msebor@redhat.com>
3129 * tree-diagnostic.c (default_tree_printer): Handle %G.
3130 * gimple-pretty-print.h (percent_G_format): Declare new function.
3131 * gimple-pretty-print.c (percent_G_format): Define.
3132 * tree-pretty-print.c (percent_K_format): Add argument.
3134 2017-08-14 Martin Sebor <msebor@redhat.com>
3136 PR translation/79998
3137 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3138 Remove a stray space.
3140 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3143 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3144 (*iordi_1_bts): Ditto.
3145 (*xordi_1_btc): Ditto.
3147 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3150 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3151 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3153 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3154 quoted strings, and make more translator-friendly.
3155 (darwin_rs6000_override_options): Likewise.
3156 (rs6000_option_override_internal): Likewise.
3157 (rs6000_return_in_memory): Fix overlong line.
3158 (init_cmulative_args): Use quoted strings, and make more
3159 translator-friendly.
3160 (rs6000_pass_by_reference): Fix overlong line.
3161 (def_builtin): Use quoted strings.
3162 (altivec_expand_predicate_builtin): Use quoted strings, and make
3163 more translator-friendly.
3164 (htm_expand_builtin): Use quoted strings.
3165 (cpu_expand_builtin): Use quoted strings, and make more
3166 translator-friendly.
3167 (altivec_expand_builtin): Likewise.
3168 (paired_expand_predicate_builtin): Likewise.
3169 (rs6000_invalid_builtin): Likewise.
3170 (builtin_function_type): Use quoted strings.
3171 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3172 more translator-friendly.
3173 (rs6000_trampoline_init): Likewise.
3174 (rs6000_handle_altivec_attribute): Likewise.
3175 (rs6000_inner_target_options): Use quoted strings.
3176 (rs6000_disable_incompatible_switches): Likewise.
3177 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3178 strings, and make more translator-friendly.
3179 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3181 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3183 PR tree-optimization/81799
3184 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3185 cond_expr to simple gimple operand.
3187 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3190 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3192 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3196 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3197 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3198 TARGET_GASISR_PROLOGUES.
3199 * config/avr/avr.c (avr_option_override): Same.
3200 (avr_pass_pre_proep::execute): Same.
3202 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3205 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3206 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3208 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3210 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3211 define_insn_and_split. Split to a memory load from 0 in
3212 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3213 using PTR mode iterator.
3214 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3215 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3216 (*add_tp_<mode>): Redefine as define_insn_and_split.
3217 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3218 address space. Merge with *add_tp_x32 using PTR mode iterator.
3219 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3220 Split to an add with a memory load from 0 in
3221 DEFAULT_TLS_SEG_REG address space.
3223 2017-08-12 Andrew Pinski <apinski@cavium.com>
3225 * config/aarch64/aarch64-option-extensions.def (rdma):
3226 Fix feature string to what Linux prints out in /proc/cpuinfo.
3228 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3231 * dwarf2out.c (modified_type_die): For C typedef types that have
3232 an ultimate origin, process the ultimate origin instead of the
3234 (gen_typedef_die): Assert that input DECLs have no ultimate
3236 (gen_type_die_with_usage): For typedef variants that have an
3237 ultimate origin, just call gen_decl_die on the original DECL.
3238 (process_scope_var): Avoid creating DIEs for local typedefs and
3239 concrete static variables.
3241 2017-08-12 Alan Modra <amodra@gmail.com>
3245 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3246 match gnu-user.h startfile.
3247 (ENDFILE_LINUX_SPEC): Similarly.
3249 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3252 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3254 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3256 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3257 * config/aarch64/aarch64.md (mov<mode>): Change.
3258 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3259 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3260 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3262 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3264 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3265 for storage order barriers.
3267 2017-08-11 Martin Liska <mliska@suse.cz>
3270 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3271 variables of void type.
3273 2017-08-11 Martin Liska <mliska@suse.cz>
3275 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3276 TARGET_SUPPORTS_ALIASES.
3277 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3278 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3279 (optimize_weakref): Likewise.
3280 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3281 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3282 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3284 2017-08-11 Martin Liska <mliska@suse.cz>
3287 * config/i386/i386.c (make_resolver_func): Do complete
3288 refactoring of the function.
3290 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3293 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3294 * config/i386/i386.c (ix86_stack_protect_guard): Use
3295 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3296 * doc/invoke.texi (x86 Options): Document
3297 -mstack-protector-guard-symbol= option.
3299 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3301 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3302 * config/i386/i386.c (ix86_split_stack_guard): New function.
3303 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3304 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3305 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3306 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3307 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3308 (split_stack_space_check): Call ix86_split_stack_guard.
3310 2017-08-10 Martin Sebor <msebor@redhat.com>
3312 * print-tree.c (print_node): Print location using the established
3314 Replace spaces with colons.
3315 (debug_raw, debug): Ditto.
3317 2017-08-10 Martin Sebor <msebor@redhat.com>
3320 * pretty-print.c (pp_format): Correct the handling of %s precision.
3322 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3325 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3327 (ix86_finalize_stack_frame_flags): This. Also clear
3328 frame_pointer_needed if -fno-omit-frame-pointer is used without
3330 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3331 with ix86_finalize_stack_frame_flags.
3332 (ix86_expand_epilogue): Likewise.
3333 (ix86_expand_split_stack_prologue): Likewise.
3334 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3336 2017-08-10 Martin Liska <mliska@suse.cz>
3339 * c-attribs.c (handle_target_attribute):
3340 Report warning for an empty string argument of target attribute.
3342 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3345 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3347 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3348 or DECL_NONLOCAL labels.
3349 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3350 or DECL_NONLOCAL labels here.
3352 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3354 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3355 to indicate when early gimple folding has been disabled.
3356 (rs6000_gimple_fold_builtin): Add debug content.
3357 (rs6000_invalid_builtin): Fix whitespace.
3358 (rs6000_expand_builtin): Fix whitespace.
3359 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3361 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3364 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3365 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3367 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3369 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3370 (qdf24xx): Likewise.
3371 * config/aarch64/aarch64.md: Include falkor.md.
3372 * config/aarch64/falkor.md: New.
3374 2017-08-09 Marek Polacek <polacek@redhat.com>
3377 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3378 * diagnostic.c (emit_diagnostic): Add a comment.
3379 (emit_diagnostic_valist): New function.
3381 2017-08-09 Marek Polacek <polacek@redhat.com>
3384 * input.c (make_location): New overload.
3385 * input.h (make_location): Declare.
3387 2017-08-08 Alan Modra <amodra@gmail.com>
3388 H.J. Lu <hongjiu.lu@intel.com>
3391 * gcc.c (NO_PIE_SPEC): Delete.
3392 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3394 (LINK_PIE_SPEC): ..to here.
3395 (LINK_COMMAND_SPEC): Support -no-pie.
3396 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3397 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3399 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3400 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3401 (ENDFILE_CRTEND_SPEC): Similarly.
3403 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3406 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3407 (mstack-protector-guard-offset=): Ditto.
3408 * config/i386/i386.c (ix86_option_override): Handle
3409 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3411 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3412 ix86_stack_protect_guard_offset variables.
3413 (TARGET_STACK_PROTECT_GUARD): Always define.
3414 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3415 and -mstack-protector-guard-offset= options.
3417 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3419 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3420 boundary case for the last candidate.
3422 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3424 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3425 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3427 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3430 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3431 * config/aarch64/aarch64-builtins.c
3432 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3433 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3434 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3436 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3437 Andrew Pinski <pinskia@gmail.com>
3440 * internal-fn.def (XORSIGN): New.
3441 * optabs.def (xorsign_optab): New.
3442 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3443 (convert_expand_mult_copysign): New.
3444 (pass_optimize_widening_mul::execute): Call
3445 convert_expand_mult_copysign.
3447 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3449 PR tree-optimization/81354
3450 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3451 Insert on edges rather than explicitly creating landing pads.
3452 (analyze_candidates_and_replace): Commit edge inserts.
3454 2017-08-08 Richard Biener <rguenther@suse.de>
3457 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3458 (expand_simple_operations): Also look through ADDR_EXPRs with
3459 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3461 2017-08-08 Richard Biener <rguenther@suse.de>
3463 PR tree-optimization/81723
3464 * tree-vect-slp.c (struct bst_traits): New hash traits.
3465 (bst_fail): New global.
3466 (vect_build_slp_tree_2): New worker, split out from ...
3467 (vect_build_slp_tree): ... this now wrapping it with using
3468 bst_fail set to cache SLP tree build fails. Properly handle
3470 (vect_analyze_slp_instance): Allocate and free bst_fail.
3472 2017-08-08 Martin Liska <mliska@suse.cz>
3475 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3476 LABEL_DECLs that can be from a different function.
3478 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3480 PR tree-optimization/81744
3481 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3482 loop's number of iterations.
3484 2017-08-08 Martin Liska <mliska@suse.cz>
3486 * asan.c: Include header files.
3487 * attribs.c (build_decl_attribute_variant): New function moved
3489 (build_type_attribute_qual_variant): Likewise.
3490 (cmp_attrib_identifiers): Likewise.
3491 (simple_cst_list_equal): Likewise.
3492 (omp_declare_simd_clauses_equal): Likewise.
3493 (attribute_value_equal): Likewise.
3494 (comp_type_attributes): Likewise.
3495 (build_type_attribute_variant): Likewise.
3496 (lookup_ident_attribute): Likewise.
3497 (remove_attribute): Likewise.
3498 (merge_attributes): Likewise.
3499 (merge_type_attributes): Likewise.
3500 (merge_decl_attributes): Likewise.
3501 (merge_dllimport_decl_attributes): Likewise.
3502 (handle_dll_attribute): Likewise.
3503 (attribute_list_equal): Likewise.
3504 (attribute_list_contained): Likewise.
3505 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3506 (lookup_attribute_by_prefix): Likewise.
3507 * bb-reorder.c: Include header files.
3508 * builtins.c: Likewise.
3509 * calls.c: Likewise.
3510 * cfgexpand.c: Likewise.
3511 * cgraph.c: Likewise.
3512 * cgraphunit.c: Likewise.
3513 * convert.c: Likewise.
3514 * dwarf2out.c: Likewise.
3515 * final.c: Likewise.
3516 * fold-const.c: Likewise.
3517 * function.c: Likewise.
3518 * gimple-expr.c: Likewise.
3519 * gimple-fold.c: Likewise.
3520 * gimple-pretty-print.c: Likewise.
3521 * gimple.c: Likewise.
3522 * gimplify.c: Likewise.
3523 * hsa-common.c: Likewise.
3524 * hsa-gen.c: Likewise.
3525 * internal-fn.c: Likewise.
3526 * ipa-chkp.c: Likewise.
3527 * ipa-cp.c: Likewise.
3528 * ipa-devirt.c: Likewise.
3529 * ipa-fnsummary.c: Likewise.
3530 * ipa-inline.c: Likewise.
3531 * ipa-visibility.c: Likewise.
3533 * lto-cgraph.c: Likewise.
3534 * omp-expand.c: Likewise.
3535 * omp-general.c: Likewise.
3536 * omp-low.c: Likewise.
3537 * omp-offload.c: Likewise.
3538 * omp-simd-clone.c: Likewise.
3539 * opts-global.c: Likewise.
3540 * passes.c: Likewise.
3541 * predict.c: Likewise.
3542 * sancov.c: Likewise.
3543 * sanopt.c: Likewise.
3544 * symtab.c: Likewise.
3545 * toplev.c: Likewise.
3546 * trans-mem.c: Likewise.
3547 * tree-chkp.c: Likewise.
3548 * tree-eh.c: Likewise.
3549 * tree-into-ssa.c: Likewise.
3550 * tree-object-size.c: Likewise.
3551 * tree-parloops.c: Likewise.
3552 * tree-profile.c: Likewise.
3553 * tree-ssa-ccp.c: Likewise.
3554 * tree-ssa-live.c: Likewise.
3555 * tree-ssa-loop.c: Likewise.
3556 * tree-ssa-sccvn.c: Likewise.
3557 * tree-ssa-structalias.c: Likewise.
3558 * tree-ssa.c: Likewise.
3559 * tree-streamer-in.c: Likewise.
3560 * tree-vectorizer.c: Likewise.
3561 * tree-vrp.c: Likewise.
3563 * ubsan.c: Likewise.
3564 * varasm.c: Likewise.
3565 * varpool.c: Likewise.
3566 * tree.c: Remove functions moved to attribs.[ch].
3568 * config/aarch64/aarch64.c: Add attrs.h header file.
3569 * config/alpha/alpha.c: Likewise.
3570 * config/arc/arc.c: Likewise.
3571 * config/arm/arm.c: Likewise.
3572 * config/avr/avr.c: Likewise.
3573 * config/bfin/bfin.c: Likewise.
3574 * config/c6x/c6x.c: Likewise.
3575 * config/cr16/cr16.c: Likewise.
3576 * config/cris/cris.c: Likewise.
3577 * config/darwin.c: Likewise.
3578 * config/epiphany/epiphany.c: Likewise.
3579 * config/fr30/fr30.c: Likewise.
3580 * config/frv/frv.c: Likewise.
3581 * config/ft32/ft32.c: Likewise.
3582 * config/h8300/h8300.c: Likewise.
3583 * config/i386/winnt.c: Likewise.
3584 * config/ia64/ia64.c: Likewise.
3585 * config/iq2000/iq2000.c: Likewise.
3586 * config/lm32/lm32.c: Likewise.
3587 * config/m32c/m32c.c: Likewise.
3588 * config/m32r/m32r.c: Likewise.
3589 * config/m68k/m68k.c: Likewise.
3590 * config/mcore/mcore.c: Likewise.
3591 * config/microblaze/microblaze.c: Likewise.
3592 * config/mips/mips.c: Likewise.
3593 * config/mmix/mmix.c: Likewise.
3594 * config/mn10300/mn10300.c: Likewise.
3595 * config/moxie/moxie.c: Likewise.
3596 * config/msp430/msp430.c: Likewise.
3597 * config/nds32/nds32-isr.c: Likewise.
3598 * config/nds32/nds32.c: Likewise.
3599 * config/nios2/nios2.c: Likewise.
3600 * config/nvptx/nvptx.c: Likewise.
3601 * config/pa/pa.c: Likewise.
3602 * config/pdp11/pdp11.c: Likewise.
3603 * config/powerpcspe/powerpcspe.c: Likewise.
3604 * config/riscv/riscv.c: Likewise.
3605 * config/rl78/rl78.c: Likewise.
3606 * config/rx/rx.c: Likewise.
3607 * config/s390/s390.c: Likewise.
3608 * config/sh/sh.c: Likewise.
3609 * config/sol2.c: Likewise.
3610 * config/sparc/sparc.c: Likewise.
3611 * config/spu/spu.c: Likewise.
3612 * config/stormy16/stormy16.c: Likewise.
3613 * config/tilegx/tilegx.c: Likewise.
3614 * config/tilepro/tilepro.c: Likewise.
3615 * config/v850/v850.c: Likewise.
3616 * config/vax/vax.c: Likewise.
3617 * config/visium/visium.c: Likewise.
3618 * config/xtensa/xtensa.c: Likewise.
3620 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3623 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3624 constraints since the -mupper-regs-* switches have been
3626 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3627 into a vector from a double word element that was extracted from
3628 another vector, and eliminate extra XXPERMDI instructions.
3629 (vsx_concat_<mode>_2): Likewise.
3630 (vsx_concat_<mode>_3): Likewise.
3631 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3632 concat to allow optimizing inserts from previous extracts.
3634 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3636 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3637 memory reference to a SSP offset in TLS address space.
3638 (ix86_print_operand) <case '@'>: Remove.
3639 (ix86_print_operand_punct_valid_p): Remove '@' code.
3640 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3642 (stack_tls_protect_set_<mode>): Remove.
3643 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3644 (stack_tls_protect_test_<mode>): Remove.
3645 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3647 2017-08-07 Olivier Hainque <hainque@adacore.com>
3650 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3652 2017-08-07 Douglas Rupp <rupp@adacore.com>
3654 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3655 variable was referenced as multidir in command.
3657 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3660 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3663 2017-08-07 Martin Liska <mliska@suse.cz>
3665 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3666 * config/rl78/rl78.c: Add include of attribs.h.
3667 * config/sh/sh.c: Likewise.
3668 * config/v850/v850.c: Likewise.
3670 2017-08-07 Tom de Vries <tom@codesourcery.com>
3673 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3675 2017-08-07 Martin Liska <mliska@suse.cz>
3677 * config/mips/mips.c: Include attribs.h.
3679 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3682 * doc/invoke.texi: Document change in behvaior for -Ofast for
3685 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3687 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3689 (aarch64_pop_regs): Likewise.
3690 (aarch64_gen_load_pair): Likewise.
3691 (aarch64_save_callee_saves): Likewise.
3692 (aarch64_restore_callee_saves): Likewise.
3694 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3696 * config/i386/i386.c: Revert the last change.
3698 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3701 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3703 (ix86_finalize_stack_frame_flags): This. Also clear
3704 frame_pointer_needed if -fno-omit-frame-pointer is used without
3706 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3707 with ix86_finalize_stack_frame_flags.
3708 (ix86_expand_epilogue): Likewise.
3709 (ix86_expand_split_stack_prologue): Likewise.
3711 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3714 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3715 to P_AES for Westmere.
3717 2017-08-07 Jonathan Yong <10walls@gmail.com>
3719 * config/i386/mingw.opt (fset-stack-executable): Removed.
3720 * config/i386/cygming.opt (fset-stack-executable): Moved
3722 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3724 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3726 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3728 2017-08-07 Marek Polacek <polacek@redhat.com>
3731 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3733 2017-08-07 Martin Liska <mliska@suse.cz>
3735 * attribs.h (canonicalize_attr_name): New function.
3736 (cmp_attribs): Move from c-format.c and adjusted.
3737 (is_attribute_p): Moved from tree.h.
3738 * tree-inline.c: Add new includes.
3739 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3740 (private_is_attribute_p): Remove.
3741 (private_lookup_attribute): Likewise.
3742 (private_lookup_attribute_by_prefix): Simplify.
3743 (remove_attribute): Use is_attribute_p.
3744 * tree.h: Remove removed declarations.
3746 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3749 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3750 instead of computing it in the function. Formatting fix.
3751 (expand_case): Don't rely on default_edge being the first edge,
3752 clear it if removing it, pass default_edge to
3753 emit_case_dispatch_table.
3754 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3757 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3759 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3760 insn in the function, emit NOP after the insn.
3762 2017-08-06 Tom de Vries <tom@codesourcery.com>
3764 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3767 2017-08-06 Tom de Vries <tom@codesourcery.com>
3769 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3772 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3774 PR tree-optimization/57371
3775 * match.pd: New pattern.
3777 2017-08-04 Marek Polacek <polacek@redhat.com>
3780 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3781 perform the computation in offset_int.
3783 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3785 PR tree-optimization/81136
3786 * tree-vectorizer.h: Include tree-hash-traits.h.
3787 (vec_base_alignments): New typedef.
3788 (vec_info): Add a base_alignments field.
3789 (vect_record_base_alignments): Declare.
3790 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3792 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3793 (create_data_ref): Add an is_conditional_in_stmt argument.
3794 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3795 the is_conditional_in_stmt field.
3796 (data_ref_loc): Add an is_conditional_in_stmt field.
3797 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3798 (find_data_references_in_stmt): Update call to create_data_ref.
3799 (graphite_find_data_references_in_stmt): Likewise.
3800 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3801 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3802 (vect_record_base_alignment): New function.
3803 (vect_record_base_alignments): Likewise.
3804 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3805 for nested statements even if we fail to compute a misalignment.
3806 Use pooled base alignments for unconditional references.
3807 (vect_find_same_alignment_drs): Compare base addresses instead
3809 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3810 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3812 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3814 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3815 Add an explicit name for the enum. Use auto_vec for slp_instances
3817 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3819 (_bb_vec_info): Add a constructor and destructor.
3820 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3821 (destroy_loop_vec_info): Delete.
3822 (vect_destroy_datarefs): Likewise.
3823 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3824 (vec_info::vec_info): New function.
3825 (vec_info::~vec_info): Likewise.
3826 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3827 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3828 destroy_loop_vec_info.
3829 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3830 (_loop_vec_info::_loop_vec_info): ...this.
3831 (destroy_loop_vec_info): Replace with...
3832 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3833 the stmt_vec_infos. Leave handling of vec_info information to its
3834 destructor. Remove explicit vector releases.
3835 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3836 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3837 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3838 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3839 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3840 (destroy_bb_vec_info): Replace with...
3841 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3842 information to its destructor.
3843 (vect_slp_analyze_bb_1): Use new and delete instead of
3844 new_bb_vec_info and destroy_bb_vec_info.
3845 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3848 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3850 * tree-data-ref.h (subscript): Add access_fn field.
3851 (data_dependence_relation): Add could_be_independent_p.
3852 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3853 (same_access_functions): Move to tree-data-ref.c.
3854 * tree-data-ref.c (ref_contains_union_access_p): New function.
3855 (access_fn_component_p): Likewise.
3856 (access_fn_components_comparable_p): Likewise.
3857 (dr_analyze_indices): Add a reference to access_fn_component_p.
3858 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3860 (constant_access_functions): Likewise.
3861 (add_other_self_distances): Likewise.
3862 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3863 (initialize_data_dependence_relation): Use XCNEW and remove
3864 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3865 of access functions that have the same type. Allow the
3866 subsequence to end with different bases in some circumstances.
3867 Record the chosen access functions in SUB_ACCESS_FN.
3868 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3869 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3870 (subscript_dependence_tester_1): Likewise dra and drb.
3871 (build_classic_dist_vector): Update calls accordingly.
3872 (subscript_dependence_tester): Likewise.
3873 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3874 DDR_COULD_BE_INDEPENDENT_P.
3875 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3876 comp_alias_ddrs instead of may_alias_ddrs.
3877 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3879 (vect_analyze_data_ref_dependence): Use it if
3880 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3881 distance vectors if that fails.
3882 (dependence_distance_ge_vf): New function.
3883 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3884 LOOP_VINFO_MAY_ALIAS_DDRS.
3886 2017-08-04 Richard Biener <rguenther@suse.de>
3889 * fold-const.c (fold_binary_loc): Properly restrict
3890 minus_var0 && minus_var1 case when associating undefined overflow
3893 2017-08-04 Tom de Vries <tom@codesourcery.com>
3895 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3897 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3899 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3900 Don't start diagnostic messages with a capital letter.
3901 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3903 (rs6000_invalid_builtin): Likewise.
3904 (rs6000_trampoline_init): Likewise.
3906 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3909 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3910 after setting changeable df flags.
3912 2017-08-03 Richard Biener <rguenther@suse.de>
3914 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3915 up if the use is in USE - X.
3917 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3919 * toplev.c (dumpfile.h): New include.
3920 (internal_error_reentered): New static function. Use it...
3921 (internal_error_function): ...here to handle reentered internal_error.
3923 2017-08-03 Richard Biener <rguenther@suse.de>
3926 * fold-const.c (split_tree): Add minus_var and minus_con
3927 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3928 here but always use minus_*.
3929 (associate_trees): Assert we never associate with MINUS_EXPR
3930 and NULL first operand. Do not recurse for PLUS_EXPR operands
3931 when associating as MINUS_EXPR either.
3932 (fold_binary_loc): Track minus_var and minus_con.
3934 2017-08-03 Tom de Vries <tom@codesourcery.com>
3937 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3938 ACCEL_COMPILER, apply finish_options on
3939 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3941 2017-08-03 Tom de Vries <tom@codesourcery.com>
3944 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3945 function_entry_patch_area_size > 0.
3947 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3950 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3951 instead of 10??LU, perform unit multiplication in wide_int,
3952 don't change alloc_object_size_limit if the limit is larger
3955 PR tree-optimization/81655
3956 PR tree-optimization/81588
3957 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3958 the case when ranges[i].low and high are 1 for unsigned type with
3962 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3963 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3965 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3967 * tree-vrp.h: Add include guard.
3969 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3972 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3973 (ud2): New insn pattern.
3974 * config/i386/i386.c (ix86_expand_epilogue):
3975 For naked functions, generate ud2 instead of trap insn.
3977 2017-08-02 Marek Polacek <polacek@redhat.com>
3980 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3982 2017-08-02 Tom de Vries <tom@codesourcery.com>
3983 Cesar Philippidis <cesar@codesourcery.com>
3985 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3986 Add missing edge probabilities.
3988 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3990 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3993 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3996 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3997 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3998 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3999 prologue_seq sequences - if any.
4001 2017-08-02 Richard Biener <rguenther@suse.de>
4003 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
4004 via vectors if supported, integer extracts via punning if supported
4005 or otherwise vector extracts.
4007 2017-08-02 Richard Biener <rguenther@suse.de>
4009 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
4011 (bitmap_insert_into_set): ... this.
4013 2017-08-02 Richard Biener <rguenther@suse.de>
4015 PR tree-optimization/81633
4017 2015-08-17 Alan Hayward <alan.hayward@arm.com>
4019 PR tree-optimization/71752
4020 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
4022 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
4024 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
4025 (machine_function::call_ms2sysv_pad_out): Remove field.
4026 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
4027 (ix86_compute_frame_layout): Likewise.
4029 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
4032 * config/i386/i386.c (ix86_set_func_type): Disallow naked
4033 attribute with interrupt attribute.
4035 2017-08-01 Andrew Pinski <apinski@cavium.com>
4037 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
4038 BIT_INSERT_EXPR's operand 1
4039 to see if the types precision matches.
4041 2017-08-01 Martin Liska <mliska@suse.cz>
4044 * builtins.c (expand_builtin_memcpy_args): Remove.
4045 (expand_builtin_memcpy): Call newly added function
4046 expand_builtin_memory_copy_args.
4047 (expand_builtin_memcpy_with_bounds): Likewise.
4048 (expand_builtin_mempcpy): Remove last argument.
4049 (expand_builtin_mempcpy_with_bounds): Likewise.
4050 (expand_builtin_memory_copy_args): New function created from
4051 expand_builtin_mempcpy_args with small modifications.
4052 (expand_builtin_mempcpy_args): Remove.
4053 (expand_builtin_stpcpy): Remove unused argument.
4054 (expand_builtin): Likewise.
4055 (expand_builtin_with_bounds): Likewise.
4057 2017-08-01 Martin Liska <mliska@suse.cz>
4060 Make mempcpy more optimal (PR middle-end/70140).
4062 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4065 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4066 __builtin_vec_cmpne verify both arguments are compatible vectors
4067 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4068 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4069 move computation of aligned to after checking the argument types.
4073 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4074 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4077 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4078 Doug Rupp <rupp@adacore.com>
4079 Olivier Hainque <hainque@adacore.com>
4081 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4082 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4084 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4085 for TARGET_OS_CPP_BUILTIN.
4086 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4087 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4089 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4090 passing required abi options to the assembler for EABI configurations.
4091 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4092 of .text.hot and .text.unlikely sections for kernel modules when
4093 using ARM style exceptions.
4094 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4095 options. Add EXTRA_CC1_SPEC.
4096 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4098 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4100 (ARM_TARGET2_DWARF_FORMAT): Define.
4101 * config/arm/t-vxworks: Adjust multilib control to removal of the
4102 Diab command line options.
4104 2017-08-01 Martin Liska <mliska@suse.cz>
4106 PR gcov-profile/81561
4107 * gcov.c (unblock): Make unblocking safe as we need to preserve
4108 index correspondence of blocks and block_lists.
4110 2017-08-01 Richard Biener <rguenther@suse.de>
4112 PR tree-optimization/81181
4113 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4114 (compute_antic): ... end of iteration here.
4116 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4118 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4119 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4120 (ftree-slp-vectorize): Likewise.
4121 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4122 can no longer be set independent of flag_tree_loop_vectorize.
4123 * omp-general.c (emp_max_vf): Likewise.
4124 * opts.c (enable_fdo_optimizations): Remove references to
4125 flag_tree_vectorize, these are now implicit.
4126 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4127 and leave it for the options machinery.
4129 2017-08-01 Martin Liska <mliska@suse.cz>
4132 * builtins.c (expand_builtin_memcpy_args): Remove.
4133 (expand_builtin_memcpy): Call newly added function
4134 expand_builtin_memory_copy_args.
4135 (expand_builtin_memcpy_with_bounds): Likewise.
4136 (expand_builtin_mempcpy): Remove last argument.
4137 (expand_builtin_mempcpy_with_bounds): Likewise.
4138 (expand_builtin_memory_copy_args): New function created from
4139 expand_builtin_mempcpy_args with small modifications.
4140 (expand_builtin_mempcpy_args): Remove.
4141 (expand_builtin_stpcpy): Remove unused argument.
4142 (expand_builtin): Likewise.
4143 (expand_builtin_with_bounds): Likewise.
4145 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4148 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4149 print "ds:" only for immediates in generic address space.
4151 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4154 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4155 (ix86_function_ok_for_sibcall): Return false for naked functions.
4157 2017-08-01 Richard Biener <rguenther@suse.de>
4159 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4160 (compute_antic): Seed worklist with exit block predecessors.
4161 * cfganal.c (dfs_find_deadend): For a cycle return the source
4162 of the edge closing it.
4164 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4166 * config/aarch64/aarch64.c
4167 (aarch64_can_const_movi_rtx_p): Move 0 check.
4169 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4171 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4172 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4174 * match.pd: Ditto in address comparison pattern.
4176 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4178 PR tree-optimization/81627
4179 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4180 closed ssa form for store-store chain.
4182 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4184 PR tree-optimization/81620
4185 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4186 for store-store chain.
4188 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4190 PR tree-optimization/81588
4191 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4192 ranges[i].in_p, invert comparison code ccode. For >/>=,
4193 swap rhs1 and rhs2 and comparison code unconditionally,
4194 for </<= don't do that. Don't swap rhs1/rhs2 again if
4195 ranges[i].in_p, instead invert comparison code ccode if
4196 opcode or oe->rank is BIT_IOR_EXPR.
4199 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4200 a direct optab to conversion optab.
4201 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4202 with GET_MODE_INNER as last argument instead of optab_handler.
4203 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4204 vector extraction if possible and optab is available.
4205 * expr.c (store_constructor): Use convert_optab_handler instead
4206 of optab_handler. Use vector initialization from smaller
4207 vectors if possible and optab is available.
4208 * tree-vect-stmts.c (vectorizable_load): Likewise.
4209 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4211 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4212 of vec_init from half-sized vectors with the same element mode.
4213 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4214 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4215 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4216 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4217 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4218 after mode in gen_vec_extract* calls.
4219 (vec_extract<mode>): Renamed to ...
4220 (vec_extract<mode><ssescalarmodelower>): ... this.
4221 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4222 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4223 element mode after mode in gen_vec_init* calls.
4224 (VEC_INIT_HALF_MODE): New mode iterator.
4225 (vec_init<mode>): Renamed to ...
4226 (vec_init<mode><ssescalarmodelower>): ... this.
4227 (vec_init<mode><ssehalfvecmodelower>): New expander.
4228 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4229 (vec_extractv2sfsf): ... this.
4230 (vec_initv2sf): Renamed to ...
4231 (vec_initv2sfsf): ... this.
4232 (vec_extractv2si): Renamed to ...
4233 (vec_extractv2sisi): ... this.
4234 (vec_initv2si): Renamed to ...
4235 (vec_initv2sisi): ... this.
4236 (vec_extractv4hi): Renamed to ...
4237 (vec_extractv4hihi): ... this.
4238 (vec_initv4hi): Renamed to ...
4239 (vec_initv4hihi): ... this.
4240 (vec_extractv8qi): Renamed to ...
4241 (vec_extractv8qiqi): ... this.
4242 (vec_initv8qi): Renamed to ...
4243 (vec_initv8qiqi): ... this.
4244 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
4245 (vec_init<mode>): Renamed to ...
4246 (vec_init<mode><VEC_base_l>): ... this.
4247 (vec_extract<mode>): Renamed to ...
4248 (vec_extract<mode><VEC_base_l>): ... this.
4249 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
4250 (vec_initv2sfsf): ... this.
4251 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
4252 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4253 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
4254 element mode after mode in gen_vec_init* calls.
4255 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4256 (vec_init<mode><Vel>): ... this.
4257 (vec_extract<mode>): Renamed to ...
4258 (vec_extract<mode><Vel>): ... this.
4259 * config/aarch64/iterators.md (Vel): New mode attribute.
4260 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4261 Add element mode after mode in gen_vec_extract* calls.
4262 * config/s390/vector.md (non_vec_l): New mode attribute.
4263 (vec_extract<mode>): Renamed to ...
4264 (vec_extract<mode><non_vec_l>): ... this.
4265 (vec_init<mode>): Renamed to ...
4266 (vec_init<mode><non_vec_l>): ... this.
4267 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4268 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4270 * config/arm/iterators.md (V_elem_l): New mode attribute.
4271 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4272 (vec_extract<mode><V_elem_l>): ... this.
4273 (vec_extractv2di): Renamed to ...
4274 (vec_extractv2didi): ... this.
4275 (vec_init<mode>): Renamed to ...
4276 (vec_init<mode><V_elem_l>): ... this.
4277 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4278 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4279 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4280 Add element mode after gen_vec_extract* calls.
4281 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4282 (vec_init<mode><unitmode>): ... this.
4283 (vec_extract<mode>): Renamed to ...
4284 (vec_extract<mode><unitmode>): ... this.
4285 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4286 (vec_init<mode><unitmode>): ... this.
4287 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4288 (vec_initv2sfsf): ... this.
4289 (vec_extractv2sf): Renamed to ...
4290 (vec_extractv2sfsf): ... this.
4291 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4292 Add element mode after gen_vec_extract* calls.
4293 * config/mips/mips.md (unitmode): New mode iterator.
4294 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4295 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4296 * config/spu/spu.md (inner_l): New mode attribute.
4297 (vec_init<mode>): Renamed to ...
4298 (vec_init<mode><inner_l>): ... this.
4299 (vec_extract<mode>): Renamed to ...
4300 (vec_extract<mode><inner_l>): ... this.
4301 * config/sparc/sparc.md (veltmode): New mode iterator.
4302 (vec_init<VMALL:mode>): Renamed to ...
4303 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4304 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4305 (vec_initv2sisi): ... this.
4306 (vec_initv2sf): Renamed to ...
4307 (vec_initv2sfsf): ... this.
4308 (vec_extractv2sf): Renamed to ...
4309 (vec_extractv2sfsf): ... this.
4310 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4311 (vec_init<mode>): Renamed to ...
4312 (vec_init<mode><VEC_base_l>): ... this.
4313 (vec_extract<mode>): Renamed to ...
4314 (vec_extract<mode><VEC_base_l>): ... this.
4315 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4316 (vec_initv2sfsf): ... this.
4317 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4318 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4319 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4320 gen_vec_init* calls.
4322 2017-08-01 Richard Biener <rguenther@suse.de>
4324 PR tree-optimization/81297
4325 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4326 TREE_OVERFLOW from INTEGER_CSTs.
4328 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4330 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4332 2017-07-31 Carl Love <cel@us.ibm.com>
4334 * config/rs6000/rs6000-c: Add support for built-in functions
4335 vector signed char vec_xl_be (signed long long, signed char *);
4336 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4337 vector signed int vec_xl_be (signed long long, signed int *);
4338 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4339 vector signed long long vec_xl_be (signed long long, signed long long *);
4340 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4341 vector signed short vec_xl_be (signed long long, signed short *);
4342 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4343 vector double vec_xl_be (signed long long, double *);
4344 vector float vec_xl_be (signed long long, float *);
4345 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4346 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4347 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4349 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4350 (altivec_expand_builtin): Add switch statement to call
4351 altivec_expand_xl_be for each builtin.
4352 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4353 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4354 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4355 __builtin_vsx_le_be_v16qi.
4356 * doc/extend.texi: Update the built-in documentation file for the
4357 new built-in functions.
4359 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4362 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4364 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4366 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4368 * config.gcc: Add z14.
4369 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4370 CPU model numbers for z13s and z14.
4371 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4373 * config/s390/s390-opts.h (enum processor_type): Rename
4374 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4375 * config/s390/s390.c (processor_table): Add field for CPU name to
4376 be passed to Binutils.
4377 (s390_asm_output_machine_for_arch): Use the new field in
4378 processor_table for Binutils.
4379 (s390_expand_builtin): Replace arch12 with z14.
4380 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4381 (s390_get_sched_attrmask): Likewise.
4382 (s390_get_unit_mask): Likewise.
4383 * config/s390/s390.opt: Add z14 to processor_type enum.
4385 2017-07-31 Martin Jambor <mjambor@suse.cz>
4388 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4389 regardless of optimization level.
4391 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4392 Martin Liska <mliska@suse.cz>
4394 * predict.def: Remove old comment and adjust probability.
4395 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4398 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4401 * config/i386/i386.c (ix86_function_naked): New function.
4402 (ix86_can_use_return_insn_p): Return false for naked functions.
4403 (ix86_expand_prologue): Skip prologue for naked functions.
4404 (ix86_expand_epilogue): Skip epilogue for naked functions
4405 and emit trap instruction.
4406 (ix86_warn_func_return): New function.
4407 (ix86_attribute_table): Add "naked" attribute specification.
4408 (TARGET_WARN_FUNC_RETURN): Define.
4409 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4411 2017-07-31 Martin Liska <mliska@suse.cz>
4413 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4414 (dump_gimple_bb_header): Always dump BB info.
4415 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4417 2017-07-31 Martin Liska <mliska@suse.cz>
4420 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4421 also with current_function_decl non-null equality.
4423 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4426 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4427 change type to the element type, instead add eltype variable and
4428 use it where we are interested in the element type.
4430 PR tree-optimization/81603
4431 * ipa-polymorphic-call.c
4432 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4433 offset arithmetic in offset_int, bail out if the resulting bit offset
4434 doesn't fit into shwi.
4436 2017-07-31 Martin Liska <mliska@suse.cz>
4438 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4439 (gimplify_save_expr): Fix comment.
4441 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4444 * config/i386/i386.c (ix86_function_arg): Update arguments for
4446 (ix86_compute_frame_layout): Set the initial stack offset to
4447 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4448 INCOMING_FRAME_SP_OFFSET.
4449 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4450 stack before exception handler returns.
4451 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4452 the 'ERROR_CODE' for exception handler.
4454 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4456 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4457 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4458 (ASM_OUTPUT_REG_POP): Ditto.
4459 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4460 instead of asm_fprintf to output pure string.
4462 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4464 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4465 to imported_module_or_decl hook.
4466 (debug_nothing_tree_tree_tree_bool): Remove.
4467 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4468 * debug.c (do_nothing_debug_hooks): Use
4469 debug_nothing_tree_tree_tree_bool_bool instead of
4470 debug_nothing_tree_tree_tree_bool.
4471 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4472 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4473 * sdbout.c (sdb_debug_hooks): Likewise.
4474 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4475 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4477 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4478 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4479 attribute, don't add anything.
4481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4483 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4484 (fold_build2_stat_loc): Likewise.
4485 (fold_build3_stat_loc): Likewise.
4486 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4487 (fold_build1_loc): Remove macro.
4488 (fold_build2_loc): Likewise.
4489 (fold_build3_loc): Likewise.
4491 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4493 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4494 (gimple_build_debug_bind_source_stat): Likewise.
4495 * gimple.h (gimple_build_debug_bind): Remove macro.
4496 (gimple_build_debug_bind_source): Likewise.
4498 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4500 * bitmap.c (bitmap_alloc): Adjust.
4501 (bitmap_gc_alloc): Likewise.
4502 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4504 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4506 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4507 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4508 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4509 (bitmap_gc_alloc_stat): Likewise.
4510 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4514 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4515 * rtl.h (shallow_copy_rtx): Remove macro.
4517 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4519 * emit-rtl.c (gen_raw_REG): Adjust.
4520 * gengenrtl.c (gendef): Likewise.
4521 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4522 * rtl.h (rtx_alloc): Remove macro.
4524 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4526 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4527 (build_tree_list_stat): Likewise.
4528 * tree.h (build_tree_list): Remove macro.
4529 (build_tree_list_vec): Likewise.
4531 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4533 * tree.c (make_vector_stat): Remove _stat from name.
4534 (build_vector_stat): Likewise.
4535 * tree.h (make_vector_stat): Remove macro.
4536 (build_vector_stat): Likewise.
4538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4540 * tree.h (build_var_debug_value): Remove prototype.
4542 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4544 * tree.c (tree_cons_stat): Remove _stat from name.
4545 * tree.h (tree_cons): Remove macro.
4547 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4549 * tree.c (build_vl_exp_stat): Remove _stat from name.
4550 * tree.h (build_vl_exp): Remove macro.
4552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4554 * tree.c (build_decl_stat): Remove _stat from name.
4555 * tree.h (build_decl): Remove macro.
4557 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4559 * gimple.c (gimple_build_with_ops_stat): Adjust.
4560 (gimple_alloc_stat): Remove _stat from name.
4561 * gimple.h (gimple_alloc): Remove macro.
4563 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4565 * tree.c (make_tree_vec_stat): Remove _stat from name.
4566 (grow_tree_vec_stat): Likewise.
4567 * tree.h (make_tree_vec_stat): Adjust prototype.
4568 (grow_tree_vec_stat): Likewise.
4569 (make_tree_vec): Remove macro.
4570 (grow_tree_vec): Likewise.
4572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4574 * fold-const.c (fold_build1_stat_loc): Adjust.
4575 (fold_build2_stat_loc): Likewise.
4576 (fold_build3_stat_loc): Likewise.
4577 * tree.c (build0_stat): Remove _stat from name.
4578 (build1_stat): Likewise.
4579 (build2_stat): Likewise.
4580 (build3_stat): Likewise.
4581 (build4_stat): Likewise.
4582 (build5_stat): Likewise.
4583 * tree.h (build1_loc): Remove macro, and rename _stat function
4585 (build2_loc): Likewise.
4586 (build3_loc): Likewise.
4587 (build4_loc): Likewise.
4588 (build5_loc): Likewise.
4590 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4592 * tree.c (make_int_cst_stat): Remove _stat from name.
4593 * tree.h (make_int_cst_stat): Adjust prototype.
4594 (make_int_cst): Remove macro.
4596 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4598 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4599 * tree.h (make_tree_binfo_stat): Adjust prototype.
4600 (make_tree_binfo): Remove.
4602 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4604 * tree.c (copy_node_stat): Rename to copy_node.
4605 (build_distinct_type_copy): Adjust.
4606 * tree.h (copy_node_stat): Adjust prototype.
4607 (copy_node): Remove macro.
4609 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4611 * tree.c (make_node_stat): rename to make_node.
4612 (build_tree_list_stat): Adjust.
4613 (build0_stat): Likewise.
4614 (build2_stat): Likewise.
4615 (build3_stat): Likewise.
4616 (build4_stat): Likewise.
4617 (build5_stat): Likewise.
4618 (build_decl_stat): Likewise.
4619 * tree.h (make_node_stat): Adjust prototype.
4620 (make_node): remove macro.
4622 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4624 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4625 (PPC_FEATURE2_SCV): Likewise.
4626 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4628 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4630 * config/aarch64/aarch64.c
4631 (aarch64_internal_mov_immediate): Add new special pattern.
4632 * config/aarch64/aarch64.md (*movdi_aarch64):
4633 Add reg/32bit const mov case.
4635 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4636 Richard Sandiford <richard.sandiford@linaro.org>
4638 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4639 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4640 Add integer and movi cases.
4641 (movi-split-hf-df-sf split, fp16): New.
4642 (enabled): Added TARGET_FP_F16INST.
4643 * config/aarch64/iterators.md (GPF_HF): New.
4644 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4646 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4648 * config/aarch64/aarch64.c
4649 (aarch64_simd_container_mode): Add prototype.
4650 (aarch64_expand_mov_immediate): Add HI support.
4651 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4652 (aarch64_can_const_movi_rtx_p): New.
4653 (aarch64_preferred_reload_class):
4654 Remove restrictions of using FP registers for certain SIMD operations.
4655 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4656 (aarch64_valid_floating_const): Add integer move validation.
4657 (aarch64_simd_imm_scalar_p): Remove.
4658 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4659 (aarch64_legitimate_constant_p): Expand list of supported cases.
4660 * config/aarch64/aarch64-protos.h
4661 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4662 (aarch64_reinterpret_float_as_int): New.
4663 (aarch64_simd_imm_scalar_p): Remove.
4664 * config/aarch64/constraints.md (Uvi): New.
4665 (Dd): Split into Ds and new Dd.
4666 * config/aarch64/aarch64.md (*movsi_aarch64):
4668 (*movdi_aarch64): Add SIMD mov case.
4670 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4672 * tree-predcom.c: (struct chain): Handle store-store chain in which
4673 stores for elimination only store loop invariant values.
4674 (execute_pred_commoning_chain): Ditto.
4675 (prepare_initializers_chain_store_elim): Ditto.
4676 (prepare_finalizers): Ditto.
4677 (is_inv_store_elimination_chain): New function.
4678 (initialize_root_vars_store_elim_1): New function.
4680 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4682 * tree-predcom.c: Revise general description of the pass.
4683 (enum chain_type): New enum type for store elimination.
4684 (struct chain): New field supporting store elimination.
4685 (struct component): Ditto.
4686 (dump_chain): Dump store-stores chain.
4687 (release_chain): Release resources.
4688 (split_data_refs_to_components): Compute and create component
4689 contains only stores for elimination.
4690 (get_chain_last_ref_at): New function.
4691 (make_invariant_chain): Initialization.
4692 (make_rooted_chain): Specify chain type in parameter and record it.
4693 (add_looparound_copies): Skip for store-stores chain.
4694 (determine_roots_comp): Compute type of chain and pass it to
4696 (initialize_root_vars_store_elim_2): New function.
4697 (finalize_eliminated_stores): New function.
4698 (remove_stmt): Handle store for elimination.
4699 (execute_pred_commoning_chain): Execute predictive commoning on
4701 (determine_unroll_factor): Skip unroll for store-stores chain.
4702 (prepare_initializers_chain_store_elim): New function.
4703 (prepare_initializers_chain): Hanlde store-store chain.
4704 (prepare_finalizers_chain, prepare_finalizers): New function.
4705 (tree_predictive_commoning_loop): Return integer value indicating
4706 if loop is unrolled or lcssa form is corrupted.
4707 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4709 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4711 * tree-predcom.c (initialize_root): Delete.
4712 (execute_pred_commoning_chain): Initialize root vars and replace
4713 reference of non-combined chain directly, rather than call above
4716 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4718 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4719 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4721 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4723 * tree-predcom.c (struct chain): New field init_seq.
4724 (release_chain): Release init_seq.
4725 (prepare_initializers_chain): Record intialization stmts in above
4727 (insert_init_seqs): New function.
4728 (tree_predictive_commoning_loop): Call insert_init_seqs.
4730 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4732 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4734 2017-07-28 Richard Biener <rguenther@suse.de>
4736 * match.pd: Remove superfluous :c.
4737 * genmatch.c (simplify::id): Add member.
4738 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4740 (current_id): New global.
4741 (dt_node::parent): Move from ...
4742 (dt_operand::parent): ... here. Add for_id member.
4743 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4744 (decision_tree::find_node): Relax order requirement when
4745 merging DT_TRUE nodes to ones inbetween the current simplify
4746 and the one we try to merge with. Add diagnostic whenever
4747 we need to enforce pattern order by not merging.
4748 (decision_tree::insert): Set current_id.
4749 (decision_tree::print_node): Dump parent node and for_id.
4750 (parser::last_id): Add member.
4751 (parser::push_simplify): Assign unique id.
4752 (parser::parser): Initialize last_id.
4754 2017-07-28 Martin Liska <mliska@suse.cz>
4757 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4758 gimple_build_debug_bind.
4760 2017-07-28 Richard Biener <rguenther@suse.de>
4762 PR tree-optimization/81502
4763 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4765 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4767 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4768 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4769 for BIT_FIELD_REF args.
4770 * fold-const.c (make_bit_field_ref): Likewise.
4771 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4773 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4776 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4777 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4778 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4779 Or it into SANITIZER_UNDEFINED.
4780 * ubsan.c: Include gimple-fold.h and varasm.h.
4781 (ubsan_expand_ptr_ifn): New function.
4782 (instrument_pointer_overflow): New function.
4783 (maybe_instrument_pointer_overflow): New function.
4784 (instrument_object_size): Formatting fix.
4785 (pass_ubsan::execute): Call instrument_pointer_overflow
4786 and maybe_instrument_pointer_overflow.
4787 * internal-fn.c (expand_UBSAN_PTR): New function.
4788 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4789 * sanitizer.def (__ubsan_handle_pointer_overflow,
4790 __ubsan_handle_pointer_overflow_abort): New builtins.
4791 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4792 * internal-fn.def (UBSAN_PTR): New internal function.
4793 * opts.c (sanitizer_opts): Add pointer-overflow.
4794 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4795 * fold-const.c (build_range_check): Compute pointer range check in
4796 integral type if pointer arithmetics would be needed. Formatting
4799 2017-07-28 Martin Liska <mliska@suse.cz>
4802 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4803 parameters that are of a variable-length.
4805 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4807 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4809 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4810 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4811 (CRT_CALL_STATIC_FUNCTION): Likewise.
4812 (ASM_DEFAULT_SPEC): New define.
4813 (ASM_SPEC32): Likewise.
4814 (ASM_SPEC64): Likewise.
4815 (ASM_SPEC_COMMON): Likewise.
4816 (ASM_SPEC): Likewise.
4817 (INVALID_64BIT): Likewise.
4818 (LINK_OS_DEFAULT_SPEC): Likewise.
4819 (LINK_OS_SPEC32): Likewise.
4820 (LINK_OS_SPEC64): Likewise.
4821 (POWERPC_LINUX): Likewise.
4822 (PTRDIFF_TYPE): Likewise.
4823 (RESTORE_FP_PREFIX): Likewise.
4824 (RESTORE_FP_SUFFIX): Likewise.
4825 (SAVE_FP_PREFIX): Likewise.
4826 (SAVE_FP_SUFFIX): Likewise.
4827 (SIZE_TYPE): Likewise.
4828 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4829 (TARGET_64BIT): Likewise.
4830 (TARGET_64BIT): Likewise.
4831 (TARGET_AIX): Likewise.
4832 (WCHAR_TYPE_SIZE): Likewise.
4833 (WCHAR_TYPE): Undefine.
4834 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4835 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4836 (CPP_OS_RTEMS_SPEC): Delete.
4837 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4838 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4840 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4842 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4844 PR tree-optimization/81578
4845 * tree-parloops.c (build_new_reduction): Bail out if
4846 reduction_code isn't one of the standard OpenMP reductions.
4847 Move the details printing after that decision.
4849 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4851 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4852 related to reload_in_progress.
4853 (splat_input_operand): Likewise.
4854 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4856 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4858 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4859 (TARGET_INSTANTIATE_DECLS): Likewise.
4860 (legitimate_indexed_address_p): Delete reload_in_progress code.
4861 (rs6000_debug_legitimate_address_p): Likewise.
4862 (rs6000_eliminate_indexed_memrefs): Likewise.
4863 (rs6000_emit_le_vsx_store): Likewise.
4864 (rs6000_emit_move_si_sf_subreg): Likewise.
4865 (rs6000_emit_move): Likewise.
4866 (register_to_reg_type): Likewise.
4867 (rs6000_pre_atomic_barrier): Likewise.
4868 (rs6000_machopic_legitimize_pic_address): Likewise.
4869 (rs6000_allocate_stack_temp): Likewise.
4870 (rs6000_address_for_fpconvert): Likewise.
4871 (rs6000_address_for_altivec): Likewise.
4872 (rs6000_secondary_memory_needed_rtx): Delete function.
4873 (rs6000_check_sdmode): Likewise.
4874 (rs6000_alloc_sdmode_stack_slot): Likewise.
4875 (rs6000_instantiate_decls): Likewise.
4876 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4877 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4878 Delete reload_in_progress.
4879 (*vec_reload_and_plus_<mptrsize>): Likewise.
4880 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4881 (vsx_div_v2di): Likewise.
4882 (vsx_udiv_v2di): Likewise.
4884 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4886 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4887 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4888 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4889 (rs6000_debug_reg_global): Delete print of LRA status.
4890 (rs6000_option_override_internal): Delete dead LRA related code.
4891 (rs6000_lra_p): Delete function.
4892 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4894 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4896 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4897 * config/riscv/rtems.h: New file.
4899 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4900 Sudakshina Das <sudi.das@arm.com>
4902 * config/aarch64/aarch64.md
4903 (define_split for and<mode>3nr_compare): Move
4904 non aarch64_logical_operand to a register.
4905 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4906 register immediate operand to a register.
4907 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4909 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4912 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4914 2017-07-27 Richard Biener <rguenther@suse.de>
4916 PR tree-optimization/81573
4917 PR tree-optimization/81494
4918 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4919 multi defuse cycle case.
4921 2017-07-27 Richard Biener <rguenther@suse.de>
4923 PR tree-optimization/81571
4924 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4927 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4929 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4930 earlier and only if MASK_FPU is set. Adjust formatting.
4932 2017-07-27 Martin Liska <mliska@suse.cz>
4934 * opt-functions.awk: Add validation of value of Init.
4935 * optc-gen.awk: Pass new argument.
4937 2017-07-27 Martin Liska <mliska@suse.cz>
4939 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4940 Fix wrong condition.
4942 2017-07-27 Martin Liska <mliska@suse.cz>
4944 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4945 BBs and edges seen by autoFDO.
4947 2017-07-27 Richard Biener <rguenther@suse.de>
4949 PR tree-optimization/81502
4950 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4951 with incompatible but same sized type.
4952 (execute_update_addresses_taken): Likewise.
4954 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4956 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4957 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4959 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4962 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4963 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4964 Change s_operand to memory_operand.
4966 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4968 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4969 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4970 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4971 Emit instructions rather than returning an expression. Handle TFmode
4972 and KFmode by casting to TImode.
4973 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4974 (rs6000_emit_le_vsx_store): Likewise.
4975 * config/rs6000/vsx.md (VSX_TI): New iterator.
4976 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4977 (*vsx_le_undo_permute_<mode>): Likewise.
4978 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4979 emit the split sequence.
4980 (*vsx_le_perm_store_<mode>): Likewise.
4982 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4984 PR tree-optimization/81555
4985 PR tree-optimization/81556
4986 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4987 if true, force CHANGED for the recursive invocation.
4988 (reassociate_bb): Remember original length of ops array, pass
4989 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4991 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4992 attributes for noipa attribute. For naked attribute use
4993 lookup_attribute first before lookup_attribute_spec.
4994 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4996 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4997 for functions with noipa attribute.
4998 (cgraph_externally_visible_p): Return true for functions with noipa
5000 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
5001 for functions with noipa attribute.
5002 * doc/extend.texi: Document noipa function attribute.
5003 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
5004 also for functions with noipa attribute.
5005 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
5007 2017-07-26 Andrew Pinski <apinski@cavium.com>
5009 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
5010 vec_unalign_load_cost and vec_unalign_store_cost.
5012 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5014 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
5015 -mvsx-small-integer option.
5016 (ISA_3_0_MASKS_IEEE): Likewise.
5017 (OTHER_VSX_VECTOR_MASKS): Likewise.
5018 (POWERPC_MASKS): Likewise.
5019 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
5020 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
5021 code, only testing for DImode being allowed in non-VSX floating
5023 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
5024 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
5026 (rs6000_option_override_internal): Delete -mvsx-small-integer.
5027 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
5028 TARGET_P8_VECTOR test.
5029 (rs6000_secondary_reload_simple_move): Likewise.
5030 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
5031 since TARGET_P9_VECTOR was already tested.
5032 (rs6000_opt_masks): Remove -mvsx-small-integer.
5033 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
5034 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5036 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
5037 test for TARGET_VEXTRACTUB was used, and that uses
5039 (p9 extract splitter): Likewise.
5040 (vsx_extract_<mode>_di_p9): Likewise.
5041 (vsx_extract_<mode>_store_p9): Likewise.
5042 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
5043 for TARGET_P9_VECTOR was used. Delete code that is now dead with
5044 the elimination of TARGET_VSX_SMALL_INTEGER.
5045 (vsx_extract_<mode>_p8): Likewise.
5046 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
5047 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
5048 (vsx_set_<mode>_p9): Likewise.
5049 (vsx_set_v4sf_p9): Likewise.
5050 (vsx_set_v4sf_p9_zero): Likewise.
5051 (vsx_insert_extract_v4sf_p9): Likewise.
5052 (vsx_insert_extract_v4sf_p9_2): Likewise.
5053 * config/rs6000/rs6000.md (sign extend splitter): Change
5054 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
5055 (floatsi<mode>2_lfiwax_mem): Likewise.
5056 (floatunssi<mode>2_lfiwzx_mem): Likewise.
5057 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
5058 since a test for TARGET_P9_VECTOR was used.
5059 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5060 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
5061 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5062 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5063 TARGET_P8_VECTOR test.
5064 (fix_trunc<mode>si2_stfiwx): Likewise.
5065 (fix_trunc<mode>si2_internal): Likewise.
5066 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5067 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5069 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5070 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5071 TARGET_P8_VECTOR test.
5072 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5073 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5074 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5076 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5077 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5078 since a test for TARGET_P9_VECTOR was used.
5079 (splitter for loading small constants): Likewise.
5081 2017-07-26 Andrew Pinski <apinski@cavium.com>
5083 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5086 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5089 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5090 (fp_valid_at): Likewise.
5092 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5094 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5095 (qdf24xx_addrcost_table): Likewise.
5096 (cortexa57_tunings): Update to use generic_branch_cost.
5097 (cortexa72_tunings): Likewise.
5098 (cortexa73_tunings): Likewise.
5099 (qdf24xx_tunings): Likewise.
5101 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5103 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5104 (thunderx2t99_branch_cost): Likewise.
5105 (cortexa35_tunings): Update to use generic_branch_cost.
5106 (cortexa53_tunings): Likewise.
5107 (cortexa57_tunings): Likewise.
5108 (cortexa72_tunings): Likewise.
5109 (cortexa73_tunings): Likewise.
5110 (thunderx2t99_tunings): Likewise.
5112 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5114 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5115 (sparc_option_override): Honour MASK_FSMULD.
5116 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5117 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5118 * config/sparc/sparc.opt (mfsmuld): New option.
5119 * doc/invoke.texi (mfsmuld): Document option.
5121 2017-07-26 Marek Polacek <polacek@redhat.com>
5124 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5126 2017-07-26 Richard Biener <rguenther@suse.de>
5128 * gimple-match-head.c (do_valueize): Return OP if valueize
5130 (get_def): New helper to get at the def stmt of a SSA name
5132 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5133 do_valueize to get at the def stmt.
5134 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5136 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5139 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5141 2017-07-26 Martin Liska <mliska@suse.cz>
5144 * function.c (expand_function_start): Make expansion of
5145 nonlocal_goto_save_area after parm_birth_insn.
5147 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5149 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5150 from all CPU target flags enable members.
5152 2017-07-26 Richard Biener <rguenther@suse.de>
5154 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5155 (decision_tree::gen): Make 'type' const.
5156 (write_predicate): Likewise.
5158 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5160 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5161 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5162 (rs6000_option_override_internal): Likewise.
5163 (rs6000_expand_vector_set): Likewise.
5164 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5165 (TARGET_UPPER_REGS_SF): Likewise.
5166 (TARGET_UPPER_REGS_DI): Likewise.
5167 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5168 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5169 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5170 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5171 (Splitters for DI constants in Altivec registers): Likewise.
5172 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5173 (vsx_set_v4sf_p9): Likewise.
5174 (vsx_set_v4sf_p9_zero): Likewise.
5175 (vsx_insert_extract_v4sf_p9): Likewise.
5176 (vsx_insert_extract_v4sf_p9_2): Likewise.
5178 2017-07-25 Carl Love <cel@us.ibm.com>
5180 * doc/extend.texi: Update the built-in documentation file for the
5181 existing built-in functions
5182 vector signed char vec_cnttz (vector signed char);
5183 vector unsigned char vec_cnttz (vector unsigned char);
5184 vector signed short vec_cnttz (vector signed short);
5185 vector unsigned short vec_cnttz (vector unsigned short);
5186 vector signed int vec_cnttz (vector signed int);
5187 vector unsigned int vec_cnttz (vector unsigned int);
5188 vector signed long long vec_cnttz (vector signed long long);
5189 vector unsigned long long vec_cnttz (vector unsigned long long);
5191 2017-07-25 Andrew Pinski <apinski@cavium.com>
5193 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5194 accesses where the use is for the first operand of a BIT_INSERT.
5196 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5199 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5200 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5202 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5204 * config/i386/gstabs.h: Delete.
5205 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5207 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5209 * config/i386/i386.c (ix86_decompose_address): Do not check for
5210 register RTX when looking at index_reg or base_reg.
5211 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5213 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5215 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5216 to update EH info here.
5218 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5220 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5222 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5224 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5226 2017-07-25 Torsten Duwe <duwe@suse.de>
5228 * common.opt: Introduce -fpatchable-function-entry
5229 command line option, and its variables function_entry_patch_area_size
5230 and function_entry_patch_area_start.
5231 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5232 including a two-value parser.
5233 * target.def (print_patchable_function_entry): New target hook.
5234 * targhooks.h (default_print_patchable_function_entry): New function.
5235 * targhooks.c (default_print_patchable_function_entry): Likewise.
5236 * toplev.c (process_options): Switch off IPA-RA if
5237 patchable function entries are being generated.
5238 * varasm.c (assemble_start_function): Look at the
5239 patchable-function-entry command line switch and current
5240 function attributes and maybe generate NOP instructions by
5241 calling the print_patchable_function_entry hook.
5242 * doc/extend.texi: Document patchable_function_entry attribute.
5243 * doc/invoke.texi: Document -fpatchable_function_entry
5244 command line option.
5245 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5247 * doc/tm.texi: Re-generate.
5249 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5252 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5253 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5255 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5257 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5259 2017-07-25 Richard Biener <rguenther@suse.de>
5261 PR tree-optimization/81455
5262 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5263 not walk in cycles when looking for guards.
5265 2017-07-25 Richard Biener <rguenther@suse.de>
5267 PR tree-optimization/81529
5268 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5269 when optimizing backedge uses.
5271 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5273 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5275 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5276 to dl_section_ref. On AIX, append an expression to subtract
5277 the size of the section length to dl_section_ref.
5279 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5281 * configure.ac: If any of the config.* scripts fail, exit 1.
5282 * configure: Regenerate.
5284 2017-07-25 Richard Biener <rguenther@suse.de>
5287 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5288 of immediate uses, be more verbose on errors.
5290 2017-07-25 Richard Biener <rguenther@suse.de>
5292 PR tree-optimization/81510
5293 * tree-vect-loop.c (vect_is_simple_reduction): When the
5294 reduction stmt is not inside the loop bail out.
5296 2017-07-25 Richard Biener <rguenther@suse.de>
5298 PR tree-optimization/81303
5299 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5300 profitability check against LOOP_VINFO_NITERSM1.
5302 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5304 * domwalk.c (cmp_bb_postorder): Simplify.
5305 (sort_bbs_postorder): New function. Use it...
5306 (dom_walker::walk): ...here to optimize common cases.
5308 2017-07-25 Martin Liska <mliska@suse.cz>
5311 * ipa-visibility.c (function_and_variable_visibility): Make the
5312 redirection just on target that supports aliasing.
5313 Fix GNU coding style.
5315 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5318 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5320 * config/arm/rtems.h: Likewise.
5321 * config/bfin/rtems.h: Likewise.
5322 * config/i386/rtemself.h: Likewise.
5323 * config/lm32/rtems.h: Likewise.
5324 * config/m32c/rtems.h: Likewise.
5325 * config/m68k/rtemself.h: Likewise.
5326 * config/microblaze/rtems.h: Likewise.
5327 * config/mips/rtems.h: Likewise.
5328 * config/moxie/rtems.h: Likewise.
5329 * config/nios2/rtems.h: Likewise.
5330 * config/powerpcspe/rtems.h: Likewise.
5331 * config/rs6000/rtems.h: Likewise.
5332 * config/rtems.h: Likewise.
5333 * config/sh/rtems.h: Likewise.
5334 * config/sh/rtemself.h: Likewise.
5335 * config/sparc/rtemself.h: Likewise.
5337 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5340 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5341 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5342 * tree-ssa-structalias.c (alias_get_name): Same.
5344 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5347 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5348 instructions if no du chain is found.
5350 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5352 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5354 2017-07-25 Richard Biener <rguenther@suse.de>
5357 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5360 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5362 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5364 (ISA_2_7_MASKS_SERVER): Likewise.
5365 (ISA_3_0_MASKS_IEEE): Likewise.
5366 (OTHER_P8_VECTOR_MASKS): Likewise.
5367 (OTHER_VSX_VECTOR_MASKS): Likewise.
5368 (POWERPC_MASKS): Likewise.
5369 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5370 duplicate list of options.
5371 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5372 explicit -mupper-regs options.
5373 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5374 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5375 alias for -mupper-regs-df.
5376 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5377 (rs6000_init_hard_regno_mode_ok): Likewise.
5378 (rs6000_option_override_internal): Likewise.
5379 (rs6000_opt_masks): Likewise.
5380 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5381 options in terms of whether -mvsx or -mpower8-vector was used.
5382 (TARGET_UPPER_REGS_DI): Likewise.
5383 (TARGET_UPPER_REGS_SF): Likewise.
5384 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5385 -mupper-regs-* options.
5387 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5389 * passes.c (emergency_dump_function): Print some empty lines and a
5390 header before the RTL dump.
5392 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5394 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5396 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5399 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5400 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5402 2017-07-24 Carl Love <cel@us.ibm.com>
5404 * config/rs6000/rs6000-c.c: Add support for built-in functions
5405 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5406 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5407 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5408 vec_extract_fp_from_shortl): Add defines for the two builtins.
5409 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5410 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5412 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5413 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5414 * doc/extend.texi: Update the built-in documentation file for the
5415 new built-in function.
5417 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5420 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5422 * doc/generic.texi: Likewise.
5423 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5424 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5426 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5428 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5429 (aarch64_mls_elt_merge<mode>): Likewise.
5431 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5433 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5434 having __cxa_atexit.
5436 2017-07-23 Michael Collison <michael.collison@arm.com>
5438 * config/arm/arm.c (arm_option_override): Deprecate
5439 use of -mstructure-size-boundary.
5440 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5441 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5443 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5446 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5447 Reduce cost estimate for direct moves.
5449 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5452 * config/i386/i386.c (ix86_option_override_internal): Disable
5453 BMI, BMI2 and TBM instructions for -m16.
5455 2017-07-21 Carl Love <cel@us.ibm.com>
5457 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5458 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5459 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5460 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5461 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5462 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5463 VMULOSW): New enum "unspec" values.
5464 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5465 altivec_vmulosw): New patterns.
5466 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5467 VMULOSW): Add definitions.
5469 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5471 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5472 (qdf24xx): Likewise.
5473 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5474 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5475 (AARCH64_FL_V8_1): Renumber.
5476 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5477 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5478 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5479 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5480 rdma to feature modifiers list.
5482 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5485 * ipa-visibility (function_and_variable_visibility): Convert
5486 recursive PLT call to direct call if appropriate.
5488 2017-07-21 Andrew Pinski <apinski@cavium.com>
5490 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5491 operand 1 to see if the types precision matches.
5492 * fold-const.c (operand_equal_p): Likewise.
5494 2017-07-21 Richard Biener <rguenther@suse.de>
5496 PR tree-optimization/81303
5497 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5498 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5499 (vect_peeling_hash_get_lowest_cost): Adjust.
5500 (vect_enhance_data_refs_alignment): Likewise. Use
5501 vect_get_peeling_costs_all_drs to compute the penalty for no
5502 peeling to match up costs.
5504 2017-07-21 Richard Biener <rguenther@suse.de>
5506 PR tree-optimization/81500
5507 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5508 we didn't identify a reduction path.
5510 2017-07-21 Tom de Vries <tom@codesourcery.com>
5511 Cesar Philippidis <cesar@codesourcery.com>
5513 PR gcov-profile/81442
5514 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5517 2017-07-21 Tom de Vries <tom@codesourcery.com>
5520 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5522 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5523 nvptx_override_options_after_change.
5525 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5527 * dwarf2out.c (output_file_names): Avoid double testing for
5530 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5532 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5534 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5536 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5538 (try_crossjump_to_edge): Do not punt on partitioned functions.
5540 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5542 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5543 Put all BBs reachable only via paths crossing cold region to cold
5545 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5547 2016-07-21 Richard Biener <rguenther@suse.de>
5549 PR tree-optimization/81303
5550 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5551 into account prologue and epilogue iterations when raising
5552 min_profitable_iters to sth at least covering one vector iteration.
5554 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5556 * config/arm/arm.c (arm_test_cpu_arch_dat):
5559 2017-07-20 Nathan Sidwell <nathan@acm.org>
5561 Remove TYPE_METHODS.
5562 * tree.h (TYPE_METHODS): Delete.
5563 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5564 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5565 (dbxout_type_methods): Scan TYPE_FIELDS.
5566 (dbxout_type): Don't check TYPE_METHODS here.
5567 * function.c (use_register_for_decl): Always ignore register for
5568 class types when not optimizing.
5569 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5570 * tree.c (free_lang_data_in_type): Stitch out member functions and
5571 templates from TYPE_FIELDS.
5572 (build_distinct_type_copy, verify_type_variant,
5573 verify_type): Member fns are on TYPE_FIELDS.
5574 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5575 * tree-pretty-print.c (dump_generic_node): Likewise.
5577 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5580 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5581 V2TImode and V4TImode.
5582 (ix86_expand_vector_extract): Likewise.
5583 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5584 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5585 (ssescalarmode): Handle V4TImode and V2TImode.
5586 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5587 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5588 (VEXTRACTI128_MODE): New mode iterator.
5589 (splitter for *vec_extractv?ti first element): New.
5590 (VEC_INIT_MODE): New mode iterator.
5591 (vec_init<mode>): Consolidate 3 expanders into one using
5592 VEC_INIT_MODE mode iterator.
5594 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5596 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5597 non_spilled_static_chain_regno_p.
5599 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5601 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5603 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5605 * bb-reorder.c (connect_traces): Allow copying of blocks within
5608 2017-07-20 Richard Biener <rguenther@suse.de>
5610 * gimple.h (gimple_phi_result): Add gphi * overload.
5611 (gimple_phi_result_ptr): Likewise.
5612 (gimple_phi_arg): Likewise. Adjust index assert to only
5613 allow actual argument accesses rather than all slots available
5615 (gimple_phi_arg_def): Add gphi * overload.
5616 * tree-phinodes.c (make_phi_node): Initialize only actual
5618 (resize_phi_node): Clear memory not covered by old node,
5619 do not initialize excess argument slots.
5620 (reserve_phi_args_for_new_edge): Initialize new argument slot
5623 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5625 PR tree-optimization/81388
5627 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5629 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5630 by removing computation of may_be_zero.
5632 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5633 Tom de Vries <tom@codesourcery.com>
5636 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5637 when gimple level profile disagrees with what RTL expander did.
5639 2017-07-20 Richard Biener <rguenther@suse.de>
5641 PR tree-optimization/61171
5642 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5643 (vect_analyze_stmt): Add slp instance parameter.
5644 (vectorizable_reduction): Likewise.
5645 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5646 (vect_is_simple_reduction): Deal with chains not detected
5647 as SLP reduction chain, specifically not properly associated
5648 chains containing a mix of plus/minus.
5649 (get_reduction_op): Remove.
5650 (get_initial_defs_for_reduction): Simplify, pass in whether
5651 this is a reduction chain, pass in the SLP node for the PHIs.
5652 (vect_create_epilog_for_reduction): Get the SLP instance as
5654 (vectorizable_reduction): Get the SLP instance as arg.
5655 During analysis remember the SLP node with the PHIs in the
5656 instance. Simplify getting at the vectorized reduction PHIs.
5657 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5658 through SLP instance.
5659 (vect_slp_analyze_operations): Likewise.
5660 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5661 (vect_transform_stmt): Likewise.
5663 2017-07-20 Tom de Vries <tom@codesourcery.com>
5665 PR tree-optimization/81489
5666 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5667 read of phi arg location to before loop that modifies phi.
5669 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5671 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5674 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5677 * except.c (execute): Fix ordering issue.
5679 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5681 PR rtl-optimization/81423
5682 * combine.c (make_compound_operation_int): Don't try to optimize
5683 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5685 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5687 PR rtl-optimization/81423
5688 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5689 with a constant that is -1 in the truncated to mode.
5691 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5693 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5694 (determine_unlikely_bbs): ... here.
5695 * predict.h (propagate_unlikely_bbs_forward): Declare.
5696 * cfgexpand.c (pass_expand::execute): Use it.
5697 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5699 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5700 propagate_unlikely_bbs_forward.
5702 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5705 * except.c (maybe_add_nop_after_section_switch): New function.
5708 2017-07-19 Tom de Vries <tom@codesourcery.com>
5710 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5712 2017-07-19 Tom de Vries <tom@codesourcery.com>
5714 * gimple.h (gimple_phi_arg): Make assert more strict.
5716 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5718 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5719 * config/rs6000/mmintrin.h: New file.
5720 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5722 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5724 PR tree-optimization/81346
5725 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5727 2017-07-19 Tom de Vries <tom@codesourcery.com>
5729 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5731 2017-07-19 Tom de Vries <tom@codesourcery.com>
5733 * config/nvptx/nvptx-modes.def: Add V2DImode.
5734 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5735 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5736 (nvptx_output_mov_insn): Handle lack of mov.b128.
5737 (nvptx_print_operand): Handle 'H' and 'L' codes.
5738 (nvptx_vector_mode_supported): Allow V2DImode.
5739 (nvptx_preferred_simd_mode): New function.
5740 (nvptx_data_alignment): New function.
5741 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5742 nvptx_preferred_simd_mode.
5743 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5745 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5747 2017-07-19 Tom de Vries <tom@codesourcery.com>
5749 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5750 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5751 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5752 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5753 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5754 (mov<VECIM>_insn): New define_insn.
5755 (define_expand "mov<VECIM>): New define_expand.
5757 2017-07-19 Tom de Vries <tom@codesourcery.com>
5759 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5761 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5763 PR tree-optimization/81346
5764 * fold-const.h (fold_div_compare, range_check_type): Declare.
5765 * fold-const.c (range_check_type): New function.
5766 (build_range_check): Use range_check_type.
5767 (fold_div_compare): No longer static, rewritten into
5768 a match.pd helper function.
5769 (fold_comparison): Don't call fold_div_compare here.
5770 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5773 2017-07-19 Nathan Sidwell <nathan@acm.org>
5775 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5776 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5777 * tree.c (find_decls_types_r, verify_type): Use
5778 TYPE_{MIN,MAX}_VALUE_RAW.
5779 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5780 (hash_tree): Likewise.
5781 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5783 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5786 2017-07-18 Tom de Vries <tom@codesourcery.com>
5789 * omp-expand.c (expand_omp_for_static_chunk): Handle
5790 equal-argument loop exit phi.
5792 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5795 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5796 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5797 operand 2 predicate.
5798 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5799 operand 2 predicate.
5800 (ror,rol -> rorx splitters): Use const_int_operand as
5801 operand 2 predicate.
5803 2017-06-18 Richard Biener <rguenther@suse.de>
5805 PR tree-optimization/81410
5806 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5807 the gap in the ! slp_perm SLP case after each group.
5809 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5812 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5815 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5818 * predict.c (set_even_probabilities): Cleanup; do not affect
5819 probabilities that are already known.
5820 (combine_predictions_for_bb): Call even when count is set.
5822 2017-07-18 Nathan Sidwell <nathan@acm.org>
5824 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5827 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5830 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5831 optimization for loop niter analysis.
5833 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5836 * config/avr/avr.c (avr_optimize_casesi): Don't use
5837 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5839 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5841 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5842 body_cost_vec from _vect_peel_extended_info.
5843 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5844 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5847 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5849 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5851 2017-07-18 Richard Biener <rguenther@suse.de>
5853 PR tree-optimization/80620
5854 PR tree-optimization/81403
5855 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5856 info when re-using a VN table entry.
5858 2017-07-18 Richard Biener <rguenther@suse.de>
5860 PR tree-optimization/81418
5861 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5862 vectype_in. Verify that with lane-reducing reduction operations
5863 we have a single def-use cycle.
5865 2017-07-17 Carl Love <cel@us.ibm.com>
5867 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5869 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5870 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5871 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5872 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5873 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5874 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5875 VMULOSW): New enum "unspec" values.
5876 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5877 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5878 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5879 altivec_vmulosw): New patterns.
5880 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5881 VMULOSW): Add definitions.
5882 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5884 * config/alpha/alpha.c: Include predict.h.
5886 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5888 * tree-vrp.c (compare_assert_loc): Fix comparison function
5889 to return predictable results.
5891 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5893 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5896 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5897 * doc/invoke.texi (mexpand-adddi): Update text.
5899 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5901 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5902 that also clobbers the CC register. The old expand code is moved
5904 (*arc_clzsi2): ... here.
5905 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5906 the CC register. The old expand code is moved to ...
5907 (arc_ctzsi2): ... here.
5909 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5911 * config/arc/arc.opt (mindexed-loads): Use initial value
5912 TARGET_INDEXED_LOADS_DEFAULT.
5913 (mauto-modify-reg): Use initial value
5914 TARGET_AUTO_MODIFY_REG_DEFAULT.
5915 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5916 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5917 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5918 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5920 2017-07-17 Martin Liska <mliska@suse.cz>
5923 * opts.c (finish_options): Do not allow -fgnu-tm
5924 w/ -fsanitize={kernel-,}address. Say sorry.
5926 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5929 * tree-loop-distribution.c (classify_partition): Only assert on
5930 numer of iterations.
5931 (merge_dep_scc_partitions): Delete prameter. Update function call.
5932 (distribute_loop): Remove code handling loop with unknown niters.
5933 (pass_loop_distribution::execute): Skip loop with unknown niters.
5935 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5938 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5939 function sort_partitions_by_post_order.
5941 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5943 PR tree-optimization/81374
5944 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5945 the max index of basic blocks, rather than number of basic blocks.
5947 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5949 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5951 (arc_legitimate_pic_operand_p): Likewise.
5952 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5954 (arc_needs_pcl_p): Likewise.
5955 (arc_legitimate_pc_offset_p): Likewise.
5956 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5957 function is also used in constrains.md.
5958 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5959 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5960 PLUS. Only return true/false in known cases, otherwise assert.
5961 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5962 is already called in arc_legitimate_constant_p.
5963 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5965 (LEGITIMATE_PIC_OPERAND_P): Use
5966 arc_raw_symbolic_reference_mentioned_p function.
5967 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5972 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5973 Andrew Burgess <andrew.burgess@embecosm.com>
5975 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5976 (arc_return_address_register): New function.
5977 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5978 (arc_handle_fndecl_attribute): Add naked attribute.
5979 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5980 (TARGET_WARN_FUNC_RETURN): Likewise.
5981 (arc_allocate_stack_slots_for_args): New function.
5982 (arc_warn_func_return): Likewise.
5983 (machine_function): Change type fn_type.
5984 (arc_compute_function_type): Consider new naked function type,
5985 change function return type.
5986 (arc_must_save_register): Adapt to handle new
5987 arc_compute_function_type's return type.
5988 (arc_expand_prologue): Likewise.
5989 (arc_expand_epilogue): Likewise.
5990 (arc_return_address_regs): Delete.
5991 (arc_return_address_register): New function.
5992 (arc_epilogue_uses): Use above function.
5993 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5994 (arc_function_type): Change encoding, add naked type.
5995 (ARC_INTERRUPT_P): Change to handle the new encoding.
5996 (ARC_FAST_INTERRUPT_P): Likewise.
5997 (ARC_NORMAL_P): Define.
5998 (ARC_NAKED_P): Likewise.
5999 (arc_compute_function_type): Delete prototype.
6000 * config/arc/arc.md (in_ret_delay_slot): Use
6001 arc_return_address_register function.
6002 (simple_return): Likewise.
6003 (p_return_i): Likewise.
6005 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6007 PR tree-optimization/81428
6008 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
6009 can't be built for those types.
6011 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6013 Remove stuff dead since r239246.
6015 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
6016 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
6017 (avr_inform_devices): Remove dead stuff.
6019 2017-07-17 Tamar Christina <tamar.christina@arm.com>
6021 * config/arm/arm_neon.h: Fix softp typo.
6023 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6025 PR tree-optimization/81365
6026 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
6027 aggregate moves onto bb predecessor edges, make sure there are no
6028 loads that could alias the lhs in between the start of bb and the
6031 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6034 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
6035 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
6036 [LSHIFTRT, outer_code = TRUNCATE]: Same.
6038 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6040 PR tree-optimization/81396
6041 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
6042 (init_symbolic_number): Initialize it to 1.
6043 (perform_symbolic_merge): Add n_ops from both operands into the new
6045 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
6046 without base_addr as useless if they need more than one operation.
6047 (bswap_replace): Handle !bswap case for NULL base_addr.
6049 2017-07-17 Tom de Vries <tom@codesourcery.com>
6052 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
6055 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
6057 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
6058 conditional builtin define __FIX_LEON3FT_B2BST.
6060 2017-07-17 Daniel Cederman <cederman@gaisler.com>
6062 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
6063 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6066 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6068 PR rtl-optimization/81424
6069 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6070 to remove potential trapping from operands if -fnon-call-exceptions.
6072 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6074 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6075 profile_proability for scalling.
6076 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6078 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6080 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6082 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6084 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6085 fixpoint arithmetics.
6087 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6089 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6090 fixpoint arithmetics.
6092 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6094 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6095 fixpoint arithmetics.
6097 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6099 * profile-count.h (profile_probability::from_reg_br_prob_note,
6100 profile_probability::to_reg_br_prob_note): New functions.
6101 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6102 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6103 * predict.c (probability_reliable_p): Update.
6104 (edge_probability_reliable_p): Update.
6105 (br_prob_note_reliable_p): Update.
6106 (invert_br_probabilities): Update.
6107 (add_reg_br_prob_note): New function.
6108 (combine_predictions_for_insn): Update.
6109 * asan.c (asan_clear_shadow): Update.
6110 * cfgbuild.c (compute_outgoing_frequencies): Update.
6111 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6112 (update_br_prob_note): Update.
6113 (rtl_verify_edges): Update.
6114 (purge_dead_edges): Update.
6115 (fixup_reorder_chain): Update.
6116 * emit-rtl.c (try_split): Update.
6117 * ifcvt.c (cond_exec_process_insns): Update.
6118 (cond_exec_process_if_block): Update.
6119 (dead_or_predicable): Update.
6120 * internal-fn.c (expand_addsub_overflow): Update.
6121 (expand_neg_overflow): Update.
6122 (expand_mul_overflow): Update.
6123 * loop-doloop.c (doloop_modify): Update.
6124 * loop-unroll.c (compare_and_jump_seq): Update.
6125 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6126 * predict.h: Update.
6127 * reorg.c (mostly_true_jump): Update.
6129 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6130 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6131 * config/arc/arc.c: (emit_unlikely_jump): Update.
6132 * config/arm/arm.c: (emit_unlikely_jump): Update.
6133 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6134 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6135 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6136 (ix86_print_operand): Update.
6137 (ix86_split_fp_branch): Update.
6138 (predict_jump): Update.
6139 * config/ia64/ia64.c (ia64_print_operand): Update.
6140 * config/mmix/mmix.c (mmix_print_operand): Update.
6141 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6142 (rs6000_expand_split_stack_prologue): Update.
6143 * config/rs6000/rs6000.c: Update.
6144 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6145 (s390_expand_vec_movstr): Update.
6146 (s390_expand_cs_tdsi): Update.
6147 (s390_expand_split_stack_prologue): Update.
6148 * config/sh/sh.c (sh_print_operand): Update.
6149 (expand_cbranchsi4): Update.
6150 (expand_cbranchdi4): Update.
6151 * config/sparc/sparc.c (output_v9branch): Update.
6152 * config/spu/spu.c (get_branch_target): Update.
6153 (ea_load_store_inline): Update.
6154 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6155 * config/tilepro/tilepro.c: Update.
6157 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6159 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6160 (gimplify_save_expr): Likewise.
6162 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6164 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6166 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6168 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6170 * timevar.def (TV_IPA_FNSUMMARY): Define.
6172 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6174 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6175 to back store errata sensitive sequence from being generated.
6176 (sqrtdf2_fix): Likewise.
6178 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6180 * tree-ssa-threadupdate.c (compute_path_counts,
6181 update_joiner_offpath_counts): Use profile_probability.
6183 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
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 Kelvin Nilsen <kelvin@gcc.gnu.org>
6193 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6194 array entries to represent __ieee128 versions of the
6195 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6196 scalar_extract_sig, and scalar_insert_exp built-in functions.
6197 (altivec_resolve_overloaded_builtin): Add special case handling
6198 for the __builtin_scalar_insert_exp function, as represented by
6199 the P9V_BUILTIN_VEC_VSIEDP constant.
6200 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6201 exponent support for __ieee128 argument.
6202 (VSESQP): Add scalar extract signature support for __ieee128
6204 (VSTDCNQP): Add scalar test negative support for __ieee128
6206 (VSIEQP): Add scalar insert exponent support for __int128 argument
6207 with __ieee128 result.
6208 (VSIEQPF): Add scalar insert exponent support for __ieee128
6209 argument with __ieee128 result.
6210 (VSTDCQP): Add scalar test data class support for __ieee128
6212 (VSTDCNQP): Add overload support for scalar test negative with
6214 (VSTDCQP): Add overload support for scalar test data class
6216 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6218 (UNSPEC_VSX_SIEXPQP): New constant.
6219 (xsxexpqp): New insn for VSX scalar extract exponent quad
6221 (xsxsigqp): New insn for VSX scalar extract significand quad
6223 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6224 precision with floating point argument.
6225 (xststdcqp): New expand for VSX scalar test data class quad
6227 (xststdcnegqp): New expand for VSX scalar test negative quad
6229 (xststdcqp): New insn to match expansions for VSX scalar test data
6230 class quad precision and VSX scalar test negative quad precision.
6231 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6232 special case operand checking to enforce that second operand of
6233 VSX scalar test data class with quad precision argument is a 7-bit
6235 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6236 prototypes and descriptions of __ieee128 versions of
6237 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6238 scalar_test_data_class, and scalar_test_neg built-in functions.
6240 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6242 PR tree-optimization/81162
6243 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6244 replace a negate with an add.
6246 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6248 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6250 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6252 * config/arm/arm-c.c (arm_cpu_builtins): Define
6253 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6255 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6257 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6258 (armv8-r): Set ARM Cortex-R52 as default CPU.
6259 * config/arm/arm-tables.opt: Regenerate.
6260 * config/arm/arm-tune.md: Regenerate.
6261 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6263 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6264 extension for -mcpu=cortex-r52.
6266 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6268 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6269 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6270 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6271 (fp-armv8): Define it as FP_ARMv8 only.
6272 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6273 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6275 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6276 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6277 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6278 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6279 than TARGET_FPU_ARMV8.
6280 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6281 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6282 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6284 * config/arm/neon.md (neon_vrint): Likewise.
6285 (neon_vcvt): Likewise.
6286 (neon_<fmaxmin_op><mode>): Likewise.
6287 (<fmaxmin><mode>3): Likewise.
6288 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6289 * config/arm/predicates.md (arm_cond_move_operator): Check against
6290 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6292 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6294 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6297 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6299 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6300 loop in comment with memset.
6302 2017-07-14 Martin Liska <mliska@suse.cz>
6304 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6305 * dwarf2out.c (is_java): Remove the function.
6306 (output_pubname): Remove usage of the function.
6307 (lower_bound_default): Remove usage of DW_LANG_Java.
6308 (gen_compile_unit_die): Likewise.
6309 * gcc.c: Remove compiler defaults for .java and .zip files.
6310 * gimple-expr.c (remove_suffix): Change as there's no longer
6311 extension than 4-letter one.
6312 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6313 (gimplify_save_expr): Likewise.
6314 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6315 as it's possible even for other languages than Java.
6316 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6317 * lto-opts.c (lto_write_options): Remove reference to Java.
6318 * opts.c (strip_off_ending): Update file extension handling.
6319 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6320 * tree-eh.c (lower_resx): Likewise.
6321 * tree.c (free_lang_data_in_type): Remove dead code.
6322 (find_decls_types_r): Likewise.
6323 (build_common_builtin_nodes): Remove Java from a comment.
6324 (verify_type): Remove dead code.
6325 * varasm.c (assemble_external): Remove Java from a comment.
6327 2017-07-14 Martin Liska <mliska@suse.cz>
6329 * opts.c (finish_options): Add quotes.
6330 (common_handle_option): Likewise.
6332 2017-07-14 Martin Liska <mliska@suse.cz>
6334 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6335 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6337 * dwarf2out.c (lower_bound_default): Do not handle
6339 (gen_compile_unit_die): Likewise.
6340 * gcc.c: Remove default extension binding for GNU Pascal.
6341 * stmt.c: Remove Pascal language from a comment.
6342 * xcoffout.c: Likewise.
6344 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6347 * diagnostic-show-locus.c (fixit_cmp): New function.
6348 (layout::layout): Sort m_fixit_hints.
6349 (column_range::column_range): Assert that the values are valid.
6350 (struct char_span): New struct.
6351 (correction::overwrite): New method.
6352 (struct source_line): New struct.
6353 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6354 calls in terms of classes source_line and char_span, and
6355 correction::overwrite.
6356 (selftest::test_overlapped_fixit_printing_2): New function.
6357 (selftest::diagnostic_show_locus_c_tests): Call it.
6359 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6362 early if there is no lhs.
6364 2017-07-13 Martin Liska <mliska@suse.cz>
6366 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6367 (gen_reference_type_die): Likewise.
6368 * stor-layout.c: Remove Pascal-related comment.
6370 2017-07-13 Martin Liska <mliska@suse.cz>
6372 * opts.c (finish_options): Add quotes to error messages.
6373 (parse_sanitizer_options): Likewise.
6375 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6377 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6379 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6381 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6383 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6385 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6387 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6389 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6392 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6393 provides the hardware capability bits, define the macro
6394 __BUILTIN_CPU_SUPPORTS__.
6395 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6396 if GLIBC does not provide the hardware capability bits. Add a
6397 gcc_unreachable call if the built-in cpu function is neither
6398 __builtin_cpu_is nor __builtin_cpu_supports.
6399 (rs6000_get_function_versions_dispatcher): Change the warning
6400 that an old GLIBC is used which does not export the capability
6401 bits to be an error.
6402 * doc/extend.texi (target_clones attribute): Document the
6403 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6404 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6405 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6406 the macros defined by GCC if the newer GLIBC is available.
6408 2017-07-12 Jeff Law <law@redhat.com>
6410 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6411 remaining includes slightly.
6412 * config/riscv/riscv-builtins.c: Include profile-count.h.
6414 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6417 * config/avr/avr.c (avr_set_current_function): In diagnostic
6418 messages: Quote keywords and (parts of) identifiers.
6419 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6422 2017-07-12 Carl Love <cel@us.ibm.com>
6424 * config/rs6000/rs6000-c.c: Add support for built-in functions
6425 vector bool char vec_revb (vector bool char);
6426 vector bool short vec_revb (vector short char);
6427 vector bool int vec_revb (vector bool int);
6428 vector bool long long vec_revb (vector bool long long);
6429 * doc/extend.texi: Update the built-in documentation file for the
6430 new built-in functions.
6432 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6434 * config/s390/s390.md: Remove movcc splitter.
6436 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6438 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6439 load/store on condition.
6441 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6444 * config/avr/avr.c (avr_encode_section_info)
6445 [progmem && !TREE_READONLY]: Error if progmem object needs
6448 2017-07-11 Michael Collison <michael.collison@arm.com>
6450 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6453 2017-07-11 Carl Love <cel@us.ibm.com>
6455 * config/rs6000/rs6000-c.c: Add support for builtins
6456 vector unsigned int vec_parity_lsbb (vector signed int);
6457 vector unsigned int vec_parity_lsbb (vector unsigned int);
6458 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6459 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6460 vector unsigned long long vec_parity_lsbb (vector signed long long);
6461 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6462 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6463 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6464 * doc/extend.texi: Update the built-in documentation file for the
6465 new built-in functions.
6467 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6469 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6470 (layout::m_primary_loc): New field.
6471 (layout::layout): Initialize new field. Move location filtering
6472 logic from here to...
6473 (layout::maybe_add_location_range): ...this new method. Add
6474 support for filtering to just the lines already specified by other
6476 (layout::will_show_line_p): New method.
6477 (gcc_rich_location::add_location_if_nearby): New method.
6478 (selftest::test_add_location_if_nearby): New test function.
6479 (selftest::diagnostic_show_locus_c_tests): Call it.
6480 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6483 2017-07-11 Tom de Vries <tom@codesourcery.com>
6485 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6486 (bb_first_real_insn): New function.
6487 (nvptx_single): Add extra initialization of broadcasted condition
6490 2017-07-11 Nathan Sidwell <nathan@acm.org>
6492 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6494 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6496 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6497 Binutils doc as TEXI will mess them up.
6498 * doc/invoke.texi (AVR Options): Same here.
6500 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6502 * config/sparc/sparc.opt (mfix-ut700): New option.
6503 (mfix-gr712rc): Likewise.
6504 (sparc_fix_b2bst): New variable.
6505 * doc/invoke.texi (SPARC options): Document them.
6506 (ARM options): Fix warnings.
6507 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6508 instructions to prevent sequences that can trigger the store-store
6509 errata for certain LEON3FT processors.
6510 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6511 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6512 * config/sparc/sparc.md (fix_b2bst): New attribute.
6513 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6515 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6518 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6520 (*rsqrtsf2_sse): Ditto.
6522 (div<mode>3): Macroize insn from divdf3 and divsf3
6523 using MODEF mode iterator.
6525 2017-07-10 Martin Sebor <msebor@redhat.com>
6527 PR tree-optimization/80397
6528 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6529 instead of testing for equality to INTEGER_TYPE.
6531 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6533 * config.gcc: Remove uclibc from arc target spec.
6535 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6537 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6539 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6542 * lto-wrapper.c (remove_option): New function.
6543 (merge_and_complain): Merge PIC/PIE options more realistically.
6545 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6547 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6551 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6552 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6553 * config.in: Regenerate.
6554 * configure: Regenerate.
6555 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6556 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6557 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6558 (TARGET_GASISR_PROLOGUES): ...target mask.
6559 * common/config/avr/avr-common.c
6560 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6561 Set -mgas-isr-prologues.
6562 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6563 INSERT_PASS_BEFORE for it.
6564 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6565 * config/avr/avr.c (avr_option_override)
6566 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6567 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6568 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6569 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6570 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6572 (make_avr_pass_pre_proep): New function.
6573 (emit_push_sfr) <treg>: Add argument to function and use it
6575 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6576 and set machine->gasisr.yes.
6577 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6578 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6579 __gcc_isr.n_pushed to .L__stack_usage.
6580 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6581 (avr_asm_final_postscan_insn): ...this new static function.
6582 * config/avr/avr.h (machine_function)
6583 <is_no_gccisr, use_L__stack_usage>: New fields.
6584 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6585 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6586 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6587 (gasisr, *gasisr): New expander and insn.
6588 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6589 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6590 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6592 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6594 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6597 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6599 Move jump-tables out of .text again.
6602 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6603 (ASM_OUTPUT_ADDR_VEC): New function.
6604 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6605 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6606 INSN_ADDRESSes as asm comment.
6607 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6608 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6609 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6610 * config/avr/avr.md (*tablejump): Adjust comment.
6611 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6612 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6614 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6615 (avr_output_addr_vec): New proto.
6616 (avr_log_t) <insn_addresses>: New field.
6618 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6621 * config/i386/i386.c (ix86_function_arg_advance): Set
6622 outgoing_args_on_stack to true if there are outgoing arguments
6624 (ix86_function_arg): Likewise.
6625 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6626 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6627 * config/i386/i386.h (machine_function): Add
6628 outgoing_args_on_stack.
6630 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6632 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6634 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6636 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6638 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6639 (REAL_H): Remove $(MACHMODE_H).
6640 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6642 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6643 $(MACHMODE_H) and double-int.h.
6644 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6646 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6649 2017-07-07 Andrew Pinski <apinski@cavium.com>
6651 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6652 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6654 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6656 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6657 Add warning if GCC was not configured to link against a GLIBC that
6658 exports the hardware capability bits.
6659 (make_resolver_func): Make resolver function private and not a
6660 COMDAT function. Create the name with clone_function_name instead
6661 of make_unique_name.
6664 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6665 correct operand in doing the split.
6667 2017-07-07 Carl Love <cel@us.ibm.com>
6669 * config/rs6000/rs6000-c: Add support for built-in function
6670 vector unsigned short vec_pack_to_short_fp32 (vector float,
6672 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6673 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6674 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6675 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6676 (convert_4f32_8i16): Add define_expand.
6677 * doc/extend.texi: Update the built-in documentation file for the
6678 new built-in function.
6680 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6682 * config/sparc/m8.md: New file.
6683 * config/sparc/sparc.md: Include m8.md.
6685 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6687 * config/sparc/sparc.opt: New option -mvis4b.
6688 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6689 (sparc_option_override): Handle VIS4B.
6690 (enum sparc_builtins): Define
6691 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6692 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6693 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6694 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6695 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6696 (check_constant_argument): New function.
6697 (sparc_vis_init_builtins): Define builtins
6698 __builtin_vis_dictunpack{8,16,32},
6699 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6700 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6701 __builtin_vis_fpcmpde{8,16,32}shl and
6702 __builtin_vis_fpcmpur{8,16,32}shl.
6703 (sparc_expand_builtin): Check that the constant operands to
6704 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6705 constant and in range.
6706 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6708 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6709 (SPARC_IMM5_P): Likewise.
6710 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6711 (enabled): Handle vis4b.
6712 (UNSPEC_DICTUNPACK): New unspec.
6713 (UNSPEC_FPCMPSHL): Likewise.
6714 (UNSPEC_FPUCMPSHL): Likewise.
6715 (UNSPEC_FPCMPDESHL): Likewise.
6716 (UNSPEC_FPCMPURSHL): Likewise.
6717 (cpu_feature): New CPU feature `vis4b'.
6718 (dictunpack{8,16,32}): New insns.
6719 (FPCSMODE): New mode iterator.
6720 (fpcscond): New code iterator.
6721 (fpcsucond): Likewise.
6722 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6723 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6724 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6725 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6726 * config/sparc/constraints.md: Define constraints `q' for unsigned
6727 2-bit integer constants and `t' for unsigned 5-bit integer
6729 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6731 (imm5_operand_dictunpack16): Likewise.
6732 (imm5_operand_dictunpack32): Likewise.
6733 (imm2_operand): Likewise.
6734 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6735 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6736 ditunpack* and fpcmp*shl builtins.
6738 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6740 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6741 * config.in: Add HAVE_AS_SPARC6 define.
6742 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6744 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6746 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6747 (CPP_CPU_SPEC): Handle m8.
6748 (ASM_CPU_SPEC): Likewise.
6749 * config/sparc/sparc-opts.h (enum processor_type): Add
6751 * config/sparc/sparc.c (m8_costs): New struct.
6752 (sparc_option_override): Handle TARGET_CPU_m8.
6753 (sparc32_initialize_trampoline): Likewise.
6754 (sparc64_initialize_trampoline): Likewise.
6755 (sparc_issue_rate): Likewise.
6756 (sparc_register_move_cost): Likewise.
6757 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6758 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6759 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6760 (CPP_CPU_SPEC): Handle M8.
6761 (ASM_CPU_SPEC): Likewise.
6762 (AS_M8_FLAG): Define.
6763 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6764 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6765 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6767 * configure: Regenerate.
6768 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6771 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6773 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6775 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6776 ("*movdi_insn_sp32"): Do not set v3pipe.
6777 ("*movsi_insn"): Likewise.
6778 ("*movdi_insn_sp64"): Likewise.
6779 ("*movsf_insn"): Likewise.
6780 ("*movdf_insn_sp32"): Likewise.
6781 ("*movdf_insn_sp64"): Likewise.
6782 ("*zero_extendsidi2_insn_sp64"): Likewise.
6783 ("*sign_extendsidi2_insn"): Likewise.
6784 ("*mov<VM32:mode>_insn"): Likewise.
6785 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6786 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6787 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6788 ("<vlop:code><VL:mode>3"): Likewise.
6789 ("*not_<vlop:code><VL:mode>3"): Likewise.
6790 ("*nand<VL:mode>_vis"): Likewise.
6791 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6792 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6793 ("one_cmpl<VL:mode>2"): Likewise.
6794 ("faligndata<VM64:mode>_vis"): Likewise.
6795 ("alignaddrsi_vis"): Likewise.
6796 ("alignaddrdi_vis"): Likweise.
6797 ("alignaddrlsi_vis"): Likewise.
6798 ("alignaddrldi_vis"): Likewise.
6799 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6800 ("bmaskdi_vis"): Likewise.
6801 ("bmasksi_vis"): Likewise.
6802 ("bshuffle<VM64:mode>_vis"): Likewise.
6803 ("cmask8<P:mode>_vis"): Likewise.
6804 ("cmask16<P:mode>_vis"): Likewise.
6805 ("cmask32<P:mode>_vis"): Likewise.
6806 ("pdistn<P:mode>_vis"): Likewise.
6807 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6809 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6811 * config/sparc/sparc.md ("subtype"): New insn attribute.
6812 ("*wrgsr_sp64"): Set insn subtype.
6813 ("*rdgsr_sp64"): Likewise.
6814 ("alignaddrsi_vis"): Likewise.
6815 ("alignaddrdi_vis"): Likewise.
6816 ("alignaddrlsi_vis"): Likewise.
6817 ("alignaddrldi_vis"): Likewise.
6818 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6819 ("fexpand_vis"): Likewise.
6820 ("fpmerge_vis"): Likewise.
6821 ("faligndata<VM64:mode>_vis"): Likewise.
6822 ("bshuffle<VM64:mode>_vis"): Likewise.
6823 ("cmask8<P:mode>_vis"): Likewise.
6824 ("cmask16<P:mode>_vis"): Likewise.
6825 ("cmask32<P:mode>_vis"): Likewise.
6826 ("fchksm16_vis"): Likewise.
6827 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6828 ("fmean16_vis"): Likewise.
6829 ("fp<plusminus_insn>64_vis"): Likewise.
6830 ("<plusminus_insn>v8qi3"): Likewise.
6831 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6832 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6833 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6834 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6835 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6836 ("*movqi_insn"): Likewise.
6837 ("*movhi_insn"): Likewise.
6838 ("*movsi_insn"): Likewise.
6839 ("movsi_pic_gotdata_op"): Likewise.
6840 ("*movdi_insn_sp32"): Likewise.
6841 ("*movdi_insn_sp64"): Likewise.
6842 ("movdi_pic_gotdata_op"): Likewise.
6843 ("*movsf_insn"): Likewise.
6844 ("*movdf_insn_sp32"): Likewise.
6845 ("*movdf_insn_sp64"): Likewise.
6846 ("*zero_extendhisi2_insn"): Likewise.
6847 ("*zero_extendqihi2_insn"): Likewise.
6848 ("*zero_extendqisi2_insn"): Likewise.
6849 ("*zero_extendqidi2_insn"): Likewise.
6850 ("*zero_extendhidi2_insn"): Likewise.
6851 ("*zero_extendsidi2_insn_sp64"): Likewise.
6852 ("ldfsr"): Likewise.
6853 ("prefetch_64"): Likewise.
6854 ("prefetch_32"): Likewise.
6855 ("tie_ld32"): Likewise.
6856 ("tie_ld64"): Likewise.
6857 ("*tldo_ldub_sp32"): Likewise.
6858 ("*tldo_ldub1_sp32"): Likewise.
6859 ("*tldo_ldub2_sp32"): Likewise.
6860 ("*tldo_ldub_sp64"): Likewise.
6861 ("*tldo_ldub1_sp64"): Likewise.
6862 ("*tldo_ldub2_sp64"): Likewise.
6863 ("*tldo_ldub3_sp64"): Likewise.
6864 ("*tldo_lduh_sp32"): Likewise.
6865 ("*tldo_lduh1_sp32"): Likewise.
6866 ("*tldo_lduh_sp64"): Likewise.
6867 ("*tldo_lduh1_sp64"): Likewise.
6868 ("*tldo_lduh2_sp64"): Likewise.
6869 ("*tldo_lduw_sp32"): Likewise.
6870 ("*tldo_lduw_sp64"): Likewise.
6871 ("*tldo_lduw1_sp64"): Likewise.
6872 ("*tldo_ldx_sp64"): Likewise.
6873 ("*mov<VM32:mode>_insn"): Likewise.
6874 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6875 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6877 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6879 * config/sparc/sparc.md ("type"): New insn type viscmp.
6880 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6882 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6883 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6884 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6885 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6887 ("n7_vis_logical_11cycle"): Likewise.
6888 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6889 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6890 * config/sparc/niagara.md ("niag_vis"): Likewise.
6891 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6892 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6894 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6896 * config/sparc/sparc.md: New instruction type `bmask'.
6897 (bmaskdi_vis): Use the `bmask' type.
6898 (bmasksi_vis): Likewise.
6899 * config/sparc/ultra3.md (us3_array): Likewise.
6900 * config/sparc/niagara7.md (n7_array): Likewise.
6901 * config/sparc/niagara4.md (n4_array): Likewise.
6902 * config/sparc/niagara2.md (niag2_vis): Likewise.
6903 (niag3_vis): Likewise.
6904 * config/sparc/niagara.md (niag_vis): Likewise.
6906 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6908 * ipa-comdats.c: Remove optimize check from gate.
6909 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6910 for functions not optimized.
6911 (ipa_fn_summary_read): Skip optimize check.
6912 (ipa_fn_summary_write): Likewise.
6913 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6915 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6917 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6918 (check_callers): Give up on caller which is not optimized.
6919 (inline_small_functions): Likewise.
6920 (ipa_inline): Do not give up when not optimizing.
6921 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6922 away unoptimizes cdtors.
6923 (whole_program_function_and_variable_visibility): Do
6924 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6925 * ipa.c (process_references): Do not check optimize.
6926 (symbol_table::remove_unreachable_nodes): Update optimize check.
6927 (set_writeonly_bit): Update optimize check.
6928 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6929 (pass_ipa_single_use::gate): Remove.
6931 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6933 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6934 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6935 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6936 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6937 permute_load, permute_store, adjust_extract, adjust_splat,
6938 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6939 replace_swap_with_copy, dump_swap_insn_table,
6940 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6941 recombine_lvx_pattern, recombine_stvx_pattern,
6942 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6943 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6944 to file rs6000-p8swap.c.
6945 * config/rs6000/rs6000-p8swap.c: New file.
6946 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6947 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6948 and rs6000*-*-* targets.
6950 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6952 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6954 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6956 * lto-wrapper.c (merge_and_complain): Do not merge
6957 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6958 fsigned_zeros, ftrapping_math, fwrapv.
6959 (append_compiler_options): Do not track these options.
6960 (append_linker_options): Likewie
6962 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6964 * cgraphunit.c (cgraph_node::finalize_function): When
6965 !flag_toplevel_reorde set no_reorder flag.
6966 (varpool_node::finalize_decl): Likewise.
6967 (symbol_table::compile): Drop no toplevel reorder path.
6969 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6971 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6972 edges; zero probability is not better than uninitialized.
6974 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6976 * asan.h (asan_sanitize_allocas_p): Declare.
6977 * asan.c (asan_sanitize_allocas_p): New function.
6978 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6979 (handle_builtin_alloca): Likewise.
6980 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6981 if !asan_sanitize_allocas_p.
6982 * params.def (asan-instrument-allocas): Add new option.
6983 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6984 * opts.c (common_handle_option): Disable allocas sanitization for
6987 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6989 * asan.c: Include gimple-fold.h.
6990 (get_last_alloca_addr): New function.
6991 (handle_builtin_stackrestore): Likewise.
6992 (handle_builtin_alloca): Likewise.
6993 (asan_emit_allocas_unpoison): Likewise.
6994 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6995 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6996 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6997 (instrument_builtin_call): Pass gimple iterator to
6998 get_mem_refs_of_builtin_call.
6999 (last_alloca_addr): New global.
7000 * asan.h (asan_emit_allocas_unpoison): Declare.
7001 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
7002 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
7003 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
7004 if function calls alloca.
7005 * gimple-fold.c (replace_call_with_value): Remove static keyword.
7006 * gimple-fold.h (replace_call_with_value): Declare.
7007 * internal-fn.c: Include asan.h.
7008 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
7009 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
7011 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7013 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
7014 (C_SELFTEST_FLAGS): New.
7015 (CPP_SELFTEST_FLAGS): New.
7016 (SELFTEST_DEPS): New, from deps of s-selftest.
7017 (C_SELFTEST_DEPS): New, from deps of s-selftest.
7018 (CPP_SELFTEST_DEPS): New.
7019 (selftest): Add dependency on s-selftest-c++.
7020 (s-selftest): Rename to...
7021 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
7022 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
7023 than SELFTEST_FLAGS.
7024 (selftest-gdb): Rename to...
7025 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
7027 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
7028 (selftest-valgrind): Rename to...
7029 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
7031 (selftest-valgrind): Reintroduce as an alias for
7032 selftest-c-valgrind.
7033 (s-selftest-c++): New.
7034 (selftest-c++-gdb): New.
7035 (selftest-c++-valgrind): New.
7037 2017-07-06 Olivier Hainque <hainque@adacore.com>
7039 * gcc.c (process_command): When deciding if undefined variables
7040 should be ignored when processing specs, accept "gcc -v" as well.
7042 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7044 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
7045 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
7047 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
7049 * config/arm/arm-cpus.in (armv8-r): Add new entry.
7050 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
7051 * config/arm/arm-tables.opt: Regenerate.
7052 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
7054 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
7056 2017-07-06 Carl Love <cel@us.ibm.com>
7058 * ChangeLog: Clean up from mid air collision
7060 2017-07-06 Carl Love <cel@us.ibm.com>
7062 * config/rs6000/rs6000-c.c: Add support for built-in functions
7063 vector signed int vec_subc (vector signed int, vector signed int);
7064 vector signed __int128 vec_subc (vector signed __int128,
7065 vector signed __int128);
7066 vector unsigned __int128 vec_subc (vector unsigned __int128,
7067 vector unsigned __int128);
7068 vector signed int vec_sube (vector signed int, vector signed int,
7070 vector unsigned int vec_sube (vector unsigned int,
7071 vector unsigned int,
7072 vector unsigned int);
7073 vector signed __int128 vec_sube (vector signed __int128,
7074 vector signed __int128,
7075 vector signed__int128);
7076 vector unsigned __int128 vec_sube (vector unsigned __int128,
7077 vector unsigned __int128,
7078 vector unsigned __int128);
7079 vector signed int vec_subec (vector signed int, vector signed int,
7081 vector unsigned int vec_subec (vector unsigned int,
7082 vector unsigned int,
7083 vector unsigned int);
7084 vector signed __int128 vec_subec (vector signed __int128,
7085 vector signed __int128,
7086 vector signed__int128);
7087 vector unsigned __int128 vec_subec (vector unsigned __int128,
7088 vector unsigned __int128,
7089 vector unsigned __int128);
7090 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7091 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7092 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7093 BU_ALTIVEC_OVERLOAD_X definitions.
7094 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7095 * doc/extend.texi: Update the built-in documentation file for the new
7098 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7101 * diagnostic-show-locus.c (layout::layout): Use start and finish
7102 spelling location for the start and finish of each range.
7103 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7104 Add unused aspect param.
7105 * input.c (expand_location_1): Add "aspect" param, and use it
7106 to access the correct part of the location.
7107 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7109 (expand_location_to_spelling_point): Likewise.
7110 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7111 param, and pass it to expand_location_1.
7113 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7115 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7116 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7117 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7118 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7119 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7120 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7121 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7122 _mm_maskz_getmant_ss): New intrinsics.
7123 (__builtin_ia32_getexpss128_mask): Changed to ...
7124 __builtin_ia32_getexpss128_round ... this.
7125 (__builtin_ia32_getexpsd128_mask): Changed to ...
7126 __builtin_ia32_getexpsd128_round ... this.
7127 * config/i386/i386-builtin-types.def
7128 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7129 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7130 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7131 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7132 __builtin_ia32_getmantss_mask_round): New builtins.
7133 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7134 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7135 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7136 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7137 * config/i386/sse.md
7138 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7139 avx512f_sgetexp<mode><mask_scalar_name>
7140 <round_saeonly_scalar_name> ... this.
7141 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7142 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7143 vgetexp<ssescalarmodesuffix>
7144 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7145 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7146 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7147 avx512f_vgetmant<mode><mask_scalar_name>
7148 <round_saeonly_scalar_name> ... this.
7149 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7150 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7151 vgetmant<ssescalarmodesuffix>
7152 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7153 %0<mask_scalar_operand4>, %1, %2
7154 <round_saeonly_scalar_mask_op4>, %3} ... this.
7155 * config/i386/subst.md (mask_scalar_operand4,
7156 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7157 round_saeonly_scalar_nimm_predicate): New subst attributes.
7159 2017-07-06 Julia Koval <julia.koval@intel.com>
7161 * config/i386/i386.c (ix86_erase_embedded_rounding):
7162 Remove code for old rounding pattern.
7164 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7166 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7168 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7170 * doc/sourcebuild.texi (Test Directives, Variants of
7171 dg-require-support): Add documentation for dg-require-stack-check.
7173 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7175 * config/i386/subst.md (mask_scalar, round_scalar,
7176 round_saeonly_scalar): New meta-templates.
7177 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7178 round_scalar_mask_operand3, round_scalar_mask_op3,
7179 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7180 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7181 round_saeonly_scalar_constraint,
7182 round_saeonly_scalar_prefix): New subst attribute.
7183 * config/i386/sse.md
7184 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7185 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7186 <round_scalar_name> ... this.
7187 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7188 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7189 <round_scalar_name> ... this.
7190 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7191 <sse>_vm<code><mode>3<mask_scalar_name>
7192 <round_saeonly_scalar_name> ... this.
7193 (v<plusminus_mnemonic><ssescalarmodesuffix>
7194 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7195 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7196 v<plusminus_mnemonic><ssescalarmodesuffix>
7197 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7198 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7199 (v<multdiv_mnemonic><ssescalarmodesuffix>
7200 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7201 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7202 v<multdiv_mnemonic><ssescalarmodesuffix>
7203 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7204 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7205 (v<maxmin_float><ssescalarmodesuffix>
7206 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7207 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7208 v<maxmin_float><ssescalarmodesuffix>
7209 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7210 %0<mask_scalar_operand3>, %1, %<iptr>2
7211 <round_saeonly_scalar_mask_op3>} ... this.
7213 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7215 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7216 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7218 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7219 Alan Hayward <alan.hayward@arm.com>
7220 David Sherwood <david.sherwood@arm.com>
7222 * combine.c (simplify_if_then_else): Remove "enum" before
7224 * compare-elim.c (can_eliminate_compare): Likewise.
7225 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7227 (aarch64_lookup_simd_builtin_type): Likewise.
7228 (aarch64_simd_builtin_type): Likewise.
7229 (aarch64_init_simd_builtin_types): Likewise.
7230 (aarch64_simd_expand_args): Likewise.
7231 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7233 (aarch64_reverse_mask): Likewise.
7234 (aarch64_simd_emit_reg_reg_move): Likewise.
7235 (aarch64_gen_adjusted_ldpstp): Likewise.
7236 (aarch64_ccmp_mode_to_code): Likewise.
7237 (aarch64_operands_ok_for_ldpstp): Likewise.
7238 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7239 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7241 (aarch64_min_divisions_for_recip_mul): Likewise.
7242 (aarch64_reassociation_width): Likewise.
7243 (aarch64_get_condition_code_1): Likewise.
7244 (aarch64_simd_emit_reg_reg_move): Likewise.
7245 (aarch64_simd_attr_length_rglist): Likewise.
7246 (aarch64_reverse_mask): Likewise.
7247 (aarch64_operands_ok_for_ldpstp): Likewise.
7248 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7249 (aarch64_gen_adjusted_ldpstp): Likewise.
7250 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7252 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7253 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7254 (arm_lookup_simd_builtin_type): Likewise.
7255 (arm_simd_builtin_type): Likewise.
7256 (arm_init_simd_builtin_types): Likewise.
7257 (arm_expand_builtin_args): Likewise.
7258 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7259 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7260 (ft32_setup_incoming_varargs): Likewise.
7261 (ft32_function_arg): Likewise.
7262 (ft32_function_arg_advance): Likewise.
7263 (ft32_pass_by_reference): Likewise.
7264 (ft32_arg_partial_bytes): Likewise.
7265 (ft32_valid_pointer_mode): Likewise.
7266 (ft32_addr_space_pointer_mode): Likewise.
7267 (ft32_addr_space_legitimate_address_p): Likewise.
7268 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7270 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7271 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7272 (iamcu_alignment): Likewise.
7273 (canonicalize_vector_int_perm): Likewise.
7274 (ix86_noce_conversion_profitable_p): Likewise.
7275 (ix86_mpx_bound_mode): Likewise.
7276 (ix86_operands_ok_for_move_multiple): Likewise.
7277 * config/microblaze/microblaze-protos.h
7278 (microblaze_expand_conditional_branch_reg): Likewise.
7279 * config/microblaze/microblaze.c
7280 (microblaze_expand_conditional_branch_reg): Likewise.
7281 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7283 (rs6000_reassociation_width): Likewise.
7284 (rs6000_invalid_binary_op): Likewise.
7285 (fusion_p9_p): Likewise.
7286 (emit_fusion_p9_load): Likewise.
7287 (emit_fusion_p9_store): Likewise.
7288 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7290 (riscv_hard_regno_mode_ok_p): Likewise.
7291 (riscv_address_insns): Likewise.
7292 (riscv_split_symbol): Likewise.
7293 (riscv_legitimize_move): Likewise.
7294 (riscv_function_value): Likewise.
7295 (riscv_hard_regno_nregs): Likewise.
7296 (riscv_expand_builtin): Likewise.
7297 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7298 (riscv_build_integer): Likewise.
7299 (riscv_split_integer): Likewise.
7300 (riscv_legitimate_constant_p): Likewise.
7301 (riscv_cannot_force_const_mem): Likewise.
7302 (riscv_regno_mode_ok_for_base_p): Likewise.
7303 (riscv_valid_base_register_p): Likewise.
7304 (riscv_valid_offset_p): Likewise.
7305 (riscv_valid_lo_sum_p): Likewise.
7306 (riscv_classify_address): Likewise.
7307 (riscv_legitimate_address_p): Likewise.
7308 (riscv_address_insns): Likewise.
7309 (riscv_load_store_insns): Likewise.
7310 (riscv_force_binary): Likewise.
7311 (riscv_split_symbol): Likewise.
7312 (riscv_force_address): Likewise.
7313 (riscv_legitimize_address): Likewise.
7314 (riscv_move_integer): Likewise.
7315 (riscv_legitimize_const_move): Likewise.
7316 (riscv_legitimize_move): Likewise.
7317 (riscv_address_cost): Likewise.
7318 (riscv_subword): Likewise.
7319 (riscv_output_move): Likewise.
7320 (riscv_canonicalize_int_order_test): Likewise.
7321 (riscv_emit_int_order_test): Likewise.
7322 (riscv_function_arg_boundary): Likewise.
7323 (riscv_pass_mode_in_fpr_p): Likewise.
7324 (riscv_pass_fpr_single): Likewise.
7325 (riscv_pass_fpr_pair): Likewise.
7326 (riscv_get_arg_info): Likewise.
7327 (riscv_function_arg): Likewise.
7328 (riscv_function_arg_advance): Likewise.
7329 (riscv_arg_partial_bytes): Likewise.
7330 (riscv_function_value): Likewise.
7331 (riscv_pass_by_reference): Likewise.
7332 (riscv_setup_incoming_varargs): Likewise.
7333 (riscv_print_operand): Likewise.
7334 (riscv_elf_select_rtx_section): Likewise.
7335 (riscv_save_restore_reg): Likewise.
7336 (riscv_for_each_saved_reg): Likewise.
7337 (riscv_register_move_cost): Likewise.
7338 (riscv_hard_regno_mode_ok_p): Likewise.
7339 (riscv_hard_regno_nregs): Likewise.
7340 (riscv_class_max_nregs): Likewise.
7341 (riscv_memory_move_cost): Likewise.
7342 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7343 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7344 (rl78_addr_space_address_mode): Likewise.
7345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7347 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7348 (rs6000_reassociation_width): Likewise.
7349 (rs6000_invalid_binary_op): Likewise.
7350 (fusion_p9_p): Likewise.
7351 (emit_fusion_p9_load): Likewise.
7352 (emit_fusion_p9_store): Likewise.
7353 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7354 (ok_for_simple_move_operands): Likewise.
7355 (ok_for_simple_move_strict_operands): Likewise.
7356 (ok_for_simple_arith_logic_operands): Likewise.
7357 (visium_legitimize_reload_address): Likewise.
7358 (visium_select_cc_mode): Likewise.
7359 (output_cbranch): Likewise.
7360 (visium_split_double_move): Likewise.
7361 (visium_expand_copysign): Likewise.
7362 (visium_expand_int_cstore): Likewise.
7363 (visium_expand_fp_cstore): Likewise.
7364 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7365 (visium_function_arg): Likewise.
7366 (visium_function_arg_advance): Likewise.
7367 (visium_libcall_value): Likewise.
7368 (visium_setup_incoming_varargs): Likewise.
7369 (visium_legitimate_constant_p): Likewise.
7370 (visium_legitimate_address_p): Likewise.
7371 (visium_legitimize_address): Likewise.
7372 (visium_secondary_reload): Likewise.
7373 (visium_register_move_cost): Likewise.
7374 (visium_memory_move_cost): Likewise.
7375 (prepare_move_operands): Likewise.
7376 (ok_for_simple_move_operands): Likewise.
7377 (ok_for_simple_move_strict_operands): Likewise.
7378 (ok_for_simple_arith_logic_operands): Likewise.
7379 (visium_function_value_1): Likewise.
7380 (rtx_ok_for_offset_p): Likewise.
7381 (visium_legitimize_reload_address): Likewise.
7382 (visium_split_double_move): Likewise.
7383 (visium_expand_copysign): Likewise.
7384 (visium_expand_int_cstore): Likewise.
7385 (visium_expand_fp_cstore): Likewise.
7386 (visium_split_cstore): Likewise.
7387 (visium_select_cc_mode): Likewise.
7388 (visium_split_cbranch): Likewise.
7389 (output_cbranch): Likewise.
7390 (visium_print_operand_address): Likewise.
7391 * expmed.c (flip_storage_order): Likewise.
7392 * expmed.h (emit_cstore): Likewise.
7393 (flip_storage_order): Likewise.
7394 * genrecog.c (validate_pattern): Likewise.
7395 * hsa-gen.c (gen_hsa_addr): Likewise.
7396 * internal-fn.c (expand_arith_overflow): Likewise.
7397 * ira-color.c (allocno_copy_cost_saving): Likewise.
7398 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7399 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7400 (process_invariant_for_inheritance): Likewise.
7401 * lra-eliminations.c (move_plus_up): Likewise.
7402 * omp-low.c (lower_oacc_reductions): Likewise.
7403 * simplify-rtx.c (simplify_subreg): Likewise.
7404 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7405 (TARGET_CHKP_BOUND_MODE): Likewise..
7406 * targhooks.c (default_chkp_bound_mode): Likewise.
7407 (default_setup_incoming_vararg_bounds): Likewise.
7408 * targhooks.h (default_chkp_bound_mode): Likewise.
7409 (default_setup_incoming_vararg_bounds): Likewise.
7410 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7411 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7412 (have_whole_vector_shift): Likewise.
7413 * tree-vect-stmts.c (vectorizable_load): Likewise.
7414 * doc/tm.texi: Regenerate.
7416 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7418 Graceful degrade if Binutils PR21472 is not available.
7421 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7422 .rodata in flash test fails.
7423 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7424 * confgure: Regenerate.
7425 * config.in: Regenerate.
7426 * config/avr/avr.c (avr_asm_named_section)
7427 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7428 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7429 (avr_asm_init_sections): Same.
7431 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7433 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7434 (fma<VH:mode>4_intrinsic): Likewise.
7435 (*fmsub<VCVTF:mode>4): Likewise.
7436 (*fmsub<VH:mode>4_intrinsic): Likewise.
7438 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7441 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7442 Don't depend on "optimize > 0".
7443 (out_movhi_r_mr, out_movqi_mr_r): Same.
7444 (out_movhi_mr_r, out_movqi_r_mr): Same.
7445 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7446 io_address_operand on "optimize > 0".
7448 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7450 * tree-loop-distribution.c: Add general explanantion on the pass.
7451 (generate_loops_for_partition): Mark distributed loop.
7452 (pg_add_dependence_edges): New parameter. Handle alias data
7453 dependence specially and record it in the parameter if asked.
7454 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7455 (init_partition_graph_vertices, add_partition_graph_edge): New.
7456 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7457 (free_partition_graph_vdata, build_partition_graph): New.
7458 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7459 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7460 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7461 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7462 (version_for_distribution_p, finalize_partitions): New.
7463 (distribute_loop): Handle alias data dependence specially. Factor
7464 out loop fusion code as functions and call these functions.
7466 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7468 * tree-loop-distribution.c (classify_partition): New parameter and
7469 better handle reduction statement.
7470 (rdg_build_partitions): Revise comment.
7471 (distribute_loop): Compute statements in all partitions and pass it
7472 to classify_partition.
7474 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7476 * tree-loop-distribution.c (enum partition_type): New.
7477 (struct partition): New field type.
7478 (partition_merge_into): Add parameter. Update partition type.
7479 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7480 (build_rdg_partition_for_vertex): Compute partition type.
7481 (rdg_build_partitions): Dump partition type.
7482 (distribute_loop): Update calls to partition_merge_into.
7484 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7486 * tree-loop-distribution.c (struct ddr_hasher): New.
7487 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7489 (classify_partition): Call get_data_dependence.
7490 (pg_add_dependence_edges): Ditto.
7491 (distribute_loop): Release data dependence hash table.
7493 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7495 * tree-loop-distribution.c (ref_base_address): Delete.
7496 (similar_memory_accesses): Rename ...
7497 (share_memory_accesses): ... to this. Check if partitions access
7498 the same memory reference.
7499 (distribute_loop): Call share_memory_accesses.
7501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7503 * tree-loop-distribution.c (struct partition): New field recording
7505 (partition_alloc, partition_free): Init and release data refs.
7506 (partition_merge_into): Merge data refs.
7507 (build_rdg_partition_for_vertex): Collect data refs for partition.
7508 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7510 (distribute_loop): Remve data refs from vertice data of partition
7513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7515 * tree-loop-distribution.c (params.h): Include header file.
7516 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7517 (datarefs_vec): New global var.
7518 (create_rdg_vertices): Use datarefs_vec directly.
7519 (free_rdg): Don't free data references.
7520 (build_rdg): Update use. Don't free data references.
7521 (distribute_loop): Compute global variable for data references.
7522 Bail out if there are too many data references.
7524 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7526 * tree-loop-distribution.c (loop_nest): New global var.
7527 (build_rdg): Use loop directly, rather than loop nest.
7528 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7530 (distribute_loop): Compute global variable loop nest. Update use.
7532 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7534 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7535 (partition_merge_into): New parameter. Dump reason for fusion.
7536 (distribute_loop): Update use of partition_merge_into.
7538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7540 * tree-loop-distribution.c (bb_top_order_index): New.
7541 (bb_top_order_index_size, bb_top_order_cmp): New.
7542 (stmts_from_loop): Use topological order.
7543 (pass_loop_distribution::execute): Compute and release topological
7544 order for basic blocks.
7546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7548 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7551 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7553 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7554 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7555 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7556 * internal-fn.def (LOOP_DIST_ALIAS): New.
7557 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7558 (fold_loop_internal_call): ... this.
7559 (vect_loop_dist_alias_call): New function.
7560 (set_uid_loop_bbs): Call fold_loop_internal_call.
7561 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7564 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7567 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7568 Require dead FLAGS_REG at the beginning of a peephole.
7570 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7573 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7574 arguments in the call to __builtin_ia32_sbb_u32.
7575 (_subborrow_u64): Swap _X and _Y arguments in the call to
7576 __builtin_ia32_sbb_u64.
7578 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7581 * tree-vrp.c (compare_assert_loc): Turn into a function template
7582 with stable template parameter. Only test if a->e is NULL,
7583 !a->e == !b->e has been verified already. Use e == NULL or
7584 e != NULL instead of e or ! e tests. If stable is true, don't use
7585 iterative_hash_expr, on the other side allow a or b or both NULL
7586 and sort the NULLs last.
7587 (process_assert_insertions): Sort using compare_assert_loc<false>
7588 instead of compare_assert_loc, later sort using
7589 compare_assert_loc<true> before calling process_assert_insertions_for
7590 in a loop. Use break instead of continue once seen NULL pointer.
7592 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7594 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7595 Cortex-R7 and Cortex-R8 processors.
7597 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7599 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7600 uninitialized while src is not.
7602 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7604 * common/config/arm/arm-common.c: Adjust include path for
7606 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7607 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7608 (arm-cpu-data.h): Likewise.
7609 (arm-cpu-cdata.h): Likewise.
7610 * config/arm/arm-cpu.h: Delete.
7611 * config/arm/arm-cpu-cdata.h: Delete.
7612 * config/arm/arm-cpu-data.h: Delete.
7614 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7616 * config/arm/arm-cpus.in (cortex-a55): New.
7617 (cortex-a75): Likewise.
7618 (cortex-a75.cortex-a55): Likewise.
7619 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7621 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7622 * config/arm/arm-cpu-cdata.h: Regenerate.
7623 * config/arm/arm-cpu-data.h: Regenerate.
7624 * config/arm/arm-cpu.h: Regenerate.
7625 * config/arm/arm-tables.opt: Regenerate.
7626 * config/arm/arm-tune.md: Regenerate.
7628 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7630 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7632 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7634 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7637 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7639 PR tree-optimization/81292
7640 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7641 full_string_p, also call adjust_related_strinfos if the adjustment
7642 is simple, otherwise invalidate related strinfos.
7644 2017-07-04 Martin Liska <mliska@suse.cz>
7647 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7648 newly created variable as DECL_IGNORED_P.
7650 2017-07-04 Martin Liska <mliska@suse.cz>
7653 * ipa-inline.c (inline_small_functions):
7654 Use xstrdup_for_dump.
7656 2017-07-04 Tom de Vries <tom@codesourcery.com>
7658 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7660 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7663 * config/darwin.c (darwin_function_switched_text_sections):
7666 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7668 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7670 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7672 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7674 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7676 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7677 min_profitable_iters, and th as inclusive lower bounds.
7678 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7679 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7680 for min_profitable_iters and min_profitable_estimate.
7681 (vect_transform_loop): Treat th as an inclusive lower bound.
7682 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7684 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7687 * config/darwin.c (darwin_function_switched_text_sections):
7688 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7689 in two pieces, and suppress the use of buf.
7691 2017-07-03 Nathan Sidwell <nathan@acm.org>
7693 * hash-table.h (hash_table_mod1): Fix indentation.
7695 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7698 * predict.c (force_edge_cold): Be more careful about propagation
7700 * profile-count.h (profile_probability::guessed,
7701 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7703 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7705 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7707 * doc/invoke.texi (rcpc architecture extension): Document it.
7709 2017-07-03 Richard Biener <rguenther@suse.de>
7711 PR tree-optimization/60510
7712 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7713 the scalar reduction PHI and use it.
7714 (vectorizable_reduction): Properly guard the single_defuse_cycle
7715 path for non-SLP reduction chains where we cannot use it.
7716 Rework reduc_def/index and vector type deduction. Rework
7717 vector operand gathering during reduction op code-gen.
7718 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7719 chains dissolve the chain and leave it to non-SLP reduction
7722 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7724 * tree-data-ref.h (dr_alignment): Declare.
7725 * tree-data-ref.c (dr_alignment): New function.
7726 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7727 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7729 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7731 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7733 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7734 and base_misalignment fields.
7735 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7736 * tree-data-ref.c: Include builtins.h.
7737 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7738 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7739 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7740 * tree-vect-data-refs.c: Include tree-cfg.h.
7741 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7742 fields instead of calculating an alignment here.
7743 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7744 innermost_loop_behavior fields.
7746 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7748 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7750 (DR_STEP_ALIGNMENT): New macro.
7751 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7752 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7753 (create_data_ref): Print it.
7754 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7755 to tell whether the step preserves vector (mis)alignment.
7756 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7757 Move the check for an integer step and generalise to all INTEGER_CST.
7758 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7759 Print the outer step alignment.
7761 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7763 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7764 with offset_alignment.
7765 (DR_ALIGNED_TO): Delete.
7766 (DR_OFFSET_ALIGNMENT): New macro.
7767 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7768 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7769 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7770 (create_data_ref): Likewise.
7771 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7772 (vect_analyze_data_refs): Likewise.
7773 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7774 creating dummy innermost behavior.
7776 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7778 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7779 with a "innermost_loop_behavior *" and refeence tree.
7780 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7781 (create_data_ref): Update call accordingly.
7782 * tree-predcom.c (find_looparound_phi): Likewise.
7784 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7786 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7787 fields with dr_wrt_vec_loop.
7788 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7789 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7790 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7791 (vect_dr_behavior): New function.
7792 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7793 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7794 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7795 track whether the step preserves the misalignment.
7796 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7797 Use vect_dr_behavior.
7798 (vect_setup_realignment): Update call accordingly.
7799 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7800 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7801 call to vect_create_addr_base_for_vector_ref.
7802 (vect_create_cond_for_align_checks): Likewise.
7803 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7804 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7805 (vect_recog_mask_conversion_pattern): Likewise.
7806 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7807 (new_stmt_vec_info): Remove redundant zeroing.
7809 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7811 * common/config/arm/arm-common.c (arm_be8_option): New function.
7812 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7813 (ISA_ARMv6): Add isa_bit_be8.
7814 * config/arm/arm.h (arm_be8_option): Add prototype.
7815 (BE8_SPEC_FUNCTION): New define.
7816 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7817 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7818 (mlittle-endian): Similarly.
7819 (mbe8, mbe32): New options.
7820 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7821 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7823 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7825 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7827 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7829 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7830 (cleanup_tree_cfg_bb): Use it.
7831 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7833 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7835 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7838 * loop-doloop.c (add_test): Update profile.
7840 2017-07-03 Martin Liska <mliska@suse.cz>
7843 * sanopt.c (rewrite_usage_of_param): New function.
7844 (sanitize_rewrite_addressable_params): Likewise.
7845 (pass_sanopt::execute): Call rewrite_usage_of_param.
7847 2017-07-03 Richard Biener <rguenther@suse.de>
7849 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7850 back to using VIEW_CONVERT_EXPR.
7852 2017-07-03 Martin Liska <mliska@suse.cz>
7855 * doc/extend.texi: Document when a resolver function is
7856 generated for target_clones.
7858 2017-07-03 Martin Liska <mliska@suse.cz>
7860 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7861 and shadow memory of auto variables which are subject of
7862 use-after-scope sanitization.
7863 (asan_expand_mark_ifn): Add do set only when is_poison.
7865 2016-07-03 Richard Biener <rguenther@suse.de>
7867 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7869 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7870 (vectorizable_reduction): Perform reduction PHI creation when
7871 visiting a reduction PHI and adjust and simplify code generation
7872 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7873 (vect_transform_loop): Visit reduction PHIs.
7874 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7875 defs into the SLP tree.
7876 (vect_build_slp_tree): Reduction defs terminate the recursion.
7877 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7879 (vect_get_vec_defs_for_stmt_copy): Export.
7880 (vect_get_vec_defs): Likewise.
7881 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7883 (vect_get_vec_defs_for_stmt_copy): Declare.
7884 (vect_get_vec_defs): Likewise.
7886 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7888 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7889 parameter with a "loop" parameter and use it instead of the
7890 loop containing DR_STMT. Don't check simple_iv when doing
7891 BB analysis. Describe the two analysis modes in the comment.
7893 2017-07-03 Tom de Vries <tom@codesourcery.com>
7895 PR tree-optimization/69468
7896 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7897 (find_same_succ_bb): Handle ignore_edge_flags.
7899 2017-07-03 Tom de Vries <tom@codesourcery.com>
7901 PR tree-optimization/81192
7902 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7904 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7906 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7908 2017-07-03 Tom de Vries <tom@codesourcery.com>
7910 PR tree-optimization/81192
7911 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7912 BB_SAME_SUCC (bb) == NULL.
7914 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7916 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7919 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7921 * dumpfile.c: Include profile-count.h
7922 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7924 (insert_cond_bb): Update profile.
7925 * tree-cfg.h (insert_cond_bb): Update prototype.
7926 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7927 * tree-dump.c: Do not include tree-cfg.
7929 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7931 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7933 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7935 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7938 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7940 * tree-complex.c (expand_complex_div_wide): update profile.
7942 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7943 Alan Hayward <alan.hayward@arm.com>
7944 David Sherwood <david.sherwood@arm.com>
7946 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7947 (CORETYPES_H): New define.
7948 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7949 (insn-modes-inline.h, s-modes-inline-h): New rules.
7950 (generated_files): Add insn-modes-inline.h.
7951 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7952 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7953 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7954 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7955 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7956 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7957 (build/gencodes.o, build/genconditions.o): Likewise.
7958 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7959 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7960 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7961 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7962 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7963 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7964 * coretypes.h: Include everything up to real.h for generators.
7965 Include insn-modes.h first. Include wide-int-print.h after
7966 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7967 * machmode.h: Don't include insn-modes.h here.
7968 * function-tests.c: Remove includes of signop.h, machmode.h,
7969 double-int.h and wide-int.h.
7971 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7973 * optc-save-gen.awk: Likewise.
7974 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7975 * godump.c: Remove include of wide-int-print.h.
7976 * pretty-print.h: Likewise.
7977 * wide-int-print.cc: Likewise.
7978 * wide-int.cc: Likewise.
7979 * hash-map-tests.c: Remove include of signop.h.
7980 * hash-set-tests.c: Likewise.
7981 * rtl-tests.c: Likewise.
7982 * mkconfig.sh: Remove include of machmode.h.
7983 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7985 (emit_insn_modes_inline_h): ...this new function. Emit the code
7986 into an insn-modes-inline.h header file, adding appropriate
7987 include guards and end comments.
7988 (emit_insn_modes_c_header): Remove include of machmode.h.
7989 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7991 (main): Handle -i flag and call emit_insn_modes_inline_h when
7994 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7996 * tree-ssa-strlen.c (strinfo): Rename the length field to
7997 nonzero_chars. Add a full_string_p field.
7998 (compare_nonzero_chars, zero_length_string_p): New functions.
7999 (get_addr_stridx): Add an offset_out parameter.
8000 Use compare_nonzero_chars.
8001 (get_stridx): Update accordingly. Use compare_nonzero_chars.
8002 (new_strinfo): Update after above changes to strinfo.
8003 (set_endptr_and_length): Set full_string_p.
8004 (get_string_length): Update after above changes to strinfo.
8005 (unshare_strinfo): Update call to new_strinfo.
8006 (maybe_invalidate): Likewise.
8007 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
8008 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
8009 as a uhwi instead of an shwi. Update after above changes to
8010 strinfo and new_strinfo.
8011 (zero_length_string): Assert that chainsi contains full strings.
8012 Use zero_length_string_p. Update call to new_strinfo.
8013 (adjust_related_strinfos): Update after above changes to strinfo.
8014 Copy full_string_p from origsi.
8015 (adjust_last_stmt): Use zero_length_string_p.
8016 (handle_builtin_strlen): Update after above changes to strinfo and
8017 new_strinfo. Install the lhs as the string length if the previous
8018 entry didn't describe a full string.
8019 (handle_builtin_strchr): Update after above changes to strinfo
8021 (handle_builtin_strcpy): Likewise.
8022 (handle_builtin_strcat): Likewise.
8023 (handle_builtin_malloc): Likewise.
8024 (handle_pointer_plus): Likewise.
8025 (handle_builtin_memcpy): Likewise. Track nonzero characters
8026 that aren't necessarily followed by a nul terminator.
8027 (handle_char_store): Likewise.
8029 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8031 PR tree-optimization/80769
8032 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
8033 for malloc and calloc. Document the new invariant that all related
8034 strinfos have delayed lengths or none do.
8035 (verify_related_strinfos): Move earlier in file.
8036 (set_endptr_and_length): New function, split out from...
8037 (get_string_length): ...here. Also set the lengths of related
8039 (zero_length_string): Assert that chainsi has known (rather than
8041 (adjust_related_strinfos): Likewise.
8043 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8045 PR tree-optimization/81136
8046 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
8047 assert that two references with the same misalignment have the same
8048 compile-time misalignment if those compile-time misalignments
8051 2017-07-01 Andi Kleen <ak@linux.intel.com>
8053 * print-tree.c (print_node): Print all attributes.
8055 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8057 * cfg.c (scale_bbs_frequencies): New function.
8058 * cfg.h (scale_bbs_frequencies): Declare it.
8059 * cfgloopanal.c (single_likely_exit): Cleanup.
8060 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
8062 (scale_loop_profile): Likewise.
8063 (loop_version): Likewise.
8064 (create_empty_loop_on_edge): Update.
8065 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8066 scale_loop_frequencies, scale_loop_profile, loopify,
8067 loop_version): Update prototypes.
8068 * modulo-sched.c (sms_schedule): Update.
8069 * predict.c (unlikely_executed_edge_p): Also check probability.
8070 (probably_never_executed_edge_p): Fix typo.
8071 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8072 * tree-parloops.c (gen_parallel_loop): Update.
8073 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8074 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8075 * tree-ssa-loop-split.c (split_loop): Update.
8076 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8077 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8078 (vect_loop_versioning): Update.
8079 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8081 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8083 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8085 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8087 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8088 to keep profile consistent.
8090 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8092 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8093 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8094 * profile-count.h (max_safe_multiplier): Make unsigned.
8095 (profile_count::guessed_zero): New.
8097 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8099 * bb-reorder.c (fix_up_crossing_landing_pad,
8100 fix_crossing_conditional_branches): Use make_single_succ_edge
8101 to keep profile consistent.
8103 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8105 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8108 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8111 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8112 the right scopes, make sure cond_jump isn't preserved between multiple
8113 iterations. Search for fallthru edge whenever there are 3+ edges and
8114 use find_fallthru_edge for it.
8116 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8118 Patch by Alexander Monakov <amonakov@ispras.ru>
8119 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8120 probabilities consistently.
8122 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8124 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8125 emit_cmp_and_jump_insns.
8127 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8130 * tree-inline.c (expand_call_inline): Combine profile statuses.
8132 2017-06-30 Andrew Pinski <apinski@cavium.com>
8134 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8135 fold_stmt returned true.
8137 2017-06-30 Nathan Sidwell <nathan@acm.org>
8139 * ggc.h (empty_string): Delete.
8140 * cfgexpand.c (expand_asm_stmt): Use plain "".
8141 * optabs.c (expand_asm_memory_barrier): Likewise.
8142 * stringpool.c (empty_string): Delete.
8143 (digit_vector, digit_string): Delete.
8144 (ggc_alloc_string): Use plain "", don't optimize single digit
8145 strings. Use ggc_alloc_atomic.
8147 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8149 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8150 comparison set and one other set, use the cost of the non-comparison
8153 2017-06-30 Nathan Sidwell <nathan@acm.org>
8155 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8158 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8160 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8161 loops. Remove now unneeded calls to gimple_switch_set_label() that
8162 just set removed labels to NULL_TREE.
8164 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8166 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8167 (set_range_info): ...here. Only call set_range_info_raw if domain
8169 (set_nonzero_bits): Call set_range_info_raw.
8170 * tree-ssanames.h (set_range_info_raw): New.
8172 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8175 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8176 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8177 of nonimmediate_operand and <store_mask_constraint> instead of m
8178 for the input operand. For V8FI iterator, always split if input
8179 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8180 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8181 <store_mask_predicate> instead of register_operand and
8182 <store_mask_constraint> instead of v for the input operand. Make
8183 sure both operands aren't MEMs for if not <mask_applied>.
8185 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8187 * lto-wrapper.c (copy_file) Close both file descriptors before
8190 2017-06-30 Martin Liska <mliska@suse.cz>
8193 * multiple_target.c (create_dispatcher_calls): Make ifunc
8194 also for function that don't have calls or are not referenced.
8196 2017-06-30 Richard Biener <rguenther@suse.de>
8198 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8199 analyze the first scalar stmt. Move vector type computation
8200 for the BB case here from ...
8201 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8202 live operation processing in the SLP case properly.
8204 2017-06-30 Richard Biener <rguenther@suse.de>
8206 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8208 2017-06-30 Martin Liska <mliska@suse.cz>
8211 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8212 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8214 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8216 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8219 2017-06-30 Richard Biener <rguenther@suse.de>
8221 PR tree-optimization/81249
8222 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8223 condition reduction result to original scalar type.
8225 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8227 * profile-count.h (enum profile_quality): Fix typos and whitespace
8230 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8232 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8233 type for branch probabilities.
8235 2017-06-29 Julian Brown <julian@codesourcery.com>
8236 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8238 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8239 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8240 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8241 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8243 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8245 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8246 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8247 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8248 CC usage from generic code to here.
8249 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8250 CC usage into the target macros.
8252 2017-06-29 Maya Rashish <coypu@sdf.org>
8254 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8257 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8259 * arm/arm-builtins.c: Include profile-count.h
8260 * except.c (sjlj_emit_function_enter): Use
8261 profile_probability::unlikely.
8263 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8265 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8266 and tocrel_offset be pointer args rather than implicitly using
8268 (legitimate_constant_pool_address_p, rs6000_emit_move,
8269 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8270 tocrel_offset and use in toc_relative_expr_p call.
8271 (print_operand, print_operand_address): Use static tocrel_base_oac
8272 and tocrel_offset_oac.
8273 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8276 2017-06-29 Maya Rashish <coypu@sdf.org>
8278 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8280 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8282 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8283 objects, take into account only the alignment of 'op0' and 'mode1' if
8286 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8288 * ccmp.c (ccmp_tree_comparison_p): New function.
8289 (ccmp_candidate_p): Update to use above function.
8290 (get_compare_parts): New function.
8291 (expand_ccmp_next): Update to use new functions.
8292 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8294 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8295 take mode as argument.
8296 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8297 * expr.c (expand_expr_real_1): Pass mode as argument.
8299 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8301 * combine.c (combine_instructions): Print insns to dump_file, together
8304 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8306 * asan.c (asan_emit_stack_protection): Update.
8307 (create_cond_insert_point): Update.
8308 * auto-profile.c (afdo_propagate_circuit): Update.
8309 * basic-block.h (struct edge_def): Turn probability to
8310 profile_probability.
8311 (EDGE_FREQUENCY): Update.
8312 * bb-reorder.c (find_traces_1_round): Update.
8313 (better_edge_p): Update.
8314 (sanitize_hot_paths): Update.
8315 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8316 (make_single_succ_edge): Update.
8317 (check_bb_profile): Update.
8318 (dump_edge_info): Update.
8319 (update_bb_profile_for_threading): Update.
8320 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8321 probabilitycount to 0.
8322 * cfgbuild.c (compute_outgoing_frequencies): Update.
8323 * cfgcleanup.c (try_forward_edges): Update.
8324 (outgoing_edges_match): Update.
8325 (try_crossjump_to_edge): Update.
8326 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8327 (expand_gimple_tailcall): Update.
8328 (construct_init_block): Use make_single_succ_edge.
8329 (construct_exit_block): Use make_single_succ_edge.
8330 * cfghooks.c (verify_flow_info): Update.
8331 (redirect_edge_succ_nodup): Update.
8332 (split_edge): Update.
8333 (account_profile_record): Update.
8334 * cfgloopanal.c (single_likely_exit): Update.
8335 * cfgloopmanip.c (scale_loop_profile): Update.
8336 (set_zero_probability): Remove.
8337 (duplicate_loop_to_header_edge): Update.
8338 * cfgloopmanip.h (loop_version): Update prototype.
8339 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8340 (force_nonfallthru_and_redirect): Update.
8341 (update_br_prob_note): Update.
8342 (rtl_verify_edges): Update.
8343 (purge_dead_edges): Update.
8344 (rtl_lv_add_condition_to_bb): Update.
8345 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8346 * cgraphunit.c (init_lowered_empty_function): Update.
8347 (cgraph_node::expand_thunk): Update.
8348 * cilk-common.c: Include profile-count.h
8349 * dojump.c (inv): Remove.
8350 (jumpifnot): Update.
8351 (jumpifnot_1): Update.
8352 (do_jump_1): Update.
8354 (do_jump_by_parts_greater_rtx): Update.
8355 (do_compare_rtx_and_jump): Update.
8356 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8357 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8358 * dwarf2cfi.c: Include profile-count.h
8359 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8360 (sjlj_emit_dispatch_table): Likewise.
8361 * explow.c: Include profile-count.h
8362 * expmed.c (emit_store_flag_force): Update.
8363 (do_cmp_and_jump): Update.
8364 * expr.c (compare_by_pieces_d::generate): Update.
8365 (compare_by_pieces_d::finish_mode): Update.
8366 (emit_block_move_via_loop): Update.
8367 (store_expr_with_bounds): Update.
8368 (store_constructor): Update.
8369 (expand_expr_real_2): Update.
8370 (expand_expr_real_1): Update.
8371 * expr.h (try_casesi, try_tablejump): Update prototypes.
8372 * gimple-pretty-print.c (dump_probability): Update.
8373 (dump_profile): New.
8374 (dump_gimple_label): Update.
8375 (dump_gimple_bb_header): Update.
8376 * graph.c (draw_cfg_node_succ_edges): Update.
8377 * hsa-gen.c (convert_switch_statements): Update.
8378 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8379 (find_if_case_1): Update.
8380 (find_if_case_2): Update.
8381 * internal-fn.c (expand_arith_overflow_result_store): Update.
8382 (expand_addsub_overflow): Update.
8383 (expand_neg_overflow): Update.
8384 (expand_mul_overflow): Update.
8385 (expand_vector_ubsan_overflow): Update.
8386 * ipa-cp.c (good_cloning_opportunity_p): Update.
8387 * ipa-split.c (split_function): Use make_single_succ_edge.
8388 * ipa-utils.c (ipa_merge_profiles): Update.
8389 * loop-doloop.c (add_test): Update.
8390 (doloop_modify): Update.
8391 * loop-unroll.c (compare_and_jump_seq): Update.
8392 (unroll_loop_runtime_iterations): Update.
8393 * lra-constraints.c (lra_inheritance): Update.
8394 * lto-streamer-in.c (input_cfg): Update.
8395 * lto-streamer-out.c (output_cfg): Update.
8396 * mcf.c (adjust_cfg_counts): Update.
8397 * modulo-sched.c (sms_schedule): Update.
8398 * omp-expand.c (expand_omp_for_init_counts): Update.
8399 (extract_omp_for_update_vars): Update.
8400 (expand_omp_ordered_sink): Update.
8401 (expand_omp_for_ordered_loops): Update.
8402 (expand_omp_for_generic): Update.
8403 (expand_omp_for_static_nochunk): Update.
8404 (expand_omp_for_static_chunk): Update.
8405 (expand_cilk_for): Update.
8406 (expand_omp_simd): Update.
8407 (expand_omp_taskloop_for_outer): Update.
8408 (expand_omp_taskloop_for_inner): Update.
8409 * omp-simd-clone.c (simd_clone_adjust): Update.
8410 * optabs.c (expand_doubleword_shift): Update.
8411 (expand_abs): Update.
8412 (emit_cmp_and_jump_insn_1): Update.
8413 (expand_compare_and_swap_loop): Update.
8414 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8415 * predict.c (predictable_edge_p): Update.
8416 (edge_probability_reliable_p): Update.
8417 (set_even_probabilities): Update.
8418 (combine_predictions_for_insn): Update.
8419 (combine_predictions_for_bb): Update.
8420 (propagate_freq): Update.
8421 (estimate_bb_frequencies): Update.
8422 (force_edge_cold): Update.
8423 * profile-count.c (profile_count::dump): Add missing space into dump.
8424 (profile_count::debug): Add newline.
8425 (profile_count::differs_from_p): Explicitly convert to unsigned.
8426 (profile_count::stream_in): Update.
8427 (profile_probability::dump): New member function.
8428 (profile_probability::debug): New member function.
8429 (profile_probability::differs_from_p): New member function.
8430 (profile_probability::differs_lot_from_p): New member function.
8431 (profile_probability::stream_in): New member function.
8432 (profile_probability::stream_out): New member function.
8433 * profile-count.h (profile_count_quality): Rename to ...
8434 (profile_quality): ... this one.
8435 (profile_probability): New.
8436 (profile_count): Update.
8437 * profile.c (compute_branch_probabilities): Update.
8438 * recog.c (peep2_attempt): Update.
8439 * sched-ebb.c (schedule_ebbs): Update.
8440 * sched-rgn.c (find_single_block_region): Update.
8441 (compute_dom_prob_ps): Update.
8442 (schedule_region): Update.
8443 * sel-sched-ir.c (compute_succs_info): Update.
8444 * stmt.c (struct case_node): Update.
8445 (do_jump_if_equal): Update.
8446 (get_outgoing_edge_probs): Update.
8447 (conditional_probability): Update.
8448 (emit_case_dispatch_table): Update.
8449 (expand_case): Update.
8450 (expand_sjlj_dispatch_table): Update.
8451 (emit_case_nodes): Update.
8452 * targhooks.c: Update.
8453 * tracer.c (better_p): Update.
8454 (find_best_successor): Update.
8455 * trans-mem.c (expand_transaction): Update.
8456 * tree-call-cdce.c: Update.
8457 * tree-cfg.c (gimple_split_edge): Upate.
8458 (move_sese_region_to_fn): Upate.
8459 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8460 * tree-eh.c (lower_resx): Upate.
8461 (cleanup_empty_eh_move_lp): Upate.
8462 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8463 * tree-inline.c (copy_edges_for_bb): Update.
8464 (copy_cfg_body): Update.
8465 * tree-parloops.c (gen_parallel_loop): Update.
8466 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8467 (gimple_gen_time_profiler): Update.
8468 * tree-ssa-dce.c (remove_dead_stmt): Update.
8469 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8470 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8471 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8472 (unloop_loops): Update.
8473 (try_peel_loop): Update.
8474 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8475 * tree-ssa-loop-split.c (connect_loops): Update.
8476 (split_loop): Update.
8477 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8478 (hoist_guard): Update.
8479 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8480 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8481 (value_replacement): Update.
8482 * tree-ssa-reassoc.c (branch_fixup): Update.
8483 * tree-ssa-tail-merge.c (replace_block_by): Update.
8484 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8485 (create_edge_and_update_destination_phis): Update.
8486 (compute_path_counts): Update.
8487 (recompute_probabilities): Update.
8488 (update_joiner_offpath_counts): Update.
8489 (freqs_to_counts_path): Update.
8490 (duplicate_thread_path): Update.
8491 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8492 (struct switch_conv_info): Update.
8493 (gen_inbound_check): Update.
8494 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8495 (vect_do_peeling): Update.
8496 (vect_loop_versioning): Update.
8497 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8498 (optimize_mask_stores): Update.
8499 * ubsan.c (ubsan_expand_null_ifn): Update.
8500 * value-prof.c (gimple_divmod_fixed_value): Update.
8501 (gimple_divmod_fixed_value_transform): Update.
8502 (gimple_mod_pow2): Update.
8503 (gimple_mod_pow2_value_transform): Update.
8504 (gimple_mod_subtract): Update.
8505 (gimple_mod_subtract_transform): Update.
8506 (gimple_ic): Update.
8507 (gimple_stringop_fixed_value): Update.
8508 (gimple_stringops_transform): Update.
8509 * value-prof.h: Update.
8511 2017-06-29 Carl Love <cel@us.ibm.com>
8513 * config/rs6000/rs6000-c.c: Add support for built-in functions
8514 vector signed int vec_signed (vector float);
8515 vector signed long long vec_signed (vector double);
8516 vector signed int vec_signed2 (vector double, vector double);
8517 vector signed int vec_signede (vector double);
8518 vector signed int vec_signedo (vector double);
8519 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8520 instruction generator.
8521 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8522 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8523 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8525 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8526 vunsignede_v2df): Add define_expands.
8527 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8528 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8529 VEC_UNSIGNEDO): Add definitions.
8530 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8531 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8532 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8533 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8534 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8535 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8536 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8537 vec_unsignede, vec_unsignedo): Add builtin defines.
8538 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8540 * doc/extend.texi: Update the built-in documentation file for the
8541 new built-in functions.
8543 2017-06-29 Richard Biener <rguenther@suse.de>
8545 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8546 reduction chains to LOOP_VINFO_REDUCTIONS.
8547 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8548 SLP reductions after processing reduction chains.
8550 2017-06-29 Nathan Sidwell <nathan@acm.org>
8552 * builtins.c (fold_builtin_FUNCTION): Use
8553 lang_hooks.decl_printable_name.
8555 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8558 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8559 with only one label.
8560 * stmt.c (expand_case): Assert NCASES is greater than one.
8562 2017-06-29 Richard Biener <rguenther@suse.de>
8564 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8566 (group_case_labels): Likewise.
8567 (find_taken_edge): Push sanity checking on val to workers...
8568 (find_taken_edge_cond_expr): ... here
8569 (find_taken_edge_switch_expr): ... and here, handle cases
8570 with just a default label.
8571 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8572 (group_case_labels): Likewise.
8573 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8574 group_case_labels does anything cleanup the CFG again.
8576 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8578 PR tree-optimization/81196
8579 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8580 exit condition comparing two IVs.
8582 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8584 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8585 profile to the dummy entry at the end of the list of architectures.
8586 * config/arm/arm-cpu-cdata.h: Regenerated.
8588 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8589 Michael Collison <michael.collison@arm.com>
8592 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8594 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8595 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8596 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8597 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8598 of shift when the shift amount is masked with constant equal to
8599 the size of the mode.
8600 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8603 2017-06-29 Martin Liska <mliska@suse.cz>
8605 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8607 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8610 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8613 (edge_growth_cache_entry::edge_growth_cache_entry): New
8615 (reset_edge_growth_cache): Update to use constructor.
8617 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8619 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8620 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8621 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8623 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8625 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8626 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8628 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8630 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8631 (*-linux-uclibc*): Add t-uclibc tmake_file.
8632 * config/t-musl: New.
8633 * config/t-uclibc: New.
8635 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8637 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8639 (gen_comm_data): Emit architectural setting of arch_prof.
8640 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8642 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8643 (armv8-m.base, armv8-m.main): Likewise.
8644 * arm-protos.h (arm_build_target): Add profile field.
8645 (arch_option): Likewise.
8646 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8648 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8649 arm_active_target.profile.
8651 2017-06-28 Richard Biener <rguenther@suse.de>
8654 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8655 TYPE_OVERFLOW_WRAPS.
8656 * match.pd (negate_expr_p): Likewise.
8657 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8658 fold_build2, not fold_binary.
8660 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8662 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8663 Convert memory address to Pmode.
8664 (aarch64_print_operand): Assert MEM operands are always Pmode.
8666 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8669 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8670 Remove redundant if.
8671 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8672 * config/arm/aarch-common-protos.h
8673 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8674 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8676 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8679 * multiple_target.c (create_dispatcher_calls): Set the default
8680 clone to be static, not public.
8682 2017-06-28 Richard Biener <rguenther@suse.de>
8684 * tree-vect-loop.c (vectorizable_reduction): Move special
8685 cond reduction IV var creation ...
8686 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8687 parameter. Use STMT_VINFO_VECTYPE.
8688 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8691 2017-06-28 Martin Liska <mliska@suse.cz>
8694 * ipa-visibility.c (non_local_p): Handle visibility.
8696 2017-06-28 Martin Liska <mliska@suse.cz>
8699 * common.opt: Add IntegerRange to various options.
8700 * opt-functions.awk (integer_range_info): New function.
8701 * optc-gen.awk: Add integer_range_info to cl_options struct.
8702 * opts-common.c (decode_cmdline_option): Handle
8703 CL_ERR_INT_RANGE_ARG.
8704 (cmdline_handle_error): Likewise.
8705 * opts.c (print_filtered_help): Show valid interval in
8706 when --help is provided.
8707 * opts.h (struct cl_option): Add range_min and range_max fields.
8708 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8710 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8712 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8713 (x * C EQ/NE y * C): New transformation.
8715 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8717 * genmultilib (combination_space): Accept '+' in option names.
8719 2017-06-28 Martin Liska <mliska@suse.cz>
8722 * asan.c (instrument_derefs): Bail out inner references
8723 that are hard register variables.
8725 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8728 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8729 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8731 2017-06-28 Richard Biener <rguenther@suse.de>
8733 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8734 (vect_get_slp_defs): Adjust.
8735 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8737 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8739 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8740 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8741 (vectorizable_reduction): Adjust.
8742 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8744 (vect_get_slp_defs): Likewise.
8745 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8746 (vectorizable_bswap): Adjust.
8747 (vectorizable_call): Likewise.
8748 (vectorizable_conversion): Likewise.
8749 (vectorizable_assignment): Likewise.
8750 (vectorizable_shift): Likewise.
8751 (vectorizable_operation): Likewise.
8752 (vectorizable_store): Likewise.
8753 (vectorizable_condition): Likewise.
8754 (vectorizable_comparison): Likewise.
8756 2017-06-28 Michael Collison <michael.collison@arm.com>
8759 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8761 (arm_gen_movmemqi): Removed unused variable 'i'.
8762 Convert 'for' loop into 'while' loop.
8763 (arm_expand_prologue): Remove last unnecessary set of insn.
8764 (thumb_pop): Remove unused variable 'pushed_words'.
8765 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8767 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8769 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8770 * config/s390/s390-protos.h: Add prototype of
8771 s390_rel_address_ok_p.
8772 * config/s390/s390.c (s390_got_symbol): New function.
8773 (s390_rel_address_ok_p): New function.
8774 (legitimize_pic_address): Use s390_rel_address_ok_p.
8775 (s390_load_got): Use s390_got_symbol.
8776 (s390_option_override): Issue error if
8777 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8778 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8780 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8782 2017-06-27 Andrew Pinski <apinski@cavium.com>
8784 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8785 (X * copysign (1.0, X)): New pattern.
8786 (X * copysign (1.0, -X)): New pattern.
8787 (copysign (-1.0, CST)): New pattern.
8789 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8791 * genmultilib (combination_space): Remove variable.
8792 Validate reuse rules against regular expression for any sequence
8793 of multilib options in any order.
8795 2017-06-27 Michael Collison <michael.collison@arm.com>
8797 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8798 call aarch64_split_simd_combine.
8799 * (aarch64_combine_internal<mode>): Delete pattern.
8800 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8801 Allow register and subreg operands.
8803 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8805 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8806 specific need, just fallback on defaults.
8807 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8809 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8810 Olivier Hainque <hainque@adacore.com>
8812 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8814 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8815 targets. Pick a default if no particular attempt applied.
8816 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8819 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8821 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8822 (x86_64-wrs-vxworks7): Likewise.
8824 2017-06-27 Marek Polacek <polacek@redhat.com>
8827 * ubsan.c (instrument_null): Check get_base_address's result for null.
8829 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8831 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8833 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8835 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8836 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8837 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8838 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8840 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8841 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8842 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8843 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8844 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8845 BUILT_IN_FEUPDATEENV): New builtins.
8846 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8847 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8848 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8849 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8851 (builtin_structptr_types): Adjust size.
8852 * tree.c (builtin_structptr_types): Add four entries.
8854 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8855 Olivier Hainque <hainque@adacore.com>
8857 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8858 (TLS_SYM): New local macro, forcing reference to __tls__ on
8859 link command lines for VxWorks 7 RTPs, triggering initialization
8861 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8862 OS features TLS support, true for RTPs on VxWorks 7.
8863 * config/vxworks.c (vxworks_override_options): Setup emutls
8866 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8868 * predict.c (test_prediction_value_range): Use -1U instead of -1
8869 to avoid narrowing conversion warning.
8870 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8871 to avoid narrowing conversion warning.
8872 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8874 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8876 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8878 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8879 64bit configurations.
8880 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8881 (SIZE_TYPE): Likewise.
8882 * config/vxworks.c (vxworks_emutls_var_fields): Use
8883 long_unsigned_type_node instead of unsigned_type_node as the offset
8884 field type, which is "pointer" mode in emutls.c.
8886 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8889 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8892 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8893 gimple_vuse copying separately from gimple_vdef copying.
8895 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8897 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8900 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8901 Olivier Hainque <hainque@adacore.com>
8903 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8904 for all vxworks7 targets.
8905 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8906 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8907 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8908 variations for VX6/VX7 and 32/64bits later on in ...
8909 (VXWORKS_LIB_SPEC): Leverage new macros.
8910 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8911 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8913 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8915 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8916 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8918 2017-06-26 Carl Love <cel@us.ibm.com>
8920 * config/rs6000/rs6000-c.c: Add support for built-in functions
8921 vector bool char vec_reve (vector bool char);
8922 vector signed char vec_reve (vector signed char);
8923 vector unsigned char vec_reve (vector unsigned char);
8924 vector bool int vec_reve (vector bool int);
8925 vector signed int vec_reve (vector signed int);
8926 vector unsigned int vec_reve (vector unsigned int);
8927 vector bool long long vec_reve (vector bool long long);
8928 vector signed long long vec_reve (vector signed long long);
8929 vector unsigned long long vec_reve (vector unsigned long long);
8930 vector bool short vec_reve (vector bool short);
8931 vector signed short vec_reve (vector signed short);
8932 vector double vec_reve (vector double);
8933 vector float vec_reve (vector float);
8934 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8935 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8936 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8937 (altivec_vreve): New pattern.
8938 * config/rs6000/altivec.h (vec_reve): New define.
8939 * doc/extend.texi (vec_rev): Update the built-in documentation file
8940 for the new built-in functions.
8942 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8944 PR tree-optimization/71815
8945 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8947 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8949 (slsr_process_phi): Likewise.
8950 (replace_uncond_cands_and_profitable_phis): Don't replace a
8951 multiply candidate with a stride of 1 (copy or cast).
8952 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8954 (lowest_cost_path): Likewise.
8955 (total_savings): Likewise.
8957 2017-06-26 Richard Biener <rguenther@suse.de>
8960 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8961 Use def_builtin_pure for all gather builtins.
8963 2017-06-26 Richard Biener <rguenther@suse.de>
8965 PR tree-optimization/81203
8966 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8969 2017-06-26 Marek Polacek <polacek@redhat.com>
8972 * doc/invoke.texi: Document -Wmultistatement-macros.
8974 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8976 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8977 arm_neon_ok_no_float_abi effective target.
8979 2017-06-26 Richard Biener <rguenther@suse.de>
8981 PR tree-optimization/80928
8982 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8983 (copy_bbs): Set BB_DUPLICATED flag early.
8984 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8986 (execute_on_shrinking_pred): Likewise.
8987 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8988 BB_DUPLICATED blocks.
8989 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8990 iterate over all PHIs considering removal of *gsi.
8992 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8994 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8997 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8999 * config/rs6000/rs6000-string.c: (expand_block_clear,
9000 do_load_for_compare, select_block_compare_mode,
9001 compute_current_alignment, expand_block_compare,
9002 expand_strncmp_align_check, expand_strn_compare,
9003 expand_block_move, rs6000_output_load_multiple)
9004 Move functions related to string/block move/compare
9006 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
9007 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
9008 for this function which is now used in two files.
9009 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
9010 * config.gcc: Add rs6000-string.o to extra_objs for
9011 targets powerpc*-*-* and rs6000*-*-*.
9013 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
9016 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
9017 32-bit, since indexed is not valid for DImode.
9018 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
9019 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
9020 (define_peephole2 for Altivec d-form load): Add 32-bit support.
9021 (define_peephole2 for Altivec d-form store): Likewise.
9024 * multiple_target.c (create_dispatcher_calls): Only create the
9025 dispatcher call if the function is the default clone of a
9028 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
9031 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
9032 a call, force the call to not be a tail call.
9034 2017-06-23 Jeff Law <law@redhat.com>
9036 * doc/contrib.texi: Add entry for Steven Pemberton's work on
9039 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
9041 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
9042 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
9043 handling for early expansion of vector shifts (sl,sr,sra,rl).
9044 (builtin_function_type): Add vector shift right instructions
9045 to the unsigned argument list.
9047 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9049 rtl-optimizatoin/79286
9050 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
9051 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
9052 trap. PIC register plus a const unspec without offset can never trap.
9054 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
9056 * tree.h (builtin_structptr_type): New type.
9057 (builtin_structptr_types): Declare new array.
9058 * tree.c (builtin_structptr_types): New array.
9059 (free_lang_data, build_common_tree_nodes): Use it.
9061 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9064 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9067 2017-06-22 Matt Turner <mattst88@gmail.com>
9069 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9070 Lake models to skylake case. Assume skylake for unknown
9071 models with clflushopt.
9073 2017-06-22 Jeff Law <law@redhat.com>
9075 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9076 frame sizes that do not satisfy aarch64_uimm12_shift.
9078 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9080 * profile-count.h (apply_probability,
9081 apply_scale, probability_in): Fix checks for zero.
9083 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9085 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9086 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9088 2016-06-22 Richard Biener <rguenther@suse.de>
9090 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9091 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9092 REDUC_MAX_EXPR support.
9093 (vectorizable_reduction): Likewise.
9094 (vect_create_epilog_for_reduction): Likewise.
9096 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9098 * match.pd (A / (1 << B) -> A >> B): New.
9099 * generic-match-head.c: Include optabs-tree.h.
9100 * gimple-match-head.c: Likewise.
9101 * optabs-tree.h (target_supports_op_p): New.
9102 * optabs-tree.c (target_supports_op_p): New.
9104 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9106 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9107 $gcc_cv_ld --help output.
9108 (gcc_cv_ld_demangle): Likewise.
9109 (gcc_cv_ld_eh_frame_hdr): Likewise.
9110 (gcc_cv_ld_pie): Likewise.
9111 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9112 (gcc_cv_ld_buildid): Likewise.
9113 (gcc_cv_ld_sysroot): Likewise.
9114 (ld_bndplt_support): Likewise.
9115 (ld_pushpopstate_support): Likewise.
9116 * configure: Regenerate.
9117 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9119 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9122 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9123 operands indexes to avoid gap between operands and match_dups.
9125 2017-06-21 Andrew Pinski <apinski@cavium.com>
9127 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9128 Increment Arith_shift and Arith_shift_reg by 1.
9129 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9131 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9132 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9133 (aarch64_strip_extend): Add new argument and test for it.
9134 (aarch64_cheap_mult_shift_p): New function.
9135 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9136 add a cost if it is true.
9137 Update calls to aarch64_strip_extend.
9138 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9140 2017-06-21 Andrew Pinski <apinski@cavium.com>
9142 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9144 (thunderxt88): Likewise.
9145 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9146 (thunderx_prefetch_tune): New variable.
9147 (thunderx2t99_prefetch_tune): Update for the correct values.
9148 (thunderxt88_tunings): New variable.
9149 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9150 generic_prefetch_tune.
9151 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9153 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9155 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9156 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9157 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9158 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9159 (aarch64_atomic_cas<mode>, GPI): Likewise.
9161 2017-06-21 Martin Liska <mliska@suse.cz>
9163 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9164 statements on cold and hot labels.
9165 * predict.c (tree_estimate_probability_bb): Remove the
9166 prediction from this place.
9168 2017-06-21 Martin Liska <mliska@suse.cz>
9170 PR tree-optimization/79489
9171 * gimplify.c (maybe_add_early_return_predict_stmt): New
9173 (gimplify_return_expr): Call the function.
9174 * predict.c (tree_estimate_probability_bb): Remove handling
9176 * predict.def: Update comment about early return predictor.
9177 * gimple-predict.h (is_gimple_predict): New function.
9178 * predict.def: Change default value of early return to 66.
9179 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9181 * passes.def: Put pass_strip_predict_hints to the beginning of
9184 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9186 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9187 FUNCTION_DECL declarations.
9188 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9190 (dwaf2out_decl): Likewise.
9191 * godump.c (go_early_global_decl): Skip call to the real debug hook
9192 for FUNCTION_DECL declarations.
9193 * passes.c (rest_of_decl_compilation): Skip call to the
9194 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9195 -fdump-go-spec is passed.
9197 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9199 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9200 Reorder for compactness.
9201 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9202 (def_builtin_pure, def_builtin_pure2): New functions.
9203 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9205 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9207 * match.pd (nop_convert): New predicate.
9208 ((A +- CST1) +- CST2): Allow some NOP conversions.
9210 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9213 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9214 with ctors/dtors if GOVD_SHARED is set.
9216 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9218 * config/aarch64/aarch64.md (movti_aarch64):
9219 Emit mov rather than orr.
9220 (movtf_aarch64): Likewise.
9221 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9222 Emit mov rather than orr.
9224 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9226 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9227 Swap alternatives, make integer dup more expensive.
9229 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9231 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9232 Return true for non-tls symbols.
9234 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9236 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9237 (cortex-a75): Likewise.
9238 (cortex-a75.cortex-a55): Likewise.
9239 * config/aarch64/aarch64-tune.md: Regenerate.
9240 * doc/invoke.texi (-mtune): Document new values for -mtune.
9242 2017-06-21 Tom de Vries <tom@codesourcery.com>
9244 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9246 (Effective-Target Keywords, Other attributes): Suggest using
9247 dg-add-options stack_size feature to get stack limit in stack_size
9248 effective target documentation.
9250 2017-06-21 Julian Brown <julian@codesourcery.com>
9251 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9253 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9254 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9255 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9257 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9258 attribute type list for neon_multiply.
9259 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9260 attribute type list for neon_multiply.
9261 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9262 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9263 attribute type list for neon_multiply.
9264 * config/arm/types.md (crypto_pmull): Add.
9265 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9266 attribute type list.
9268 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9270 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9273 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9275 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9276 to make sure not to dereference a NULL cost_classes_ptr pointer.
9278 2017-06-20 Carl Love <cel@us.ibm.com>
9280 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9281 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9282 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9284 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9285 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9286 VMULOSW): New enum "unspec" values.
9287 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9288 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9289 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9290 altivec_vmulosw): New patterns.
9291 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9292 VMULOSW): Add definitions.
9294 2017-06-20 Julia Koval <julia.koval@intel.com>
9296 * config/i386/i386.c: Fix rounding expand for new pattern.
9297 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9299 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9301 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9302 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9304 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9306 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9309 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9311 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9312 architecture, then by implementer ID.
9313 * config/aarch64/aarch64-tune.md: Regenerate.
9315 2017-06-20 Richard Biener <rguenther@suse.de>
9318 * fold-const.c (split_tree): Fold to type before negating.
9320 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9322 * diagnostic-show-locus.c
9323 (selftest::test_fixit_deletion_affecting_newline): New function.
9324 (selftest::diagnostic_show_locus_c_tests): Call it.
9326 2017-06-20 Andreas Schwab <schwab@suse.de>
9329 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9332 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9334 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9335 __ARM_FEATURE_COPROC according to support.
9337 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9339 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9340 Rewritten to avoid overflow for > 32-bit pointers.
9343 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9344 by removing enum keyword.
9345 (ubsan_type_descriptor): Likewise. Formatting fix.
9348 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9349 splitter): Require TARGET_SSE2 in the condition.
9351 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9354 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9355 for doing vector set of SFmode on ISA 3.0.
9356 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9357 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9359 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9360 SFmode value into a V4SF variable that was extracted from another
9361 V4SF variable without converting the element to double precision
9362 and back to single precision vector format.
9363 (vsx_insert_extract_v4sf_p9_2): Likewise.
9365 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9367 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9368 in UWHI to avoid undefined overflow.
9371 * ubsan.h (enum ubsan_encode_value_phase): New.
9372 (ubsan_encode_value): Change second argument to
9373 enum ubsan_encode_value_phase with default value of
9374 UBSAN_ENCODE_VALUE_GENERIC.
9375 * ubsan.c (ubsan_encode_value): Change second argument to
9376 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9377 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9378 create_tmp_var_raw instead of create_tmp_var and use a
9380 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9381 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9382 ubsan_encode_value callers.
9385 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9386 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9387 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9389 2017-06-19 Richard Biener <rguenther@suse.de>
9392 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9393 estimates if we changed anything.
9395 2017-06-19 Richard Biener <rguenther@suse.de>
9397 PR tree-optimization/80887
9398 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9399 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9400 simplified lookups, then reset mprts_hook.
9401 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9403 (try_to_simplify): Likewise.
9405 2017-06-19 Martin Liska <mliska@suse.cz>
9408 * gimplify.c (gimplify_switch_expr):
9409 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9411 2017-06-19 Martin Liska <mliska@suse.cz>
9413 * doc/install.texi: Document that PGO runs in 4 stages.
9415 2017-06-19 Martin Liska <mliska@suse.cz>
9418 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9419 to dispatcher function name.
9420 * multiple_target.c (replace_function_decl): New function.
9421 (create_dispatcher_calls): Redirect both edges and references.
9423 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9425 * profile-count.c (profile_count::dump): Dump quality.
9426 (profile_count::differs_from_p): Update for unsigned val.
9427 * profile-count.h (profile_count_quality): New enum.
9428 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9430 2017-06-19 Richard Biener <rguenther@suse.de>
9432 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9433 struct function as arg.
9434 (estimate_numbers_of_iterations): Export overload with loop arg.
9435 (free_numbers_of_iterations_estimates_loop): Use an overload of
9436 free_numbers_of_iterations_estimates instead.
9437 * tree-cfg.c (remove_bb): Adjust.
9438 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9439 * tree-parloops.c (gen_parallel_loop): Likewise.
9440 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9442 (tree_unroll_loops_completely): Likewise.
9443 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9444 Use an overload instead and export.
9445 (estimated_loop_iterations): Adjust.
9446 (max_loop_iterations): Likewise.
9447 (likely_max_loop_iterations): Likewise.
9448 (estimate_numbers_of_iterations): Take struct function as arg
9450 (loop_exits_before_overflow): Adjust.
9451 (free_numbers_of_iterations_estimates_loop): Use an overload.
9452 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9453 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9455 2017-06-19 Richard Biener <rguenther@suse.de>
9458 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9459 RANGE_EXPR conservatively.
9461 2017-06-16 Carl Love <cel@us.ibm.com>
9463 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9464 definitions for vec_float, vec_float2, vec_floato,
9465 vec_floate built-ins.
9466 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9467 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9469 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9470 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9471 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9472 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9473 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9474 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9475 vec_floato): Add builtin defines.
9476 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9477 Update the built-in documentation file for the new built-in
9480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9482 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9483 (mthumb): Mark as the negative of -marm.
9485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9487 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9489 (ARM Options, -mtune): Document that this accepts the same
9490 extension options as -mcpu.
9491 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9495 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9496 permitted extensions.
9498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9500 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9501 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9502 (armv8-m.main): Add option +nodsp.
9503 * config/arm/arm-cpu-cdata.h: Regenerated.
9505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9507 * config/arm/t-fuchsia: New file.
9508 * config.gcc (arm*-*-fuchsia*): Use it.
9510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9512 * config/arm/t-symbian: Rewrite for new option infrastructure.
9514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9516 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9517 (MULTILIB_REQUIRED): Likewise.
9519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9521 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9522 (MULTILIB_RESUE): Likewise.
9523 (MULTILIB_MATCHES): Likewise.
9524 (MULTLIB_REQUIRED): Likewise.
9526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9528 * config/arm/t-rtems: Rewrite for new option framework.
9530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9532 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9533 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9534 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9535 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9536 * config/arm/t-multilib: ... here.
9537 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9538 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9539 armv7-a and armv8*-a when A-profile libraries have not been built.
9540 * config/arm/t-rmprofile: Rewrite.
9542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9544 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9545 with a backslash. Remove the backslash after substituting unescaped
9547 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9551 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9552 the driver to pass through the default setting of -mfloat-abi.
9553 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9555 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9556 (all_feat_combs): New rule.
9557 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9559 * config/arm/t-aprofile: Rewrite.
9561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9563 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9564 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9565 fpu is not specified by the user/command-line.
9566 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9567 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9568 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9569 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9570 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9571 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9575 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9576 * config/arm/t-arm-elf: Rewritten.
9578 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9580 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9581 have some floating-point instructions.
9582 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9583 (TARGET_MAYBE_HARD_FLOAT): New macro.
9584 * config/arm/arm-builtins.c (arm_init_builtins): Use
9585 TARGET_MAYBE_HARD_FLOAT.
9586 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9590 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9591 (configargs.h): Include it.
9592 (arm_print_hint_for_fpu_option): New function.
9593 (arm_parse_fpu_option): New function.
9594 (candidate_extension): New class.
9595 (arm_canon_for_multilib): New function.
9596 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9597 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9598 (ARCH_CANONICAL_SPECS): New macro.
9599 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9603 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9604 are set after handling multilib fragments. Set target_cpu_default2
9607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9609 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9611 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9612 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9614 (check_arch): Likewise.
9615 * config/arm/arm.c (arm_configure_build_target): Handle
9616 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9617 options in the default.
9619 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9621 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9622 when an option is an alias of another.
9623 * config/arm/parsecpu.awk (optalias): New parser token.
9624 (gen_comm_data): Mark non-alias options as such. Emit entries
9625 for extension aliases.
9626 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9627 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9628 (armv6kz, armv6zk, armv6t2): Likewise.
9629 (armv7): Make vfpv3-d16 an alias.
9630 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9632 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9633 Sort in canonical order.
9634 (armv8-a): Sort in canonical order.
9635 (armv8.1-a, armv8.2-a): Likewise.
9636 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9638 (cortex-a9): Sort in canonical order.
9639 * config/arm/arm.c (selftests.h): Include it.
9640 (arm_test_cpu_arch_data): New function.
9641 (arm_run_self_tests): New function.
9642 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9643 (targetm): Move declaration to the end of the file.
9644 * arm-cpu-cdata.h: Regenerated.
9646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9648 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9649 call to target_mode_check describing the type of option passed.
9650 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9651 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9652 arm_parse_cpu_option_name to match parameters against list of
9654 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9655 arm_arch_core_flags data structure.
9656 * config/arm/arm-cpu_cdata.h: Regenerated.
9658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9660 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9662 (arm_print_hint_for_cpu_option): Likewise.
9663 (arm_print_hint_for_arch_option): Likewise.
9664 (arm_parse_cpu_option_name): Likewise.
9665 (arm_parse_arch_option_name): Likewise.
9666 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9667 of entries in the all_fpus list.
9668 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9669 (arm_parse_cpu_option_name): Declare.
9670 (arm_parse_arch_option_name): Declare.
9671 (arm_parse_option_features): Declare.
9672 (arm_intialize_isa): Declare.
9673 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9675 (gen_comm_data): ... here. Make definitions non-static.
9676 * config/arm/arm-cpu-data.h: Regenerated.
9677 * config/arm/arm-cpu-cdata.h: Regenerated.
9679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9681 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9682 (cpu_arch_extension): New structure.
9683 (cpu_arch_option, arch_option, cpu_option): New structures.
9684 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9686 (gen_data): Generate new format data tables.
9687 * config/arm/arm.c (cpu_tune): New structure.
9688 (cpu_option, processors): Delete.
9689 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9690 (arm_print_hint_for_cpu_option): ... this and ...
9691 (arm_print_hint_for_arch_option): ... this.
9692 (arm_parse_arch_cpu_name): Delete. Replace with ...
9693 (arm_parse_cpu_option_name): ... this and ...
9694 (arm_parse_arch_option_name): ... this.
9695 (arm_unrecognized_feature): Change type of target parameter to
9697 (arm_parse_arch_cpu_features): Delete. Replace with ...
9698 (arm_parse_option_features): ... this.
9699 (arm_configure_build_target): Rework to use new configuration data
9701 (arm_print_tune_info): Rework for new configuration data tables.
9702 * config/arm/arm-cpu-data.h: Regenerated.
9703 * config/arm/arm-cpu.h: Regenerated.
9705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9707 * Makefile.in (OBJS): Move sbitmap.o from here ...
9708 (OBJS-libcommon): ... to here.
9710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9712 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9713 (ISA_ALL_CRYPTO): New macro.
9714 (ISA_ALL_SIMD): New macro
9715 (ISA_ALL_FP): New macro.
9716 * config/arm/arm.c (fpu_bitlist): Update initializer.
9717 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9719 (arm9e): Add fpu. Add option for nofp
9720 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9721 (arm926ej-s, arm1026ej-s): Likewise.
9722 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9723 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9724 neon-fp16, neon-vfpv4, nofp and nosimd.
9725 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9726 (cortex-a8): Add fpu. Add option for nofp.
9727 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9728 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9729 (cortex-r4f): Add fpu.
9730 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9731 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9733 (cortex-r8): Likewise.
9734 (cortex-m4): Add fpu. Add option for nofp.
9735 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9736 (cortex-a17.cortex-a7): Likewise.
9737 (cortex-a32): Add fpu. Add options for crypto and nofp.
9738 (cortex-a35, cortex-a53): Likewise.
9739 (cortex-a57): Add fpu. Add option for crypto.
9740 (cortex-a72, cortex-a73): Likewise.
9741 (exynos-m1): Likewise.
9742 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9743 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9744 (cortex-m33): Add fpu. Add option for nofp.
9745 * config/arm/arm-cpu-cdata.h: Regenerated
9746 * config/arm/arm-cpu-data.h: Regenerated.
9748 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9750 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9751 (armv5te, armv5tej): Likewise.
9752 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9753 (armv7): Add options fp and vfpv3-d16.
9754 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9755 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9757 (armv7ve): Likewise.
9758 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9759 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9760 (armv8-a): Add nocrypto option.
9761 (armv8.1-a, armv8.2-a): Likewise.
9762 (armv8-m.main): add options fp, fp.dp and nofp.
9764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9766 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9768 (armv8-a+crc): Delete.
9769 (armv8.1-a): Add options simd, crypto and nofp.
9770 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9771 (armv8.2-a+fp16): Delete.
9772 (armv8-m.main): Add option dsp.
9773 (armv8-m.main+dsp): Delete.
9774 (cortex-a8): Add fpu. Add nofp option.
9775 (cortex-a9): Add fpu. Add nofp and nosimd options.
9776 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9777 link to main cpu and architecture data structures.
9778 (gen_comm_data): Only put isa attributes from the main architecture
9780 (option): New statement for architecture and CPU entries.
9781 * arm.c (struct cpu_option): New structure.
9782 (struct processors): Add entry for options.
9783 (arm_unrecognized_feature): New function.
9784 (arm_parse_arch_cpu_name): Ignore any characters after the first
9786 (arm_parse_arch_cpu_feature): New function.
9787 (arm_configure_build_target): Separate out any CPU and architecture
9788 features and parse separately. Don't error out if -mfpu=auto is
9789 used with only an architecture string.
9790 (arm_print_asm_arch_directives): New function.
9791 (arm_file_start): Call it.
9792 * config/arm/arm-cpu-cdata.h: Regenerated.
9793 * config/arm/arm-cpu-data.h: Likewise.
9794 * config/arm/arm-tables.opt: Likewise.
9796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9798 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9799 assembler when it is not -mfpu=auto.
9801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9803 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9804 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9805 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9806 (ASM_CPU_SPEC): Rewrite.
9807 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9808 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9809 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9810 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9811 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9812 copied string is NUL-terminated. Also strip any characters prefixed
9814 (arm_rewrite_selected_arch): New function.
9815 (arm_rewrite_march): New function.
9817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9819 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9820 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9821 (march, mcpu, mtune): Convert to string-based options.
9822 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9823 (arm_parse_arch_cpu_name): New function.
9824 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9825 identify selected architecture or CPU.
9826 (arm_option_save): New function.
9827 (TARGET_OPTION_SAVE): Redefine.
9828 (arm_option_restore): Restore string options.
9829 (arm_option_print): Print string options.
9831 2017-06-16 Martin Sebor <msebor@redhat.com>
9833 PR tree-optimization/80933
9834 PR tree-optimization/80934
9835 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9836 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9837 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9838 (gimple_fold_builtin): Call them.
9840 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9842 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9843 as unlikely; update profile.
9845 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9847 * predict.c (force_edge_cold): Handle declaring edges impossible
9850 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9852 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9854 (try_unroll_loop_completely): Fix reporting.
9856 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9858 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9860 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9863 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9864 if given a non-constant argument for an intrinsic which requires a
9867 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9869 * profile.c (compare_freqs): New function.
9870 (branch_prob): Sort edge list.
9871 (find_spanning_tree): Assume that the list is priority sorted.
9873 2017-06-16 Richard Biener <rguenther@suse.de>
9875 PR tree-optimization/81090
9876 * passes.def (pass_record_bounds): Remove.
9877 * tree-pass.h (make_pass_record_bounds): Likewise.
9878 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9879 make_pass_record_bounds): Likewise.
9880 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9881 not free niter estimates at the beginning but at the end.
9882 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9884 2017-06-16 Richard Biener <rguenther@suse.de>
9886 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9887 initializer to workaround ICE in host GCC 4.8.
9889 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9891 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9893 (clone_inlined_nodes): Update.
9895 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9897 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9898 prefetch settings, and enable prefetching by default at -O3.
9900 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9902 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9903 Set flag_prefetch_loop_arrays according to tuning data.
9905 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9907 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9909 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9910 [Unrelated to main purpose of the patch] Place the pointer field last
9911 to enable type checking errors when tune structure are wrongly merged.
9912 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9913 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9914 (thunderx2t99_prefetch_tune): New tune constants.
9915 (tune_params *_tunings): Update all tunings (no functional change).
9916 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9917 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9918 from tunings structures.
9920 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9923 * ubsan.c (instrument_null): Add T argument, use it instead
9924 of computing it based on IS_LHS.
9925 (instrument_object_size): Likewise.
9926 (pass_ubsan::execute): Adjust instrument_null and
9927 instrument_object_size callers to pass gimple_get_lhs or
9928 gimple_assign_rhs1 result to it. Use instrument_null instead of
9929 calling get_base_address and instrument_mem_ref. Handle
9930 aggregate call arguments for object-size sanitization.
9932 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9934 PR tree-optimization/81089
9935 * tree-vrp.c (is_masked_range_test): Validate operands of
9938 2017-06-15 Martin Sebor <msebor@redhat.com>
9941 * dumpfile.c (dump_register): Avoid calling memset to initialize
9942 a class with a default ctor.
9943 * gcc.c (struct compiler): Remove const qualification.
9944 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9945 * hash-table.h: Ditto.
9946 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9948 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9949 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9951 * params.h (struct param_info): Make struct members non-const.
9952 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9953 with default initialization.
9954 * vec.h (vec_copy_construct, vec_default_construct): New helper
9956 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9957 with vec_copy_construct.
9958 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9959 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9960 * doc/invoke.texi (-Wclass-memaccess): Document.
9962 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9964 * emit-rtl.h (is_leaf): Update comment about local
9967 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9970 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9971 for a variable to have a section before checking if the section has a
9973 Set section to.persistent if persistent attribute is set.
9974 Warn if .persistent attribute is used on an automatic variable.
9976 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9978 PR rtl-optimization/80474
9979 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9981 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9983 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9986 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9988 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9989 (casesi): Emit bounds checking as RTL.
9990 (casesi_internal_mips16_<mode>): Remove bounds checking.
9992 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9994 * config/xtensa/xtensa.c (xtensa_option_override): Append
9995 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9996 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9997 xtensa_doloop_hooks): Define unconditionally.
9998 (xtensa_reorg_loops): Only call reorg_loops in the presence of
10000 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
10001 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
10002 for it in xtensa_option_override.
10003 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
10004 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
10006 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
10008 * doc/cppopts.texi: Document '-' special value to -MF.
10010 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
10012 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
10013 (cortex_a53_fconst): Likewise.
10014 (cortex_a53_fpmul): Likewise.
10015 (cortex_a53_f_load_64): Likewise.
10016 (cortex_a53_f_load_many): Likewise.
10017 (cortex_a53_advsimd_alu): Likewise.
10018 (cortex_a53_advsimd_alu_q): Likewise.
10019 (cortex_a53_advsimd_mul): Likewise.
10020 (cortex_a53_advsimd_mul_q): Likewise.
10021 (fpmac bypass): Add new bypass for fpmac-fpmac case.
10022 Add missing fmul, r2f_cvt and fconst cases.
10024 2017-06-14 Richard Biener <rguenther@suse.de>
10026 PR middle-end/81088
10027 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
10029 (fold_binary_loc): When associating do not treat pre-existing
10030 TREE_OVERFLOW on literal constants as a reason to allow
10031 TREE_OVERFLOW on associated literal constants.
10033 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10035 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
10036 (MASK_FEATURES): New macro.
10037 * config/sparc/sparc.c (sparc_option_override): Remove the special
10038 handling of -mfpu and generalize it to all MASK_FEATURES switches.
10040 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10042 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
10043 a division of 0 if non-call exceptions are enabled.
10045 2017-06-14 Andrew Pinski <apinski@cavium.com>
10046 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10049 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
10050 Improve vector initialization code gen for only variable case.
10052 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10054 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
10056 2017-06-14 Richard Biener <rguenther@suse.de>
10058 PR tree-optimization/81083
10059 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
10062 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10064 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10065 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10066 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10067 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10068 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10069 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10071 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10073 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10074 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10076 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10078 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10080 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10082 * config/rs6000/t-rtems: Don't handle SPE.
10084 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10086 * config/rs6000/t-linux: Don't handle SPE.
10088 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10090 * config/rs6000/eabispe.h: Delete file.
10092 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10094 * config/rs6000/t-spe: Delete file.
10096 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10098 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10099 (rs6000_legitimate_offset_address_p): Return false for anything in
10100 V2SImode or V2SFmode.
10102 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10104 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10105 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10106 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10108 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10109 (rs6000_legitimate_offset_address_p): Ditto.
10110 (rs6000_emit_move): Ditto.
10111 (rs6000_init_builtins): Remove V4HI_type_node.
10113 2017-06-13 Martin Liska <mliska@suse.cz>
10116 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10117 (gate_asan): Likewise.
10118 * asan.h (asan_no_sanitize_address_p): Remove the function.
10119 (sanitize_flags_p): New function.
10120 * builtins.def: Fix coding style.
10121 * common.opt: Use renamed enum value.
10122 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10123 * doc/extend.texi: Document no_sanitize attribute.
10124 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10125 to SANITIZE_UNDEFINED_NONDEFAULT.
10126 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10127 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10128 Use sanitize_flags_p.
10129 * gimplify.c (gimplify_function_tree): Likewise.
10130 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10131 * opts.c (parse_no_sanitize_attribute): New function.
10132 (common_handle_option): Use renamed enum value.
10133 * opts.h (parse_no_sanitize_attribute): Declare.
10134 * tree.c (sanitize_flags_p): New function.
10135 * tree.h: Declared here.
10136 * tsan.c: Use sanitize_flags_p.
10137 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10138 (instrument_mem_ref): Likewise.
10139 (instrument_bool_enum_load): Likewise.
10140 (do_ubsan_in_current_function): Remove the function.
10141 (pass_ubsan::execute): Use sanitize_flags_p.
10142 * ubsan.h: Remove do_ubsan_in_current_function
10143 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10144 (dump_function_to_file): Use it here.
10146 2017-06-13 Martin Jambor <mjambor@suse.cz>
10148 PR tree-optimization/80803
10149 PR tree-optimization/81063
10150 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10151 (propagate_subaccesses_across_link): Enqueue subtree whenever
10152 necessary instead of relying on the caller.
10154 2017-06-13 Martin Jambor <mjambor@suse.cz>
10156 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10157 that have a first_link.
10158 (sort_and_splice_var_accesses): Do not check first_link before
10160 (subtree_mark_written_and_enqueue): Likewise.
10161 (propagate_all_subaccesses): Likewise and do not stop at first
10162 parent with a first_link.
10164 2017-06-13 Martin Jambor <mjambor@suse.cz>
10166 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10169 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10171 * match.pd: New pattern.
10173 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10175 * tree-vrp.c (is_masked_range_test): New function.
10176 (register_edge_assert_for): Determine ranges for
10179 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10181 PR tree-optimization/67328
10182 * fold-const.c (maskable_range_p): New function.
10183 (build_range_check): Generate bittests if possible.
10185 2017-06-13 Martin Liska <mliska@suse.cz>
10187 * gimple-pretty-print.c (dump_probability): Add new argument.
10188 (dump_edge_probability): Dump both probability and count.
10189 (dump_gimple_label): Likewise.
10190 (dump_gimple_bb_header): Likewise.
10192 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10195 * config/avr/avr-devices.c: Fix indentation.
10196 * config/avr/gen-avr-mmcu-specs.c: Dito.
10198 2017-06-13 Richard Biener <rguenther@suse.de>
10200 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10201 instead get vector type from stmt_info.
10202 (vectorizable_reduction): Adjust. Remove dead code.
10204 2017-06-13 Richard Biener <rguenther@suse.de>
10206 PR middle-end/81065
10207 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10208 case of C * (x * C2 + C3).
10209 (fold_addr_of_array_ref_difference): Properly fold index difference.
10211 2017-06-12 David S. Miller <davem@davemloft.net>
10214 * config/sparc/sparc.md (return expander): Emit frame blockage if
10215 function uses alloca.
10217 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10219 * combine.c (make_field_assignment): Check len rather than the mode
10220 precision when calling force_to_mode.
10222 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10224 Support multilibs and devices that see flash in RAM address range.
10227 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10228 (avr_mcu_t) <flash_pm_offset>: New field.
10229 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10230 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10231 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10232 (AVR_TINY_PM_OFFSET): Remove macro.
10233 * config/avr/avr.opt (-mshort-calls): New option.
10234 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10235 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10236 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10237 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10238 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10239 instead of avr_arch->have_jmp_call.
10240 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10241 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10242 avr_arch->flash_pm_offset to define.
10243 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10244 new field flash_pm_offset. Add entry for avrxmega3.
10245 (avr_texinfo): Add entry for avrxmega3.
10246 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10247 attiny212, attiny214,
10248 attiny412, attiny414, attiny416, attiny417,
10249 attiny814, attiny816, attiny817,
10250 attiny1614, attiny1616, attiny1617,
10251 attiny3214, attiny3216, attiny3217.
10252 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10253 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10254 (avr_print_operand_address) [AVR_TINY]: Same.
10255 (avr_asm_init_sections) <readonly_data_section>: Only patch
10256 callback if avr_arch->flash_pm_offset = 0.
10257 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10258 for rodata if avr_arch->flash_pm_offset != 0.
10259 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10260 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10261 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10262 (m_options): Append opt_rcall.
10263 (m_dirnames): Append dir_rcall.
10264 * config/avr/t-multilib: Regenerate.
10266 * configure.ac [target=avr]: Check whether avrxmega3 default
10267 linker description file works as needed.
10268 * configure: Regenerate.
10269 * doc/avr-mmcu.texi: Regenerate.
10270 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10271 <__AVR_ARCH__>: Document avrxmega3 and 103.
10272 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10273 <__AVR_SHORT_CALLS__>: Document it.
10274 <__AVR_PM_BASE_ADDRESS__>: Document it.
10275 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10276 (AVR Variable Attributes) <progmem>: Document this is
10277 not needed for avrxmega3.
10278 (AVR Named Address Spaces) <__flash>: Dito.
10280 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10282 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10284 2017-06-12 Doug Rupp <rupp@adacore.com>
10286 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10287 Append vxworks-stdint.h to the tm_file list.
10288 * config/vxworks-stdint.h: New file.
10290 2017-06-12 Martin Liska <mliska@suse.cz>
10292 PR tree-optimization/81041
10293 * tree-profile.c (gimple_gen_ic_func_profiler):
10294 Create an extra BB in profile-generate
10295 (gimple_gen_time_profiler): Likewise.
10297 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10299 PR tree-optimization/81003
10300 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10301 (update_range_test): Use it instead of force_gimple_operand_gsi.
10303 2017-06-12 Richard Biener <rguenther@suse.de>
10305 PR tree-optimization/81053
10306 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10307 with backedge value not defined in loop. Simplify def stmt
10310 2017-06-11 Tom de Vries <tom@codesourcery.com>
10313 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10315 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10316 nvptx_cannot_force_const_mem.
10318 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10320 * opts.c (finish_options): Move test for flag_split_stack after
10321 it has been initialized.
10323 2017-06-11 Jason Merrill <jason@redhat.com>
10325 * tree.h (id_equal): New.
10326 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10327 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10328 instead of strcmp of IDENTIFIER_POINTER.
10330 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10332 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10333 (mark_all_inlined_calls_cdtor): Fix formating.
10334 (inline_transform): Rescale profile before inlining.
10336 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10338 * cgraph.h (cgraph_edge::clone): Update prototype.
10339 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10340 (cgraph_node::create_clone): Update.
10341 (cgraph_node::create_version_clone): Update.
10342 * tree-inline.c (copy_bb): Update.
10343 (expand_call_inline): Update.
10345 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10347 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10348 factored out from ...
10349 (rs6000_emit_prologue): ... here.
10351 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10353 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10354 factored out from ...
10355 (rs6000_emit_prologue): ... here.
10357 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10359 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10361 (handle_missing_profiles): Fix computation of tp_first_run.
10362 (counts_to_freqs): Do not touch freqs when count is 0.
10364 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10366 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10369 2017-06-10 Tom de Vries <tom@codesourcery.com>
10371 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10372 attributes): Document signal effective target.
10374 2017-06-10 Tom de Vries <tom@codesourcery.com>
10376 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10377 Document effective target stack_size.
10379 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10381 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10382 to the edit_context if they can be auto-applied.
10384 2017-06-9 Ian Lance Taylor <iant@golang.org>
10386 * opts.c (finish_options): If -fsplit-stack, disable implicit
10387 -forder-blocks-and-partition.
10388 * doc/invoke.texi (Optimize Options): Document that when using
10389 -fsplit-stack -forder-blocks-and-partition is not implicitly
10392 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10394 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10395 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10396 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10397 * builtins.def (abort, trap, unreachable): Declare cold.
10398 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10399 * tree-core.h (ECF_COLD): New.
10400 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10401 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10403 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10405 * predict.c (unlikely_executed_stmt_p): Cleanup.
10407 2017-06-09 Richard Biener <rguenther@suse.de>
10409 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10410 model if the ref is always written to.
10412 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10414 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10416 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10418 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10421 2017-06-09 Tom de Vries <tom@codesourcery.com>
10424 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10425 "target cannot support label values" when encountering LABEL_REF.
10427 2017-06-09 Martin Liska <mliska@suse.cz>
10429 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10430 (gimple_gen_ic_func_profiler): Emit direct comparison
10431 of __gcov_indirect_call_callee with NULL.
10432 (gimple_gen_time_profiler): Change probability from
10433 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10435 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10437 * profile.c (edge_gcov_counts): Turn to pointer.
10438 (compute_branch_probabilities, compute_branch_probabilities): Update.
10439 (branch_prob): Do not clear edge_gcov_count.
10440 * profile.h (edge_gcov_counts): Turn to pointer.
10441 (edge_gcov_count): Update.
10443 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10445 * gimple.h (gimple_check_failed): Mark cold.
10447 2017-06-09 Richard Biener <rguenther@suse.de>
10449 PR tree-optimization/66623
10450 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10451 refactor check_reduction into two parts, properly computing
10452 whether we have to check reduction validity for outer loop
10455 2017-06-09 Richard Biener <rguenther@suse.de>
10457 PR tree-optimization/79483
10458 * graphite-scop-detection.c (order): New global.
10459 (get_order): Compute bb to order mapping that satisfies code
10460 generation constraints.
10461 (cmp_pbbs): New helper.
10462 (build_scops): Start domwalk at entry block, sort generated
10465 2017-06-09 Richard Biener <rguenther@suse.de>
10467 PR middle-end/81007
10468 * ipa-polymorphic-call.c
10469 (ipa_polymorphic_call_context::restrict_to_inner_class):
10470 Skip FIELD_DECLs with error_mark_node type.
10471 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10474 2017-06-09 Martin Liska <mliska@suse.cz>
10476 * predict.c (struct branch_predictor): New struct.
10477 (test_prediction_value_range): New test.
10478 (predict_c_tests): New function.
10479 * selftest-run-tests.c (selftest::run_tests): Run the function.
10480 * selftest.h: Declare new tests.
10482 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10485 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10486 gen_add3_insn did not fail.
10487 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10488 r0, construct that number in a temporary reg and add that reg to r0.
10489 If asked to put the result in r0 as well, fail.
10491 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10494 for early expansion of vec_eqv.
10496 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10498 PR middle-end/81005
10499 * ubsan.c (instrument_null): Avoid pointless code temporary.
10500 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10502 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10506 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10509 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10512 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10514 * predict.c (unlikely_executed_edge_p): Move ahead.
10515 (probably_never_executed_edge_p): Use it.
10517 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10519 PR middle-end/79988
10520 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10521 gimple_call_builtin_p call.
10523 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10525 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10526 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10527 rtl_check_failed_type2, rtl_check_failed_code1,
10528 rtl_check_failed_code2, rtl_check_failed_code_mode,
10529 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10530 rtvec_check_failed_bounds, rtl_check_failed_flag,
10531 _fatal_insn_not_found, _fatal_insn): Likewise.
10532 * tree.h (tree_contains_struct_check_failed,
10533 tree_check_failed, tree_not_check_failed,
10534 tree_class_check_failed, tree_range_check_failed,
10535 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10536 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10537 tree_operand_check_failed, omp_clause_check_failed,
10538 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10543 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10544 flag_branch_probabilities.
10545 * ipa-inline.c (edge_badness): Likewise.
10546 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10547 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10548 * predict.c (maybe_hot_frequency_p): Likewise.
10549 (probably_never_executed): Likewise.
10550 * sched-ebb.c (schedule_ebbs): Likewise.
10551 * sched-rgn.c (find_single_block_region): Likewise.
10552 * tracer.c (tail_duplicate): Likewise.
10554 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10556 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10557 longer requires x_flag_profile_use.
10559 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10561 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10562 instead of flag_reorder_blocks_and_partition.
10563 * dbxout.c (dbxout_function_end): Likewise.
10564 * dwarf2out.c (gen_subprogram_die): Likewise.
10565 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10566 * hw-doloop.c (reorg_loops): Likewise.
10567 * varasm.c (assemble_start_function,
10568 assemble_end_function): Likewise.
10569 (decide_function_section): Do not check for
10570 flag_reorder_blocks_and_partition.
10572 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10574 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10576 (chkp_get_hard_register_fake_addr_expr): Ditto.
10577 (chkp_build_addr_expr): Add check for hard reg case.
10578 (chkp_parse_array_and_component_ref): Ditto.
10579 (chkp_find_bounds_1): Ditto.
10580 (chkp_process_stmt): Don't generate bounds store for
10583 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10585 * predict.c (maybe_hot_bb_p): Do not check profile status.
10586 (maybe_hot_edge_p): Likewise.
10587 (probably_never_executed): Check for zero counts even if profile
10589 (unlikely_executed_edge_p): New function.
10590 (unlikely_executed_stmt_p): New function.
10591 (unlikely_executed_bb_p): New function.
10592 (set_even_probabilities): Use unlikely predicates.
10593 (combine_predictions_for_bb): Likewise.
10594 (predict_paths_for_bb): Likewise.
10595 (predict_paths_leading_to_edge): Likewise.
10596 (determine_unlikely_bbs): New function.
10597 (estimate_bb_frequencies): Use it.
10598 (compute_function_frequency): Use zero counts even if profile is
10600 * profile-count.h: Fix typo.
10602 2017-08-08 Julia Koval <julia.koval@intel.com>
10604 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10605 _mm512_mask_cvtsepi16_storeu_epi8,
10606 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10607 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10608 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10609 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10610 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10611 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10612 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10613 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10614 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10615 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10616 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10617 __builtin_ia32_pmovuswb256mem_mask,
10618 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10619 __builtin_ia32_pmovwb512mem_mask): New builtins.
10621 2017-08-08 Julia Koval <julia.koval@intel.com>
10623 PR target/73350,80862
10624 * config/i386/subst.md (round): Fix round pattern.
10625 * config/i386/i386.c (ix86_erase_embedded_rounding):
10626 Fix erasing rounding for the fixed pattern.
10628 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10630 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10632 2017-06-08 Martin Liska <mliska@suse.cz>
10634 PR gcov-profile/80911
10635 * gcov.c (block_info::block_info): New constructor.
10637 2017-06-07 Carl Love <cel@us.ibm.com>
10639 * config/rs6000/rs6000-c: The return type of the following
10640 built-in functions was implemented as int not long long. Fix sign
10641 of return value for the unsigned version of vec_mulo and vec_mule.
10642 vector unsigned long long vec_bperm (vector unsigned long long,
10643 vector unsigned char)
10644 vector signed long long vec_mule (vector signed int,
10646 vector unsigned long long vec_mule (vector unsigned int,
10647 vector unsigned int)
10648 vector signed long long vec_mulo (vector signed int,
10650 vector unsigned long long vec_mulo (vector unsigned int,
10651 vector unsigned int)
10652 * doc/extend.texi: Fix the documentation for the built-in
10655 2017-06-07 Carl Love <cel@us.ibm.com>
10658 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10661 2017-06-07 Carl Love <cel@us.ibm.com>
10663 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10664 support, Generate doublehv for signed int/float for BE case only.
10666 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10668 * doc/invoke.texi (mcx16): Rewrite.
10670 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10672 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10673 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10674 *mov<mode>_softfloat, and an anonymous splitter): Use
10675 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10679 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10681 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10682 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10683 (rs6000_debug_reg_global): Adjust.
10684 (rs6000_init_hard_regno_mode_ok): Adjust.
10685 (rs6000_dbx_register_number): Adjust.
10686 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10687 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10688 Remove SPE_ACC and SPEFSCR.
10689 (REG_ALLOC_ORDER): Ditto.
10690 (FRAME_POINTER_REGNUM): Change to 111.
10691 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10692 (REG_CLASS_NAMES): Ditto.
10693 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10694 (REGISTER_NAMES): Ditto.
10695 (ADDITIONAL_REG_NAMES): Ditto.
10696 (rs6000_reg_names): Ditto.
10697 * config/rs6000/rs6000.md: Renumber some register number
10700 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10702 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10704 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10705 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10707 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10708 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10709 Delete the SPE high registers.
10710 (REG_ALLOC_ORDER): Ditto.
10711 (enum reg_class): Remove SPE_HIGH_REGS.
10712 (REG_CLASS_NAMES): Ditto.
10713 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10714 (REGISTER_NAMES): Ditto.
10715 (rs6000_reg_names): Ditto.
10716 * doc/tm.texi.in: Remove SPE as example.
10717 * doc/tm.texi: Regenerate.
10719 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10721 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10722 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10723 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10724 * config/rs6000/rs6000.md (type): Remove "brinc".
10726 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10728 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10729 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10730 * config/rs6000/linuxspe.h: Delete file.
10731 * config/rs6000/rs6000.md: Don't include spe.md.
10732 * config/rs6000/spe.h: Delete file.
10733 * config/rs6000/spe.md: Delete file.
10734 * config/rs6000/t-rs6000: Remove spe.md.
10736 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10738 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10739 (reg_or_none500mem_operand): Delete.
10740 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10741 instead of reg_or_none500mem_operand.
10743 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10745 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10746 handling of SPE flags.
10747 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10749 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10751 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10753 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10754 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10755 paired_divv2sf3): Similar.
10756 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10757 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10758 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10760 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10761 Rename the paired_* instruction patterns.
10762 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10764 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10765 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10766 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10767 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10768 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10769 PAIRED_VECTOR_MODE.
10770 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10771 (spe_func_has_64bit_regs_p): Delete.
10772 (spe_expand_predicate_builtin): Delete.
10773 (spe_expand_evsel_builtin): Delete.
10774 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10775 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10776 (invalid_e500_subreg): Delete.
10777 (rs6000_legitimize_address): Always force_reg op2 as well, for
10778 paired single memory accesses.
10779 (rs6000_member_type_forces_blk): Delete.
10780 (rs6000_spe_function_arg): Delete.
10781 (rs6000_expand_unop_builtin): Delete SPE handling.
10782 (rs6000_expand_binop_builtin): Ditto.
10783 (spe_expand_stv_builtin): Delete.
10784 (bdesc_2arg_spe): Delete.
10785 (spe_expand_builtin): Delete.
10786 (spe_expand_predicate_builtin): Delete.
10787 (spe_expand_evsel_builtin): Delete.
10788 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10789 (spe_init_builtins): Delete.
10790 (spe_func_has_64bit_regs_p): Delete.
10791 (savres_routine_name): Delete "info" parameter. Adjust callers.
10792 (rs6000_emit_stack_reset): Ditto.
10793 (rs6000_dwarf_register_span): Delete.
10794 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10795 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10796 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10798 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10800 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10801 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10802 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10803 mulv2sf3, divv2sf3): Delete expanders.
10805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10807 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10809 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10811 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10812 * config/rs6000/rs6000.c: Ditto.
10814 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10816 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10817 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10818 comparison_operator.
10820 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10822 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10823 * config/rs6000/rs6000.opt: Ditto.
10824 * config/rs6000/t-rtems: Ditto.
10826 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10828 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10829 TARGET_E500_SINGLE by 0, simplify.
10830 * config/rs6000/rs6000.c: Ditto.
10831 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10832 (spe_build_register_parallel): Delete.
10833 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10834 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10835 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10836 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10837 (E500_CONVERT): Delete.
10838 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10842 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10843 * config/rs6000/dfp.md: Ditto.
10844 (negdd2, *negdd2_fpr): Merge.
10845 (absdd2, *absdd2_fpr): Merge.
10846 (negtd2, *negtd2_fpr): Merge.
10847 (abstd2, *abstd2_fpr): Merge.
10848 * config/rs6000/e500.h: Delete file.
10849 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10850 TARGET_FPRS by 1 and simplify.
10851 * config/rs6000/rs6000-c.c: Ditto.
10852 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10853 TARGET_DF_SPE by 0.
10854 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10856 * config/rs6000/rs6000.md: Ditto.
10857 (floatdidf2, *floatdidf2_fpr): Merge.
10858 (move_from_CR_gt_bit): Delete.
10859 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10860 (E500_CR_IOR_COMPARE): Delete.
10861 (All patterns that require !TARGET_FPRS): Delete.
10862 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10864 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10866 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10868 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10870 * graphds.c (add_edge): Intitialize edge's attached data.
10871 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10872 pointer parameter. Call pointed function on each edge during
10873 graph traversing. Skip traversing the edge when the function
10875 (graphds_dfs, graphds_scc): Ditto.
10876 (for_each_edge): New parameter. Pass the new parameter to callback
10878 * graphds.h (skip_edge_callback): New function pointer type.
10879 (graphds_dfs, graphds_scc): New function pointer parameter.
10880 (graphds_edge_callback, for_each_edge): New parameter.
10882 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10884 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10885 out code checking if runtime alias check is possible to below ...
10886 Call the new function.
10887 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10888 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10890 2017-06-07 Marek Polacek <polacek@redhat.com>
10893 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10894 TYPE_OVERFLOW_WRAPS check.
10896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10898 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10899 if versioning is required.
10900 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10901 peeling with the check for versioning.
10903 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10905 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10906 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10907 Set true to new parameter if new ssa variable is defined.
10908 (vect_gen_vector_loop_niters): Refactor. Set range information
10909 for the new vector loop bound variable.
10910 (vect_do_peeling): Ditto.
10912 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10914 * tree-affine.c (ssa.h): Include header file.
10915 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10916 has wrapping overflow behavior.
10918 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10920 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10922 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10924 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10925 (tree_to_aff_combination): ... here.
10927 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10929 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10930 reg_pressure model function.
10931 (ivopts_global_cost_for_size): Delete.
10932 (determine_set_costs, iv_ca_recount_cost): Call new model function
10933 ivopts_estimate_reg_pressure.
10935 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10937 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10938 expensive than udiv. Remove floating point cases from mod.
10940 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10942 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10943 Increase idiv cost.
10945 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10947 * config/aarch64/aarch64.md
10948 (copysignsf3): Fix mask generation.
10950 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10952 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10954 (class dump_manager): Add register_dumps method.
10955 * dumpfile.c: Include langhooks.h.
10956 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10957 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10958 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10959 (dump_manager::dump_register): Start with 512 entries instead of 32.
10960 (dump_manager::register_dumps): New method.
10961 * toplev.c (general_init): Instead of invoking register_dumps
10962 langhook, invoke register_dumps method on the dump manager.
10963 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10966 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10968 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10969 Say that # requires an associated define_split to exist, and that
10970 the define_split must be suitable for use after register allocation.
10972 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10974 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10975 (compute_outgoing_frequencies): Also initialize zero counts.
10976 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10977 around loops; preserve more of profile when nothing changes.
10979 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10981 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10983 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10984 * config/arm/arm-cpu-cdata.h: Regenerate.
10985 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10986 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10987 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10988 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10989 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10991 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10993 * config/arm/t-rmprofile: Likewise.
10994 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10996 2017-06-06 David S. Miller <davem@davemloft.net>
10999 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
11000 blockage if function uses alloca.
11002 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11004 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
11005 New "uid" fields to hold pretty-print IDs of group and ref.
11006 Memory references are now identified as <group_id>:<ref_id>
11007 instead of using [random] addresses.
11008 (dump_mem_details): Simplify, no functional change.
11009 (dump_mem_ref): Simplify and make output more concise.
11010 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
11011 (find_or_create_group): Initialize group uid.
11012 (record_ref): Initialize ref uid. Improve debug output.
11013 (prune_group_by_reuse, should_issue_prefetch_p,)
11014 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
11015 (mark_nontemporal_store, determine_loop_nest_reuse):
11016 Improve debug output.
11018 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11020 * dbgcnt.def (prefetch): New debug counter.
11021 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
11022 (schedule_prefetches): Stop issueing prefetches if debug counter
11025 2017-06-06 Tom de Vries <tom@codesourcery.com>
11027 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
11028 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
11030 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11032 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
11033 Use aarch64_reg_or_zero predicate for operand 4.
11034 (aarch64_compare_and_swap<mode> define_insn_and_split):
11035 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
11036 (aarch64_store_exclusive<mode>): Likewise for operand 2.
11038 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11040 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
11041 (arm_compute_save_core_reg_mask): This.
11042 (thumb1_compute_save_reg_mask): Rename into ...
11043 (thumb1_compute_save_core_reg_mask): This.
11044 (arm_compute_save_reg0_reg12_mask): Adapt comment.
11045 (arm_compute_frame_layout): Likewise.
11047 2017-06-06 Richard Biener <rguenther@suse.de>
11049 PR tree-optimization/80974
11050 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
11051 keep or clear leaders SSA info.
11053 2017-06-06 Tom de Vries <tom@codesourcery.com>
11055 * config/nvptx/nvptx.c (split_mode_p): New function.
11056 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
11058 2017-06-06 Tom de Vries <tom@codesourcery.com>
11060 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
11062 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11065 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11068 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11070 * shrink-wrap.c (handle_simple_exit): Update profile.
11071 (try_shrink_wrapping): Upate profile.
11073 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11075 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11076 (tree_guess_outgoing_edge_probabilities): New.
11077 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11078 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11080 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11082 * ipa-split.c (split_function): Initialize return bb profile.
11084 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11086 * profile.c (compute_branch_probabilities): Also initialize
11087 EXIT_BLOCK profile.
11089 2017-06-06 Richard Biener <rguenther@suse.de>
11091 PR tree-optimization/80928
11092 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11093 (vect_analyze_loop_operations): Properly guard analysis for
11095 (vect_transform_loop): Likewise.
11096 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11097 (vect_model_induction_cost): Do not cost for pure SLP.
11098 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11099 of induction in inner loop vectorization.
11100 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11101 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11102 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11104 (vect_analyze_slp_cost_1): Cost induction.
11105 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11106 (vect_get_slp_vect_defs): Likewise.
11107 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11108 (vect_transform_stmt): Handle SLP reductions.
11109 * tree-vectorizer.h (vectorizable_induction): Adjust.
11111 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11113 * config/rs6000/rs6000.c (make_resolver_func): Update
11114 init_lowered_empty_function call.
11116 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11118 * doc/invoke.texi: Document the -fprofile-abs-path option.
11119 * common.opt (fprofile-abs-path): New option.
11120 * gcov-io.h (gcov_write_filename): Declare.
11121 * gcov-io.c (gcov_write_filename): New function.
11122 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11123 * profile.c (output_location): Likewise.
11125 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11127 * shring-wrap.c: Revert accidental commit.
11129 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11131 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11133 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11135 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11137 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11138 profile in callgraph edge.
11139 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11140 (apply_scale): Likewise.
11141 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11142 Also scale profile when inlining function with zero profile.
11143 (initialize_cfun): Update exit block profile even when it is zero.
11144 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11145 when profile is read.
11147 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11149 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11150 (CLONE_*): New constants to define the processors we can generate
11151 code for with the target_clone attribute.
11152 (rs6000_clone_map): New array to identify which clone processors
11153 the current program is running on.
11154 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11155 target_clone attribute.
11156 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11157 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11158 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11159 (cpu_expand_builtin): Add support for target_clone attribute.
11160 (rs6000_valid_attribute_p): Allow "default" attribute.
11161 (get_decl_name): New debug function to simplify printing the
11162 current function name in debugging statements.
11163 (rs6000_clone_priority): New functions to support the target_clone
11164 attribute, and be able to generate code to switch between ISA 2.05
11165 through ISA 3.0 (power6 through power9).
11166 (rs6000_compare_version_priority): Likewise.
11167 (rs6000_get_function_versions_dispatcher): Likewise.
11168 (make_resolver_func): Likewise.
11169 (add_condition_to_bb): Likewise.
11170 (dispatch_function_versions): Likewise.
11171 (rs6000_generate_version_dispatcher_body): Likewise.
11172 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11173 (fusion_gpr_load_p): Fix a spacing issue.
11174 * doc/extend.texi (Common Function Attributes): Document that the
11175 PowerPC supports the target_clone attribute.
11177 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11179 * config/arm/arm.h: explain F symbol found in description of ARM
11180 register allocation in its legend.
11182 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11184 * config/mips/frame-header-opt.c: Include profile-count.h.
11185 * config/riscv/riscv.c: Include profile-count.h
11187 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11189 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11191 (sm_set_flag_if_changed): Add bbs field.
11192 (execute_sm_if_changed_flag_set): Pass BBS.
11193 (execute_sm): Update.
11195 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11197 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11200 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11202 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11203 (peephole2): New peephole2 to emit the above.
11204 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11206 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11208 * config/aarch64/aarch64.c (define_peephole2 above
11209 *sub_<shift>_<mode>): New peephole.
11211 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11213 * config/i386/i386.c (make_resolver_func): Update.
11214 * Makefile.in: Add profile-count.h and profile-count.o
11215 * auto-profile.c (afdo_indirect_call): Update to new API.
11216 (afdo_set_bb_count): Update.
11217 (afdo_propagate_edge): Update.
11218 (afdo_propagate_circuit): Update.
11219 (afdo_calculate_branch_prob): Update.
11220 (afdo_annotate_cfg): Update.
11221 * basic-block.h: Include profile-count.h
11222 (struct edge_def): Turn count to profile_count.
11223 (struct basic_block_def): Likewie.
11224 (REG_BR_PROB_BASE): Move to profile-count.h
11225 (RDIV): Move to profile-count.h
11226 * bb-reorder.c (max_entry_count): Turn to profile_count.
11227 (find_traces): Update.
11228 (rotate_loop):Update.
11229 (connect_traces):Update.
11230 (sanitize_hot_paths):Update.
11231 * bt-load.c (migrate_btr_defs): Update.
11232 * cfg.c (RDIV): Remove.
11233 (init_flow): Use alloc_block.
11234 (alloc_block): Uninitialize count.
11235 (unchecked_make_edge): Uninitialize count.
11236 (check_bb_profile): Update.
11237 (dump_edge_info): Update.
11238 (dump_bb_info): Update.
11239 (update_bb_profile_for_threading): Update.
11240 (scale_bbs_frequencies_int): Update.
11241 (scale_bbs_frequencies_gcov_type): Update.
11242 (scale_bbs_frequencies_profile_count): New.
11243 * cfg.h (update_bb_profile_for_threading): Update.
11244 (scale_bbs_frequencies_profile_count): Declare.
11245 * cfgbuild.c (compute_outgoing_frequencies): Update.
11246 (find_many_sub_basic_blocks): Update.
11247 * cfgcleanup.c (try_forward_edges): Update.
11248 (try_crossjump_to_edge): Update.
11249 * cfgexpand.c (expand_gimple_tailcall): Update.
11250 (construct_exit_block): Update.
11251 * cfghooks.c (verify_flow_info): Update.
11252 (dump_bb_for_graph): Update.
11253 (split_edge): Update.
11254 (make_forwarder_block): Update.
11255 (duplicate_block): Update.
11256 (account_profile_record): Update.
11257 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11258 (get_estimated_loop_iterations): Update.
11259 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11260 (single_likely_exit): Update.
11261 * cfgloopmanip.c (scale_loop_profile): Update.
11263 (set_zero_probability): Update.
11264 (lv_adjust_loop_entry_edge): Update.
11265 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11266 (purge_dead_edges): Update.
11267 (rtl_account_profile_record): Update.
11268 * cgraph.c (cgraph_node::create): Uninitialize count.
11269 (symbol_table::create_edge): Uninitialize count.
11270 (cgraph_update_edges_for_call_stmt_node): Update.
11271 (cgraph_edge::dump_edge_flags): Update.
11272 (cgraph_node::dump): Update.
11273 (cgraph_edge::maybe_hot_p): Update.
11274 * cgraph.h: Include profile-count.h
11275 (create_clone), create_edge, create_indirect_edge): Update.
11276 (cgraph_node): Turn count to profile_count.
11277 (cgraph_edge0: Likewise.
11278 (make_speculative, clone): Update.
11279 (create_edge): Update.
11280 (init_lowered_empty_function): Update.
11281 * cgraphclones.c (cgraph_edge::clone): Update.
11282 (duplicate_thunk_for_node): Update.
11283 (cgraph_node::create_clone): Update.
11284 * cgraphunit.c (cgraph_node::analyze): Update.
11285 (cgraph_node::expand_thunk): Update.
11286 * final.c (dump_basic_block_info): Update.
11287 * gimple-streamer-in.c (input_bb): Update.
11288 * gimple-streamer-out.c (output_bb): Update.
11289 * graphite.c (print_global_statistics): Update.
11290 (print_graphite_scop_statistics): Update.
11291 * hsa-brig.c: Include basic-block.h.
11292 * hsa-dump.c: Include basic-block.h.
11293 * hsa-gen.c (T sum_slice): Update.
11294 (convert_switch_statements):Update.
11295 * hsa-regalloc.c: Include basic-block.h.
11296 * ipa-chkp.c (chkp_produce_thunks): Update.
11297 * ipa-cp.c (struct caller_statistics): Update.
11298 (init_caller_stats): Update.
11299 (gather_caller_stats): Update.
11300 (ipcp_cloning_candidate_p): Update.
11301 (good_cloning_opportunity_p): Update.
11302 (get_info_about_necessary_edges): Update.
11303 (dump_profile_updates): Update.
11304 (update_profiling_info): Update.
11305 (update_specialized_profile): Update.
11306 (perhaps_add_new_callers): Update.
11307 (decide_about_value): Update.
11308 (ipa_cp_c_finalize): Update.
11309 * ipa-devirt.c (struct odr_type_warn_count): Update.
11310 (struct decl_warn_count): Update.
11311 (struct final_warning_record): Update.
11312 (possible_polymorphic_call_targets): Update.
11313 (ipa_devirt): Update.
11314 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11315 * ipa-icf.c (sem_function::merge): Update.
11316 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11317 * ipa-inline.c (compute_uninlined_call_time): Update.
11318 (compute_inlined_call_time): Update.
11319 (want_inline_small_function_p): Update.
11320 (want_inline_self_recursive_call_p): Update.
11321 (edge_badness): Update.
11322 (lookup_recursive_calls): Update.
11323 (recursive_inlining): Update.
11324 (inline_small_functions): Update.
11325 (dump_overall_stats): Update.
11326 (dump_inline_stats): Update.
11327 * ipa-profile.c (ipa_profile_generate_summary): Update.
11328 (ipa_propagate_frequency): Update.
11329 (ipa_profile): Update.
11330 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11331 * ipa-utils.c (ipa_merge_profiles): Update.
11332 * loop-doloop.c (doloop_modify): Update.
11333 * loop-unroll.c (report_unroll): Update.
11334 (unroll_loop_runtime_iterations): Update.
11335 * lto-cgraph.c (lto_output_edge): Update.
11336 (lto_output_node): Update.
11337 (input_node): Update.
11338 (input_edge): Update.
11339 (merge_profile_summaries): Update.
11340 * lto-streamer-in.c (input_cfg): Update.
11341 * lto-streamer-out.c (output_cfg): Update.
11342 * mcf.c (create_fixup_graph): Update.
11343 (adjust_cfg_counts): Update.
11344 (sum_edge_counts): Update.
11345 * modulo-sched.c (sms_schedule): Update.
11346 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11347 * predict.c (maybe_hot_count_p): Update.
11348 (probably_never_executed): Update.
11349 (dump_prediction): Update.
11350 (combine_predictions_for_bb): Update.
11351 (propagate_freq): Update.
11352 (handle_missing_profiles): Update.
11353 (counts_to_freqs): Update.
11354 (rebuild_frequencies): Update.
11355 (force_edge_cold): Update.
11356 * predict.h: Include profile-count.h
11357 (maybe_hot_count_p, counts_to_freqs): UPdate.
11358 * print-rtl-function.c: Do not include cfg.h
11359 * print-rtl.c: Include basic-block.h
11360 * profile-count.c: New file.
11361 * profile-count.h: New file.
11362 * profile.c (is_edge_inconsistent): Update.
11363 (correct_negative_edge_counts): Update.
11364 (is_inconsistent): Update.
11365 (set_bb_counts): Update.
11366 (read_profile_edge_counts): Update.
11367 (compute_frequency_overlap): Update.
11368 (compute_branch_probabilities): Update; Initialize and deinitialize
11370 (branch_prob): Update.
11371 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11372 (edge_gcov_count): New.
11373 (bb_gcov_count): New.
11374 * shrink-wrap.c (try_shrink_wrapping): Update.
11375 * tracer.c (better_p): Update.
11376 * trans-mem.c (expand_transaction): Update.
11377 (ipa_tm_insert_irr_call): Update.
11378 (ipa_tm_insert_gettmclone_call): Update.
11379 * tree-call-cdce.c: Update.
11380 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11381 (gimple_duplicate_sese_tail): Update.
11382 (gimple_account_profile_record): Update.
11383 (execute_fixup_cfg): Update.
11384 * tree-inline.c (copy_bb): Update.
11385 (copy_edges_for_bb): Update.
11386 (initialize_cfun): Update.
11387 (freqs_to_counts): Update.
11388 (copy_cfg_body): Update.
11389 (expand_call_inline): Update.
11390 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11391 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11392 (try_unroll_loop_completely): Update.
11393 (try_peel_loop): Update.
11394 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11395 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11396 * tree-ssa-loop-split.c (connect_loops): Update.
11397 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11398 * tree-ssa-reassoc.c (branch_fixup): Update.
11399 * tree-ssa-tail-merge.c (replace_block_by): Update.
11400 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11401 (compute_path_counts): Update.
11402 (update_profile): Update.
11403 (recompute_probabilities): Update.
11404 (update_joiner_offpath_counts): Update.
11405 (estimated_freqs_path): Update.
11406 (freqs_to_counts_path): Update.
11407 (clear_counts_path): Update.
11408 (ssa_fix_duplicate_block_edges): Update.
11409 (duplicate_thread_path): Update.
11410 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11411 (struct switch_conv_info): Update.
11412 * tree-tailcall.c (decrease_profile): Update.
11413 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11414 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11415 * value-prof.c (check_counter): Update.
11416 (gimple_divmod_fixed_value): Update.
11417 (gimple_mod_pow2): Update.
11418 (gimple_mod_subtract): Update.
11419 (gimple_ic_transform): Update.
11420 (gimple_stringop_fixed_value): Update.
11421 * value-prof.h (gimple_ic): Update.
11423 2017-06-02 Carl Love <cel@us.ibm.com>
11425 * config/rs6000/rs6000-c: Add support for built-in functions
11426 vector double vec_doublee (vector signed int);
11427 vector double vec_doublee (vector unsigned int);
11428 vector double vec_doublee (vector float);
11429 vector double vec_doubleh (vector signed int);
11430 vector double vec_doubleh (vector unsigned int);
11431 vector double vec_doubleh (vector float);
11432 vector double vec_doublel (vector signed int);
11433 vector double vec_doublel (vector unsigned int);
11434 vector double vec_doublel (vector float);
11435 vector double vec_doubleo (vector signed int);
11436 vector double vec_doubleo (vector unsigned int);
11437 vector double vec_doubleo (vector float);.
11438 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11439 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11441 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11442 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11443 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11445 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11446 vec_doublel, vec_doubleh.
11447 * doc/extend.texi: Update the built-in documentation file for the
11448 new built-in functions.
11450 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11453 * ipa-inline-analysis.c (free_growth_caches): Set
11454 edge_removal_hook_holder to NULL after removing it.
11456 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11458 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11459 comparision with zero.
11461 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11462 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11463 for early expansion of vec_min and vec_max builtins.
11464 (builtin_function_type): Add min/max unsigned variants to those
11465 identified as having unsigned arguments.
11467 2017-06-02 Olivier Hainque <hainque@adacore.com>
11469 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11471 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11473 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11474 Use VALL_F16 iterator rather than VALL.
11476 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11478 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11479 Emit CBNZ inside loop when doing a strong exchange and comparing
11480 against zero. Generate the CC flags after the loop.
11482 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11484 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11485 (dl_section_ref): New.
11486 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11487 On AIX, append an expression to subtract the size of the
11488 section length to dl_section_ref.
11490 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11493 for early expansion of vector absolute builtins.
11495 2017-06-02 Richard Biener <rguenther@suse.de>
11497 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11498 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11500 2017-06-02 Richard Biener <rguenther@suse.de>
11502 PR tree-optimization/80948
11503 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11504 stmt order as well.
11506 2017-06-02 Richard Biener <rguenther@suse.de>
11508 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11510 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11511 for inductions as relevant.
11513 2017-06-02 Richard Biener <rguenther@suse.de>
11515 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11516 (vectorizable_induction): ... this. Remove dead code.
11518 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11520 * builtins. (expand_builtin_alloca): Remove second parameter and
11521 infer its value from the first parameter instead.
11522 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11524 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11526 PR rtl-optimization/80903
11527 * loop-doloop.c (add_test): Unshare sequence.
11529 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11531 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11533 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11535 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11537 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11538 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11539 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11540 (xlogue_layout::get_stub_rtx): Make static.
11541 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11542 (xlogue_layout::compute_stub_managed_regs): Rename to...
11543 (xlogue_layout::count_stub_managed_regs): ...this.
11544 (xlogue_layout::is_stub_managed_reg): New function.
11545 (xlogue_layout::m_stub_names): Rename to...
11546 (xlogue_layout::s_stub_names): ...this, make static.
11547 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11548 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11549 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11550 xlogue_layout::s_stub_names): Instantiate statics.
11551 (stub_managed_regs): Remove.
11552 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11553 (disable_call_ms2sysv_xlogues): Rename to...
11554 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11555 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11557 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11558 change after reload_completed.
11559 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11561 (ix86_expand_prologue): Likewise.
11562 (ix86_expand_epilogue): Likewise.
11563 (ix86_expand_split_stack_prologue): Likewise.
11564 (ix86_compute_frame_layout): Remove frame parameter ...
11565 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11566 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11568 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11569 (ix86_frame): Move from here ...
11570 * config/i386/i386.h (ix86_frame): ... to here.
11571 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11572 complete ix86_frame data structure instead. Remove some_ld_name.
11574 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11576 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11577 symbols that hold a DECL_VALUE_EXPR.
11579 2017-06-01 Martin Jambor <mjambor@suse.cz>
11581 PR tree-optimization/80898
11582 * tree-sra.c (process_subtree_disqualification): Removed.
11583 (disqualify_candidate): Do not acll
11584 process_subtree_disqualification.
11585 (subtree_mark_written_and_enqueue): New function.
11586 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11587 RHS has been disqualified and re-queue LHS if necessary. Apart
11588 from that, ignore disqualified RHS.
11590 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11592 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11593 address fetch for z10 or later.
11595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11597 * config/arc/arc.md (tst_movb): Add guard when splitting.
11599 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11601 * config/arc/arc.c (arc_can_eliminate): Test against
11602 arc_frame_pointer_needed.
11604 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11606 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11607 to prevent store reordering.
11608 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11609 (type): Add block type.
11610 (stack_tie): Define special instruction to be used in
11613 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11615 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11616 constraint. It is not valid for the pattern.
11617 (noncommutative_binary_comparison): Likewise.
11619 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11621 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11624 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11626 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11627 be used by the reg-alloc.
11629 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11631 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11632 reg-alloc when having mul64 or mul32x16 instructions.
11633 (mulsidi3): Likewise.
11634 (umulsidi3): Likewise.
11635 (mulsi32x16): New pattern.
11636 (mulsi64): Likewise.
11637 (mulsidi64): Likewise.
11638 (umulsidi64): Likewise.
11639 (MUL32x16_REG): Define.
11640 (mul64_600): Use MUL32x16_REG.
11641 (mac64_600): Likewise.
11642 (umul64_600): Likewise.
11643 (umac64_600): Likewise.
11645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11647 * config/arc/arc.md (mulsi3_700): Make it commutative.
11649 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11651 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11653 (*sign_extendsidi2_insn): Likewise for movstosw.
11655 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11657 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11658 the type of the input discriminant value. Convert the
11659 discriminant value of signedness vary.
11661 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11663 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11664 Add to -Wall section.
11666 2017-06-01 Richard Biener <rguenther@suse.de>
11668 PR middle-end/66313
11669 * fold-const.c (fold_plusminus_mult_expr): If the factored
11670 factor may be zero use a wrapping type for the inner operation.
11671 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11672 and handle moved defs.
11673 (process_assignment): Properly guard the unary op case. Return a
11674 tri-state indicating that moving the stmt before the call may allow
11675 to continue. Pass through to_move.
11676 (find_tail_calls): Handle moving unrelated defs before
11679 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11682 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11683 splitter result in the canonical way.
11685 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11687 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11688 also for 32bit target. Update insn attributes.
11689 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11691 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11693 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11694 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11695 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11696 (_mm_maskz_min_ss): New intrinsics.
11698 2017-05-31 Martin Liska <mliska@suse.cz>
11700 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11701 Change comment style to one we normally use.
11702 (vectorizable_reduction): Likewise.
11703 (vectorizable_induction): Likewise.
11704 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11705 (vectorizable_call): Likewise.
11706 (vectorizable_simd_clone_call): Likewise.
11707 (vectorizable_conversion): Likewise.
11708 (vectorizable_assignment): Likewise.
11709 (vectorizable_shift): Likewise.
11710 (vectorizable_operation): Likewise.
11711 (vectorizable_store): Likewise.
11712 (vectorizable_load): Likewise.
11713 * tree-vectorizer.h: Likewise.
11715 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11717 * passes.c (emergency_dump_function): New.
11718 * tree-pass.h (emergency_dump_function): Declare.
11719 * plugin.c (plugins_internal_error_function): Remove.
11720 * plugin.h (plugins_internal_error_function): Remove declaration.
11721 * toplev.c (internal_error_function): New static function. Use it...
11722 (general_init): ...here.
11724 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11726 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11727 (arc_rtx_costs): Add costs for new patterns.
11728 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11729 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11731 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11733 * tree-ssa-strlen.c (get_next_strinfo): New function.
11734 (get_stridx_plus_constant): Use it.
11735 (zero_length_string): Likewise.
11736 (adjust_related_strinfos): Likewise.
11737 (adjust_last_stmt): Likewise.
11739 2017-05-31 Richard Biener <rguenther@suse.de>
11742 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11743 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11745 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11747 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11748 loop_vinfo argument and use of dependence distance vectors.
11749 Check instead whether the two references differ only in their
11750 initial value and assume that they have the same alignment if the
11751 difference is a multiple of the vector alignment.
11752 (vect_analyze_data_refs_alignment): Update call accordingly.
11754 2017-05-31 Martin Liska <mliska@suse.cz>
11757 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11759 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11761 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11762 (create_intersect_range_checks): Move from ...
11763 * tree-data-ref.c (create_intersect_range_checks_index)
11764 (create_intersect_range_checks): ... to here.
11765 (create_runtime_alias_checks): New function factored from ...
11766 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11767 here. Call above function.
11768 * tree-data-ref.h (create_runtime_alias_checks): New function.
11770 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11772 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11773 segment length for dr_b and compute it in wide_int.
11775 2017-05-31 Richard Biener <rguenther@suse.de>
11777 PR tree-optimization/80906
11778 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11779 and pass through iv_map.
11780 (copy_bb_and_scalar_dependences): Adjust.
11781 (translate_pending_phi_nodes): Likewise.
11782 (copy_loop_close_phi_args): Handle code-generating IVs instead
11785 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11787 * diagnostic-color.c (color_dict): Add "type-diff".
11788 (parse_gcc_colors): Update comment.
11789 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11790 -fdiagnostics-show-template-tree and -fno-elide-type.
11791 (GCC_COLORS): Add type-diff to example.
11793 (-fdiagnostics-show-template-tree): New.
11794 (-fno-elide-type): New.
11795 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11796 the pp_format_decoder callback. Call any m_format_postprocessor's
11798 (pretty_printer::pretty_printer): Initialize
11799 m_format_postprocessor.
11800 (pretty_printer::~pretty_printer): Delete any
11801 m_format_postprocessor.
11802 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11803 (class format_postprocessor): New class.
11804 (struct pretty_printer::format_decoder): Document the new parameters.
11805 (struct pretty_printer::m_format_postprocessor): New field.
11806 * tree-diagnostic.c (default_tree_printer): Update for new
11807 bool and const char ** params.
11808 * tree-diagnostic.h (default_tree_printer): Likewise.
11810 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11812 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11813 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11814 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11815 rs6000_gen_cell_microcode code.
11816 (rs6000_final_prescan_insn): Delete.
11817 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11818 "warn-cell-microcode" entries.
11819 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11820 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11821 throughout. Change cc_reg_not_micro_cr0_operand to
11822 cc_reg_not_cr0_operand throughout.
11823 (*extendhi<mode>2_noload): Delete.
11824 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11825 (mwarn-cell-microcode): Delete.
11826 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11827 -mgen-cell-microcode and -mwarn-cell-microcode.
11829 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11832 * config/i386/constraints.md (Yd): New constraint.
11834 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11835 and (?Yd, r) alternatives. Update insn attributes.
11836 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11837 and (?*Yd, r) alternatives. Update insn attributes.
11838 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11840 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11842 * gimplify.c (gimplify_modify_expr): Don't create a
11843 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11846 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11848 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11850 2017-05-30 Richard Biener <rguenther@suse.de>
11852 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11853 and reduc_def fields.
11854 (STMT_VINFO_REDUC_TYPE): New define.
11855 (STMT_VINFO_REDUC_DEF): Likewise.
11856 (vect_force_simple_reduction): Adjust prototype.
11857 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11858 (vect_is_simple_reduction): Remove check_reduction argument.
11859 (vect_force_simple_reduction): Adjust and set
11860 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11861 (vectorizable_reduction): Do not re-do reduction analysis
11862 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11863 * tree-parloops.c (gather_scalar_reductions): Adjust.
11865 2017-05-30 Richard Biener <rguenther@suse.de>
11867 PR middle-end/80901
11868 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11871 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11873 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11874 Introduce unknown_misalignment parameter and remove vf.
11875 (vect_peeling_hash_get_lowest_cost):
11876 Pass unknown_misalignment parameter.
11877 (vect_enhance_data_refs_alignment):
11878 Fix unsupportable data ref treatment.
11880 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11882 * tree-vect-data-refs.c (vect_get_data_access_cost):
11883 Workaround for SLP handling.
11884 (vect_enhance_data_refs_alignment):
11885 Compute costs for doing no peeling at all, compare to the best
11886 peeling costs so far and avoid peeling if cheaper.
11888 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11890 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11891 Return peeling info and set costs to zero for unlimited cost
11893 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11894 with unknown misalignment. Compute and costs for unknown
11895 misalignment, compare them to the costs for known misalignment
11896 and choose the cheapest for peeling.
11898 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11900 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11901 (vect_get_peeling_costs_all_drs): Create function.
11902 (vect_peeling_hash_get_lowest_cost):
11903 Use vect_get_peeling_costs_all_drs.
11904 (vect_peeling_supportable): Create function.
11905 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11907 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11909 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11910 DR_HAS_NEGATIVE_STEP.
11911 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11912 (vect_enhance_data_refs_alignment): Use.
11913 (vect_duplicate_ssa_name_ptr_info): Use.
11914 * tree-vectorizer.h (dr_misalignment): Use.
11915 (known_alignment_for_access_p): Use.
11917 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11920 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11922 (has_section_name): New function.
11924 2017-05-30 Martin Liska <mliska@suse.cz>
11927 * auto-profile.c (get_function_decl_from_block): Fix
11930 2017-05-30 Richard Biener <rguenther@suse.de>
11932 PR middle-end/80876
11933 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11935 2017-05-30 Martin Liska <mliska@suse.cz>
11937 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11938 * dumpfile.h (struct dump_file_info): Remove ctors.
11940 2017-05-30 Martin Liska <mliska@suse.cz>
11942 * predict.def: Fix GNU coding style.
11944 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11946 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11947 Mark 'to' argument with ATTRIBUTE_UNUSED.
11949 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11951 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11952 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11953 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11956 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11958 * doc/install.texi (Options specification): Restore entry of
11959 --enable-sjlj-exceptions.
11961 2017-05-27 Michael Eager <eager@eagercon.com>
11964 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11966 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11968 * config/microblaze/microblaze.h
11969 (FIXED_REGISTERS): Update in macro.
11970 (CALL_USED_REGISTERS): Update in macro.
11972 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11974 * doc/install.texi: Add links to macOS binary distributions.
11976 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11980 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11982 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11984 2017-05-26 Martin Liska <mliska@suse.cz>
11986 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11988 2017-05-26 Martin Liska <mliska@suse.cz>
11990 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11991 always leading ';; '.
11992 (dump_bb_info): Likewise.
11993 (brief_dump_cfg): Likewise.
11994 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11995 * dumpfile.c: Remove usage of TDF_VERBOSE.
11996 * dumpfile.h (enum dump_kind): Likewise.
11997 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11998 * print-tree.c (debug_verbose): Remove.
11999 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
12000 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
12001 * tree-diagnostic.c (default_tree_printer): Replace
12002 TDF_DIAGNOSTIC with TDF_SLIM.
12004 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12006 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
12007 in parameter loop, rather than loop_vinfo.
12008 (create_intersect_range_checks): Ditto.
12009 (vect_create_cond_for_alias_checks): Update call to above functions.
12011 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12013 PR tree-optimization/80815
12014 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
12015 for merging runtime alias checks. Handle negative DR_STEPs.
12017 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12019 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
12021 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
12022 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
12023 out code pruning runtime alias checks.
12024 * tree-data-ref.c (prune_runtime_alias_test_list): New function
12025 factored out from above.
12026 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12028 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12030 (prune_runtime_alias_test_list): New decalaration.
12032 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12034 * tree-vect-data-refs.c (compare_tree): Rename and move ...
12035 * tree-data-ref.c (data_ref_compare_tree): ... to here.
12036 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
12037 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
12038 (operator==, comp_dr_with_seg_len_pair): Ditto.
12039 (vect_prune_runtime_alias_test_list): Ditto.
12041 2017-05-26 Martin Liska <mliska@suse.cz>
12044 * params.def: Bound partial-inlining-entry-probability param.
12046 2017-05-26 Marek Polacek <polacek@redhat.com>
12049 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
12052 2017-05-26 Richard Biener <rguenther@suse.de>
12054 PR tree-optimization/80842
12055 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
12058 2017-05-26 Richard Biener <rguenther@suse.de>
12060 PR tree-optimization/80844
12061 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
12063 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12065 * doc/md.texi (Machine Constraints): Update x86 family
12066 machine constraints section to match 'config/i386/constraints.md'.
12068 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12070 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12072 2017-05-25 Nathan Sidwell <nathan@acm.org>
12074 * doc/invoke.texi (--enable-languages): Update documentation.
12076 2017-05-25 Martin Liska <mliska@suse.cz>
12078 * dumpfile.c: Add TDF_FOLDING.
12079 * dumpfile.h (enum dump_kind): Likewise.
12080 * genmatch.c (dt_simplify::gen_1): Use it.
12082 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12084 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12086 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12088 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12089 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12091 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12093 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12094 * match.pd (X == C): Rewrite it here.
12095 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12096 with_certain_nonzero_bits2): New predicates.
12097 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12099 2017-05-24 Nathan Sidwell <nathan@acm.org>
12101 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12104 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12107 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12109 * config/powerpcspe: New port. Files are copied from the rs6000
12110 port, with "rs6000" in filenames replaced by "powerpcspe".
12112 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12114 PR rtl-optimization/80754
12115 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12117 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12119 * config/sparc/sparc.md (length): Return the correct value for -mflat
12120 sibcalls to match output_sibcall.
12122 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12126 * config/rs6000/rs6000.c (struct machine_function): Add new field
12128 (rs6000_get_separate_components): Init that field, use it.
12129 (rs6000_components_for_bb): Use the field.
12131 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12133 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12135 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12137 PR middle-end/80823
12138 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12140 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12143 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12144 address against address_operand predicate.
12145 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12147 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12149 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12150 parameters passed indirectly.
12152 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12154 * config/i386/i386.md (*movdi_internal): Remove SSE4
12155 alternative 18 (?r, *v). Update insn attributes.
12156 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12157 Update insn attributes.
12158 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12159 Update insn attributes.
12160 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12161 alternative 1 (r, v). Remove isa attribute.
12162 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12163 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12164 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12166 2017-05-23 Tom de Vries <tom@codesourcery.com>
12168 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12171 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12173 * cgraphunit.c (symbol_table::process_new_functions): Update.
12174 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12175 (inline_generate_summary): Rename to ...
12176 (ipa_fn_summary_generate): ... this one.
12177 (inline_read_summary): Rename to ...
12178 (ipa_fn_summary_read): ... this one.
12179 (inline_write_summary): Rename to ...
12180 (ipa_fn_summary_write): ... this one.
12181 (inline_free_summary): Rename to ...
12182 (ipa_free_fn_summary): ... this one.
12183 (pass_data_local_fn_summary, pass_local_fn_summary,
12184 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12185 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12186 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12187 make_pass_ipa_fn_summary): New.
12188 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12189 inline_write_summary, inline_free_summary): Remove.
12190 (ipa_free_fn_summary) : New.
12191 * ipa-inline.c (ipa_inline): Update.
12192 (pass_ipa_inline): Do not generate summaries.
12193 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12195 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12196 and add pass_ipa_fn_summary.
12197 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12199 (make_pass_inline_parameters): Remove.
12201 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12203 * omp-low.c (struct omp_context): Remove "default_kind" member.
12206 * omp-offload.c (execute_oacc_device_lower): Remove the
12207 parallelism dimensions function attributes for unparallelized
12208 OpenACC kernels constructs.
12210 2017-05-23 Martin Liska <mliska@suse.cz>
12212 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12214 (cgraph_edge::make_speculative): Likewise.
12215 (cgraph_edge::resolve_speculation): Likewise.
12216 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12217 (cgraph_node::dump): Likewise.
12218 * cgraph.h: Likewise.
12219 * cgraphunit.c (analyze_functions): Likewise.
12220 (symbol_table::compile): Likewise.
12221 * ipa-cp.c (print_all_lattices): Likewise.
12222 (determine_versionability): Likewise.
12223 (initialize_node_lattices): Likewise.
12224 (ipcp_verify_propagated_values): Likewise.
12225 (estimate_local_effects): Likewise.
12226 (update_profiling_info): Likewise.
12227 (create_specialized_node): Likewise.
12228 (perhaps_add_new_callers): Likewise.
12229 (decide_about_value): Likewise.
12230 (decide_whether_version_node): Likewise.
12231 (identify_dead_nodes): Likewise.
12232 (ipcp_store_bits_results): Likewise.
12233 * ipa-devirt.c (dump_targets): Likewise.
12234 (ipa_devirt): Likewise.
12235 * ipa-icf.c (sem_item::dump): Likewise.
12236 (sem_function::equals): Likewise.
12237 (sem_variable::equals): Likewise.
12238 (sem_item_optimizer::read_section): Likewise.
12239 (sem_item_optimizer::execute): Likewise.
12240 (congruence_class::dump): Likewise.
12241 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12242 (dump_inline_summary): Likewise.
12243 (estimate_node_size_and_time): Likewise.
12244 (inline_analyze_function): Likewise.
12245 * ipa-inline-transform.c (inline_call): Likewise.
12246 * ipa-inline.c (report_inline_failed_reason): Likewise.
12247 (want_early_inline_function_p): Likewise.
12248 (edge_badness): Likewise.
12249 (update_edge_key): Likewise.
12250 (inline_small_functions): Likewise.
12251 * ipa-profile.c (ipa_profile): Likewise.
12252 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12253 (ipa_make_edge_direct_to_target): Likewise.
12254 (remove_described_reference): Likewise.
12255 (ipa_impossible_devirt_target): Likewise.
12256 (propagate_controlled_uses): Likewise.
12257 (ipa_print_node_params): Likewise.
12258 (ipcp_transform_function): Likewise.
12259 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12260 (propagate_pure_const): Likewise.
12261 * ipa-reference.c (generate_summary): Likewise.
12262 (read_write_all_from_decl): Likewise.
12263 (propagate): Likewise.
12264 (ipa_reference_read_optimization_summary): Likewise.
12265 * ipa-utils.c (ipa_merge_profiles): Likewise.
12266 * ipa.c (walk_polymorphic_call_targets): Likewise.
12267 (symbol_table::remove_unreachable_nodes): Likewise.
12268 (ipa_single_use): Likewise.
12269 * passes.c (execute_todo): Likewise.
12270 * predict.c (drop_profile): Likewise.
12271 * symtab.c (symtab_node::get_dump_name): New function.
12272 (symtab_node::dump_name): Likewise.
12273 (symtab_node::dump_asm_name): Likewise.
12274 (symtab_node::dump_references): Likewise.
12275 (symtab_node::dump_referring): Likewise.
12276 (symtab_node::dump_base): Likewise.
12277 (symtab_node::debug_symtab): Likewise.
12278 * tree-sra.c (convert_callers_for_node): Likewise.
12279 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12280 * value-prof.c (init_node_map): Likewise.
12282 2017-05-23 Martin Liska <mliska@suse.cz>
12284 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12285 and symtab_node::debug_symtab to symbol_table::debug.
12286 * cgraphunit.c (analyze_functions): Use the renamed function.
12287 (symbol_table::compile): Likewise.
12288 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12289 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12290 * passes.c (execute_todo): Likewise.
12291 * symtab.c (symbol_table::dump): New function.
12292 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12294 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12296 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12297 that nonconst implies exec.
12299 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12301 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12302 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12303 (inline_edge_summary_vec): Turn into ...
12304 (ipa_call_summaries): ... this one.
12305 (redirect_to_unreachable, edge_set_predicate,
12306 evaluate_properties_for_edge, inline_summary_alloc,
12307 reset_ipa_call_summary, reset_inline_summary,
12308 inline_summary_t::duplicate): Update.
12309 (inline_edge_duplication_hook): Turn to ...
12310 (ipa_call_summary_t::duplicate): ... this one.
12311 (inline_edge_removal_hook): Turn to ...
12312 (ipa_call_summary_t::remove): ... this one.
12313 (dump_inline_edge_summary): Turn to ...
12314 (dump_ipa_call_summary): ... this one.
12315 (estimate_function_body_sizes): Update.
12316 (inline_update_callee_summaries): Update.
12317 (remap_edge_change_prob): Update.
12318 (remap_edge_summaries): Update.
12319 (inline_merge_summary): Update.
12320 (do_estimate_edge_time): Update.
12321 (inline_generate_summary): Update.
12322 (inline_read_section): Update.
12323 (inline_read_summary): Update.
12324 (inline_free_summary): Update.
12325 * ipa-inline.c (can_inline_edge_p): Update.
12326 (compute_inlined_call_time): Update.
12327 (want_inline_small_function_p): Update.
12328 (edge_badness): Update.
12329 (early_inliner): Update.
12330 * ipa-inline.h (inline_edge_summary): Turn to ...
12331 (ipa_call_summary): ... this one.
12332 (ipa_call_summary_t): New class.
12333 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12334 (ipa_call_summaries): New.
12335 (inline_edge_summary): Remove.
12336 (estimate_edge_growth): Update.
12337 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12338 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12339 * ipa-split.c (execute_split_functions): Update.
12340 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12342 2017-05-23 Tom de Vries <tom@codesourcery.com>
12344 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12345 attributes): Document rdrand effective target.
12347 2017-05-23 Tom de Vries <tom@codesourcery.com>
12349 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12350 attributes): Sort alphabetically.
12352 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12354 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12356 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12359 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12360 V2DF/V2DI splat into two separate patterns, one that handles
12361 registers, and the other that only handles memory. Drop support
12362 for splatting from a GPR on ISA 2.07 and then splitting the
12363 splat into direct move and splat.
12364 (vsx_splat_<mode>_reg): Likewise.
12365 (vsx_splat_<mode>_mem): Likewise.
12367 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12369 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12371 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12373 PR middle-end/80809
12374 * omp-low.c (finish_taskreg_remap): New function.
12375 (finish_taskreg_scan): If unit size of ctx->record_type
12376 is non-constant, unshare the size expression and replace
12377 decls in it with possible outer var refs.
12379 PR middle-end/80809
12380 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12381 GOVD_SHARED rather than GOVD_PRIVATE with it.
12382 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12383 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12385 PR middle-end/80853
12386 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12387 as last argument to build_outer_var_ref for pointer bases of array
12388 section reductions.
12390 2017-05-19 Martin Sebor <msebor@redhat.com>
12392 * print-tree.c (print_node): Print DECL_READ_P flag.
12394 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12396 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12397 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12398 * cgraph.c: Likewise.
12399 * cgraphunit.c: Likewise.
12400 * gengtype.c: Likewise.
12401 * ipa-cp.c: Likewise.
12402 * ipa-devirt.c: Likewise.
12403 * ipa-icf.c: Likewise.
12404 * ipa-predicate.c: Likewise.
12405 * ipa-profile.c: Likewise.
12406 * ipa-prop.c: Likewise.
12407 * ipa-split.c: Likewise.
12409 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12410 edge_predicate_pool, dump_inline_hints,
12411 inline_summary::account_size_time, redirect_to_unreachable,
12412 edge_set_predicate, set_hint_predicate,
12413 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12414 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12415 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12416 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12417 ipa_call_summary_t::remove, initialize_growth_caches,
12418 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12419 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12420 mark_modified, unmodified_parm_1, unmodified_parm,
12421 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12422 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12423 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12424 will_be_nonconstant_predicate, record_modified_bb_info,
12425 get_minimal_bb, record_modified, param_change_prob,
12426 phi_result_unknown_predicate, predicate_for_phi_result,
12427 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12428 estimate_function_body_sizes, compute_inline_parameters,
12429 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12430 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12431 inline_update_callee_summaries, remap_edge_change_prob,
12432 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12433 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12434 inline_analyze_function, inline_summary_t::insert,
12435 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12436 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12437 inline_free_summary): Move to ipa-fnsummary.h
12438 (predicate_t): Remove.
12439 * ipa-fnsummary.c: New file.
12440 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12441 (enum inline_hints_vals, inline_hints, agg_position_info,
12442 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12443 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12444 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12445 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12446 inline_read_summary, inline_write_summary, inline_free_summary,
12447 inline_analyze_function, initialize_inline_failed,
12448 inline_merge_summary, inline_update_overall_summary,
12449 compute_inline_parameters): Move to ipa-fnsummary.h
12450 * ipa-fnsummary.h: New file.
12451 * ipa-inline-transform.h: Include ipa-inline.h.
12452 * ipa-inline.c: LIkewise.
12454 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12456 * ipa-inline.c (edge_badness): Use inlined_time instead of
12457 inline_summaries->get.
12459 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12461 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12463 2017-05-22 Nathan Sidwell <nathan@acm.org>
12465 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12466 (fdump-lang): Document 'raw' option.
12467 * dumpfile.h (TDI_tu): Delete.
12468 * dumpfile.c (dump_files): Remove translation-unit.
12469 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12471 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12473 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12474 command option from $(AWK) call.
12475 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12476 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12477 [FORMAT]: Remove handling of variable.
12478 * config/avr/t-multilib: Regenerate.
12480 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12482 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12484 (dump_inline_summary): Do not print self_time.
12485 (estimate_function_body_sizes): Do not set self_time.
12486 (compute_inline_parameters): Likewise.
12487 (inline_read_section, inline_write_summary): Do not stream self_time.
12488 * ipa-inline.h (inline_summary): Drop self_time.
12490 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12492 * ipa-inline-analysis.c (account_size_time): Rename to ...
12493 (inline_summary::account_size_time): ... this one.
12494 (reset_ipa_call_summary): Turn to ...
12495 (ipa_call_summary::reset): ... this one.
12496 (reset_inline_summary): Turn to ...
12497 (inline_summary::reset): ... this one.
12498 (inline_summary_t::remove): Update.
12499 (inline_summary_t::duplicate): Update.
12500 (ipa_call_summary_t::remove): Update.
12501 (dump_inline_summary): Update.
12502 (estimate_function_body_sizes): Update.
12503 (compute_inline_parameters): Update.
12504 (estimate_node_size_and_time): Update.
12505 (inline_merge_summary): Update.
12506 (inline_update_overall_summary): Update.
12507 (inline_read_section): Update.
12508 (inline_write_summary): Update.
12509 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12510 add account_size_time and reset member functions.
12511 (ipa_call_summary): Add reset function.
12512 * ipa-predicate.h (predicate::operator &): Constify.
12514 2017-05-22 Richard Biener <rguenther@suse.de>
12516 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12518 2017-05-19 Jason Merrill <jason@redhat.com>
12520 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12522 2017-05-19 Marek Polacek <polacek@redhat.com>
12525 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12526 TYPE_OVERFLOW_WRAPS checks.
12528 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12530 * tree-core.h (enum omp_clause_default_kind): Add
12531 "OMP_CLAUSE_DEFAULT_PRESENT".
12532 * tree-pretty-print.c (dump_omp_clause): Handle it.
12533 * gimplify.c (enum gimplify_omp_var_data): Add
12534 "GOVD_MAP_FORCE_PRESENT".
12535 (gimplify_adjust_omp_clauses_1): Map it to
12536 "GOMP_MAP_FORCE_PRESENT".
12537 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12539 * gimplify.c (oacc_default_clause): Clarify.
12541 2017-05-19 Nathan Sidwell <nathan@acm.org>
12543 LANG_HOOK_REGISTER_DUMPS
12544 * toplev.c (general_init): Call register dump lang hook.
12545 * doc/invoke.texi: Document -fdump-lang option family.
12546 * dumpfile.c (dump_files): Remove class dump here.
12547 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12548 * dumpfile.h (tree_dump_index): Remove TDI_class.
12549 * langhooks-def.h (lhd_register_dumps): Declare.
12550 (LANG_HOOKS_REGISTER_DUMPS): Define.
12551 (LANG_HOOKS_INITIALIZER): Add it.
12552 * langhooks.c (lhd_register_dumps): Define.
12553 * langhooks.h (struct lang_hooks): Add register_dumps.
12555 2017-05-19 Nathan Sidwell <nathan@acm.org>
12557 * context.h (context::set_passes): New.
12558 * context.c (context::context): Do not create pass manager.
12559 * toplev.c (general_init): Create pass manager here.
12561 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12563 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12564 use this splitter if two add or or instructions would also work for
12565 the constant we want to generate.
12567 2017-05-19 Richard Biener <rguenther@suse.de>
12570 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12571 predicate evaluation.
12573 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12575 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12577 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12580 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12582 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12583 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12584 (inline_edge_summary_vec): Turn into ...
12585 (ipa_call_summaries): ... this one.
12586 (redirect_to_unreachable, edge_set_predicate,
12587 evaluate_properties_for_edge, inline_summary_alloc,
12588 reset_ipa_call_summary, reset_inline_summary,
12589 inline_summary_t::duplicate): Update.
12590 (inline_edge_duplication_hook): Turn to ...
12591 (ipa_call_summary_t::duplicate): ... this one.
12592 (inline_edge_removal_hook): Turn to ...
12593 (ipa_call_summary_t::remove): ... this one.
12594 (dump_inline_edge_summary): Turn to ...
12595 (dump_ipa_call_summary): ... this one.
12596 (estimate_function_body_sizes): Update.
12597 (inline_update_callee_summaries): Update.
12598 (remap_edge_change_prob): Update.
12599 (remap_edge_summaries): Update.
12600 (inline_merge_summary): Update.
12601 (do_estimate_edge_time): Update.
12602 (inline_generate_summary): Update.
12603 (inline_read_section): Update.
12604 (inline_read_summary): Update.
12605 (inline_free_summary): Update.
12606 * ipa-inline.c (can_inline_edge_p): Update.
12607 (compute_inlined_call_time): Update.
12608 (want_inline_small_function_p): Update.
12609 (edge_badness): Update.
12610 (early_inliner): Update.
12611 * ipa-inline.h (inline_edge_summary): Turn to ...
12612 (ipa_call_summary): ... this one.
12613 (ipa_call_summary_t): New class.
12614 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12615 (ipa_call_summaries): New.
12616 (inline_edge_summary): Remove.
12617 (estimate_edge_growth): Update.
12618 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12619 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12620 * ipa-split.c (execute_split_functions): Update.
12621 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12623 2017-05-19 Richard Biener <rguenther@suse.de>
12625 PR middle-end/80764
12626 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12628 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12630 * config/rs6000/rs6000.c (struct machine_function): Add field
12631 fpr_is_wrapped_separately.
12632 (rs6000_get_separate_components): Use 64 components. Handle the
12633 new FPR components.
12634 (rs6000_components_for_bb): Handle the FPR components.
12635 (rs6000_emit_prologue_components): Handle the FPR components.
12636 (rs6000_emit_epilogue_components): Handle the FPR components.
12637 (rs6000_set_handled_components): Handle the FPR components.
12638 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12639 that are already separately shrink-wrapped.
12640 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12641 that are already separately shrink-wrapped.
12643 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12646 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12649 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12650 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12651 cases where the register allocator uses a move and an offsettable
12652 memory operation to/from a FPR register on ISA 2.06/2.07.
12653 (define_peephole2 for Altivec d-form store): Likewise.
12655 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12658 * config/i386/mmx.md (*mov<mode>_internal): Enable
12659 alternatives 11, 12, 13 and 14 also for 32bit targets.
12660 Remove alternatives 15, 16, 17 and 18.
12661 * config/i386/sse.md (vec_concatv2di): Change
12662 alternative (!x, *y) to (x, ?!*Yn).
12664 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12666 * dumpfile.h (enum dump_kind): Remove stray comma.
12668 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12670 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12671 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12672 predicate::num_conditions
12673 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12674 (CHANGED): turn into predicate::changed.
12675 (agg_position_info): Move to ipa-predicate.h
12676 (add_condition, predicate::add_clause, predicate::operator &=,
12677 predicate::or_with, predicate::evaluate, predicate::probability,
12678 dump_condition, dump_clause, predicate::dump,
12679 predicate::remap_after_duplication, predicate::remap_after_inlining,
12680 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12681 (evaluate_conditions_for_known_args): Update.
12682 (set_cond_stmt_execution_predicate): Update.
12683 * ipa-inline.h: Include ipa-predicate.h
12684 (condition, inline_param_summary, conditions, agg_position_info,
12685 predicate): Move to ipa-predicate.h
12686 * ipa-predicate.c: New file.
12687 * ipa-predicate.h: New file.
12689 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12691 * final.c (leaf_function_p): Check we are not in a sequence.
12693 2017-05-18 Martin Liska <mliska@suse.cz>
12695 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12696 * dumpfile.c (dump_register): Use new enum dump_kind.
12697 (get_dump_file_name): Likewise.
12698 (dump_enable_all): Likewise.
12699 (dump_switch_p_1): Likewise.
12700 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12701 * dumpfile.h (enum dump_kind): New enum type.
12702 (struct dump_file_info): Create constructor and
12703 format fields and comments.
12704 * passes.c (pass_manager::register_one_dump_file):
12706 * statistics.c (statistics_early_init): Likewise.
12707 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12708 TDF_TREE with TDF_SLIM.
12709 (gather_memory_references_ref): Likewise.
12711 2017-05-18 Martin Liska <mliska@suse.cz>
12713 * vec.h (struct vnull): Use it.
12715 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12717 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12718 (true_predicate, false_predicate, true_predicate_p,
12719 false_predicate_p): Remove.
12720 (single_cond_predicate, not_inlined_predicate): Turn to member function
12722 (add_condition): Update.
12723 (add_clause): Turn to...
12724 (predicate::add_clause): ... this one; update; allow passing NULL
12726 (and_predicates): Turn to ...
12727 (predicate::operator &=): ... this one.
12728 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12729 (or_predicates): Turn to ...
12730 (predicate::or_with): ... this one.
12731 (evaluate_predicate): Turn to ...
12732 (predicate::evaluate): ... this one.
12733 (predicate_probability): Turn to ...
12734 (predicate::probability): ... this one.
12735 (dump_condition): Update.
12736 (dump_predicate): Turn to ...
12737 (predicate::dump): ... this one.
12738 (account_size_time): Update.
12739 (edge_set_predicate): Update.
12740 (set_hint_predicate): UPdate.
12741 (evaluate_conditions_for_known_args): Update.
12742 (evaluate_properties_for_edge): Update.
12743 (remap_predicate_after_duplication): Turn to...
12744 (predicate::remap_after_duplication): ... this one.
12745 (remap_hint_predicate_after_duplication): Update.
12746 (inline_summary_t::duplicate): UPdate.
12747 (dump_inline_edge_summary): Update.
12748 (dump_inline_summary): Update.
12749 (set_cond_stmt_execution_predicate): Update.
12750 (set_switch_stmt_execution_predicate): Update.
12751 (compute_bb_predicates): Update.
12752 (will_be_nonconstant_expr_predicate): Update.
12753 (will_be_nonconstant_predicate): Update.
12754 (phi_result_unknown_predicate): Update.
12755 (predicate_for_phi_result): Update.
12756 (array_index_predicate): Update.
12757 (estimate_function_body_sizes): Update.
12758 (estimate_node_size_and_time): Update.
12759 (estimate_ipcp_clone_size_and_time): Update.
12760 (remap_predicate): Rename to ...
12761 (predicate::remap_after_inlining): ... this one.
12762 (remap_hint_predicate): Update.
12763 (inline_merge_summary): Update.
12764 (inline_update_overall_summary): Update.
12765 (estimate_size_after_inlining): Update.
12766 (read_predicate): Rename to ...
12767 (predicate::stream_in): ... this one.
12768 (read_inline_edge_summary): Update.
12769 (write_predicate): Rename to ...
12770 (predicate::stream_out): ... this one.
12771 (write_inline_edge_summary): Update.
12772 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12773 (clause_t): Turn to uint32_t
12774 (predicate): Turn to class; implement constructor and operators
12776 (size_time_entry): Update.
12777 (inline_summary): Update.
12778 (inline_edge_summary): Update.
12780 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12782 * fold-const.c (fold_binary_loc): Move transformation...
12783 * match.pd (C - X CMP X): ... here.
12785 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12787 * config/sparc/sparc.c (sparc_option_override): Set function
12788 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12789 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12791 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12793 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12795 2017-05-18 Marek Polacek <polacek@redhat.com>
12798 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12799 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12800 gimple_assign_load_p.
12802 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12804 PR middle-end/80692
12805 * real.c (do_compare): Give decimal_do_compare preference over
12806 comparing just the signs.
12808 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12810 * doc/md.texi (Canonicalization of Instructions): Describe the
12811 canonical form of instructions that inherently set a condition
12814 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12816 PR middle-end/80775
12817 * tree-cfg.c: Move deletion of unreachable case statements to after
12818 the merging of consecutive case labels.
12820 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12822 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12823 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12824 restoring of callee-saved registers.
12826 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12828 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12829 * config/visium/visium.c (single_set_and_flags): Likewise.
12830 * config/visium/visium.md (Substitutions): Likewise.
12832 2017-05-17 Martin Liska <mliska@suse.cz>
12834 * cfg.c: Introduce dump_flags_t type and
12835 use it instead of int type.
12837 * cfghooks.c: Likewise.
12838 * cfghooks.h (struct cfg_hooks): Likewise.
12839 * cfgrtl.c: Likewise.
12840 * cfgrtl.h: Likewise.
12841 * cgraph.c (cgraph_node::get_body): Likewise.
12842 * coretypes.h: Likewise.
12843 * domwalk.c: Likewise.
12844 * domwalk.h: Likewise.
12845 * dumpfile.c (struct dump_option_value_info): Likewise.
12846 (dump_enable_all): Likewise.
12847 (dump_switch_p_1): Likewise.
12848 (opt_info_switch_p): Likewise.
12849 * dumpfile.h (enum tree_dump_index): Likewise.
12850 (struct dump_file_info): Likewise.
12851 * genemit.c: Likewise.
12852 * generic-match-head.c: Likewise.
12853 * gengtype.c (open_base_files): Likewise.
12854 * gimple-pretty-print.c: Likewise.
12855 * gimple-pretty-print.h: Likewise.
12856 * graph.c (print_graph_cfg): Likewise.
12857 * graphite-scop-detection.c (dot_all_sese): Likewise.
12858 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12859 * loop-unroll.c (report_unroll): Likewise.
12860 * passes.c (pass_manager::register_one_dump_file): Likewise.
12861 * print-tree.c: Likewise.
12862 * statistics.c: Likewise.
12863 * tree-cfg.c: Likewise.
12864 * tree-cfg.h: Likewise.
12865 * tree-dfa.c: Likewise.
12866 * tree-dfa.h: Likewise.
12867 * tree-dump.c (dump_function): Likewise.
12868 * tree-dump.h (struct dump_info): Likewise.
12869 * tree-pretty-print.c: Likewise.
12870 * tree-pretty-print.h: Likewise.
12871 * tree-ssa-live.c: Likewise.
12872 * tree-ssa-live.h: Likewise.
12873 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12874 * tree-vect-loop.c: Likewise.
12875 * tree-vect-slp.c: Likewise.
12877 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12878 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12880 PR tree-optimization/80457
12881 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12882 of all arguments to a statement as scalar_to_vec operations.
12883 (vectorizable_call): Adjust call to vect_model_simple_cost for
12885 (vectorizable_conversion): Likewise.
12886 (vectorizable_assignment): Likewise.
12887 (vectorizable_shift): Likewise.
12888 (vectorizable_operation): Likewise.
12889 (vectorizable_comparison): Likewise.
12890 (vect_is_simple_cond): Record the def types for operands.
12891 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12892 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12893 for statement argument count.
12895 2017-05-16 Carl Love <cel@us.ibm.com>
12897 * config/rs6000/rs6000-c: Add support for built-in functions
12898 vector unsigned long long vec_bperm (vector unsigned long long,
12899 vector unsigned char)
12900 vector signed long long vec_mule (vector signed int,
12902 vector unsigned long long vec_mule (vector unsigned int,
12903 vector unsigned int)
12904 vector signed long long vec_mulo (vector signed int,
12906 vector unsigned long long vec_mulo (vector unsigned int,
12907 vector unsigned int)
12908 vector signed char vec_sldw (vector signed char,
12909 vector signed char,
12911 vector unsigned char vec_sldw (vector unsigned char,
12912 vector unsigned char,
12914 vector signed short vec_sldw (vector signed short,
12915 vector signed short,
12917 vector unsigned short vec_sldw (vector unsigned short,
12918 vector unsigned short,
12920 vector signed int vec_sldw (vector signed int,
12923 vector unsigned int vec_sldw (vector unsigned int,
12924 vector unsigned int,
12926 vector signed long long vec_sldw (vector signed long long,
12927 vector signed long long,
12929 vector unsigned long long vec_sldw (vector unsigned long long,
12930 vector unsigned long long,
12932 * config/rs6000/rs6000-c: Add support for built-in functions
12933 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12934 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12935 * doc/extend.texi: Update the built-in documentation for the
12936 new built-in functions.
12938 2017-05-16 Marek Polacek <polacek@redhat.com>
12942 * tree.c (save_expr): Don't fold the expression.
12944 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12946 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12947 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12948 and (?*y,m). Update insn attributes.
12950 2017-05-16 Martin Liska <mliska@suse.cz>
12952 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12953 flags argument of print_gimple_stmt, print_gimple_expr,
12954 print_generic_stmt and print_generic_expr.
12955 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12956 * coretypes.h: Likewise.
12957 * except.c (dump_eh_tree): Likewise.
12958 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12959 * gimple-pretty-print.h: Likewise.
12960 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12961 (backprop::push_to_worklist): Likewise.
12962 (backprop::pop_from_worklist): Likewise.
12963 (backprop::process_use): Likewise.
12964 (backprop::intersect_uses): Likewise.
12965 (note_replacement): Likewise.
12966 * gimple-ssa-store-merging.c
12967 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12968 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12969 (pass_store_merging::execute): Likewise.
12970 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12971 (ssa_base_cand_dump_callback): Likewise.
12972 (dump_incr_vec): Likewise.
12973 (replace_refs): Likewise.
12974 (replace_mult_candidate): Likewise.
12975 (create_add_on_incoming_edge): Likewise.
12976 (create_phi_basis): Likewise.
12977 (insert_initializers): Likewise.
12978 (all_phi_incrs_profitable): Likewise.
12979 (introduce_cast_before_cand): Likewise.
12980 (replace_one_candidate): Likewise.
12981 * gimplify.c (gimplify_expr): Likewise.
12982 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12983 (set_rename): Likewise.
12984 (rename_uses): Likewise.
12985 (copy_loop_phi_nodes): Likewise.
12986 (add_close_phis_to_merge_points): Likewise.
12987 (copy_loop_close_phi_args): Likewise.
12988 (copy_cond_phi_args): Likewise.
12989 (graphite_copy_stmts_from_block): Likewise.
12990 (translate_pending_phi_nodes): Likewise.
12991 * graphite-poly.c (print_pdr): Likewise.
12992 (dump_gbb_cases): Likewise.
12993 (dump_gbb_conditions): Likewise.
12994 (print_scop_params): Likewise.
12995 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12996 (build_cross_bb_scalars_use): Likewise.
12997 (gather_bbs::before_dom_children): Likewise.
12998 * hsa-dump.c (dump_hsa_immed): Likewise.
12999 * ipa-cp.c (print_ipcp_constant_value): Likewise.
13000 (get_replacement_map): Likewise.
13001 * ipa-inline-analysis.c (dump_condition): Likewise.
13002 (estimate_function_body_sizes): Likewise.
13003 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
13004 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
13005 * ipa-prop.c (ipa_dump_param): Likewise.
13006 (ipa_print_node_jump_functions_for_edge): Likewise.
13007 (ipa_modify_call_arguments): Likewise.
13008 (ipa_modify_expr): Likewise.
13009 (ipa_dump_param_adjustments): Likewise.
13010 (ipa_dump_agg_replacement_values): Likewise.
13011 (ipcp_modif_dom_walker::before_dom_children): Likewise.
13012 * ipa-pure-const.c (check_stmt): Likewise.
13013 (pass_nothrow::execute): Likewise.
13014 * ipa-split.c (execute_split_functions): Likewise.
13015 * omp-offload.c (dump_oacc_loop_part): Likewise.
13016 (dump_oacc_loop): Likewise.
13017 * trans-mem.c (tm_log_emit): Likewise.
13018 (tm_memopt_accumulate_memops): Likewise.
13019 (dump_tm_memopt_set): Likewise.
13020 (dump_tm_memopt_transform): Likewise.
13021 * tree-cfg.c (gimple_verify_flow_info): Likewise.
13022 (print_loop): Likewise.
13023 * tree-chkp-opt.c (chkp_print_addr): Likewise.
13024 (chkp_gather_checks_info): Likewise.
13025 (chkp_get_check_result): Likewise.
13026 (chkp_remove_check_if_pass): Likewise.
13027 (chkp_use_outer_bounds_if_possible): Likewise.
13028 (chkp_reduce_bounds_lifetime): Likewise.
13029 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
13030 (chkp_mark_completed_bounds): Likewise.
13031 (chkp_register_incomplete_bounds): Likewise.
13032 (chkp_mark_invalid_bounds): Likewise.
13033 (chkp_maybe_copy_and_register_bounds): Likewise.
13034 (chkp_build_returned_bound): Likewise.
13035 (chkp_get_bound_for_parm): Likewise.
13036 (chkp_build_bndldx): Likewise.
13037 (chkp_get_bounds_by_definition): Likewise.
13038 (chkp_generate_extern_var_bounds): Likewise.
13039 (chkp_get_bounds_for_decl_addr): Likewise.
13040 * tree-chrec.c (chrec_apply): Likewise.
13041 * tree-data-ref.c (dump_data_reference): Likewise.
13042 (dump_subscript): Likewise.
13043 (dump_data_dependence_relation): Likewise.
13044 (analyze_overlapping_iterations): Likewise.
13045 * tree-inline.c (expand_call_inline): Likewise.
13046 (tree_function_versioning): Likewise.
13047 * tree-into-ssa.c (dump_defs_stack): Likewise.
13048 (dump_currdefs): Likewise.
13049 (dump_names_replaced_by): Likewise.
13050 (dump_update_ssa): Likewise.
13051 (update_ssa): Likewise.
13052 * tree-object-size.c (pass_object_sizes::execute): Likewise.
13053 * tree-parloops.c (build_new_reduction): Likewise.
13054 (try_create_reduction_list): Likewise.
13055 (ref_conflicts_with_region): Likewise.
13056 (oacc_entry_exit_ok_1): Likewise.
13057 (oacc_entry_exit_single_gang): Likewise.
13058 * tree-pretty-print.h: Likewise.
13059 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
13060 (get_scalar_evolution): Likewise.
13061 (add_to_evolution): Likewise.
13062 (get_loop_exit_condition): Likewise.
13063 (analyze_evolution_in_loop): Likewise.
13064 (analyze_initial_condition): Likewise.
13065 (analyze_scalar_evolution): Likewise.
13066 (instantiate_scev): Likewise.
13067 (number_of_latch_executions): Likewise.
13068 (gather_chrec_stats): Likewise.
13069 (final_value_replacement_loop): Likewise.
13070 (scev_const_prop): Likewise.
13071 * tree-sra.c (dump_access): Likewise.
13072 (disqualify_candidate): Likewise.
13073 (create_access): Likewise.
13074 (reject): Likewise.
13075 (maybe_add_sra_candidate): Likewise.
13076 (create_access_replacement): Likewise.
13077 (analyze_access_subtree): Likewise.
13078 (analyze_all_variable_accesses): Likewise.
13079 (sra_modify_assign): Likewise.
13080 (initialize_constant_pool_replacements): Likewise.
13081 (find_param_candidates): Likewise.
13082 (decide_one_param_reduction): Likewise.
13083 (replace_removed_params_ssa_names): Likewise.
13084 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13085 * tree-ssa-copy.c (dump_copy_of): Likewise.
13086 (copy_prop_visit_cond_stmt): Likewise.
13087 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13088 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13089 (record_equivalences_from_stmt): Likewise.
13090 * tree-ssa-dse.c (compute_trims): Likewise.
13091 (delete_dead_call): Likewise.
13092 (delete_dead_assignment): Likewise.
13093 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13094 (forward_propagate_into_cond): Likewise.
13095 (pass_forwprop::execute): Likewise.
13096 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13097 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13099 (move_computations_worker): Likewise.
13100 (execute_sm): Likewise.
13101 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13102 (remove_exits_and_undefined_stmts): Likewise.
13103 (remove_redundant_iv_tests): Likewise.
13104 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13105 (adjust_iv_update_pos): Likewise.
13106 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13107 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13108 (value_replacement): Likewise.
13109 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13110 * tree-ssa-pre.c (print_pre_expr): Likewise.
13111 (get_representative_for): Likewise.
13112 (create_expression_by_pieces): Likewise.
13113 (insert_into_preds_of_block): Likewise.
13114 (eliminate_insert): Likewise.
13115 (eliminate_dom_walker::before_dom_children): Likewise.
13116 (eliminate): Likewise.
13117 (remove_dead_inserted_code): Likewise.
13118 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13119 * tree-ssa-reassoc.c (get_rank): Likewise.
13120 (eliminate_duplicate_pair): Likewise.
13121 (eliminate_plus_minus_pair): Likewise.
13122 (eliminate_not_pairs): Likewise.
13123 (undistribute_ops_list): Likewise.
13124 (eliminate_redundant_comparison): Likewise.
13125 (update_range_test): Likewise.
13126 (optimize_range_tests_var_bound): Likewise.
13127 (optimize_vec_cond_expr): Likewise.
13128 (rewrite_expr_tree): Likewise.
13129 (rewrite_expr_tree_parallel): Likewise.
13130 (linearize_expr): Likewise.
13131 (break_up_subtract): Likewise.
13132 (linearize_expr_tree): Likewise.
13133 (attempt_builtin_powi): Likewise.
13134 (attempt_builtin_copysign): Likewise.
13135 (transform_stmt_to_copy): Likewise.
13136 (transform_stmt_to_multiply): Likewise.
13137 (dump_ops_vector): Likewise.
13138 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13139 (print_scc): Likewise.
13140 (set_ssa_val_to): Likewise.
13141 (visit_reference_op_store): Likewise.
13142 (visit_use): Likewise.
13143 (sccvn_dom_walker::before_dom_children): Likewise.
13144 (run_scc_vn): Likewise.
13145 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13147 (expr_hash_elt::print): Likewise.
13148 (const_and_copies::pop_to_marker): Likewise.
13149 (const_and_copies::record_const_or_copy_raw): Likewise.
13150 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13151 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13152 (dump_predicates): Likewise.
13153 (find_uninit_use): Likewise.
13154 (warn_uninitialized_phi): Likewise.
13155 (pass_late_warn_uninitialized::execute): Likewise.
13156 * tree-ssa.c (verify_vssa): Likewise.
13157 (verify_ssa): Likewise.
13158 (maybe_optimize_var): Likewise.
13159 * tree-vrp.c (dump_value_range): Likewise.
13160 (dump_all_value_ranges): Likewise.
13161 (dump_asserts_for): Likewise.
13162 (register_edge_assert_for_2): Likewise.
13163 (vrp_visit_cond_stmt): Likewise.
13164 (vrp_visit_switch_stmt): Likewise.
13165 (vrp_visit_stmt): Likewise.
13166 (vrp_visit_phi_node): Likewise.
13167 (simplify_cond_using_ranges_1): Likewise.
13168 (fold_predicate_in): Likewise.
13169 (evrp_dom_walker::before_dom_children): Likewise.
13170 (evrp_dom_walker::push_value_range): Likewise.
13171 (evrp_dom_walker::pop_value_range): Likewise.
13172 (execute_early_vrp): Likewise.
13174 2017-05-16 Richard Biener <rguenther@suse.de>
13176 * dwarf2out.c (loc_list_from_tree_1): Do not create
13177 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13179 2017-05-16 Richard Biener <rguenther@suse.de>
13181 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13183 (note_variable_value_in_expr): If we resolved the decl ref
13184 do not push to the stack.
13186 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13188 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13189 operations in fast-math mode.
13190 (vaddq_f16): Likewise.
13191 (vmul_f16): Likewise.
13192 (vmulq_f16): Likewise.
13193 (vsub_f16): Likewise.
13194 (vsubq_f16): Likewise.
13195 * config/arm/neon.md (add<mode>3): New.
13197 (fma:<VH:mode>3): New. Also remove outdated comment.
13200 2017-05-16 Martin Liska <mliska@suse.cz>
13204 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13205 (odr_types_equivalent_p): Likewise.
13207 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13209 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13211 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13214 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13215 non-interunit SSE move alternatives with '?'.
13216 (zero-extendsidi peephole2): New peephole to skip intermediate
13217 general register in SSE zero-extend sequence.
13219 2017-05-15 Jeff Law <law@redhat.com>
13221 * reorg.c (relax_delay_slots): Create a new variable to hold
13222 the temporary target rather than clobbering TARGET_LABEL.
13224 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13225 missing argument to extract_bit_field call.
13226 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13228 2017-05-15 Martin Liska <mliska@suse.cz>
13231 * gcc.c (process_command): Do not allow empty argument of -o option.
13233 2017-05-15 Renlin Li <renlin.li@arm.com>
13235 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13236 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13237 * config/aarch64/constraints.md (Usf): Add long call check.
13238 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13239 (call_value): Likewise.
13240 (sibcall): Likewise.
13241 (sibcall_value): Likewise.
13243 (call_value_insn): New.
13244 (sibcall_insn): Update rtx pattern.
13245 (sibcall_value_insn): Likewise.
13246 (call_internal): Remove.
13247 (call_value_internal): Likewise.
13248 (sibcall_internal): Likewise.
13249 (sibcall_value_internal): Likewise.
13250 (call_reg): Likewise.
13251 (call_symbol): Likewise.
13252 (call_value_reg): Likewise.
13253 (call_value_symbol): Likewise.
13255 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13258 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13260 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13262 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13263 compatible with CCGOCmode and with CCZmode.
13265 2017-05-14 Martin Sebor <msebor@redhat.com>
13267 PR middle-end/77671
13268 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13269 (gimple_fold_builtin_snprintf): Same.
13270 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13271 (gimple_fold_builtin_snprintf): Same.
13272 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13273 of character types.
13274 (is_call_safe): New function.
13275 (try_substitute_return_value): Call it.
13276 (try_simplify_call): New function.
13277 (pass_sprintf_length::handle_gimple_call): Call it.
13279 2017-05-14 Martin Sebor <msebor@redhat.com>
13281 PR middle-end/80669
13282 * builtins.c (expand_builtin_stpncpy): Simplify.
13284 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13286 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13287 * config/i386/i386.h
13288 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13289 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13290 (struct machine_function): Add new members call_ms2sysv,
13291 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13292 (struct machine_frame_state): New fields sp_realigned and
13293 sp_realigned_offset.
13294 * config/i386/i386.c
13295 (enum xlogue_stub): New enum.
13296 (enum xlogue_stub_sets): New enum.
13297 (class xlogue_layout): New class.
13298 (struct ix86_frame): New fields stack_realign_allocate_offset,
13299 stack_realign_offset and outlined_save_offset. Modify comments to
13300 detail stack layout when using out-of-line stubs.
13301 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13302 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13303 -mcall-ms2sysv-xlogues.
13304 (stub_managed_regs): New static variable.
13305 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13306 registers managed by out-of-line stub.
13307 (disable_call_ms2sysv_xlogues): New function.
13308 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13309 m->call_ms2sysv when appropriate and compute frame layout for
13311 (sp_valid_at, fp_valid_at): New inline functions.
13312 (choose_basereg): New function.
13313 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13315 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13316 Use align parameter of choose_baseaddr to generated aligned SSE movs
13318 (pro_epilogue_adjust_stack): Modify to track
13319 machine_frame_state::sp_realigned.
13320 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13321 (ix86_nsaved_sseregs): Likewise.
13322 (ix86_emit_save_regs): Likewise.
13323 (ix86_emit_save_regs_using_mov): Likewise.
13324 (ix86_emit_save_sse_regs_using_mov): Likewise.
13325 (get_scratch_register_on_entry): Likewise.
13326 (gen_frame_set): New function.
13327 (gen_frame_load): Likewise.
13328 (gen_frame_store): Likewise.
13329 (emit_outlined_ms2sysv_save): Likewise.
13330 (emit_outlined_ms2sysv_restore): Likewise.
13331 (ix86_expand_prologue): Modify stack re-alignment code and call
13332 emit_outlined_ms2sysv_save when appropriate.
13333 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13334 parameter rtx_insn *insn, which allows the function to be used to only
13335 generate the notes.
13336 (ix86_expand_epilogue): Modify validity checks of frame and stack
13337 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13338 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13339 * config/i386/predicates.md
13340 (save_multiple): New predicate.
13341 (restore_multiple): Likewise.
13342 * config/i386/sse.md
13343 (save_multiple<mode>): New pattern.
13344 (save_multiple_realign<mode>): Likewise.
13345 (restore_multiple<mode>): Likewise.
13346 (restore_multiple_and_return<mode>): Likewise.
13347 (restore_multiple_leave_return<mode>): Likewise.
13348 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13350 2017-05-14 Julia Koval <julia.koval@intel.com>
13352 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13353 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13354 (__builtin_ia32_xsetbv): New builtins.
13355 * config/i386/i386.c (ix86_expand_special_args_builtin):
13357 (ix86_expand_builtin): Special expand for new intrinsics.
13358 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13359 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13360 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13362 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13364 * cfganal.c (inverted_post_order_compute): Change argument type
13366 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13367 * df-core.c (rest_of_handle_df_initialize): Adjust.
13368 (rest_of_handle_df_finish): Likewise.
13369 (df_analyze_1): Likewise.
13370 (df_analyze): Likewise.
13371 (loop_inverted_post_order_compute): Change argument to be a vec *.
13372 (df_analyze_loop): Adjust.
13373 (df_get_n_blocks): Likewise.
13374 (df_get_postorder): Likewise.
13375 * df.h (struct df_d): Change field to be a vec.
13376 * lcm.c (compute_laterin): Adjust.
13377 (compute_available): Likewise.
13378 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13379 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13380 * tree-ssa-pre.c (compute_antic): Likewise.
13382 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13384 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13385 (depth_first_search::depth_first_search): Change structure init
13386 function to this constructor.
13387 (depth_first_search::add_bb): Rename function to this member.
13388 (depth_first_search::execute): Likewise.
13389 (flow_dfs_compute_reverse_finish): Adjust.
13391 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13393 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13394 (longest_simple_path): Likewise.
13395 * shrink-wrap.c (spread_components): Likewise.
13396 (disqualify_problematic_components): Likewise.
13397 (emit_common_heads_for_components): Likewise.
13398 (emit_common_tails_for_components): Likewise.
13399 (insert_prologue_epilogue_for_components): Likewise.
13401 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13403 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13406 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13408 * df-core.c (df_set_blocks): Start using auto_bitmap.
13409 (df_compact_blocks): Likewise.
13410 * df-problems.c (df_rd_confluence_n): Likewise.
13411 * df-scan.c (df_insn_rescan_all): Likewise.
13412 (df_process_deferred_rescans): Likewise.
13413 (df_update_entry_block_defs): Likewise.
13414 (df_update_exit_block_uses): Likewise.
13415 (df_entry_block_bitmap_verify): Likewise.
13416 (df_exit_block_bitmap_verify): Likewise.
13417 (df_scan_verify): Likewise.
13418 * lra-constraints.c (lra_constraints): Likewise.
13419 (undo_optional_reloads): Likewise.
13420 (lra_undo_inheritance): Likewise.
13421 * lra-remat.c (calculate_gen_cands): Likewise.
13422 (do_remat): Likewise.
13423 * lra-spills.c (assign_spill_hard_regs): Likewise.
13424 (spill_pseudos): Likewise.
13425 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13426 (bitmap_set_subtract_values): Likewise.
13428 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13430 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13431 management with auto_bitmap.
13432 (fix_inter_tick): Likewise.
13433 (fix_recovery_deps): Likewise.
13434 * ira.c (add_store_equivs): Likewise.
13435 (find_moveable_pseudos): Likewise.
13436 (split_live_ranges_for_shrink_wrap): Likewise.
13437 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13438 (rtx_reuse_manager::seen_def_p): Likewise.
13439 (rtx_reuse_manager::set_seen_def): Likewise.
13440 * print-rtl.h (class rtx_reuse_manager): Likewise.
13442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13444 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13446 (migrate_btr_def): Likewise.
13447 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13448 * df-core.c (loop_post_order_compute): Likewise.
13449 (loop_inverted_post_order_compute): Likewise.
13450 * hsa-common.h: Likewise.
13451 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13452 * init-regs.c (initialize_uninitialized_regs): Likewise.
13453 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13454 (inline_small_functions): Likewise.
13455 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13456 * ira.c (combine_and_move_insns): Likewise.
13457 (build_insn_chain): Likewise.
13458 * loop-invariant.c (find_invariants): Likewise.
13459 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13460 * predict.c (tree_predict_by_opcode): Likewise.
13461 (predict_paths_leading_to): Likewise.
13462 (predict_paths_leading_to_edge): Likewise.
13463 (estimate_loops_at_level): Likewise.
13464 (estimate_loops): Likewise.
13465 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13466 (spread_components): Likewise.
13467 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13468 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13469 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13470 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13471 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13472 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13473 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13474 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13475 (mark_threaded_blocks): Likewise.
13476 (thread_through_all_blocks): Likewise.
13477 * tree-ssa.c (verify_ssa): Likewise.
13478 (execute_update_addresses_taken): Likewise.
13479 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13481 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13483 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13485 (post_order_compute): Likewise.
13486 (inverted_post_order_compute): Likewise.
13487 (pre_and_rev_post_order_compute_fn): Likewise.
13489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13491 * genrecog.c (int_set::int_set): Explicitly construct our
13492 auto_vec base class.
13493 * vec.h (auto_vec::auto_vec): New constructor.
13495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13497 * bitmap.h (class auto_bitmap): New constructor taking
13498 bitmap_obstack * argument.
13500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13502 * bitmap.h (class auto_bitmap): Change type of m_bits to
13503 bitmap_head, and adjust ctor / dtor and member operators.
13505 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13507 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13508 when returned register mode doesn't match original mode.
13510 2017-05-12 Jeff Law <law@redhat.com>
13511 Jakub Jelinek <jakub@redhat.com>
13513 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13514 we look for cc setter after the compare-elim changes.
13515 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13516 within the vector to match what compare-elim now expects.
13517 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13518 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13520 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13521 after the compare-elim changes.
13522 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13523 the vector to match what compare-elim now expects.
13524 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13525 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13526 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13527 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13528 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13530 * config/visium/visium.c (single_set_and_flags): Fix where
13531 we look for cc setter after the compare-elim changes.
13532 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13533 with the vector to match what compare-elim now expects.
13534 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13535 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13536 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13537 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13538 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13539 (neg<mode>2_insn_set_overflow): Likewise.
13541 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13543 PR middle-end/79794
13544 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13545 maybe_expand_insn call, set ops[0].target. If still set after call,
13546 set alt_rtl. Add extra arg to recursive calls.
13547 (extract_bit_field): Add alt_rtl argument. Pass to
13549 * expmed.h (extract_bit_field): Fix prototype.
13550 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13551 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13552 to extract_bit_field_calls.
13553 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13554 Pass alt_rtl to extract_bit_field calls.
13555 * calls.c (store_unaligned_arguments_into_psuedos)
13556 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13557 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13559 * optabs.h (struct expand_operand): Add target bitfield.
13561 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13563 * compare-elim.c (try_eliminate_compare): Canonicalize
13564 operation with embedded compare to
13565 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13566 (set (reg) (operation)].
13568 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13570 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13573 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13574 cost of adding a carry flag for ADC instruction.
13575 [case MINUS]: Ignore the cost of subtracting a carry flag
13576 for SBB instruction.
13578 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13580 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13582 * config/rs6000/bmiintrin.h: New file.
13583 * config/rs6000/bmi2intrin.h: New file.
13584 * config/rs6000/x86intrin.h: New file.
13586 2017-05-12 Jeff Law <law@redhat.com>
13588 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13591 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13593 PR middle-end/80707
13594 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13596 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13598 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13599 early expansion of vector divide builtins.
13600 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13601 builtins identified as having unsigned arguments.
13603 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13605 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13606 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13607 expansion of vector logical operations (and, andc, or, xor,
13610 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13612 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13613 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13615 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13617 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13618 early GIMPLE expansion of vector multiplies.
13620 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13622 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13623 TARGET_HAVE_MOVT conditional.
13624 (movt splitter): Likewise.
13626 2017-05-12 Richard Biener <rguenther@suse.de>
13628 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13629 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13630 Fold all stmts not inplace.
13632 2017-05-12 Richard Biener <rguenther@suse.de>
13634 PR tree-optimization/80713
13635 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13636 inserted_exprs bit for not removed stmts.
13638 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13640 PR middle-end/69921
13641 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13642 parallelized" attribute for parallelized OpenACC kernels.
13643 * omp-offload.c (execute_oacc_device_lower): Use it.
13645 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13646 Set "oacc kernels" attribute.
13647 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13648 parameter. Adjust all users.
13649 (oacc_fn_attrib_kernels_p): Remove function.
13650 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13651 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13652 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13653 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13654 assert "oacc kernels" attribute is set.
13656 2017-05-11 Carl Love <cel@us.ibm.com>
13658 * config/rs6000/rs6000-c: Add support for built-in functions
13659 vector unsigned char vec_popcnt (vector signed char)
13660 vector unsigned char vec_popcnt (vector unsigned char)
13661 vector unsigned short vec_popcnt (vector signed short)
13662 vector unsigned short vec_popcnt (vector unsigned short)
13663 vector unsigned int vec_popcnt (vector signed int)
13664 vector unsigned int vec_popcnt (vector unsigned int)
13665 vector unsigned long long vec_popcnt (vector signed long long)
13666 vector unsigned long long vec_popcnt (vector unsigned long long)
13667 vector signed long long vec_slo (vector signed long long,
13668 vector signed char)
13669 vector signed long long vec_slo (vector signed long long,
13670 vector unsigned char)
13671 vector unsigned long long vec_slo (vector unsigned long long,
13672 vector signed char)
13673 vector unsigned long long vec_slo (vector unsigned long long,
13674 vector unsigned char)
13675 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13676 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13677 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13678 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13679 * doc/extend.texi: Update the built-in documentation file for the
13680 new built-in functions.
13682 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13684 * attribs.h (sorted_attr_string): Move machine independent
13685 functions for target clone support from the i386 port to common
13686 code. Rename ix86_function_versions to common_function_versions.
13687 Rename make_name to make_unique_name.
13688 (common_function_versions): Likewise.
13689 (make_unique_name): Likewise.
13690 (make_dispatcher_decl): Likewise.
13691 (is_function_default_version): Likewise.
13692 * attribs.c (attr_strcmp): Likewise.
13693 (sorted_attr_string): Likewise.
13694 (common_function_versions): Likewise.
13695 (make_unique_name): Likewise.
13696 (make_dispatcher_decl): Likewise.
13697 (is_function_default_version): Likewise.
13698 * config/i386/i386.c (attr_strcmp): Likewise.
13699 (sorted_attr_string): Likewise.
13700 (ix86_function_versions): Likewise.
13701 (make_name): Likewise.
13702 (make_dispatcher_decl): Likewise.
13703 (is_function_default_version): Likewise.
13704 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13706 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13709 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13710 Account for direct move costs for vec_construct of integer
13713 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13716 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13717 (UNSPEC_STX_ATOMIC): Ditto.
13718 (loaddi_via_sse): New insn.
13719 (storedi_via_sse): Ditto.
13720 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13721 Update corresponding peephole2 patterns.
13722 (atomic_storedi_fpu): Ditto.
13724 2017-05-11 Julia Koval <julia.koval@intel.com>
13726 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13727 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13729 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13730 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13731 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13733 2017-05-11 Nathan Sidwell <nathan@acm.org>
13735 * graphite-poly.c: Include dumpfile.h.
13737 * dumpfle.h (dump_function): Declare here ...
13738 * tree-dump.h (dump_function): ... not here.
13739 * dumpfile.c: #include tree-cfg.h.
13740 (dump_function): Move here from ...
13741 * tree-dump.c (dump_function): ... here.
13742 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13743 * graphite-poly.c: Don't include tree-dump.h.
13744 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13745 * print-tree.c: Likewise.
13746 * stor-layout.c: Likewise.
13747 * tree-nested.c: Likewise.
13749 * dumpfile.c (dump_start): Use TDF_FLAGS.
13750 (dump_enable_all): Fix TDF_KIND check thinko.
13752 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13754 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13755 array entries to represent two legal parameterizations of the
13756 overloaded __builtin_cmpb function, as represented by the
13757 P6_OV_BUILTIN_CMPB constant.
13758 (altivec_resolve_overloaded_builtin): Add special case handling
13759 for the __builtin_cmpb function, as represented by the
13760 P6_OV_BUILTIN_CMPB constant.
13761 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13762 (BU_P6_64BIT_2): New macro.
13763 (BU_P6_OVERLOAD_2): New macro
13764 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13765 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13766 (CMPB): Add overload support to represent both 32-bit and 64-bit
13767 compare-bytes function.
13768 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13769 support for TARGET_CMPB.
13770 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13771 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13772 documentation of the __builtin_cmpb overloaded built-in function.
13774 2017-05-11 Richard Biener <rguenther@suse.de>
13776 PR tree-optimization/80705
13777 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13778 bases are not vectorizable.
13780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13782 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13783 when counting register pressure.
13785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13787 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13788 (iv_ca_more_deps): Renamed to ...
13789 (iv_ca_compare_deps): ... this.
13790 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13792 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13794 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13796 (determine_group_iv_costs): ... here.
13797 (find_inv_vars_cb): Record inv var if it's not recorded before.
13799 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13801 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13802 (get_shiftadd_cost): Ditto.
13804 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13806 * tree-ssa-address.c: Include header file.
13807 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13809 (add_to_parts): Refactor.
13810 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13811 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13814 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13816 PR tree-optimization/53090
13817 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13819 (extract_cond_operands): Detect condition with IV on both sides
13820 and return COMP_IV_EXPR_2.
13821 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13822 (rewrite_use_compare): Simplify by removing call to function
13823 extract_cond_operands.
13825 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13827 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13828 (extract_cond_operands): Detect condition comparing against non-
13829 invariant bound and return appropriate enum value.
13830 (find_interesting_uses_cond): Update use of extract_cond_operands.
13831 Handle its return value accordingly.
13832 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13834 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13836 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13837 nonlinear iv_use computation in loop invariant sensitive way.
13839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13841 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13842 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13844 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13846 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13847 (dump_cand): Support iv_cand.inv_exprs.
13848 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13850 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13855 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13857 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13858 as local function. Include necessary header files.
13859 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13863 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13865 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13867 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13868 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13869 RSHIFT_EXPR and BIT_NOT_EXPR.
13871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13873 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13874 (adjust_setup_cost): New parameter supporting round up adjustment.
13875 (struct address_cost_data): Delete.
13876 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13877 (split_address_cost, ptr_difference_cost): Delete.
13878 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13879 (struct ainc_cost_data): New struct.
13880 (get_address_cost_ainc): New function.
13881 (get_address_cost, get_computation_cost): Reimplement.
13882 (determine_group_iv_cost_address): Record inv_expr for all uses of
13884 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13885 (iv_ca_has_deps): Reimplemented to ...
13886 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13888 (iv_ca_extend): Call iv_ca_more_deps.
13890 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13892 * tree-ssa-address.c (struct mem_address): Move to header file.
13893 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13894 * tree-ssa-address.h (struct mem_address): Move from C file.
13895 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13897 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13899 * tree-affine.h (aff_combination_type): New interface.
13900 (aff_combination_zero_p): Remove static.
13901 (aff_combination_const_p): New interface.
13902 (aff_combination_singleton_var_p): New interfaces.
13904 2017-05-11 Richard Biener <rguenther@suse.de>
13906 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13907 Skip unreachable blocks and destinations.
13908 (eliminate): Move stmt removal and fixup ...
13909 (fini_eliminate): ... here. Skip inserted exprs.
13910 (pass_pre::execute): Move fini_pre after fini_eliminate.
13911 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13912 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13913 PRE to get rid of dead code that has invalid SSA form and
13914 split critical edges again.
13916 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13918 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13920 2017-05-11 Richard Biener <rguenther@suse.de>
13922 * passes.c (execute_function_todo): Verify loops if they are
13923 said to be up-to-date.
13924 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13925 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13927 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13930 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13931 handle calling assemble_external ourself.
13934 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13935 modes with zero size. Enhance comment.
13937 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13939 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13940 built-ins for vec_xl and vec_xst with short and char pointer
13943 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13945 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13946 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13947 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13948 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13949 (_mm_maskz_min_round_ss): New intrinsics.
13950 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13951 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13952 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13953 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13954 (__builtin_ia32_minss_mask_round): New builtins.
13955 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13956 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13957 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13959 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13960 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13962 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13965 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13967 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13968 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13969 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13970 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13971 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13972 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13973 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13974 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13975 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13976 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13977 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13978 (__builtin_ia32_mulss_mask_round): New builtins.
13979 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13980 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13981 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13983 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13984 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13986 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13989 2017-05-10 Julia Koval <julia.koval@intel.com>
13991 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13992 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13993 (_mm256_setr_m128i): New intrinsics.
13995 2017-05-10 Julia Koval <julia.koval@intel.com>
13997 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13998 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13999 (_mm_maskz_rcp14_ss): New intrinsics.
14000 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
14001 (__builtin_ia32_rcp14ss_mask): New builtins.
14002 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
14004 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
14006 PR tree-optimization/51513
14007 * tree-cfg.c (gimple_seq_unreachable_p): New function.
14008 (assert_unreachable_fallthru_edge_p): Use it.
14009 (group_case_labels_stmt): Likewise.
14010 * tree-cfg.h: Prototype it.
14011 * stmt.c: Include cfghooks.h and tree-cfg.h.
14012 (emit_case_dispatch_table) <gap_label>: New local variable.
14013 Use it to fill dispatch table gaps.
14014 Test for default_label before updating probabilities.
14015 (expand_case) <default_label>: Remove unneeded initialization.
14016 Test for unreachable default case statement and remove its edge.
14017 Set default_label accordingly.
14018 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
14020 2017-05-10 Carl Love <cel@us.ibm.com>
14022 * config/rs6000/rs6000-c: Add support for built-in functions
14023 vector signed char vec_neg (vector signed char)
14024 vector signed short int vec_neg (vector short int)
14025 vector signed int vec_neg (vector signed int)
14026 vector signed long long vec_neg (vector signed long long)
14027 vector float vec_neg (vector float)
14028 vector double vec_neg (vector double)
14029 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
14031 * config/rs6000/altivec.h: Add define for vec_neg
14032 * doc/extend.texi: Update the built-in documentation for the
14033 new built-in functions.
14035 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14037 PR tree-optimization/77644
14038 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
14040 2017-05-10 Nathan Sidwell <nathan@acm.org>
14042 * dumpfile.h (TDI_lang_all): New.
14043 (TDF_KIND): New. Renumber others
14044 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
14046 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
14048 (get_dump_file_name): Adjust suffix generation.
14049 (dump_enable_all): Use TDF_KIND.
14050 * doc/invoke.texi (-fdump-lang-all): Document.
14052 * dumpfile.h: Tabify.
14054 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
14057 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
14058 Move member access before delete.
14060 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
14062 * tree-inline.c (expand_call_inline): Split block at stmt
14065 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14068 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14069 are now unused after splitting mov{sf,sd}_hardfloat.
14070 (f32_lr2): Likewise.
14071 (f32_lm): Likewise.
14072 (f32_lm2): Likewise.
14073 (f32_li): Likewise.
14074 (f32_li2): Likewise.
14075 (f32_lv): Likewise.
14076 (f32_sr): Likewise.
14077 (f32_sr2): Likewise.
14078 (f32_sm): Likewise.
14079 (f32_sm2): Likewise.
14080 (f32_si): Likewise.
14081 (f32_si2): Likewise.
14082 (f32_sv): Likewise.
14083 (f32_dm): Likewise.
14084 (f32_vsx): Likewise.
14085 (f32_av): Likewise.
14086 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14087 For movsf, order stores so the VSX stores occur before the GPR
14088 store which encourages the register allocator to use a traditional
14089 FPR instead of a GPR. For movsd, order the stores so that the GPR
14090 store comes before the VSX stores to allow the power6 to work.
14091 This is due to the power6 not having a 32-bit integer store
14092 instruction from a FPR.
14093 (movsf_hardfloat): Likewise.
14094 (movsd_hardfloat): Likewise.
14096 2017-05-09 Martin Sebor <msebor@redhat.com>
14098 PR translation/80280
14099 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14102 PR translation/80280
14103 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14104 data member added in r247778.
14105 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14107 2017-05-09 Nathan Sidwell <nathan@acm.org>
14109 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14111 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14114 2017-05-09 Marek Polacek <polacek@redhat.com>
14116 * doc/invoke.texi: Fix typo.
14118 2017-05-09 Richard Biener <rguenther@suse.de>
14120 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14121 (vrp_val_is_min): Likewise.
14122 (set_value_range_to_value): Likewise.
14123 (set_value_range_to_nonnegative): Likewise.
14124 (gimple_assign_nonzero_p): Likewise.
14125 (gimple_stmt_nonzero_p): Likewise.
14126 (vrp_int_const_binop): Likewise. Remove unreachable case.
14127 (adjust_range_with_scev): Adjust comments.
14128 (compare_range_with_value): Likewise.
14129 (extract_range_from_phi_node): Likewise.
14130 (test_for_singularity): Likewise.
14132 2017-05-09 Richard Biener <rguenther@suse.de>
14134 * tree-vrp.c (get_single_symbol): Add assert that we don't
14135 get overflowed constants as invariant part.
14136 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14137 checks. Use wi::cmp instead of recursing for integer constants.
14138 (compare_values): Just ignore whether we assumed undefined
14139 overflow instead of failing the compare.
14140 (extract_range_for_var_from_comparison_expr): Add comment before the
14141 TREE_NO_WARNING sets.
14142 (test_for_singularity): Likewise.
14143 (extract_range_from_comparison): Do not disable optimization
14144 when we assumed undefined overflow.
14145 (extract_range_basic): Remove init of unused var.
14147 2017-05-09 Richard Biener <rguenther@suse.de>
14149 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14150 (extract_range_from_multiplicative_op_1): Adjust.
14151 (extract_range_from_binary_expr_1): Use int_const_binop.
14153 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14156 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14157 rs6000_store_data_bypass_p in seven define_bypass directives and
14158 in several comments.
14159 * config/rs6000/rs6000-protos.h: Add prototype for
14160 rs6000_store_data_bypass_p function.
14161 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14162 function implements slightly different (rs6000-specific) semantics
14163 than store_data_bypass_p, returning false rather than aborting
14164 with assertion error when arguments do not satisfy the
14165 requirements of store data bypass.
14166 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14167 rs6000_store_data_bypass_p.
14169 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14171 * config/xtensa/xtensa-protos.h
14172 (xtensa_initial_elimination_offset): New declaration.
14173 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14174 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14175 macro definition, add case for FRAME_POINTER_REGNUM when
14176 FRAME_GROWS_DOWNWARD.
14177 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14178 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14179 xtensa_initial_elimination_offset.
14181 2017-05-08 Nathan Sidwell <nathan@acm.org>
14183 * doc/invoke.texi: Alphabetize -fdump options.
14185 2017-05-08 Martin Sebor <msebor@redhat.com>
14187 PR translation/80280
14188 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14190 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14192 * target.def (compute_frame_layout): New optional target hook.
14193 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14194 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14195 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14197 * reload1.c (verify_initial_elim_offsets): Likewise.
14198 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14199 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14200 (arm_get_frame_offsets): Split up into this ...
14201 (arm_compute_frame_layout): ... and this function.
14203 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14205 * config/aarch64/constraints.md (Usa): New constraint.
14206 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14208 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14210 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14211 with_multilib_list after it has been checked.
14213 2017-05-08 Richard Biener <rguenther@suse.de>
14215 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14216 (bitmap_set_subtract_values): Likewise.
14218 2017-05-08 Richard Biener <rguenther@suse.de>
14220 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14221 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14223 (gimple_stmt_nonzero_warnv_p): Rename to ...
14224 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14226 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14227 (extract_range_basic): Adjust, do not disable propagation on
14228 strict overflow sensitive simplification.
14229 (vrp_visit_cond_stmt): Likewise.
14231 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14233 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14234 body size unconditionally.
14236 2017-05-07 Jeff Law <law@redhat.com>
14239 2017-05-06 Jeff Law <law@redhat.com>
14240 PR tree-optimization/78496
14241 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14244 PR tree-optimization/78496
14245 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14246 (simplify_stmt_using_ranges): Call it.
14247 (vrp_dom_walker::before_dom_children): Extract equivalences
14248 from an ASSERT_EXPR with an equality comparison against a
14251 2017-05-06 Jeff Law <law@redhat.com>
14253 PR tree-optimization/78496
14254 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14257 PR tree-optimization/78496
14258 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14259 (simplify_stmt_using_ranges): Call it.
14260 (vrp_dom_walker::before_dom_children): Extract equivalences
14261 from an ASSERT_EXPR with an equality comparison against a
14264 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14266 * lra-constraints.c (lra_copy_reg_equiv): New function.
14267 (split_reg): Use it to copy equivalence information from the
14268 original register to the spill register.
14270 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14272 PR rtl-optimization/75964
14273 * simplify-rtx.c (simplify_const_relational_operation): Remove
14274 invalid handling of comparisons of integer ABS.
14276 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14278 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14279 initialize to zero.
14280 (init_regs): Remove declaration.
14281 (function_arg_advance_32): Initialize error_p as boolean variable.
14283 2017-05-05 Nathan Sidwell <nathan@acm.org>
14285 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14286 lines. Use for (;;).
14288 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14290 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14291 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14292 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14293 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14294 VF=2 that require versioning.
14296 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14298 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14301 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14303 * diagnostic.h (diagnostic_override_option_index): Convert from
14304 macro to inline function.
14306 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14308 * diagnostic.c (last_module_changed_p): New function.
14309 (set_last_module): New function.
14310 (diagnostic_report_current_module): Convert macro usage to
14311 the above functions.
14312 * diagnostic.h (diagnostic_context::last_module): Strengthen
14313 from const line_map * to const line_map_ordinary *.
14314 (diagnostic_last_module_changed): Delete macro.
14315 (diagnostic_set_last_module): Delete macro.
14317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14319 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14320 with diagnostic_report_diagnostic.
14321 (diagnostic_n_impl_richloc): Likewise.
14322 * diagnostic.h (report_diagnostic): Delete macro.
14323 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14324 with diagnostic_report_diagnostic.
14325 * substring-locations.c (format_warning_va): Likewise.
14327 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14329 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14330 save/restor of format_spec. Move option-printing code to...
14331 (print_option_information): ...this new function, and
14332 reimplement by simply printing to the pretty_printer,
14333 rather than appending to the format string.
14335 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14337 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14338 handling logic into...
14339 (update_effective_level_from_pragmas): ...this new function.
14341 2017-05-04 Andrew Waterman <andrew@sifive.com>
14343 * config/riscv/riscv.opt (mstrict-align): New option.
14344 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14345 (SLOW_UNALIGNED_ACCESS): Define.
14346 (riscv_slow_unaligned_access): Declare.
14347 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14349 (riscv_slow_unaligned_access): New variable.
14350 (rocket_tune_info): Set slow_unaligned_access to true.
14351 (optimize_size_tune_info): Set slow_unaligned_access to false.
14352 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14353 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14354 (riscv_option_override): Set riscv_slow_unaligned_access.
14355 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14357 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14359 * config/riscv/riscv.md: Unify indentation.
14361 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14366 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14368 (extendsi<mode>2): Add support for doing sign extension via
14369 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14370 don't have ISA 3.0 instructions.
14371 (extendsi<mode>2 splitter): Likewise.
14372 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14373 generate the normal insns since SImode can now go in vector
14374 registers. Disallow the special UNSPECs needed for previous
14375 machines to hide SImode being used. Add new insns
14376 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14377 (fix_trunc<mode>si2_stfiwx): Likewise.
14378 (fix_trunc<mode>si2_internal): Likewise.
14379 (fixuns_trunc<mode>si2): Likewise.
14380 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14381 (fctiw<u>z_<mode>_smallint): Likewise.
14382 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14383 of floating point to 32-bit integer from doing a direct move to
14384 the GPR registers to do a store.
14385 (fctiwz_<mode>): Break long line.
14387 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14389 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14390 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14391 (addr_list, addr_offset_valid_p): New.
14392 (split_address_groups): Check offset validity with above function.
14393 (gt-tree-ssa-loop-ivopts.h): Include header file.
14395 2017-05-05 Nathan Sidwell <nathan@acm.org>
14397 * config.gcc (arm*-*-*): Add missing 'fi'.
14399 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14401 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14402 included in -fopt-info does not matter.
14403 * doc/optinfo.texi (-fopt-info): Fix description of default
14404 behavour. Explicitly say order of options included in -fopt-info
14407 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14409 * config.gcc: Allow combinations of aprofile and rmprofile values for
14410 --with-multilib-list.
14411 * config/arm/t-multilib: New file.
14412 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14413 variables. Remove setting of ISA and floating-point ABI in
14414 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14415 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14416 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14417 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14419 * config/arm/t-rmprofile: Likewise except for the matches changes.
14420 * doc/install.texi (--with-multilib-list): Document the combination of
14421 aprofile and rmprofile values and warn about pitfalls in doing that.
14423 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14425 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14426 (movdi_aarch64): Likewise.
14428 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14430 PR tree-optimization/80632
14431 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14433 (build_arrays): Initialize it for virtual phis.
14434 (fix_phi_nodes): Use it for virtual phis.
14436 PR tree-optimization/80558
14437 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14438 [x, y] op z into [x op, y op z] for op & or | if conditions
14441 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14442 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14445 * config/arm/arm.md (use_literal_pool): Remove.
14446 (64-bit immediate split): No longer takes cost into consideration
14447 if arm_disable_literal_pool is enabled.
14448 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14449 used when arm_disable_literal_pool is enabled.
14450 (arm_max_const_double_inline_cost): Remove use of
14451 arm_disable_literal_pool.
14452 (push_minipool_fix): Add assert.
14453 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14454 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14455 (no_literal_pool_sf_immediate): New.
14457 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14459 PR tree-optimization/80613
14460 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14461 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14463 2017-05-05 Richard Biener <rguenther@suse.de>
14465 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14467 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14469 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14470 of this flag from insn conditions due to removal from r247495.
14472 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14474 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14476 (arm_early_store_addr_dep_ptr): Likewise.
14477 * config/arm/aarch-common-protos.h
14478 (arm_early_load_addr_dep_ptr): Add prototype.
14479 (arm_early_store_addr_dep_ptr): Likewise.
14480 * config/arm/cortex-a53.md: Add new bypasses.
14482 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14484 * tree.c (next_type_uid): Change type to unsigned.
14485 (type_hash_canon): Decrement back next_type_uid if
14486 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14487 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14490 2017-05-04 Martin Sebor <msebor@redhat.com>
14492 * builtins.c: Fix a trivial typo in a comment.
14494 PR middle-end/79234
14495 * builtins.c (check_sizes): Adjust to handle reading past the end.
14496 Avoid printing excessive upper bound of ranges. Use %E to print
14497 tree nodes instead of converting them to %wu.
14498 (expand_builtin_memchr): New function.
14499 (compute_dest_size): Rename...
14500 (compute_objsize): ...to this.
14501 (expand_builtin_memcpy): Adjust.
14502 (expand_builtin_mempcpy): Adjust.
14503 (expand_builtin_strcat): Adjust.
14504 (expand_builtin_strcpy): Adjust.
14505 (check_strncat_sizes): Adjust.
14506 (expand_builtin_strncat): Adjust.
14507 (expand_builtin_strncpy): Adjust and simplify.
14508 (expand_builtin_memset): Adjust.
14509 (expand_builtin_bzero): Adjust.
14510 (expand_builtin_memcmp): Adjust.
14511 (expand_builtin): Handle memcmp.
14512 (maybe_emit_chk_warning): Check strncat just once.
14514 2017-05-04 Martin Sebor <msebor@redhat.com>
14516 PR preprocessor/79214
14517 PR middle-end/79222
14518 PR middle-end/79223
14519 * builtins.c (check_sizes): Add inlining context and issue
14520 warnings even when -Wno-system-headers is set.
14521 (check_strncat_sizes): Same.
14522 (expand_builtin_strncat): Same.
14523 (expand_builtin_memmove): New function.
14524 (expand_builtin_stpncpy): Same.
14525 (expand_builtin): Handle memmove and stpncpy.
14527 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14529 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14530 which is not used any more.
14532 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14534 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14536 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14538 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14539 (cortexa53_tunings): Likewise.
14540 (cortexa57_tunings): Likewise.
14541 (cortexa72_tunings): Likewise.
14542 (cortexa73_tunings): Likewise.
14544 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14546 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14547 Set loop alignment to 8.
14549 2017-05-04 Martin Sebor <msebor@redhat.com>
14551 PR translation/80280
14552 * builtins.c (expand_builtin_object_size): Add missing quoting to
14553 %D and like directives.
14554 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14555 (hsa_type_for_tree_type): Same.
14556 (verify_function_arguments): Same.
14557 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14558 * varasm.c (get_section): Same.
14561 2017-05-04 Martin Sebor <msebor@redhat.com>
14563 PR translation/80280
14564 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14566 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14568 * config/aarch64/aarch64.c (generic_addrcost_table):
14569 Change HI/TI mode setting.
14571 2017-05-04 Martin Jambor <mjambor@suse.cz>
14573 PR tree-optimization/80622
14574 * tree-sra.c (comes_initialized_p): New function.
14575 (build_accesses_from_assign): Only set write lazily when
14576 comes_initialized_p is false.
14577 (analyze_access_subtree): Use comes_initialized_p.
14578 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14579 instead of testing for PARM_DECL.
14581 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14583 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14584 constraint on operand 0 to allow more general addressing modes.
14585 Adjust output template.
14586 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14588 * config/aarch64/aarch64-protos.h
14589 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14590 * config/aarch64/constraints.md (Dp): New address constraint.
14591 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14594 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14596 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14597 update use of estimate_ipcp_clone_size_and_time.
14598 (estimate_local_effects): Update use of
14599 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14600 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14601 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14602 Return nonspecialized time.
14604 2017-05-04 Richard Biener <rguenther@suse.de>
14606 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14607 for the last VUSE which def dominates the PHI. Directly call
14609 (get_continuation_for_phi_1): Remove.
14611 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14613 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14614 to explain the use of truncating division. Cap the number of
14615 iterations to the maximum given by nb_iterations_upper_bound,
14618 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14620 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14621 * configure: Regenerate.
14622 * config.in: Regenerate.
14623 * config/i386/driver-mingw32.c: new file.
14624 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14625 * config.host: Link driver-mingw32.o on MinGW host.
14626 * doc/install.texi: Document new --enable-mingw-wildcard configure
14629 2017-05-04 Marek Polacek <polacek@redhat.com>
14631 PR tree-optimization/80612
14632 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14634 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14635 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14637 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14638 (movt splitter): Likewise.
14639 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14640 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14641 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14642 block for Thumb-1 with MOVT.
14643 (thumb2_legitimate_address_p): Move code block ...
14644 (can_avoid_literal_pool_for_label_p): ... into this new function.
14645 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14647 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14648 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14649 "M-profile targets with the MOVT instruction".
14651 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14653 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14654 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14655 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14657 2017-05-04 Martin Liska <mliska@suse.cz>
14659 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14660 variable cond_code.
14662 2017-05-04 Richard Biener <rguenther@suse.de>
14664 * tree.c (array_at_struct_end_p): Handle arrays at struct
14665 end with flexarrays more conservatively. Refactor and treat
14666 arrays of arrays or aggregates more strict. Fix
14667 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14668 * tree.c (array_at_struct_end_p): Adjust prototype.
14669 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14670 * gimple-fold.c (get_range_strlen): Likewise.
14671 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14673 2017-05-04 Richard Biener <rguenther@suse.de>
14675 PR tree-optimization/31130
14676 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14678 (supports_overflow_infinity): Likewise.
14679 (is_negative_overflow_infinity): Likewise.
14680 (is_positive_overflow_infinity): Likewise.
14681 (is_overflow_infinity): Likewise.
14682 (stmt_overflow_infinity): Likewise.
14683 (overflow_infinity_range_p): Likewise.
14684 (usable_range_p): Remove as always returning true.
14685 (make_overflow_infinity): Remove.
14686 (negative_overflow_infinity): Likewise.
14687 (positive_overflow_infinity): Likewise.
14688 (avoid_overflow_infinity): Likewise.
14689 (set_value_range): Adjust accordingly.
14690 (set_value_range_to_nonnegative): Likewise, remove now unused
14691 overflow_infinity arg.
14692 (vrp_operand_equal_p): Adjust.
14693 (update_value_range): Likewise.
14694 (range_int_cst_singleton_p): Likewise.
14695 (operand_less_p): Likewise.
14696 (compare_values_warnv): Likewise.
14697 (extract_range_for_var_from_comparison_expr): Likewise.
14698 (vrp_int_const_binop): Likewise.
14699 (zero_nonzero_bits_from_vr): Likewise.
14700 (extract_range_from_multiplicative_op_1): Likewise.
14701 (extract_range_from_binary_expr_1): Likewise.
14702 (extract_range_from_unary_expr): Likewise.
14703 (extract_range_from_comparison): Likewise.
14704 (extract_range_basic): Likewise.
14705 (adjust_range_with_scev): Likewise.
14706 (compare_ranges): Likewise.
14707 (compare_range_with_value): Likewise.
14708 (dump_value_range): Likewise.
14709 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14711 (simplify_cond_using_ranges): Adjust.
14713 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14715 * brig-builtins.def: Added a builtin for class_f64.
14716 * builtin-types.def: Added a builtin type needed by class_f64.
14718 2017-05-03 Jason Merrill <jason@redhat.com>
14720 * timevar.def: Add TV_CONSTEXPR.
14722 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14724 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14726 2017-05-03 Martin Jambor <mjambor@suse.cz>
14728 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14729 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14730 * ipa-cp.c (ipcp_propagate_stage): Do not call
14731 ipa_update_after_lto_read.
14732 * ipa-inline.c (ipa_inline): Likewise.
14734 2017-05-03 Martin Jambor <mjambor@suse.cz>
14736 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14737 tag. Added a default constructor and a destructor.
14738 (ipa_edge_args_sum_t): New class;
14739 (ipa_edge_args_sum): Declare.
14740 (ipa_edge_args_vector): Remove declaration.
14741 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14742 (ipa_free_edge_args_substructures): Remove declaration.
14743 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14744 (ipa_edge_args_info_available_for_edge_p): Likewise.
14745 * ipa-prop.c (ipa_edge_args_vector): Removed.
14746 (edge_removal_hook_holder): Likewise.
14747 (edge_duplication_hook_holder): Likewise.
14748 (ipa_edge_args_sum): New variable.
14749 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14750 ipa_edge_args_vector.
14751 (ipa_free_edge_args_substructures): Likewise.
14752 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14753 ipa_edge_args_vector.
14754 (ipa_edge_removal_hook): Turned into method
14755 ipa_edge_args_sum_t::remove.
14756 (ipa_edge_duplication_hook): Turned into method
14757 ipa_edge_args_sum_t::duplicate.
14758 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14759 registering edge hooks.
14760 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14761 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14762 ipa_edge_args_sum instead of ipa_edge_args_vector.
14763 * ipa-profile.c (ipa_profile): Likewise.
14765 2017-05-03 Martin Jambor <mjambor@suse.cz>
14767 * symbol-summary.h (function_summary): New method exists.
14768 (function_summary::symtab_removal): Deallocate through release.
14769 (call_summary): New class.
14770 (gt_ggc_mx): New overload.
14771 (gt_pch_nx): Likewise.
14772 (gt_pch_nx): Likewise.
14774 2017-05-03 Jeff Law <law@redhat.com>
14776 PR tree-optimization/78496
14777 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14778 from simplify_cond_using_ranges. Split off code to walk
14779 backwards through casts into ...
14780 (simplify_cond_using_ranges_2): New function.
14781 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14782 (execute_vrp): After identifying jump threads, call
14783 simplify_cond_using_ranges_2.
14785 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14788 * ipa-inline.h (inline_summary): Add ctor.
14789 (create_ggc): Do not use ggc_cleared_alloc.
14791 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14792 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14794 * gcc.c (handle_braces): Support escaping in switch matching
14796 * doc/invoke.texi (Spec Files): Document it.
14797 Remove superfluous @code markup in items.
14799 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14801 * diagnostic-show-locus.c (struct column_range): New struct.
14802 (get_affected_columns): New function.
14803 (get_printed_columns): New function.
14804 (struct correction): New struct.
14805 (correction::ensure_capacity): New function.
14806 (correction::ensure_terminated): New function.
14807 (struct line_corrections): New struct.
14808 (line_corrections::~line_corrections): New dtor.
14809 (line_corrections::add_hint): New function.
14810 (layout::print_trailing_fixits): Reimplement in terms of the new
14812 (selftest::test_overlapped_fixit_printing): New function.
14813 (selftest::diagnostic_show_locus_c_tests): Call it.
14815 2017-05-03 Nathan Sidwell <nathan@acm.org>
14817 Canonicalize canonical type hashing
14818 * tree.h (type_hash_canon_hash): Declare.
14819 * tree.c (type_hash_list, attribute_hash_list): Move into
14820 type_hash_canon_hash.
14821 (build_type_attribute_qual_variant): Break out hash code calc into
14822 type_hash_canon_hash.
14823 (type_hash_canon_hash): New. Generic type hash computation.
14824 (build_range_type_1, build_array_type_1, build_function_type,
14825 build_method_type_directly, build_offset_type, build_complex_type,
14826 make_vector_type): Call it.
14828 2017-05-03 Richard Biener <rguenther@suse.de>
14830 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14831 When all DRs have unknown misaligned do not always peel
14832 when there is a store but apply the same costing model as if
14833 there were only loads.
14835 2017-05-03 Richard Biener <rguenther@suse.de>
14838 PR tree-optimization/80492
14839 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14840 compare_base_decls returning dont-know properly.
14842 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14844 * config/arm/iterators.md (CCSI): New mode iterator.
14845 (arch): New mode attribute.
14846 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14847 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14848 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14849 code iterator for success result mode.
14850 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14851 the corresponding new insn generators.
14853 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14856 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14857 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14859 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14861 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14862 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14863 (DDR_A): Wrap DDR argument in brackets.
14864 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14865 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14866 (DDR_REVERSED_P): Likewise.
14868 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14870 PR tree-optimization/79472
14871 * tree-switch-conversion.c (struct switch_conv_info): Add
14872 contiguous_range and default_case_nonstandard fields.
14873 (collect_switch_conv_info): Compute contiguous_range and
14874 default_case_nonstandard fields, don't clear final_bb if
14875 contiguous_range and only the default case doesn't have the required
14877 (check_all_empty_except_final): Set default_case_nonstandard instead
14878 of failing if contiguous_range and the default case doesn't have empty
14880 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14881 and only the default case doesn't have the required constants. Skip
14883 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14884 if default_case_nonstandard.
14885 (build_constructors): Build constant 1 just once. Assert that default
14886 values aren't inserted in between cases if contiguous_range. Skip
14888 (build_arrays): Skip virtual phis.
14889 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14890 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14891 Handle virtual phis.
14892 (gen_inbound_check): Handle default_case_nonstandard case.
14893 (process_switch): Adjust check_final_bb caller. Call
14894 gather_default_values with the first non-default case instead of
14895 default case if default_case_nonstandard.
14897 2017-05-02 Nathan Sidwell <nathan@acm.org>
14899 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14900 check. Fix formatting.
14902 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14904 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14905 errors when comparing specialized and unspecialized times.
14907 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14909 * diagnostic-show-locus.c
14910 (layout::should_print_annotation_line_p): Make private.
14911 (layout::print_annotation_line): Make private.
14912 (layout::annotation_line_showed_range_p): Make private.
14913 (layout::show_ruler): Make private.
14914 (layout::print_source_line): Make private. Pass in line and
14915 line_width, rather than calling location_get_source_line. Drop
14917 (layout::print_leading_fixits): New method.
14918 (layout::print_any_fixits): Rename to...
14919 (layout::print_trailing_fixits): ...this, and make private.
14920 Don't print newline fixits.
14921 (diagnostic_show_locus): Move logic for printing one row into...
14922 (layout::print_line): ...this new function. Move the
14923 location_get_source_line call and error-handling from
14924 print_source_line to here. Call print_leading_fixits, and rename
14925 print_any_fixits to print_trailing_fixits.
14926 (selftest::test_fixit_insert_containing_newline): Update now that
14927 newlines are partially supported.
14928 (selftest::test_fixit_insert_containing_newline_2): New test.
14929 (selftest::test_fixit_replace_containing_newline): Update comments.
14930 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14931 * edit-context.c (class added_line): New class.
14932 (class edited_line): Describe newline handling in comment.
14933 (edited_line::actually_edited_p): New method.
14934 (edited_line::print_content): Delete redundant decl.
14935 (edited_line::m_predecessors): New field.
14936 (edited_file::print_content): Call edited_line::print_content.
14937 (edited_file::print_diff): Update to support newlines.
14938 (edited_file::print_diff_hunk): Likewise.
14939 (edited_file::print_run_of_changed_lines): New function.
14940 (edited_file::print_diff_line): Convert to...
14941 (print_diff_line): ...this.
14942 (edited_file::get_effective_line_count): New function.
14943 (edited_line::edited_line): Initialize new field m_predecessors.
14944 (edited_line::~edited_line): Clean up m_predecessors.
14945 (edited_line::apply_fixit): Handle newlines.
14946 (edited_line::get_effective_line_count): New function.
14947 (edited_line::print_content): New function.
14948 (edited_line::print_diff_lines): New function.
14949 (selftest::test_applying_fixits_insert_containing_newline): New
14951 (selftest::test_applying_fixits_replace_containing_newline): New
14953 (selftest::insert_line): New function.
14954 (selftest::test_applying_fixits_multiple_lines): Add example of
14956 (selftest::edit_context_c_tests): Call the new tests.
14958 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14960 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14961 parameter cand. Update dump information.
14962 (get_computation_cost): Update uses.
14964 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14966 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14967 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14968 (get_computation_at, rewrite_use_address): Update use of
14969 get_computation_aff.
14971 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14973 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14974 (get_computation): Delete.
14975 (get_computation_cost): Implement like get_computation_cost_at.
14976 Use get_computation_at.
14977 (get_computation_cost_at): Delete.
14978 (rewrite_use_nonlinear_expr): Use get_computation_at.
14979 (rewrite_use_compare, remove_unused_ivs): Ditto.
14981 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14983 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14987 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14988 (ivopts_global_cost_for_size): Rename parameter and update uses.
14989 (iv_ca_recount_cost): Update uses.
14990 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14991 candidates seperately in n_invs and n_cands.
14992 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14996 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14997 (find_inv_vars_cb): New.
14998 (find_depends): Renamed to ...
14999 (find_inv_vars): ... this.
15000 (add_candidate_1, force_var_cost): Call find_inv_vars.
15001 (split_address_cost, determine_group_iv_cost_cond): Ditto.
15003 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15005 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
15006 inv_vars. Add inv_exprs.
15007 (struct iv_cand): Rename depends_on to inv_vars.
15008 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
15009 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
15010 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
15011 (dump_cand): Dump inv_vars.
15012 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
15013 (record_invariant, find_depends, add_candidate_1): Ditto.
15014 (set_group_iv_cost, force_var_cost): Ditto.
15015 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
15016 (get_computation_cost_at, get_computation_cost): Ditto.
15017 (determine_group_iv_cost_generic): Ditto.
15018 (determine_group_iv_cost_address): Ditto.
15019 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
15020 (determine_group_iv_costs): Ditto.
15021 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
15022 (iv_ca_set_remove_invariants): Renamed to ...
15023 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
15024 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
15025 (iv_ca_set_add_invariants): Renamed to ...
15026 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
15027 (iv_ca_set_cp): Use iv_ca_set_add_invs.
15028 (iv_ca_has_deps): Support inv_vars and inv_exprs.
15029 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
15030 (create_new_ivs): Remove useless dump.
15032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15034 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
15036 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
15037 (iv_ca_set_no_cp, create_new_iv): Ditto.
15039 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15041 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15043 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15045 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
15046 function tree_check2.
15048 2017-05-02 Martin Liska <mliska@suse.cz>
15050 * doc/gcov.texi: Add missing preposition.
15051 * gcov.c (function_info::function_info): Properly fill up
15052 all member variables.
15054 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15056 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
15058 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15060 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
15062 2017-05-02 Martin Liska <mliska@suse.cz>
15065 * lto-streamer-in.c (lto_read_tree_1): Remove
15066 LTO_STREAMER_DEBUG.
15067 * lto-streamer.c (struct tree_hash_entry): Likewise.
15068 (struct tree_entry_hasher): Likewise.
15069 (tree_entry_hasher::hash): Likewise.
15070 (tree_entry_hasher::equal): Likewise.
15071 (lto_streamer_init): Likewise.
15072 (lto_orig_address_map): Likewise.
15073 (lto_orig_address_get): Likewise.
15074 (lto_orig_address_remove): Likewise.
15075 * lto-streamer.h: Likewise.
15076 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15077 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15079 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15081 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15082 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15083 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15084 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15085 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15086 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15087 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15088 (mm_maskz_sub_ss): New intrinsics.
15089 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15090 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15091 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15092 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15093 (__builtin_ia32_subss_mask_round): New builtins.
15094 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15095 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15096 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15098 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15099 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15101 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15104 2017-05-02 Martin Jambor <mjambor@suse.cz>
15106 PR tree-optimization/78687
15107 * tree-sra.c (access): New field parent.
15108 (process_subtree_disqualification): New function.
15109 (disqualify_candidate): Call it.
15110 (build_accesses_from_assign): Reset write flag if creating an
15112 (build_access_subtree): Fill in parent field and also prpagate
15113 down grp_write flag.
15114 (create_artificial_child_access): New parameter set_grp_write, set
15115 grp_write to its value.
15116 (propagate_subaccesses_across_link): Also propagate grp_write flag
15118 (propagate_all_subaccesses): Push the closest parent back to work
15119 queue if add_access_to_work_queue returned true.
15121 2017-05-02 Richard Biener <rguenther@suse.de>
15123 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15124 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15125 -fstrict-overflow documentation.
15126 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15127 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15128 flag_strict_overflow.
15129 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15130 * lto-opts.c (lto_write_options): Do not stream it.
15131 * lto-wrapper.c (merge_and_complain): Do not handle it.
15132 * opts.c (default_options_table): Do not set -fstrict-overflow.
15133 (finish_options): Likewise do not clear it when sanitizing.
15134 * simplify-rtx.c (simplify_const_relational_operation): Do not
15135 test flag_strict_overflow.
15137 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15139 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15140 using enabled attribute.
15141 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15142 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15143 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15144 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15145 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15146 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15147 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15148 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15149 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15150 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15152 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15154 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15156 2017-05-02 Richard Biener <rguenther@suse.de>
15158 PR tree-optimization/80591
15160 2017-04-10 Richard Biener <rguenther@suse.de>
15162 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15165 2017-05-02 Richard Biener <rguenther@suse.de>
15167 PR tree-optimization/80549
15168 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15169 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15170 headers if they do not have a preheader.
15172 2017-05-02 Martin Liska <mliska@suse.cz>
15175 * common.opt: Fix typo.
15176 * doc/invoke.texi: Likewise.
15178 2017-05-01 Jan Beulich <jbeulich@suse.com>
15180 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15181 swapping, add (x,x,m,x,n) alternative.
15183 2017-05-01 Nathan Sidwell <nathan@acm.org>
15185 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15186 unnecessary unadjusted_alignment check.
15188 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15191 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15193 * gimplify.c (gimplify_cilk_detach): New function.
15194 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15195 * tree-core.h: Document EXPR_CILK_SPAWN.
15196 * tree.h (EXPR_CILK_SPAWN): Define.
15198 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15200 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15201 to use new fixit_hint representation, using the "replace" logic.
15202 (get_line_span_for_fixit_hint): Likewise.
15203 (layout::print_any_fixits): Likewise.
15204 (selftest::test_one_liner_many_fixits): Rename to...
15205 (selftest::test_one_liner_many_fixits_1): ...this, and update
15206 comment and expected output to reflect that the multiple fix-it
15207 hints are now consolidated into one insertion.
15208 (selftest::test_one_liner_many_fixits_2): New test.
15209 (selftest::test_diagnostic_show_locus_one_liner): Update for
15211 (selftest::test_fixit_consolidation): Update for fix-it API
15213 * diagnostic.c (print_parseable_fixits): Likewise.
15214 * edit-context.c (edited_line::m_line_events): Convert from
15215 auto_vec <line_event *> to auto_vec <line_event>.
15216 (class line_event): Convert from abstract base class to a concrete
15217 class, taking over the role of replace_event.
15218 (class insert_event): Delete.
15219 (class replace_event): Rename to class line_event. Convert to
15221 (edit_context::add_fixits): Reimplement.
15222 (edit_context::apply_insert): Delete.
15223 (edit_context::apply_replace): Rename to...
15224 (edit_context::apply_fixit): ...this. Convert to half-open range.
15225 (edited_file::apply_insert): Delete.
15226 (edited_file::apply_replace): Rename to...
15227 (edited_file::apply_fixit): ...this.
15228 (edited_line::~edited_line): Drop deletion of events.
15229 (edited_line::apply_insert): Delete.
15230 (edited_line::apply_replace): Rename to...
15231 (edited_line::apply_fixit): ...this. Convert to half-open range.
15232 Update for change to type of m_line_events.
15233 * edit-context.h (edit_context::apply_insert): Delete.
15234 (edit_context::apply_replace): Rename to...
15235 (edit_context::apply_fixit): ...this.
15237 2017-05-01 Martin Sebor <msebor@redhat.com>
15239 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15242 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15245 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15246 __get_cpuid_max returns 0.
15247 (__get_cpuid_count): Ditto.
15249 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15251 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15252 replacement expression is another instance of one of its arguments.
15254 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15257 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15258 check for stack push/pop autoinc.
15259 * config/i386/i386.c (ix86_agi_dependent): Return false
15260 if the only reason why modified_in_p returned true is that
15261 addr is SP based and set_insn is a push or pop.
15263 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15265 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15268 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15271 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15272 (account_size_time): Use two predicates - exec_pred and
15274 (evaluate_conditions_for_known_args): Compute both clause and
15276 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15277 (inline_summary_t::duplicate): Update.
15278 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15280 (compute_inline_parameters): Likewise.
15281 (estimate_edge_size_and_time): Update caluclation of time.
15282 (estimate_node_size_and_time): Compute both time and nonspecialized
15284 (estimate_ipcp_clone_size_and_time): Update.
15285 (inline_merge_summary): Update.
15286 (do_estimate_edge_time): Update.
15287 (do_estimate_edge_size): Update.
15288 (do_estimate_edge_hints): Update.
15289 (inline_read_section, inline_write_summary): Stream both new predicates.
15290 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15292 (compute_inlined_call_time): Cleanup.
15293 (big_speedup_p): Update.
15294 (edge_badness): Update.
15295 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15296 (size_time_entry): Replace predicate by exec_predicate and
15297 nonconst_predicate.
15298 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15299 (estimate_edge_time): Return also nonspec_time.
15300 (reset_edge_growth_cache): Update.
15302 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15304 PR rtl-optimization/80491
15305 * ifcvt.c (noce_process_if_block): When looking for x setter
15306 with missing else_bb, don't check only the insn right before
15307 cond_earliest, but look for the last insn that x is modified in
15308 within the same bb.
15310 PR rtl-optimization/80491
15311 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15313 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15315 PR tree-optimization/80487
15316 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15318 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15320 PR tree-optimization/79697
15321 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15322 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15323 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15325 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15326 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15328 2017-04-28 Martin Sebor <msebor@redhat.com>
15330 PR tree-optimization/80523
15331 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15332 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15334 (maybe_warn, format_directive, parse_directive): Use new functions.
15335 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15337 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15339 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15341 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15343 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15344 target_header_dir): Set correctly.
15345 * configure: Regenerated.
15346 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15347 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15348 instead of SYSTEM_HEADER_DIR.
15350 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15352 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15353 (estimate_local_effects): Likewise.
15354 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15355 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15356 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15357 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15358 do_estimate_edge_time, estimate_edge_time): Likewise.
15359 * ipa-inline-analysis.c (estimate_node_size_and_time,
15360 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15361 (estimate_time_after_inlining): Remove.
15363 2017-04-28 Martin Liska <mliska@suse.cz>
15365 * doc/gcov.texi: Enhance documentation of gcov.
15367 2017-04-28 Martin Liska <mliska@suse.cz>
15369 * doc/gcov.texi: Sort options in alphabetic order.
15370 * doc/gcov-dump.texi: Likewise.
15371 * doc/gcov-tool.texi: Likewise.
15372 * gcov.c (print_usage): Likewise.
15373 * gcov-dump.c (print_usage): Likewise.
15374 * gcov-tool.c (print_merge_usage_message): Likewise.
15375 (print_rewrite_usage_message): Likewise.
15376 (print_overlap_usage_message): Likewise.
15378 2017-04-28 Martin Liska <mliska@suse.cz>
15380 PR gcov-profile/53915
15381 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15383 2017-04-28 Martin Liska <mliska@suse.cz>
15385 PR gcov-profile/79891
15386 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15387 is marked by compiler as living on a line.
15388 (get_cycles_count): Remove usage of the union.
15389 (output_intermediate_file): Likewise.
15390 (find_source): Fix GNU coding style.
15391 (accumulate_line_counts): Remove old non-all block mode.
15392 (output_lines): Remove usage of the union.
15393 * profile.c (output_location): Include all BBs, even if
15394 belonging to a same line (and file) as a previous BB.
15396 2017-04-28 Martin Liska <mliska@suse.cz>
15398 * gcov.c (process_args): Handle new argument 'w'.
15399 (read_graph_file): Assign ID to BBs.
15400 (output_branch_count): Display BB # if verbose flag is set.
15401 (output_lines): Likewise for arcs.
15402 (print_usage): Add '--verbose' option help.
15403 * doc/gcov.texi: Document --verbose (-w) option.
15405 2017-04-28 Martin Liska <mliska@suse.cz>
15407 * gcov.c (struct block_location_info): New struct.
15408 (process_file): Fill up the new structure.
15409 (read_graph_file): Replace usage of encoding by the newly added
15411 (add_line_counts): Likewise.
15412 (accumulate_line_counts): Remove usage of the union.
15413 (function_info::function_info): New function.
15414 (function_info::~function_info): Likewise.
15415 (process_file): Call delete instead of release_function.
15416 (release_function): Release the function.
15417 (release_structures): Call delete instead of release_function.
15418 (solve_flow_graph): Replace usage of num_blocks.
15419 (find_exception_blocks): Likewise.
15420 (output_lines): Fix GNU coding style.
15422 2017-04-28 Martin Liska <mliska@suse.cz>
15425 * coverage.c (coverage_remove_note_file): New function.
15426 * coverage.h: Declare the function.
15427 * toplev.c (finalize): Clean if an error has been seen.
15429 2017-04-28 Martin Liska <mliska@suse.cz>
15431 PR gcov-profile/80031
15432 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15433 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15434 * gcov.c (read_graph_file): Read just number of blocks.
15435 * profile.c (branch_prob): Do not stream 0 flags per a basic
15438 2017-04-28 Martin Liska <mliska@suse.cz>
15440 * gcov-dump.c (tag_*): Add new argument to declarations.
15441 (dump_gcov_file): Likewise.
15442 (tag_blocks): Add and use new argument depth.
15443 (tag_arcs): Likewise.
15444 (tag_lines): Likewise.
15445 (tag_counters): Likewise.
15446 (tag_summary): Likewise.
15447 (dump_working_sets): Use depth to do a proper indentation.
15449 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15452 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15453 * symtab.c (symtab_node::debug_symtab): Move definition here.
15455 2017-04-28 Richard Biener <rguenther@suse.de>
15457 * lto-streamer.h (LTO_major_version): Bump to 7.
15459 2017-04-28 Richard Biener <rguenther@suse.de>
15461 * tree-vrp.c (assert_info): New struct.
15462 (add_assert_info): New helper.
15463 (register_edge_assert_for_2): Refactor to add asserts to a vector
15465 (register_edge_assert_for_1): Likewise.
15466 (register_edge_assert_for): Likewise.
15467 (finish_register_edge_assert_for): New helper actually registering
15468 asserts where live on edge.
15469 (find_conditional_asserts): Adjust.
15470 (find_switch_asserts): Likewise.
15471 (evrp_dom_walker::try_find_new_range): Generalize.
15472 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15474 2017-04-27 Marek Polacek <polacek@redhat.com>
15477 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15478 arg10 and arg11 to itype.
15480 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15482 * doc/extend.texi (Object Size Checking): Improve grammar.
15484 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15487 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15488 that the logic for permitting reciprocal estimates matches that
15491 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15494 * tree.c (type_cache_hasher::equal): Only compare
15495 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15496 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15497 non-aggregate element types.
15498 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15499 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15501 2017-04-27 Richard Biener <rguenther@suse.de>
15503 PR middle-end/80533
15504 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15505 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15506 keeping a reference to a trailing array.
15508 2017-04-27 Richard Biener <rguenther@suse.de>
15510 PR middle-end/80539
15511 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15512 being in loop-closed SSA form conservatively.
15513 (chrec_fold_multiply_poly_poly): Likewise.
15515 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15517 PR middle-end/79665
15518 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15519 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15521 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15524 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15525 (aarch64_function_arg_alignment): Return unsigned int again, but still
15526 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15527 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15528 Don't emit -Wpsabi note.
15529 (aarch64_function_arg_boundary): Likewise.
15530 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15533 2017-04-26 Nathan Sidwell <nathan@acm.org>
15535 * tree.h (crc32_unsigned_n): Declare.
15536 (crc32_unsigned, crc32_unsigned): Make inline.
15537 * tree.c (crc32_unsigned_bits): Replace with ...
15538 (crc32_unsigned_n): ... this.
15539 (crc32_unsigned, crc32_byte): Remove.
15540 (crc32_string): Remove unnecessary braces.
15542 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15544 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15545 * ipa-inline-analysis.c (MAX_TIME): Remove.
15546 (account_size_time): Use sreal for time.
15547 (dump_inline_summary): Update.
15548 (estimate_function_body_sizes): Update.
15549 (estimate_edge_size_and_time): Update.
15550 (estimate_calls_size_and_time): Update.
15551 (estimate_node_size_and_time): Update.
15552 (inline_merge_summary): Update.
15553 (inline_update_overall_summary): Update.
15554 (estimate_time_after_inlining): Update.
15555 (inline_read_section): Update.
15556 (inline_write_summary): Update.
15557 * ipa-inline.c (compute_uninlined_call_time): Update.
15558 (compute_inlined_call_time): Update.
15559 (recursive_inlining): Update.
15560 (inline_small_functions): Update.
15561 (dump_overall_stats): Update.
15562 * ipa-inline.h: Include sreal.h.
15563 (size_time_entry): Turn time to sreal.
15564 (inline_summary): Turn self_time nad time to sreal.
15566 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15568 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15570 (sreal::stream_out, sreal::stream_in): New.
15571 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15573 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15575 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15578 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15581 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15583 (dimode_scalar_chain::compute_convert_gain): Ditto.
15584 (dimode_scalar_chain::make_vector_copies): Ditto.
15585 (dimode_scalar_chain::convert_reg): Ditto.
15586 (dimode_scalar_chain::convert_insn): Ditto.
15587 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15588 (VI248_AVX512BW_1): New mode iterator.
15589 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15590 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15593 2017-04-25 Martin Sebor <msebor@redhat.com>
15595 PR tree-optimization/80497
15596 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15597 constants are representable in HOST_WIDE_INT.
15598 (parse_directive): Ditto.
15600 2017-04-25 Martin Sebor <msebor@redhat.com>
15603 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15604 (new_zero_array): Adjust signature.
15605 (dom_info::dom_init): Used unsigned rather that size_t.
15606 (dom_info::dom_info): Same.
15608 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15609 Jakub Jelinek <jakub@redhat.com>
15612 * config/arm/arm.c: Include gimple.h.
15613 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15614 returns negative, increment ncrn only if it returned positive.
15615 (arm_needs_doubleword_align): Return int instead of bool,
15616 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15617 members, but if there is any such non-FIELD_DECL
15618 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15619 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15620 returns negative, increment nregs only if it returned positive.
15621 (arm_setup_incoming_varargs): Likewise.
15622 (arm_function_arg_boundary): Emit -Wpsabi note if
15623 arm_needs_doubleword_align returns negative, return
15624 DOUBLEWORD_ALIGNMENT only if it returned positive.
15626 2017-04-25 Marek Polacek <polacek@redhat.com>
15629 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15630 first argument to type.
15632 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15635 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15636 type checks to test for compatibility instead of equality.
15638 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15639 Jakub Jelinek <jakub@redhat.com>
15642 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15644 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15645 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15646 the alignment computation, but return their maximum in warn_alignment.
15647 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15648 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15650 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15651 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15654 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15656 * config/arc/simdext.md (dmpyh): Fix typo.
15658 2017-04-25 Richard Biener <rguenther@suse.de>
15660 PR tree-optimization/80492
15661 * alias.c (compare_base_decls): Handle registers with asm
15662 specification conservatively.
15663 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15664 compare_base_decls returning dont-know properly.
15666 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15668 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15669 (legitimate_offset_address_p): New function.
15670 (arc_legitimate_address_p): Use above function.
15672 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15674 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15676 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15678 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15679 ACCH registers whenever they are available.
15681 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15683 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15684 double regs fix when not used.
15686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15688 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15690 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15691 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15693 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15695 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15696 when emitting short ADD instructions.
15698 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15700 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15702 (cmpsi_cc_c_insn): Likewise.
15703 (cbranchsi4_scratch): Compute proper instruction length using
15704 compact_hreg_operand.
15705 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15707 2017-04-25 Richard Biener <rguenther@suse.de>
15709 PR middle-end/80509
15710 * passes.c (pass_manager::pass_manager): Initialize
15711 m_name_to_pass_map.
15713 2017-04-25 Richard Biener <rguenther@suse.de>
15715 PR tree-optimization/79201
15716 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15718 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15721 * config/s390/vector.md: Split MEM->GPR vector moves for
15722 non-s_operand addresses.
15724 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15727 * config/s390/predicates.md (reload_const_wide_int_operand): New
15729 * config/s390/s390.md ("movti"): Remove d/P alternative.
15730 ("movti_bigconst"): New pattern definition.
15732 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15735 * s390-protos.h (s390_expand_cs_hqi): Removed.
15736 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15737 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15738 modes as well as CCZ1mode and CCZmode.
15739 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15740 signature of s390_emit_compare_and_swap.
15741 (s390_expand_cs_hqi): Likewise, make static.
15742 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15743 compare-and-swap, in some cases.
15744 (s390_expand_cs): Wrapper function.
15745 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15747 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15748 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15749 patterns for small and large integers. Forbid symref memory operands.
15750 Move expander to s390.c. Require cc register.
15751 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15752 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15753 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15754 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15755 symref memory operands. Remove CC mode and call s390_match_ccmode
15757 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15759 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15761 * config/s390/s390.md (define_peephole2): New peephole to help
15762 combining the load-and-test pattern with volatile memory.
15764 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15766 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15767 with CCZmode for TARGET_Z196.
15769 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15771 PR rtl-optimization/80501
15772 * combine.c (make_compound_operation_int): Set subreg_code to SET
15773 even for AND with mask of the sign bit of mode.
15775 PR rtl-optimization/80500
15776 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15777 sum's initial value.
15779 2017-04-25 Julian Brown <julian@codesourcery.com>
15780 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15782 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15784 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15786 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15788 2017-04-25 Julian Brown <julian@codesourcery.com>
15789 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15791 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15792 (thunderx2t99_sha): New Reservation.
15794 2017-04-25 Julian Brown <julian@codesourcery.com>
15795 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15797 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15798 type for 1-element load.
15800 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15802 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15804 2017-04-24 Martin Jambor <mjambor@suse.cz>
15806 PR tree-optimization/80293
15807 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15808 char arrays not totally scalarizable if it is false.
15809 (analyze_all_variable_accesses): Pass correct value in the new
15810 parameter. Add a statistics counter.
15812 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15814 PR middle-end/79931
15815 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15817 2017-04-24 Richard Biener <rguenther@suse.de>
15819 PR tree-optimization/80494
15820 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15821 out for complex types.
15823 2017-04-24 Richard Biener <rguenther@suse.de>
15825 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15826 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15827 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15828 (DFS): Adjust and never fail.
15829 (sccvn_dom_walker::fail): Remove.
15830 (sccvn_dom_walker::before_dom_children): Adjust.
15831 (run_scc_vn): Likewise and never fail.
15832 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15833 (pass_fre::execute): Likewise.
15835 2017-04-24 Richard Biener <rguenther@suse.de>
15837 PR tree-optimization/79725
15838 * tree-ssa-sink.c (statement_sink_location): Return whether
15839 failure reason was zero uses. Move that check later.
15840 (sink_code_in_bb): Deal with zero uses by removing the stmt
15843 2017-04-24 Richard Biener <rguenther@suse.de>
15846 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15847 pointer-based references.
15849 2017-04-24 Richard Biener <rguenther@suse.de>
15852 * pass_manager.h (pass_manager::operator new): Remove.
15853 (pass_manager::operator delete): Likewise.
15854 * passes.c (pass_manager::operator new): Remove.
15855 (pass_manager::operator delete): Likewise.
15856 (pass_manager::pass_manager): Zero individual pass members.
15858 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15861 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15862 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15863 Check "XEXP (src, 1)" operand here.
15864 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15865 Check "XEXP (src, 1)" operand here.
15866 (dimode_scalar_chain::make_vector_copies): Detect count register
15867 of a shift instruction. Zero extend count register from QImode
15868 to DImode to satisfy vector shift pattern count operand predicate.
15869 Substitute vector shift count operand with a DImode copy.
15870 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15873 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15875 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15876 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15877 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15878 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15879 (UNSPEC_NOREX_MEM): Remove definition.
15881 2017-04-21 Richard Biener <rguenther@suse.de>
15883 PR tree-optimization/79547
15884 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15885 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15886 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15887 without any constraints.
15889 2017-04-21 Richard Biener <rguenther@suse.de>
15891 PR tree-optimization/78847
15892 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15894 2017-04-21 Richard Biener <rguenther@suse.de>
15896 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15897 (build_distinct_type_copy): Likewise.
15898 (build_variant_type_copy): Likewise.
15899 * tree.c (build_qualified_type): Pass down mem-stat info.
15900 (build_distinct_type_copy): Likewise.
15901 (build_variant_type_copy): Likewise.
15903 2017-04-21 Richard Biener <rguenther@suse.de>
15905 PR tree-optimization/80237
15906 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15908 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15909 for a simplified result.
15911 2016-04-21 Richard Biener <rguenther@suse.de>
15913 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15914 sth as strict as a simple_iv but a chrec without symbols and an
15915 operand defined in the loop we are peeling (and not some subloop).
15916 (propagate_constants_for_unrolling): Propagate all constants.
15918 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15921 * config/i386/i386.c (print_reg): Remove assert for disalowed
15922 regno values, call output_operand_lossage instead.
15924 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15927 * config/i386/constraints.md (Yc): New register constraint.
15928 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15929 Use Yc constraint for alternative 2 of operand 0. Remove
15930 preferred_for_speed attribute.
15932 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15934 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15935 lastprivate clauses in SIMT case.
15937 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15939 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15941 2017-04-20 Richard Biener <rguenther@suse.de>
15943 PR tree-optimization/57796
15944 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15945 as N scalar stores.
15946 (vect_model_load_cost): Cost gathers as N scalar loads.
15948 2017-04-20 Richard Biener <rguenther@suse.de>
15950 * ggc-page.c (ggc_allocated_p): Rename to ...
15951 (safe_lookup_page_table_entry): ... this and return the lookup
15953 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15955 2017-04-20 Richard Biener <rguenther@suse.de>
15957 PR tree-optimization/80453
15958 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15959 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15960 from the conditions.
15961 (vn_phi_eq): Pass them down.
15962 (vn_phi_lookup): Record them.
15963 (vn_phi_insert): Likewise.
15965 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15967 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15968 uninitialized variable warning to avoid buffer overrun.
15970 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15973 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15974 is suppressed for '{ 0 }' in C.
15976 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15978 * BASE-VER: Set to 8.0.0.
15980 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15982 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15983 priority .init_array and .fini_array section with SECTION_NOTYPE
15986 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15988 PR middle-end/80423
15989 * tree.h (build_array_type): Add typeless_storage default argument.
15990 * tree.c (type_cache_hasher::equal): Also compare
15991 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15992 (build_array_type): Add typeless_storage argument, set
15993 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15995 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15996 (build_array_type): Likewise. Add typeless_storage argument.
15998 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15999 Jakub Jelinek <jakub@redhat.com>
16001 PR tree-optimization/80426
16002 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
16003 operation on symbolic operands, also compute the overflow for the
16004 invariant part when the operation degenerates into a negation.
16006 2017-04-19 Jakub Jelinek <jakub@redhat.com>
16009 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
16010 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
16013 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
16015 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
16018 * config/avr/avr.c (tree.h): Include it.
16019 (cgraph.h): Include it.
16020 (avr_encode_section_info): Don't warn for uninitialized progmem
16021 variable if it's just an alias.
16023 2017-04-19 Richard Biener <rguenther@suse.de>
16026 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
16027 when needed by AutoPGO.
16029 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
16032 * doc/lto.texi: Remove an extra 'that'.
16034 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
16036 PR rtl-optimization/80429
16037 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
16038 are only used in debug insns.
16040 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16041 Vladimir Makarov <vmakarov@redhat.com>
16043 * config/sparc/predicates.md (input_operand): Add comment. Return
16044 true for any memory operand when LRA is in progress.
16045 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
16047 2017-04-18 Jeff Law <law@redhat.com>
16050 * mips.md ({return,simple_return}_internal): Do not overwrite
16053 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16055 PR tree-optimization/80443
16056 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
16057 instead of adding 1, subtract -1 and similarly instead of subtracting
16060 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
16062 PR rtl-optimization/80357
16063 * haifa-sched.c (tmp_bitmap): New variable.
16064 (model_recompute): Handle duplicate use records.
16065 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16066 (free_global_sched_pressure_data): Free it.
16068 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16071 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16072 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16073 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16074 instead of SYSTEM_HEADER_DIR.
16076 2017-04-18 Jeff Law <law@redhat.com>
16078 PR middle-end/80422
16079 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16080 predecessors after walking up the insn chain.
16082 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16085 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16086 sizetype type into debug info.
16088 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16091 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16092 unneeded test for TARGET_UPPER_REGS_SF.
16093 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16095 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16098 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16099 instead of gsi_after_labels.
16101 2017-04-18 Jeff Law <law@redhat.com>
16103 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16107 2017-04-13 Jeff Law <law@redhat.com>
16108 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16109 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16111 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16114 * config/avr/avr.c (intl.h): Include it.
16115 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16117 2017-04-18 Martin Liska <mliska@suse.cz>
16119 PR gcov-profile/78783
16120 * gcov-tool.c (gcov_output_files): Validate that destination
16121 file is either removed by the tool or by a user.
16123 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16124 Guy Benyei <guybe@mellanox.com>
16126 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16127 block, and do not negate it, the stored id is already negative.
16129 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16131 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16133 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16136 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16137 masks of options that should be turned off if the VSX vector
16138 options are turned off.
16139 (OTHER_P8_VECTOR_MASKS): Likewise.
16140 (OTHER_VSX_VECTOR_MASKS): Likewise.
16141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16142 rs6000_disable_incompatible_switches to validate no type switches
16144 (rs6000_incompatible_switch): New function to disallow turning on
16145 other vector options if -mno-vsx, -mno-power8-vector, or
16146 -mno-power9-vector are specified.
16148 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16150 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16152 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16154 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16155 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16156 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16157 (ARG_POINTER_CFA_OFFSET): Likewise.
16159 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16161 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16162 conditions to take advantage of various optimizations.
16164 2017-04-13 Jeff Law <law@redhat.com>
16166 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16167 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16168 (zero_extendsidi2_dext): Likewise.
16170 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16173 * fold-const.c (fold_ternary_loc): Revert
16174 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16177 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16179 PR rtl-optimization/80343
16180 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16181 new scratch pseudo.
16183 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16186 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16187 to ubsan_encode_value.
16189 2017-04-13 Jeff Law <law@redhat.com>
16191 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16192 appearing in DEBUG_INSNs.
16194 2017-04-13 Martin Liska <mliska@suse.cz>
16196 PR gcov-profile/80413
16197 * gcov-io.c (gcov_write_string): Copy to buffer just when
16198 allocated size is greater than zero.
16200 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16203 * dwarf2out.c (decls_for_scope): Ignore declarations of
16204 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16206 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16209 * ipa-visibility.c (non_local_p): Fix typos.
16210 (localize_node): When localizing symbol in same comdat group,
16211 dissolve the group only when we know external symbols are going
16213 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16215 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16217 PR tree-optimization/79390
16218 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16219 order does not result in usable sequence, retry with reversed operand
16225 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16226 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16227 op0 instead of fold_convert_loc (loc, type, arg0).
16229 2017-04-12 Jeff Law <law@redhat.com>
16231 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16232 has a delay slot in the generated code.
16234 * config/cris/cris.md (cris_preferred_reload_class): Return
16235 GENNONACR_REGS rather than GENERAL_REGS.
16237 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16240 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16241 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16242 signedness of the result type.
16244 2017-04-12 Richard Biener <rguenther@suse.de>
16245 Jeff Law <law@redhat.com>
16247 PR tree-optimization/80359
16248 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16249 trim stores to TARGET_MEM_REFs.
16251 2017-04-12 Richard Biener <rguenther@suse.de>
16253 PR tree-optimization/79390
16254 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16255 threading case even more.
16257 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16260 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16261 for quad_address_p for TImode, instead of just not indexed_address.
16263 2017-04-12 Richard Biener <rguenther@suse.de>
16264 Bernd Edlinger <bernd.edlinger@hotmail.de>
16266 PR middle-end/79671
16267 * alias.c (component_uses_parent_alias_set_from): Handle
16268 TYPE_TYPELESS_STORAGE.
16269 (get_alias_set): Likewise.
16270 * tree-core.h (tree_type_common): Add typeless_storage flag.
16271 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16272 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16273 for types containing members with TYPE_TYPELESS_STORAGE.
16274 (place_field): Likewise.
16275 (layout_type): Likewise for ARRAY_TYPE.
16276 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16277 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16278 TYPE_TYPELESS_STORAGE.
16279 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16281 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16284 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16285 first argument to type.
16287 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16291 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16292 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16293 (rs6000_expand_binop_builtin): Likewise.
16294 (rs6000_expand_ternop_builtin): Likewise; also add missing
16295 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16296 vshasigma built-ins.
16297 * doc/extend.texi: Document that vec_xxpermdi's third argument
16298 must be a constant.
16300 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16302 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16303 Use shift_const cost parameter when calculating gain of STV shifts.
16305 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16307 PR rtl-optimization/70478
16308 * lra-constraints.c (process_alt_operands): Check memory for
16309 disfavoring memory insn operand.
16311 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16313 PR middle-end/80100
16314 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16315 left shift in unsigned HOST_WIDE_INT type.
16317 PR rtl-optimization/80385
16318 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16319 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16322 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16323 if they have any depend clauses.
16325 2017-04-11 Martin Liska <mliska@suse.cz>
16328 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16329 * ipa-split.c (split_function): Create a local comdat symbol
16330 if caller is in a comdat group.
16332 2017-04-11 Martin Liska <mliska@suse.cz>
16335 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16338 2017-04-11 Martin Sebor <msebor@redhat.com>
16340 PR middle-end/80364
16341 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16342 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16344 (directive::set_width, directive::set_precision, format_character):
16346 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16349 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16352 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16353 conflict, set target->arch_name instead of target->cpu_name.
16355 2017-04-11 Richard Biener <rguenther@suse.de>
16357 PR tree-optimization/80374
16358 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16359 build_zero_cst, remove fold_convertible_p check again.
16361 2017-04-11 Martin Liska <mliska@suse.cz>
16364 * ubsan.c (instrument_object_size): Do not instrument register
16367 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16370 * config/i386/i386-builtin-types.def
16371 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16372 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16373 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16374 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16375 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16376 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16377 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16378 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16379 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16380 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16381 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16382 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16383 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16384 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16385 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16386 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16387 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16388 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16389 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16390 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16391 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16392 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16393 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16394 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16395 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16396 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16397 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16398 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16399 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16400 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16401 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16402 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16403 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16404 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16405 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16406 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16407 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16408 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16409 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16410 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16411 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16412 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16413 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16414 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16415 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16416 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16418 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16419 flag to second_arg_count, handle 4 argument function type _COUNT
16420 aliases, handle second_arg_count on second argument rather than last.
16422 2017-04-10 Jeff Law <law@redhat.com>
16424 PR tree-optimization/80374
16425 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16426 record anything if we can not convert integer_zero_node to the
16429 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16432 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16433 Enhance special handling given to the TARGET_P9_MINMAX option in
16434 relation to certain other options.
16436 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16438 PR tree-optimization/80153
16439 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16440 remove POINTER_PLUS_EXPR's base part directly, rather than through
16443 2017-04-10 Richard Biener <rguenther@suse.de>
16444 Bin Cheng <bin.cheng@arm.com>
16446 PR tree-optimization/80153
16447 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16448 the first element of pointer type aff_tree. Build result expr in
16450 (add_elt_to_tree): Convert to type unconditionally. Remove other
16451 fold_convert calls.
16452 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16453 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16455 2017-04-10 Richard Biener <rguenther@suse.de>
16457 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16460 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16462 PR rtl-optimization/70478
16463 * lra-constraints.c (curr_small_class_check): New.
16464 (update_and_check_small_class_inputs): New.
16465 (process_alt_operands): Update curr_small_class_check. Disfavor
16466 alternative insn memory operands. Check available regs for small
16469 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16472 * config/mips/mips.opt (-mvirt): Update description.
16473 * doc/invoke.texi (-mvirt): Likewise.
16475 2017-04-10 Richard Biener <rguenther@suse.de>
16477 PR middle-end/80362
16478 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16479 looking for NEGATE_EXPR in -A / -B to A / B folding.
16481 2017-04-10 Martin Liska <mliska@suse.cz>
16483 PR gcov-profile/80224
16484 * gcov.c (print_usage): Fix usage string.
16485 (get_gcov_intermediate_filename): Remove.
16486 (output_gcov_file): Use both for normal and intermediate format.
16487 (generate_results): Do not initialize special file for
16488 intermediate format.
16490 2017-04-10 Richard Biener <rguenther@suse.de>
16492 PR tree-optimization/80304
16493 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16496 2017-04-10 Nathan Sidwell <nathan@acm.org>
16499 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16500 (rs6000_init_builtins): Use it.
16502 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16504 * config/arm/arm.md (<mrc>): Add mode to SET source.
16505 (<mrrc>): Likewise.
16507 2017-04-10 Richard Biener <rguenther@suse.de>
16509 PR middle-end/80344
16510 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16512 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16515 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16516 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16517 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16518 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16519 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16520 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16521 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16522 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16523 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16524 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16525 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16526 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16527 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16528 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16529 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16530 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16531 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16532 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16533 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16534 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16535 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16536 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16537 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16539 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16541 PR rtl-optimization/70478
16542 * lra-constraints.c: Reverse the last patch.
16544 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16546 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16547 Add comment for WCHAR_T.
16549 2017-04-08 Martin Liska <mliska@suse.cz>
16552 2017-04-07 Martin Liska <mliska@suse.cz>
16555 * ipa-split.c (split_function): Add function part to a same comdat
16558 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16561 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16563 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16565 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16566 (vec_pairing): ...to this.
16567 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16568 (rs6000_sched_init): Adjust for name change.
16569 (struct rs6000_sched_context): Likewise.
16570 (rs6000_init_sched_context): Likewise.
16571 (rs6000_set_sched_context): Likewise.
16573 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16579 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16581 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16582 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16583 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16585 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16587 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16589 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16591 PR rtl-optimization/70703
16592 * ira-color.c (update_conflict_hard_regno_costs): Use
16593 int64_t instead of HOST_WIDE_INT.
16595 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16597 PR rtl-optimization/70478
16598 * lra-constraints.c (process_alt_operands): Disfavor alternative
16599 insn memory operands.
16601 2017-04-07 Jeff Law <law@redhat.com>
16603 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16604 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16606 2017-04-07 Martin Liska <mliska@suse.cz>
16609 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16610 Show error message instead of an ICE.
16612 2017-04-07 Martin Liska <mliska@suse.cz>
16615 * ipa-split.c (split_function): Add function part to a same comdat
16618 2017-04-07 Richard Biener <rguenther@suse.de>
16620 PR middle-end/80341
16621 * tree.c (get_unwidened): Also handle ! for_type case for
16623 * convert.c (do_narrow): Split out from ...
16624 (convert_to_integer_1): ... here. Do not pass final truncation
16625 type to get_unwidened for TRUNC_DIV_EXPR.
16627 2017-04-07 Richard Biener <rguenther@suse.de>
16629 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16631 (aff_combination_const): Adjust.
16632 (aff_combination_scale): Likewise.
16633 (aff_combination_add_elt): Likewise.
16634 (aff_combination_add_cst): Likewise.
16635 (aff_combination_convert): Likewise.
16636 (add_elt_to_tree): Likewise. Remove unused argument.
16637 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16639 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16641 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16643 * config/arm/arm.c (arm_default_short_enums): Use
16644 ARM_DEFAULT_SHORT_ENUMS.
16645 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16647 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16650 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16651 members with redundant out-of-class redeclaration.
16653 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16656 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16657 * config/i386/i386.md (*zero_extendsidi2):
16658 Add (?*x,*x) and (?*v,*v) alternatives.
16660 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16663 * config/i386/i386.c (ix86_expand_builtin)
16664 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16665 mode from insn data. Convert operands to insn operand mode.
16666 Copy operands that don't satisfy insn predicate to a register.
16668 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16670 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16673 2017-04-06 Richard Biener <rguenther@suse.de>
16675 PR tree-optimization/80334
16676 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16677 preserve alignment of accesses.
16679 2017-04-06 Richard Biener <rguenther@suse.de>
16681 PR tree-optimization/80262
16682 * tree-sra.c (build_ref_for_offset): Preserve address-space
16684 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16685 Drop useless address-space information on MEM_REF offsets.
16687 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16689 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16691 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16693 PR rtl-optimization/70703
16694 * ira-color.c (update_conflict_hard_regno_costs): Use
16695 HOST_WIDE_INT instead of long.
16697 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16700 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16701 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16703 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16704 for x86_64 target. Handle -m3dnowa option.
16706 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16708 PR rtl-optimization/70703
16709 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16710 (update_conflict_hard_regno_costs): Use long instead of unsigned
16711 arithmetic for cost calculation.
16713 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16714 Bernd Edlinger <bernd.edlinger@hotmail.de>
16717 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16720 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16723 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16724 ptr_mode with Pmode throughout.
16725 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16726 into probe_stack_range and use DImode.
16728 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16731 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16732 call_eh_return is true.
16734 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16736 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16737 Initialize last_match_fntype_index.
16739 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16742 * tree-nvr.c: Include internal-fn.h.
16743 (pass_return_slot::execute): Ignore internal calls without
16746 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16747 Richard Biener <rguenther@suse.de>
16750 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16751 captures used multiple times, except for the last use.
16752 * generic-match-head.c: Include gimplify.h.
16754 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16756 PR tree-optimization/79390
16757 * target.h (struct noce_if_info): Declare.
16758 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16759 * target.def (noce_conversion_profitable_p): New target hook.
16760 * ifcvt.h (struct noce_if_info): New type, moved from ...
16761 * ifcvt.c (struct noce_if_info): ... here.
16762 (noce_conversion_profitable_p): Renamed to ...
16763 (default_noce_conversion_profitable_p): ... this. No longer
16765 (noce_try_store_flag_constants, noce_try_addcc,
16766 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16767 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16768 instead of noce_conversion_profitable_p.
16769 * config/i386/i386.c: Include ifcvt.h.
16770 (ix86_option_override_internal): Don't override
16771 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16772 (ix86_noce_conversion_profitable_p): New function.
16773 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16774 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16775 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16776 * doc/tm.texi: Regenerated.
16778 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16780 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16783 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16786 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16787 instructions for small multiply cores.
16789 2017-04-04 Jeff Law <law@redhat.com>
16791 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16793 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16794 that are not set by the loop over the elements.
16796 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16799 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16800 int mode, convert_modes it to mode as unsigned, otherwise use
16801 lowpart_subreg to mode rather than SImode.
16802 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16803 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16804 Use DImode instead of SImode for the shift count operand.
16805 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16808 2017-04-04 Richard Biener <rguenther@suse.de>
16810 PR middle-end/80281
16811 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16812 arithmetic done for the negate or the plus. Simplify.
16813 (A - (-B) -> A + B): Likewise.
16814 * fold-const.c (split_tree): Make sure to not negate pointers.
16816 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16818 PR rtl-optimization/60818
16819 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16820 a compare of comparisons with the thing compared if this results
16821 in a different machine mode.
16823 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16825 * alias.c (base_alias_check): Fix typo in comment.
16826 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16827 * cgraphunit.c (symbol_table::compile): Likewise.
16828 * collect2.c (maybe_run_lto_and_relink): Likewise.
16829 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16830 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16831 * config/avr/avr.c (avr_map_op_t): Likewise.
16832 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16833 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16834 * config/epiphany/epiphany.md (movcc): Likewise.
16835 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16836 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16838 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16839 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16840 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16841 * config/sh/sh.c (sh_rtx_costs): Likewise.
16842 * fold-const.c (fold_truth_andor): Likewise.
16843 * genautomata.c (collapse_flag): Likewise.
16844 * gengtype.h (struct type::u::s): Likewise.
16845 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16846 * input.c (FORMAT_AMOUNT): Likewise.
16847 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16848 (known_aggs_to_agg_replacement_list): Likewise.
16849 * ipa-inline-analysis.c: Likewise.
16850 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16851 * ipa-polymorphic-call.c
16852 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16853 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16854 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16856 * modulo-sched.c (apply_reg_moves): Likewise.
16857 * omp-expand.c (build_omp_regions_1): Likewise.
16858 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16859 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16860 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16861 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16862 * value-prof.c: Likewise.
16863 * var-tracking.c (val_reset): Likewise.
16865 2017-04-03 Richard Biener <rguenther@suse.de>
16867 PR tree-optimization/80275
16868 * fold-const.c (split_address_to_core_and_offset): Handle
16871 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16873 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16874 descriptors is at least equal to that of functions.
16876 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16878 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16880 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16883 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16884 (mov<IMOD4:mode>): New expander.
16885 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16887 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16889 PR rtl-optimization/79405
16890 * fwprop.c (propagations_left): New variable.
16891 (forward_propagate_into): Decrement it.
16892 (fwprop_init): Initialize it.
16893 (fw_prop): If the variable has reached zero, stop propagating.
16894 (fwprop_addr): Ditto.
16896 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16899 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16900 a FUNCTION_DECL, pass it as decl instead of origin to
16903 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16905 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16908 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16911 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16912 TARGET_VSX_SMALL_INTEGER.
16914 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16916 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16917 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16919 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16921 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16922 extraction from odd-numbered MSA register.
16924 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16926 PR middle-end/80173
16927 * expmed.c (store_bit_field_1): Don't attempt to create
16928 a word subreg out of hard registers wider than word if they
16929 have HARD_REGNO_NREGS of 1 for their mode.
16931 PR middle-end/80163
16932 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16933 conversions to integer types wider than word and pointer.
16936 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16937 (rtx_equal_for_cselib_p): Pass 0 to it.
16938 * cselib.c (cselib_hasher::equal): Likewise.
16939 (rtx_equal_for_cselib_1): Add depth argument. If depth
16940 is 128, don't look up VALUE locs and punt. Increment
16941 depth in recursive calls when walking VALUE locs.
16943 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16945 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16946 (make_gcov_file_name): Use the canonical path name for generating
16948 (read_line): Fix handling of files with ascii null bytes.
16950 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16952 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16953 to initialise a vector register instead
16954 of using a const_int.
16956 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16958 PR translation/80189
16959 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16960 diagnostic messages.
16962 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16965 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16966 (dfp_diex_<mode>): Update mode of operand 1.
16967 * doc/extend.texi (dxex, dxexq): Document change to return type.
16968 (diex, diexq): Document change to argument type.
16970 2017-03-30 Martin Jambor <mjambor@suse.cz>
16973 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16974 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16975 it reflects the signature changes performed at the callee side.
16976 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16977 to cgraph_build_function_type_skip_args.
16978 (build_function_decl_skip_args): Adjust call to the above function.
16980 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16983 * config/i386/sse.md
16984 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16985 register as dest whenever it is a MEM not rtx_equal_p to the
16986 corresponding dup operand, and when forcing into reg move the
16987 reg into the memory afterwards.
16988 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16989 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16990 for the force_reg mode.
16991 (avx512vl_vextractf128<mode>): Use register as dest either
16992 always when a MEM, or when it is a MEM not rtx_equal_p to the
16993 corresponding dup operand, or even not when it is a CONST_VECTOR
16994 depending on the mode and lo vs. hi.
16995 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16997 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16998 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16999 Likewise. Require that operands[2] is even.
17000 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
17001 Remove extraneous parens. Require that operands[2] is a multiple
17003 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
17004 operands[0] is a MEM if <mask_applied>, the predicates/constraints
17005 disallow memory then.
17007 2017-03-30 Richard Biener <rguenther@suse.de>
17009 PR tree-optimization/77498
17010 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
17011 to non-constants over backedges.
17013 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
17015 PR rtl-optimization/80233
17016 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
17017 as last_combined_insn. Do not test for BARRIER_P separately.
17019 2017-03-29 Andreas Schwab <schwab@suse.de>
17022 * calls.c (prepare_call_address): Convert funexp to Pmode before
17023 copying to temp reg.
17025 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17027 PR tree-optimization/80158
17028 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
17029 Handle possible future case of more than one alternate
17031 (replace_rhs_if_not_dup): Likewise.
17032 (replace_one_candidate): Likewise.
17034 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
17036 PR rtl-optimization/80193
17037 * ira.c (ira): Do not check allocation for LRA.
17039 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
17041 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
17042 (nvptx_output_simt_exit): Declare.
17043 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
17044 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
17045 (init_softstack_frame): Move initialization of crtl->is_leaf to...
17046 (nvptx_declare_function_name): ...here. Emit declaration of local
17047 memory space buffer for omp_simt_enter insn.
17048 (nvptx_output_unisimt_switch): New.
17049 (nvptx_output_softstack_switch): New.
17050 (nvptx_output_simt_enter): New.
17051 (nvptx_output_simt_exit): New.
17052 * config/nvptx/nvptx.h (struct machine_function): New fields
17053 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
17054 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
17055 (UNSPECV_SIMT_EXIT): Ditto.
17056 (omp_simt_enter_insn): New insn.
17057 (omp_simt_enter): New expansion.
17058 (omp_simt_exit): New insn.
17059 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
17061 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
17062 (expand_GOMP_SIMT_ENTER_ALLOC): New.
17063 (expand_GOMP_SIMT_EXIT): New.
17064 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17065 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17066 (GOMP_SIMT_EXIT): Ditto.
17067 * target-insns.def (omp_simt_enter): New insn.
17068 (omp_simt_exit): Ditto.
17069 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17071 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17072 (lower_rec_input_clauses): Likewise.
17073 (lower_lastprivate_clauses): Handle SIMT privatization.
17075 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17076 (ompdevlow_adjust_simt_enter): New.
17077 (find_simtpriv_var_op): New.
17078 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17079 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17081 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17082 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17083 (copy_decl_for_dup_finish): Ditto.
17085 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17087 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17090 * config/i386/i386.c (ix86_option_override_internal): Always
17091 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17093 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17095 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17097 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17099 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17100 mark new edge's irreducible flag accordign to it.
17101 (vect_do_peeling): Check loop preheader edge's irreducible flag
17102 and pass it to function slpeel_add_loop_guard.
17104 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17106 PR tree-optimization/80218
17107 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17108 Update block frequencies and counts.
17110 2017-03-28 Richard Biener <rguenther@suse.de>
17112 PR tree-optimization/78644
17113 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17114 of a simplification result we may not use it at all.
17116 2017-03-28 Richard Biener <rguenther@suse.de>
17119 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17120 without arguments, generate default definition of a SSA name.
17122 2017-03-28 Richard Biener <rguenther@suse.de>
17124 PR middle-end/80222
17125 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17126 TYPE_REF_CAN_ALIAS_ALL references.
17127 * fold-const.c (fold_indirect_ref_1): Likewise.
17129 2017-03-28 Martin Liska <mliska@suse.cz>
17132 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17133 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17135 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17136 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17138 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17139 (EXTRA_SPECS): Define.
17140 (SUBTARGET_EXTRA_SPECS): Likewise.
17141 (SUBTARGET_CPP_SPEC): Likewise.
17142 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17143 SUBTARGET_EXTRA_SPECS.
17144 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17146 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17148 * config/arc/simdext.md (vst64_insn): Update pattern.
17149 (vld32wh_insn): Likewise.
17150 (vld32wl_insn): Likewise.
17151 (vld64_insn): Likewise.
17152 (vld32_insn): Likewise.
17154 2017-03-28 Marek Polacek <polacek@redhat.com>
17157 * fold-const.c (fold_comparison): Use protected_set_expr_location
17158 instead of SET_EXPR_LOCATION.
17160 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17162 * tree.c (add_expr): Avoid name lookup warning.
17164 2017-03-27 Jeff Law <law@redhat.com>
17166 PR tree-optimization/80216
17167 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17168 function name. Limit recursion depth.
17169 (record_temporary_equivalences): Corresponding changes.
17171 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17173 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17176 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17179 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17181 * cfgcleanup.c (reg_note_cfa_p): New array.
17182 (insns_have_identical_cfa_notes): New function.
17183 (old_insns_match_p): Don't cross-jump in between /f
17184 and non-/f instructions. If both i1 and i2 are frame related,
17185 verify all CFA notes, their order and content.
17187 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17190 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17191 HImode and SImode with zero extend to DImode to one insn.
17192 (bswap<mode>2_extenddi): Likewise.
17193 (bswapsi2_extenddi): Likewise.
17194 (bswaphi2_extendsi): Likewise.
17195 (bswaphi2): Combine bswap HImode and SImode into one insn.
17196 Separate memory insns from swapping register.
17197 (bswapsi2): Likewise.
17198 (bswap<mode>2): Likewise.
17199 (bswaphi2_internal): Delete, no longer used.
17200 (bswapsi2_internal): Likewise.
17201 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17202 store, and gpr<-gpr swap insns.
17203 (bswap<mode>2_store): Likewise.
17204 (bswaphi2_reg): Register only splitter, combine with the splitter.
17205 (bswaphi2 splitter): Likewise.
17206 (bswapsi2_reg): Likewise.
17207 (bswapsi2 splitter): Likewise.
17208 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17209 the insns into load, store, and register/register insns.
17210 (bswapdi2_ldbrx): Likewise.
17211 (bswapdi2_load): Likewise.
17212 (bswapdi2_store): Likewise.
17213 (bswapdi2_reg): Likewise.
17215 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17217 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17218 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17220 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17223 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17225 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17226 special handling for target option conflicts between dform
17227 options (-mpower9-dform, -mpower9-dform-vector,
17228 -mpower9-dform-scalar) and -mno-direct-move.
17230 2017-03-27 Richard Biener <rguenther@suse.de>
17232 PR tree-optimization/80181
17233 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17235 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17237 * config/arc/predicates.md (move_double_src_operand): Replace the
17238 call to move_double_src_operand with a call to address_operand.
17240 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17242 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17243 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17244 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17246 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17248 * config/arc/predicates.md (long_immediate_loadstore_operand):
17249 Consider scaled addresses cases.
17251 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17253 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17254 restored when in interrupt.
17255 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17256 doesn't have delay slot.
17258 2017-03-27 Richard Biener <rguenther@suse.de>
17261 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17262 inlined thunk clones.
17264 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17267 * asan.c (instrument_derefs): Copy over last operand from
17268 original COMPONENT_REF to the new COMPONENT_REF with
17269 DECL_BIT_FIELD_REPRESENTATIVE.
17270 * ubsan.c (instrument_object_size): Likewise.
17272 2017-03-27 Richard Biener <rguenther@suse.de>
17274 PR tree-optimization/80170
17275 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17276 sure DR/SCEV didnt fold in constants we do not see when looking
17277 at the reference base alignment.
17279 2017-03-27 Richard Biener <rguenther@suse.de>
17281 PR middle-end/80171
17282 * gimple-fold.c (fold_ctor_reference): Properly guard against
17283 NULL return value from canonicalize_constructor_val.
17285 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17288 * config/i386/i386.c (ix86_expand_builtin)
17289 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17290 flags reg setting and flags reg using instructions.
17291 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17292 clobbering instructions to zero extend op2.
17294 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17296 * doc/install.texi (Configuration) <--with-aix-soname>:
17297 Update link to AIX ld.
17299 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17301 PR rtl-optimization/80160
17302 PR rtl-optimization/80159
17303 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17304 reg_alternate_class into account.
17306 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17309 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17310 to consider in curr_insn_transform.
17312 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17314 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17315 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17316 and emit_mode_inner.
17318 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17320 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17321 argument to the overloaded builtin variants. Use the new flag to
17322 deprecate certain builtin variants.
17323 * config/s390/s390-builtin-types.def: Add new builtin types.
17324 * config/s390/s390-builtins.h: Support new flags field for
17325 overloaded builtins.
17326 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17327 (s390_macro_to_expand): Enable vector float data type.
17328 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17329 builtins by incrementing the __VEC__ version number.
17330 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17332 (s390_resolve_overloaded_builtin): Emit error messages depending
17333 on the builtin flags.
17334 * config/s390/s390.c (s390_expand_builtin): Support additional
17335 flags argument. Change error message to match the messages
17336 emitted in s390-c.c.
17337 * config/s390/s390.md: New UNSPEC_* constants.
17338 (op_type): Add new instruction types.
17339 * config/s390/vecintrin.h: Add new builtins and test data class
17341 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17342 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17343 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17344 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17345 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17346 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17348 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17349 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17350 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17351 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17353 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17354 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17355 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17356 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17358 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17359 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17360 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17361 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17362 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17363 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17364 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17366 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17367 ("vec_scatter_element<V_HW_4:mode>_DI")
17368 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17369 ("vec_fpint<mode>", "vflls")
17370 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17371 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17372 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17373 ("*vec_cmphe<mode>_cc"): ... these.
17375 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17376 mode constant instead of magic value.
17378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17380 * config/s390/s390.c (s390_expand_vec_compare): Support other
17381 vector floating point modes than just V2DF.
17382 (s390_expand_vcond): Likewise.
17383 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17384 (s390_cannot_change_mode_class): Prevent mode changes between TF
17385 and V1TF in vector registers.
17386 * config/s390/s390.md (DF, SF): New mode attributes.
17387 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17388 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17389 SFmode support for VRs.
17390 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17392 (VFT, VF_HW): New mode iterators.
17393 (vw, sdx): New mode attributes.
17394 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17395 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17396 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17397 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17398 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17399 also the new vector floating point modes. Renaming to ...
17401 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17402 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17403 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17404 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17405 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17406 ("vec_unordered<mode>"): ... these.
17408 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17409 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17410 ("*vec_extendv2df"): New insn definitions.
17412 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17414 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17415 ("mulditi3_2", "*muldi3_sign"): New patterns.
17416 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17417 rename the pattern definition.
17419 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17421 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17423 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17427 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17428 instruction if possible.
17429 * config/s390/vector.md (vec_halfnumelts): New mode
17431 ("*vec_vllezlf<mode>"): New pattern.
17433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17435 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17436 ("popcountv4si2", "popcountv2di2"): Rename to ...
17437 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17438 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17440 ("popcount<mode>2_vxe"): New pattern.
17442 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17444 * common/config/s390/s390-common.c (processor_flags_table): Add
17446 * config.gcc: Add arch12.
17447 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17448 Default to arch12 for unknown CPU model numbers.
17449 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17450 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17451 PROCESSOR_max sanity check.
17452 * config/s390/s390-opts.h (enum processor_type): Add
17454 * config/s390/s390.c (processor_table): Add arch12.
17455 (s390_expand_builtin): Add check for B_VXE flag.
17456 (s390_issue_rate): Add PROCESSOR_ARCH12.
17457 (s390_get_sched_attrmask): Likewise.
17458 (s390_get_unit_mask): Likewise.
17459 (s390_sched_score): Enable z13 scheduling for arch12.
17460 (s390_sched_reorder): Likewise.
17461 (s390_sched_variable_issue): Likewise.
17462 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17464 (s390_tune_attr): Use z13 scheduling also for arch12.
17465 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17466 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17467 (TARGET_VXE_P): New macros.
17468 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17469 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17470 * config/s390/s390.opt: Add arch12 as processor_type.
17472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17474 * config/s390/s390.md
17475 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17476 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17477 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17479 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17480 Rename expanders to ...
17482 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17483 ("fixuns_truncdddi2_emu"): ... these.
17485 ("fixuns_trunc<mode>si2_emu"): New expander.
17487 ("*fixuns_truncdfdi2_z13"): Rename to ...
17488 ("*fixuns_truncdfdi2_vx"): ... this.
17490 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17492 * config/s390/2964.md: Remove the single element vector compare
17493 instructions which are no longer used.
17494 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17496 (s390_canonicalize_comparison): Remove handling of DFmode
17498 (s390_expand_vec_compare_scalar): Remove function.
17499 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17500 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17502 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17506 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17508 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17509 will anyway by matched by mov<mode>_64dfp.
17511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17513 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17514 vlef/vstef. Add missing operand to vleif.
17516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17518 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17519 pair for all vector types with 64 bit elements.
17520 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17521 * config/s390/vector.md (V_HW_64): ... here.
17522 (V_128_NOSINGLE): New mode iterator.
17523 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17524 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17525 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17526 ("*vec_load_pairv2di"): Change to ...
17527 ("*vec_load_pair<mode>"): ... this one.
17529 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17531 * config/s390/constraints.md: Add comments.
17532 (jKK): Reject element sizes > 8 bytes.
17533 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17535 * config/s390/s390.md: Add the s_operand checks formerly in
17536 s390_split_ok_p to various splitters where they are still
17538 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17539 for 128 bit vectors. Plus two splitters.
17541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17543 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17549 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17550 error if the boundary argument is not constant.
17552 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17554 PR rtl-optimization/80112
17555 * loop-doloop.c (doloop_condition_get): Don't check condition
17556 if cmp isn't SET with IF_THEN_ELSE src.
17558 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17560 PR tree-optimization/80158
17561 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17562 replacing a candidate statement, also replace it for the
17563 candidate's alternate interpretation.
17564 (replace_rhs_if_not_dup): Likewise.
17565 (replace_one_candidate): Likewise.
17567 2017-03-24 Richard Biener <rguenther@suse.de>
17569 PR tree-optimization/80167
17570 * graphite-isl-ast-to-gimple.c
17571 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17573 (translate_isl_ast_to_gimple::get_rename): Likewise.
17575 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17578 handling of certain combinations of target options, including the
17579 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17580 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17582 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17585 * config/arm/arm.md (*load_multiple): Add reload_completed to
17586 matching condition.
17588 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17589 Richard Biener <rguenth@suse.de>
17591 PR tree-optimization/79908
17592 PR tree-optimization/80136
17593 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17594 been cast away, gimplify_and_add suffices.
17596 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17598 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17600 2017-03-23 Richard Biener <rguenther@suse.de>
17602 PR tree-optimization/80032
17603 * gimplify.c (gimple_push_cleanup): Forced unconditional
17604 cleanups still have to go to the conditional_cleanups
17607 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17609 PR tree-optimization/80072
17610 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17612 (next_operand_entry_id): Change type to unsigned int.
17613 (sort_by_operand_rank): Make sure to return the right return value
17614 even if unsigned fields are bigger than INT_MAX.
17615 (struct oecount): Change cnt and id type to unsigned int.
17616 (oecount_hasher::equal): Formatting fix.
17617 (oecount_cmp): Make sure to return the right return value
17618 even if unsigned fields are bigger than INT_MAX.
17619 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17622 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17623 TREE_READONLY on result if writing it more than once.
17626 * doc/invoke.texi (-fsanitize=thread): Document that with
17627 -fnon-call-exceptions atomics are not able to throw
17631 * tsan.c: Include tree-eh.h.
17632 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17633 maybe_clean_or_replace_eh_stmt where needed.
17634 (instrument_memory_accesses): Add cfg_changed argument.
17635 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17636 if it returned true.
17637 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17639 PR rtl-optimization/63191
17640 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17641 wrapper function, moved the whole old content into ...
17642 (ix86_delegitimize_address_1): ... this. New inline function.
17643 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17644 true as last argument instead of ix86_delegitimize_address.
17646 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17648 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17649 cortexa57_branch_cost.
17651 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17653 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17655 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17658 * doc/md.texi (Constraints): Document wA constraint.
17659 * config/rs6000/constraints.md (wA): New.
17660 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17661 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17662 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17663 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17665 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17668 * gimplify.c (is_oacc_declared): New function.
17669 (oacc_default_clause): Use it to set default flags for acc declared
17670 variables inside parallel regions.
17671 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17672 declared variables.
17673 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17674 declare attribute to any decl as necessary.
17676 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17679 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17680 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17681 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17682 (arm_arch_lpae): This.
17683 * config/arm/arm.c (arm_arch7ve): Rename into ...
17684 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17685 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17688 2017-03-22 Martin Liska <mliska@suse.cz>
17691 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17692 error message instead of an ICE.
17694 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17696 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17698 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17700 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17705 2017-03-21 Martin Sebor <msebor@redhat.com>
17707 * doc/extend.texi: Use "cannot" instead of "can't."
17708 * doc/hostconfig.texi: Same.
17709 * doc/install.texi: Same.
17710 * doc/invoke.texi: Same.
17711 * doc/loop.texi: Same.
17712 * doc/md.texi: Same.
17713 * doc/objc.texi: Same.
17714 * doc/rtl.texi: Same.
17715 * doc/tm.texi: Same.
17716 * doc/tm.texi.in: Same.
17717 * doc/trouble.texi: Same.
17719 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17722 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17723 (collect_checksum_attributes): Set it.
17724 (die_checksum_ordered): Use it.
17726 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17728 PR tree-optimization/79908
17729 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17730 change: For a VA_ARG whose LHS has been cast away, use
17731 force_gimple_operand to construct the side effects.
17733 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17735 PR translation/80001
17736 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17737 more amenable to translation.
17738 (oacc_loop_auto_partitions): Likewise.
17740 2017-03-21 Marek Polacek <polacek@redhat.com>
17741 Martin Sebor <msebor@redhat.com>
17743 PR tree-optimization/80109
17744 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17745 on INTEGRAL_TYPE_P.
17747 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17748 Segher Boessenkool <segher@kernel.crashing.org>
17751 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17752 check reg_used_between_p between insn and one of succ or succ2
17753 depending on if succ is artificial insn not inserted into insn
17756 2017-03-21 Martin Liska <mliska@suse.cz>
17758 PR gcov-profile/80081
17759 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17760 * doc/gcc.texi: Include gcov-dump stuff.
17761 * doc/gcov-dump.texi: New file.
17763 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17765 PR rtl-optimization/79150
17766 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17767 conditional jump, if the jump is the last insn of the loop.
17769 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17770 Richard Biener <rguenth@suse.de>
17772 PR tree-optimization/79908
17773 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17774 been cast away, use force_gimple_operand to construct the side
17777 2017-03-21 Martin Liska <mliska@suse.cz>
17779 PR libfortran/79956
17780 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17783 2017-03-21 Brad Spengler <spender@grsecurity.net>
17786 * plugin.c (htab_hash_plugin): New function.
17787 (add_new_plugin): Use it and adjust.
17788 (parse_plugin_arg_opt): Adjust.
17789 (init_one_plugin): Likewise.
17791 2017-03-21 Richard Biener <rguenther@suse.de>
17793 PR tree-optimization/80032
17794 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17795 if set force the cleanup to happen unconditionally.
17796 (gimplify_target_expr): Push inserted clobbers with force_uncond
17797 to avoid them being removed by control-dependent DCE.
17799 2017-03-21 Richard Biener <rguenther@suse.de>
17801 PR tree-optimization/80122
17802 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17803 va_arg_pack_len when the inlined call stmt requires pack
17805 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17807 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17810 * tsan.c (instrument_builtin_call): If the memory model argument
17811 is not a constant, assume it is valid.
17814 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17817 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17819 PR rtl-optimization/79910
17820 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17821 if its dest is used by an insn before I2 (other than the combined
17822 insns themselves, which are properly handled already).
17824 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17827 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17829 * combine.c (record_used_regs): New static function.
17830 (try_combine): Handle situations where there is an additional
17831 instruction between I2 and I3 which needs to have a LOG_LINK
17835 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17837 * combine.c (try_combine): Delete redundant i1 test. Call
17838 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17840 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17843 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17844 alternatives 13/14.
17846 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17848 PR tree-optimization/80054
17849 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17850 the optimization if a PHI or any of its arguments is not dominated
17851 by the candidate's basis. Use gphi* rather than gimple* as
17853 (replace_profitable_candidates): Clean up a gimple* variable that
17854 should be a gphi* variable.
17856 2017-03-20 Martin Sebor <msebor@redhat.com>
17859 * doc/extend.texi (attribute constructor): Document present limitation.
17861 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17864 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17865 __POWER9_VECTOR__ #ifdef control, change template definition to
17866 use Power9-specific built-in function.
17867 (vec_any_eq): Likewise.
17868 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17869 to control outcomes from this test.
17870 (vector_ae_<mode>p): For VEC_F modes, likewise.
17872 2017-03-20 Ian Lance Taylor <iant@google.com>
17874 * config/i386/i386.c (ix86_function_regparm): Save an extra
17875 register for -fsplit-stack with DECL_STATIC_CHAIN.
17877 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17880 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17881 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17883 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17885 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17887 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17890 2017-03-20 Marek Polacek <polacek@redhat.com>
17893 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17895 2017-03-20 Richard Biener <rguenther@suse.de>
17897 PR tree-optimization/80113
17898 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17899 allocate extra SSA name for PHI def.
17900 (add_close_phis_to_outer_loops): Likewise.
17901 (add_close_phis_to_merge_points): Likewise.
17902 (copy_loop_close_phi_args): Likewise.
17903 (copy_cond_phi_nodes): Likewise.
17905 2017-03-20 Martin Liska <mliska@suse.cz>
17907 PR middle-end/79753
17908 * tree-chkp.c (chkp_build_returned_bound): Do not build
17909 returned bounds for a LHS that's not a BOUNDED_P type.
17911 2017-03-20 Martin Liska <mliska@suse.cz>
17915 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17916 COMPLEX_CST and VECTOR_CST.
17918 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17921 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17922 target operand. A new splitter adds the clobber statement in case
17923 the target operand is dead anyway.
17925 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17927 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17928 to age-old versions of binutils and glibc.
17930 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17932 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17934 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17936 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17938 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17940 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17941 requirement for binutils 2.13.
17943 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17945 * combine.c (try_combine): Delete redundant i1 test. Call
17946 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17948 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17950 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17951 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17953 <riscv64-*-elf>: Re-arrange section
17954 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17955 <riscv32-*-linux>: Likewise.
17956 <riscv64-*-elf>: Likewise
17957 <riscv64-*-linux>: Likewise.
17959 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17962 * aarch64.opt(verbose-cost-dump): Fix typo.
17964 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17967 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17968 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17970 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17972 * reload.c (find_reloads): When reloading a nonoffsettable address,
17973 use RELOAD_OTHER for it and its address reloads.
17975 PR rtl-optimization/79910
17976 * combine.c (record_used_regs): New static function.
17977 (try_combine): Handle situations where there is an additional
17978 instruction between I2 and I3 which needs to have a LOG_LINK
17981 2017-03-17 Jeff Law <law@redhat.com>
17983 PR tree-optimization/71437
17984 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17985 conditional in the hash table first.
17986 (vrp_dom_walker::before_dom_children): Extract condition from
17987 ASSERT_EXPR. Record condition, its inverion and any implied
17988 conditions as well.
17990 2017-03-17 Marek Polacek <polacek@redhat.com>
17991 Markus Trippelsdorf <markus@trippelsdorf.de>
17993 PR tree-optimization/80079
17994 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17997 2017-03-17 Richard Biener <rguenther@suse.de>
17999 PR middle-end/80075
18000 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
18001 Properly verify the LHS before the RHS possibly claims to be
18003 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
18006 2017-03-17 Martin Jambor <mjambor@suse.cz>
18008 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
18009 (List of -O2 options): Likewise.
18010 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
18013 2017-03-17 Tom de Vries <tom@codesourcery.com>
18015 * gcov-dump.c (print_usage): Print bug_report_url.
18017 2017-03-17 Richard Biener <rguenther@suse.de>
18019 PR middle-end/80050
18020 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
18021 (parser::peek): Likewise.
18023 2017-03-17 Richard Biener <rguenther@suse.de>
18025 PR tree-optimization/80048
18026 * sese.c (free_sese_info): Properly release rename_map and
18027 copied_bb_map elements.
18029 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
18031 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
18032 Add linked-list forward and backlinks. Insert on
18033 construction, remove on destruction.
18034 (class pass_store_merging): Add m_stores_head field.
18035 (pass_store_merging::terminate_and_process_all_chains):
18036 Iterate over m_stores_head list.
18037 (pass_store_merging::terminate_all_aliasing_chains):
18039 (pass_store_merging::execute): Check for debug stmts first.
18040 Push new chains onto the m_stores_head stack.
18042 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
18045 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
18046 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
18047 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
18049 2017-03-16 Jeff Law <law@redhat.com>
18051 PR tree-optimization/71437
18052 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
18053 member function. Implementation moved into after_dom_children
18054 member function and into the threader's thread_outgoing_edges
18056 (dom_opt_dom_walker::after_dom_children): Simplify by moving
18057 some code into new thread_outgoing_edges.
18058 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
18059 definition. Simplify marker handling (do it here). Assume we always
18060 have the available expression and the const/copies tables.
18061 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
18063 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18064 * tree-vrp.c (equiv_stack): No longer file scoped.
18065 (vrp_dom_walker): New class.
18066 (vrp_dom_walker::before_dom_children): New member function.
18067 (vrp_dom_walker::after_dom_children): Likewise.
18068 (identify_jump_threads): Setup domwalker. Use it rather than
18069 walking edges in a random order by hand. Simplify setup/finalization.
18070 (finalize_jump_threads): Remove.
18071 (vrp_finalize): Do not call identify_jump_threads here.
18072 (execute_vrp): Do it here instead and call thread_through_all_blocks
18075 PR tree-optimization/71437
18076 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18078 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18080 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18081 (dom_opt_dom_walker::thread_across_edge): Remove
18082 handle_dominating_asserts argument. All callers changed.
18083 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18084 changes. Remove calls to lhs_of_dominating_assert. Other
18085 uses of handle_dominating_asserts turn into unconditional code
18086 (simplify_control_stmt_condition_1): Likewise.
18087 (simplify_control_stmt_condition): Likewise.
18088 (thread_through_normal_block, thread_across_edge): Likewise.
18089 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18090 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18091 object if it is not an SSA_NAME.
18092 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18093 before calling into the VRP specific simplifiers.
18094 (identify_jump_threads): Remove handle_dominating_asserts
18097 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18100 * tree-diagnostic.c (default_tree_printer): No longer static.
18101 * tree-diagnostic.h (default_tree_printer): New prototype.
18103 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18105 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18106 Change ins into fmov.
18108 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18110 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18111 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18112 Use h_con constraint for operand 1.
18113 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18114 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18116 2017-03-15 Jeff Law <law@redhat.com>
18118 PR tree-optimization/71437
18119 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18120 (record_temporary_equivalences): Use it.
18122 PR tree-optimization/71437
18123 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18124 tree-ssa-scopedtables.
18125 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18126 (record_conditions, record_cond, vuse_eq): Likewise.
18127 (record_edge_info): Adjust to API tweak of record_conditions.
18128 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18129 (record_temporary_equivalences, optimize_stmt): Likewise.
18130 (eliminate_redundant_computations): Likewise.
18131 (record_equivalences_from_stmt): Likewise.
18132 * tree-ssa-scopedtables.c: Include options.h and params.h.
18133 (vuse_eq): New function, moved from tree-ssa-dom.c
18134 (build_and_record_new_cond): Likewise.
18135 (record_conditions): Likewise. Accept vector of conditions rather
18136 than edge_equivalence structure for first argument.
18137 for the first argument.
18138 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18139 from tree-ssa-dom.c.
18140 (avail_exprs_stack::record_cond): Likewise.
18141 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18142 from tree-ssa-dom.c.
18143 (avail_exprs_stack): Add new member functions lookup_avail_expr
18145 (record_conditions): Declare.
18147 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18150 * lra-constraints.c (process_alt_operands): Increase reject for
18151 reloading an input/output operand.
18153 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18156 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18157 insns to convert from signed/unsigned char/short to IEEE 128-bit
18159 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18161 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18164 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18165 subreg of inner mode for values already in registers.
18167 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18169 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18170 iteration reg is used after the loop.
18172 2017-03-14 Martin Sebor <msebor@redhat.com>
18174 PR tree-optimization/79800
18175 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18176 precision in negative-positive range.
18177 (format_floating): Call non-const overload with adjusted precision.
18179 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18182 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18185 2017-03-14 Martin Sebor <msebor@redhat.com>
18187 PR middle-end/80020
18188 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18189 * builtins.def (aligned_alloc): Use it.
18192 * Makefile.in (GTFILES): Add calls.c.
18193 * calls.c: Include "gt-calls.h".
18195 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18197 PR rtl-optimization/79728
18198 * regs.h (struct target_regs): New field
18199 x_contains_allocatable_regs_of_mode.
18200 (contains_allocatable_regs_of_mode): New macro.
18201 * reginfo.c (init_reg_sets_1): Initialize it, and change
18202 contains_reg_of_mode so it includes global regs as well.
18203 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18204 rather than contains_regs_of_mode.
18206 2017-03-14 Martin Liska <mliska@suse.cz>
18208 * doc/invoke.texi: Document options that can't be combined with
18209 -fcheck-pointer-bounds.
18211 2017-03-14 Martin Liska <mliska@suse.cz>
18213 PR middle-end/79831
18214 * doc/invoke.texi (-Wchkp): Document the option.
18216 2017-03-14 Martin Liska <mliska@suse.cz>
18218 * Makefile.in: Install gcov-dump.
18220 2017-03-14 Martin Liska <mliska@suse.cz>
18222 * multiple_target.c (expand_target_clones): Bail out for
18223 an invalid attribute.
18225 2017-03-14 Richard Biener <rguenther@suse.de>
18227 * alias.c (struct alias_set_entry): Pack properly.
18228 * cfgloop.h (struct loop): Likewise.
18229 * cse.c (struct set): Likewise.
18230 * ipa-utils.c (struct searchc_env): Likewise.
18231 * loop-invariant.c (struct invariant): Likewise.
18232 * lra-remat.c (struct cand): Likewise.
18233 * recog.c (struct change_t): Likewise.
18234 * rtl.h (struct address_info): Likewise.
18235 * symbol-summary.h (function_summary): Likewise.
18236 * tree-loop-distribution.c (struct partition): Likewise.
18237 * tree-object-size.c (struct object_size_info): Likewise.
18238 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18239 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18240 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18241 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18242 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18243 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18244 (struct _stmt_vec_info): Likewise.
18246 2017-03-14 Martin Liska <mliska@suse.cz>
18249 * multiple_target.c (create_dispatcher_calls): Check that
18250 a target can create a function dispatcher.
18252 2017-03-14 Martin Liska <mliska@suse.cz>
18255 * multiple_target.c (expand_target_clones): Drop local.local
18256 flag for default implementation.
18258 2017-03-14 Richard Biener <rguenther@suse.de>
18260 PR tree-optimization/80030
18261 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18263 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18265 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18266 gcc_fallthrough() instead of __attribute__((fallthrough));
18268 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18270 * doc/gcc.texi: Remove "up" link to (DIR).
18271 * doc/gccint.texi: Ditto.
18273 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18275 * doc/install.texi (Specific) <avr>: Remove reference to
18278 2017-03-13 Jeff Law <law@redhat.com>
18280 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18281 attribute rather than comments.
18283 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18284 match_scratch operand is highest.
18286 2017-03-13 Martin Liska <mliska@suse.cz>
18288 PR middle-end/78339
18289 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18290 is a CHKP clone, use original declaration.
18292 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18294 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18295 (arc_conditional_register_usage): Use a different allocation order
18296 when optimizing for size.
18297 * common/config/arc/arc-common.c (arc_option_optimization_table):
18298 Section anchors default on when optimizing for size.
18300 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18302 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18304 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18306 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18307 * config/arc/arc.md (cpu_facility): Add cd variant.
18308 (*movqi_insn): Add code density variant.
18309 (*movhi_insn): Likewise.
18310 (*movqi_insn): Likewise.
18311 (*addsi3_mixed): Likewise.
18312 (subsi3_insn): Likewise.
18314 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18316 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18318 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18320 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18321 expressions with MINUS and UNARY ops.
18323 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18326 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18328 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18329 between vec_select and vector argument.
18330 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18331 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18332 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18333 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18334 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18335 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18337 2017-03-13 Richard Biener <rguenther@suse.de>
18340 * params.def (vect-max-peeling-for-alignment): Fix typo.
18342 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18344 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18345 issue that only occurred with binutils below 2.18.
18347 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18349 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18350 refer to binutils 2.11/2.12 minimum.
18352 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18354 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18355 ftp.kernel.org and simplify binutils requirement.
18357 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18359 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18361 (Optimize Options): Ditto. Also remove redundancy.
18363 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18365 PR translation/79848
18366 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18368 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18369 to G_ to avoid double translation.
18371 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18373 PR translation/79923
18374 * auto-profile.c (get_combined_location): Convert leading
18375 character of diagnostics to lower case and remove trailing period.
18376 (read_profile): Likewise for various diagnostics.
18377 * config/arm/arm.c (arm_option_override): Remove trailing period
18378 from various diagnostics.
18379 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18380 (msp430_expand_delay_cycles): Likewise.
18382 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18385 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18386 full command-line argument, rather than just "str".
18387 (aarch64_validate_march): Likewise.
18388 (aarch64_validate_mtune): Likewise.
18390 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18392 PR rtl-optimization/78911
18393 * lra-assigns.c (must_not_spill_p): New function.
18394 (spill_for): Use it.
18396 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18398 PR tree-optimization/79981
18399 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18400 ATOMIC_COMPARE_EXCHANGE ifn result.
18401 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18402 IFN_ATOMIC_COMPARE_EXCHANGE.
18404 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18407 * opts.c (parse_sanitizer_options): Add missing question mark to
18408 "did you mean" message.
18410 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18412 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18414 (VMULEUH_UNS): Likewise.
18415 (VMULOUB_UNS): Likewise.
18416 (VMULOUH_UNS): Likewise.
18417 * config/rs6000/rs6000.c (builtin_function_type): Remove
18418 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18420 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18423 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18424 x with result of extra_parsing_for_operand_code_0.
18425 (function_reader::extra_parsing_for_operand_code_0): Convert
18426 return type from void to rtx, returning x. When reading
18427 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18428 larger size containing struct block_symbol.
18430 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18432 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18433 -mfloat128-hardware without -m64.
18435 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18438 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18439 entries to the case statement that marks unsigned arguments to
18440 overloaded functions.
18442 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18444 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18445 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18447 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18450 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18451 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18453 2017-03-10 Martin Liska <mliska@suse.cz>
18457 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18458 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18461 2017-03-10 Olivier Hainque <hainque@adacore.com>
18463 * tree-switch-conversion (array_value_type): Start by resetting
18464 candidate type to it's main variant.
18466 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18468 PR rtl-optimization/79909
18469 * combine.c (try_combine): Use simplify_replace_rtx on individual
18470 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18471 of the whole CALL_INSN_FUNCTION_USAGE.
18473 PR tree-optimization/79972
18474 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18475 get_range_info on SSA_NAMEs. Formatting fixes.
18477 2017-03-10 Richard Biener <rguenther@suse.de>
18478 Jakub Jelinek <jakub@redhat.com>
18480 PR tree-optimization/77975
18481 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18482 edge to be constant.
18483 (get_val_for): For constant x return it. Formatting fix.
18484 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18485 would use the same bases as the current one.
18487 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18489 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18490 instead of vec_select for V1TImode.
18491 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18493 (VSX_LE_128): Add V1TI to this mode iterator.
18494 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18495 (*vsx_le_perm_store_<mode>): Likewise.
18496 (pre-reload splitter for VSX stores): Likewise.
18497 (post-reload splitter for VSX stores): Likewise.
18498 (*vsx_xxpermdi2_le_<mode>): Likewise.
18499 (*vsx_lxvd2x2_le_<mode>): Likewise.
18500 (*vsx_stxvd2x2_le_<mode>): Likewise.
18502 2017-03-09 Michael Eager <eager@eagercon.com>
18504 Correct failures with --enable-checking=yes,rtl.
18506 * config/microblaze/microblaze.c (microblaze_expand_shift):
18507 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18508 test for const0_rtx.
18509 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18510 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18512 2017-03-09 Richard Biener <rguenther@suse.de>
18514 PR tree-optimization/79977
18515 * graphite-scop-detection.c (scop_detection::merge_sese):
18516 Handle the case of extra exits to blocks dominating the entry.
18518 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18520 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18523 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18525 PR rtl-optimization/79949
18526 * lra-constraints.c (process_alt_operands): Check memory when
18527 trying to predict a cycle. Print about the overall increase.
18529 2017-03-09 Richard Biener <rguenther@suse.de>
18531 PR middle-end/79971
18532 * gimple-expr.c (useless_type_conversion_p): Preserve
18533 TYPE_SATURATING for fixed-point types.
18535 2017-03-09 Richard Biener <rguenther@suse.de>
18538 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18539 alignment of BLKmode params.
18541 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18544 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18545 (VALL_NO_V2Q): Likewise.
18547 * config/aarch64/aarch64-simd.md
18548 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18550 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18551 VALL_NO_V2Q mode iterator.
18552 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18554 2017-03-09 Martin Liska <mliska@suse.cz>
18556 PR tree-optimization/79631
18557 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18558 tree_int_cst_sign_bit just for INTEGER constants.
18560 2017-03-09 Martin Liska <mliska@suse.cz>
18564 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18567 2017-03-09 Marek Polacek <polacek@redhat.com>
18570 * tree.c (inchash::add_expr): Handle TREE_VEC.
18572 2017-03-09 Martin Liska <mliska@suse.cz>
18575 (chkp_narrow_size_and_offset): New function.
18576 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18577 (void chkp_parse_bit_field_ref): New function.
18578 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18579 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18581 2017-03-09 Martin Liska <mliska@suse.cz>
18584 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18585 (chkp_find_bounds_1): Remove gcc_unreachable.
18587 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18590 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18591 BUILT_IN_SYNC*, determine the access type from the size suffix and
18592 always build a MEM_REF with that type. Handle forgotten
18593 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18596 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18597 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18598 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18599 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18600 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18601 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18602 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18603 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18604 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18605 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18606 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18607 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18608 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18609 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18610 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18611 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18612 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18613 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18614 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18615 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18616 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18617 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18618 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18619 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18620 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18621 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18622 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18623 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18624 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18625 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18626 definitions outside of __OPTIMIZE__ guarded section.
18629 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18630 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18631 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18632 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18635 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18637 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18638 ("vfenez<mode>"): Add missing constraints.
18640 2017-03-08 Martin Sebor <msebor@redhat.com>
18643 * config/nds32/nds32.c (nds32_option_override):
18644 Fix misspelled diagnostic.
18646 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18649 * gimplify.c (gimplify_omp_for): Replace index var in outer
18650 taskloop statement with an artificial variable and add
18651 OMP_CLAUSE_PRIVATE clause for it.
18653 2017-03-08 Richard Biener <rguenther@suse.de>
18655 PR tree-optimization/79955
18656 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18657 for accesses that are completely outside of the variable.
18659 2017-03-08 Andrew Haley <aph@redhat.com>
18661 PR tree-optimization/79943
18662 * tree-ssa-loop-split.c (compute_new_first_bound): When
18663 calculating the new upper bound, (END-BEG) should be added, not
18666 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18668 * config/avr/avr.md (setmemhi): Make sure match_dup
18669 operand number comes before match_scratch.
18671 2017-03-08 Richard Biener <rguenther@suse.de>
18673 PR tree-optimization/79920
18674 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18675 with ncopies == 1 to ...
18676 (vect_transform_slp_perm_load): ... here. Properly compute
18677 all element loads by iterating VF times over the group. Do
18678 not handle ncopies (computed in a broken way) in
18679 vect_create_mask_and_perm.
18681 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18684 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18685 is a uniform vector, use uniform_vector_p return value instead of
18686 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18688 2017-03-07 Marek Polacek <polacek@redhat.com>
18690 PR middle-end/79809
18691 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18692 (alloca_call_type): Likewise.
18694 2017-03-07 Martin Liska <mliska@suse.cz>
18696 * gcov.c (process_args): Put comment to correct location.
18698 2017-03-07 Martin Liska <mliska@suse.cz>
18700 PR middle-end/68270
18701 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18702 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18703 (chkp_narrow_bounds_for_field): Likewise.
18704 (chkp_parse_array_and_component_ref): Pass one more argument to
18707 2017-03-07 Richard Biener <rguenther@suse.de>
18709 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18712 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18714 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18715 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18717 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18720 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18721 to end of description.
18722 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18724 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18726 PR rtl-optimization/79901
18727 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18729 (*avx512f_<code><mode>3<mask_name>): ... this.
18730 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18731 iterator instead of VI8_AVX2_AVX512BW.
18733 PR rtl-optimization/79901
18734 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18735 min/max expander, expand it using expand_vec_cond_expr.
18738 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18741 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18744 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18745 to void * for PCH reasons.
18746 * dwarf2out.c (output_loc_operands, output_die): Cast
18747 v.val_vec.array to unsigned char *.
18749 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18752 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18755 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18757 PR rtl-optimization/79571
18758 * lra-constraints.c (process_alt_operands): Calculate static
18759 reject and subtract it from overall when only addresses will be
18762 2017-03-06 Julia Koval <julia.koval@intel.com>
18765 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18766 incoming stack boundary to 128 for 64-bit targets.
18768 2017-03-06 Richard Biener <rguenther@suse.de>
18770 PR tree-optimization/79894
18771 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18772 to NULL after folding it.
18774 2017-03-06 Richard Biener <rguenther@suse.de>
18776 PR tree-optimization/79824
18777 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18778 check disabling peeling for gaps.
18780 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18782 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18783 attributes): Document gettimeofday.
18785 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18787 * config/s390/s390.c (s390_option_override_internal): Set
18788 PARAM_MIN_VECT_LOOP_BOUND
18790 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18792 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18793 * config/s390/s390.md: Likewise.
18795 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18798 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18799 (<avx2_avx512>_perm<mode>): Rename to ...
18800 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18802 (<avx512>_perm<mode>_mask): Rename to ...
18803 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18805 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18806 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18807 instead of VI8F_256_512.
18808 (avx512f_perm<mode>): New define_expand.
18809 (avx512f_perm<mode>_mask): Likewise.
18810 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18811 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18813 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18815 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18816 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18818 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18820 2017-03-06 Martin Liska <mliska@suse.cz>
18823 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18824 when having a SSA NAME w/o VAR_DECL assigned to it.
18826 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18828 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18829 msa_dpsub_<su>_d): Fix MODE for vec_select.
18831 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18833 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18835 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18837 2017-03-06 Richard Biener <rguenther@suse.de>
18839 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18840 * plugin.c (register_plugin_info): Likewise.
18841 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18843 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18845 * config/i386/sse.md (sse_storehps, sse_storelps,
18846 avx_<castmode><avxsizesuffix>_<castmode>,
18847 avx512f_<castmode><avxsizesuffix>_<castmode>,
18848 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18849 in condition that at least one operand is not a MEM.
18851 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18853 PR middle-end/79805
18854 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18855 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18857 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18858 gimple_call_nothrow_p flag based on whether original builtin can throw.
18859 If it can, emit following stmts on the fallthrough edge.
18860 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18861 don't create new bb if inserting just debug stmts on the edge, try to
18862 insert them on the fallthru bb or just reset debug stmts.
18864 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18867 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18868 restore recog_data (including the operand rtxes inside it) around
18869 the call to get_insn_template.
18871 2017-03-03 Martin Sebor <msebor@redhat.com>
18873 PR tree-optimization/79699
18874 * context.c (context::~context): Free MPFR caches to avoid
18875 a memory leak on program exit.
18877 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18879 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18880 Use wide_int::ulow () instead of .elt (0).
18882 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18884 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18885 (*pushxf): Limit oF constraint to 32bit targets and add oC
18886 constraint for 64bit targets.
18887 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18888 (*pushdf): Change rmF constraint to rmC.
18890 2017-03-03 Martin Liska <mliska@suse.cz>
18892 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18893 Remove unused variable.
18895 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18898 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18899 is a memory operand, increase num_memory.
18900 (ix86_expand_args_builtin): Likewise.
18902 2017-03-03 Jan Hubicka <jh@suse.cz>
18905 * ipa-devirt.c (maybe_record_node): Properly handle
18906 __cxa_pure_virtual visibility.
18908 2017-03-03 Martin Liska <mliska@suse.cz>
18910 PR tree-optimization/79803
18911 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18913 (pass_loop_prefetch::execute): Disabled optimization if an
18914 assumption about L1 cache size is not met.
18916 2017-03-03 Martin Liska <mliska@suse.cz>
18918 PR rtl-optimization/79574
18919 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18920 (hash_scan_set): Likewise.
18921 (dump_hash_table): Likewise.
18922 (hoist_code): Likewise.
18924 2017-03-03 Richard Biener <rguenther@suse.de>
18926 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18927 in place of uhigh ().
18928 (fixed_convert_from_real): Likewise.
18930 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18933 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18935 2017-03-03 Richard Biener <rguenther@suse.de>
18937 PR middle-end/79818
18938 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18939 TYPE_OVERFLOW_UNDEFINED check.
18941 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18943 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18945 (vector_ae_<mode>_p): Likewise.
18946 (vector_nez_<mode>_p): Likewise.
18947 (vector_ne_v2di_p): Likewise.
18948 (vector_ae_v2di_p): Likewise.
18949 (vector_ne_<mode>_p): Likewise.
18950 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18952 (vsx_tsqrt<mode>2_fe): Likewise.
18954 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18957 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18959 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18961 PR rtl-optimization/79780
18962 * cprop.c (one_cprop_pass): When second and further conditional trap
18963 in a single basic block is turned into an unconditional trap, turn it
18964 into a deleted note to avoid RTL verification failures.
18966 2017-03-02 Richard Biener <rguenther@suse.de>
18968 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18970 2017-03-02 Richard Biener <rguenther@suse.de>
18972 PR tree-optimization/79345
18974 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18975 param and abort the walk, returning -1 if it is hit.
18976 (walk_aliased_vdefs): Take a limit param and pass it on.
18977 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18978 defaulting to 0 and return a signed int.
18979 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18980 (check_defs): New helper.
18981 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18982 about uninitialized memory.
18983 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18984 bogus uninitialized warning.
18985 (fixed_convert_from_real): Likewise.
18987 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18989 PR tree-optimization/66768
18990 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18991 iv_use if base object can't be determined.
18993 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18995 PR tree-optimization/79345
18996 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18997 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18998 (get_pattern_stats): Initialize it.
18999 * genemit.c (gen_expand): Verify match_scratch numbers come after
19000 match_operand/match_dup numbers.
19001 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
19002 match_scratch numbers.
19003 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
19005 * config/s390/s390.md (trunctdsd2): Likewise.
19007 2017-03-02 Richard Biener <rguenther@suse.de>
19009 * wide-int.h (wide_int_storage::operator=): Implement in terms
19012 2017-03-02 Richard Biener <rguenther@suse.de>
19014 PR tree-optimization/79777
19015 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
19016 the to insert expression to sth existing.
19018 2017-03-01 Martin Sebor <msebor@redhat.com>
19020 PR middle-end/79692
19021 * gimple-ssa-sprintf.c
19022 (directive::known_width_and_precision): New function.
19023 (format_integer): Use it.
19024 (get_mpfr_format_length): Consider the full range of precision
19025 when computing %g output with the # flag. Set the likely byte
19026 count to 3 rather than 1 when precision is indeterminate.
19027 (format_floating): Correct the lower bound of precision.
19029 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19031 * doc/invoke.texi: Document default code model for 64-bit Linux.
19033 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19036 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
19037 udiv rather than div since input pattern is unsigned.
19039 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
19041 * config/i386/i386.c (print_reg): Warn for values of
19042 unsupported size in integer register.
19044 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
19047 * config/rs6000/predicates.md (current_file_function_operand): Do
19048 not allow self calls to be local if the function is replaceable.
19050 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
19053 * config/rs6000/altivec.h (vec_ctz and others): Change the
19054 preprocessor macro that controls conditional compilation from
19055 _ARCH_PWR9 to __POWER9_VECTOR__.
19056 (vec_all_ne): Change parameterization of __altivec_scalar_pred
19057 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
19058 control (instead of _ARCH_PWR9 control) so that template
19059 definition uses power9-specific function.
19060 (vec_any_eq): Likewise.
19061 (vec_all_ne): Change macro definition to use a power9-specific
19062 expansion under #ifdef __POWER9_VECTOR__ control (instead of
19063 _ARCH_PWR9 control).
19064 (vec_any_eq) Likewise.
19065 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19066 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19067 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19068 support for xvcmpnedp instruction.
19069 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19070 macro expansion so that Power9 implementation of vec_all_ne does
19071 not use the AltiVec predicate framework.
19072 (VCMPNEH_P): Likewise.
19073 (VCMPNEW_P): Likewise.
19074 (VCMPNED_P): Likewise.
19075 (VCMPNEFP_P): Likewise.
19076 (VCMPNEDP_P): Likewise.
19077 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19078 implementation of vec_any_eq to not use AltiVec predicate
19080 (VCMPAEH_P): Likewise.
19081 (VCMPAEW_P): Likewise.
19082 (VCMPAED_P): Likewise.
19083 (VCMPAEFP_P): Likewise.
19084 (VCMPAEDP_P): Likewise.
19085 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19086 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19087 not use the AltiVec predicate framework.
19088 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19089 of vec_any_eq to not use AltiVec predicate framework.
19090 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19091 support for predefined __POWER9_VECTOR__ macro to indicate that
19092 Power9 instruction selection is enabled.
19093 (altivec_overloaded_builtins): Remove extraneous
19094 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19095 function argument types RS6000_BTI_bool_V16QI and
19096 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19097 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19098 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19099 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19100 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19101 Power9 for implementations of vec_cmpne. Change the signature for
19102 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19103 (representing vec_all_ne) to remove the previously described first
19104 argument of type RS6000_BTI_INTSI, as this was an artifact of
19105 reliance on the AltiVec predicate framework, which is no longer
19106 used in the implementation of these functions. Add
19107 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19108 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19109 since, unlike the AltiVec predicate framework implementation, we
19110 do not share function descriptors between vec_alle and vec_anyeq.
19111 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19112 set of modes that receive special treatment even when
19113 TARGET_P9_VECTOR is true. The special treatment emits code that
19114 does not depend on Power9 instructions.
19115 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19116 define_expand to not rely on AltiVec predicate framework.
19117 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19119 (vector_ne_v2di_p): Change this define_expand to not rely on
19120 AltiVec predicate framework.
19121 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19123 (vector_ne_<mode>_p): Change this define_expand to not rely on
19124 AltiVec predicate framework.
19125 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19127 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19128 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19129 define_insn pattern.
19130 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19131 define_insn pattern because the xvcmpne<VSs>. instruction is not
19133 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19134 instruction is not supported.
19136 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19138 * config/nvptx/nvptx.c: Include intl.h.
19140 2017-03-01 Martin Jambor <mjambor@suse.cz>
19143 * ipa-prop.h (ipa_bits): Removed field known.
19144 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19145 to pointers. Adjusted their comments to warn about their sharing.
19146 (ipcp_transformation_summary): Change bits to a vector of pointers.
19147 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19148 (ipa_get_ipa_bits_for_value): Declare.
19149 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19150 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19151 (ipa_bits_hash_table): Likewise.
19152 (ipa_vr_ggc_hash_traits): Likewise.
19153 (ipa_vr_hash_table): Likewise.
19154 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19155 being pointers and vr_known being removed.
19156 (ipa_set_jf_unknown): Likewise.
19157 (ipa_get_ipa_bits_for_value): New function.
19158 (ipa_set_jfunc_bits): Likewise.
19159 (ipa_get_value_range): New overloaded functions.
19160 (ipa_set_jfunc_vr): Likewise.
19161 (ipa_compute_jump_functions_for_edge): Use the above functions to
19162 construct bits and vr parts of jump functions.
19163 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19164 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19166 (ipcp_grow_transformations_if_necessary): Also allocate
19167 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19169 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19170 them. Fix too long lines.
19171 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19172 vr_known being removed.
19173 (ipa_read_jump_function): Use new setter functions to construct bits
19174 and vr parts of jump functions or set them to NULL.
19175 (write_ipcp_transformation_info): Adjust for bits being pointers.
19176 (read_ipcp_transformation_info): Likewise.
19177 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19179 Include gt-ipa-prop.h.
19180 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19182 (ipcp_store_bits_results): Likewise.
19183 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19184 Do not write to existing jump functions but use a temporary instead.
19186 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19189 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19190 attempt to use its first operand as BIT_FIELD_REF base.
19192 2017-03-01 Richard Biener <rguenther@suse.de>
19194 PR middle-end/79721
19195 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19196 interpolating formula in wrapping arithmetic.
19197 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19199 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19201 PR tree-optimization/79734
19202 * tree-vect-generic.c (expand_vector_condition): Optimize
19203 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19204 Handle VEC_COND_EXPR where comparison has different inner width from
19205 type's inner width.
19207 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19209 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19210 markup, and similar issues. Remove @opindex entries for things
19211 that aren't options. Add missing -mmpy-option entries.
19213 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19215 PR tree-optimization/79737
19216 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19217 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19218 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19219 instead of byte_size. Formatting fix.
19220 (shift_bytes_in_array_right): Formatting fix.
19222 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19225 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19226 condition on optimize for the leaf function test.
19228 2017-02-28 Martin Liska <mliska@suse.cz>
19231 * read-rtl-function.c (function_reader::handle_unknown_directive):
19232 Bail out when one uses -flto.
19234 2017-02-28 Martin Liska <mliska@suse.cz>
19236 * common.opt: Replace space with tabular for options of <number>
19238 * config/i386/i386.opt: Show <number> value for
19239 -mlarge-data-threshold.
19240 * opts.c (print_filtered_help): Do not display number in hexadecimal
19243 2017-02-28 Martin Liska <mliska@suse.cz>
19245 * common.opt: Fix --help=option -Q for options which are of
19248 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19250 * config/i386/i386.c (print_reg): Error out for values
19251 of 8-bit size in invalid integer register.
19253 2017-02-28 Martin Sebor <msebor@redhat.com>
19255 PR tree-optimization/79691
19256 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19258 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19261 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19262 gcc_unreachable with output_operand_lossage.
19264 2017-02-28 Richard Biener <rguenther@suse.de>
19266 PR tree-optimization/79740
19267 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19269 (visit_nary_op): Insert the nary into the hashtable if we
19270 pattern-matched sth.
19271 * tree-ssa-pre.c (eliminate_insert): Robustify.
19273 2017-02-28 Richard Biener <rguenther@suse.de>
19275 PR middle-end/79731
19276 * fold-const.c (decode_field_reference): Reject out-of-bound
19279 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19281 * config/i386/i386.c: Include intl.h.
19282 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19283 instead of just cond ? "..." : "...".
19284 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19285 * coverage.c (read_counts_file): Likewise.
19286 * omp-offload.c: Include intl.h.
19287 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19288 of just cond ? "..." : "...".
19289 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19290 of just cond ? "..." : "...".
19292 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19295 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19297 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19298 'tune for' CPU name.
19299 * config/arm/arm-cpu-data.h: Regenerated.
19301 2017-02-28 Richard Biener <rguenther@suse.de>
19303 PR tree-optimization/79732
19304 * tree-inline.c (expand_call_inline): Do not shadow var.
19306 2017-02-28 Richard Biener <rguenther@suse.de>
19308 PR tree-optimization/79723
19309 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19310 address-space properly.
19312 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19314 * doc/optinfo.texi (Optimization groups): Fix option used for
19316 * doc/invoke.texi (-fopt-info): Document "omp".
19317 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19318 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19319 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19320 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19321 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19323 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19325 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19326 OPTGROUP_OMP with "omp".
19328 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19331 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19332 for arithmetic shift of unsigned V2DI.
19334 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19336 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19337 arc/linux.h headers.
19338 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19339 (LINK_SPEC): Likewise.
19340 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19341 (EXTRA_SPECS): Likewise.
19342 (STARTFILE_SPEC): Likewise.
19343 (ENDFILE_SPEC): Likewise.
19344 (LIB_SPEC): Likewise.
19345 (TARGET_SDATA_DEFAULT): Likewise.
19346 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19347 (MULTILIB_DEFAULTS): Likewise.
19348 (DWARF2_UNWIND_INFO): Likewise.
19349 * config/arc/big.h: New file.
19350 * config/arc/elf.h: Likewise.
19351 * config/arc/linux.h: Likewise.
19352 * config/arc/t-uClibc: Remove.
19354 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19356 PR tree-optimization/77536
19357 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19358 (tree_transform_and_unroll_loop): Use above function to compute the
19359 estimated niter of unrolled loop and use it when scaling profile.
19360 Also use count info rather than frequency if it's non-zero.
19361 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19362 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19363 (vect_transform_loop): Call above function.
19365 2017-02-27 Richard Biener <rguenther@suse.de>
19367 PR tree-optimization/45397
19368 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19369 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19370 (visit_nary_op): Add pattern matching for CSEing sign-changed
19371 or truncated operations with wider ones.
19373 2017-02-27 Richard Biener <rguenther@suse.de>
19375 PR tree-optimization/79690
19376 * tree-vect-stmts.c (vectorizable_store): Use vector type
19377 built from the DR with address-space.
19379 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19381 * doc/invoke.texi (Optimize Options): Refine the description
19382 of asan-use-after-return.
19384 2017-02-25 Alan Modra <amodra@gmail.com>
19386 PR rtl-optimization/79584
19387 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19388 base, not ad->base_term, the reg within base. Remove assertion
19389 that ad->base == ad->base_term. Replace gen_int_mode using
19390 bogus mode with const0_rtx.
19392 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19394 PR middle-end/79396
19395 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19396 FMA_EXPR like tcc_binary or tcc_unary.
19398 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19401 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19403 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19404 (output_loc_operands): Handle DW_OP_call_ref and
19405 DW_OP_GNU_variable_value.
19406 (struct variable_value_struct): New type.
19407 (struct variable_value_hasher): Likewise.
19408 (variable_value_hash): New variable.
19409 (string_types): Remove.
19410 (copy_loc_descr): New function.
19411 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19412 (prepend_loc_descr_to_each): New function.
19413 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19414 instead of add_loc_descr_to_each if the first argument is single
19415 location list and the second has multiple.
19416 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19417 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19418 when looking for variable value which doesn't have other location info.
19419 (loc_list_from_tree): Formatting fix.
19420 (gen_array_type_die): Simplify DW_AT_string_length handling.
19421 (adjust_string_types): Remove.
19422 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19423 string_types. Call resolve_variable_values.
19424 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19425 (resolve_addr_in_expr): Likewise. Add A argument.
19426 (copy_deref_exprloc): Remove deref argument. Adjust for the
19427 original expression being DW_OP_GNU_variable_value with optionally
19428 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19429 optionally after it.
19430 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19431 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19432 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19433 (variable_value_hasher::hash, variable_value_hasher::equal): New
19435 (resolve_variable_value_in_expr, resolve_variable_value,
19436 resolve_variable_values, note_variable_value_in_expr,
19437 note_variable_value): New functions.
19438 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19441 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19444 * opts.h (handle_generated_option): Add GENERATED_P argument.
19445 * opts-common.c (handle_option): Adjust function comment.
19446 (handle_generated_option): Add GENERATED_P argument, pass it to
19448 (control_warning_option): Pass false to handle_generated_option
19450 * opts.c (maybe_default_option): Pass true to handle_generated_option
19452 * optc-gen.awk: Likewise.
19454 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19456 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19457 a REG, look at the REG it is a SUBREG of.
19458 (splitter for cmpeqsi_t): Ditto.
19460 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19462 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19463 the special USEs with the pattern of the insn, not the insn itself.
19465 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19468 * doc/invoke.texi: Document -mload-store-pairs.
19470 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19471 Sandra Loosemore <sandra@codesourcery.com>
19473 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19474 argument isn't a CONST_INT.
19475 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19476 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19477 (nios2_validate_compare): Bypass alternate compare logic if *op2
19478 is not a CONST_INT.
19479 (ldstwm_operation_p): Return false if first_base is not a REG or
19480 if first_offset is not a CONST_INT.
19482 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19484 * config/cris/cris.md: Use correct operand in a define_peephole2.
19486 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19488 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19490 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19492 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19493 this_insn if it is an INSN or JUMP_INSN.
19494 (force_offsettable): Look at base, not at addr.
19495 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19496 on things that aren't necessarily CONST_INTs.
19498 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19500 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19501 -mfpmath=sse is the default also for x86-32 targets with SSE2
19502 instruction set when @option{-ffast-math} is enabled
19504 2017-02-24 Jeff Law <law@redhat.com>
19506 PR rtl-optimizatoin/79286
19507 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19510 2017-02-24 Richard Biener <rguenther@suse.de>
19512 PR tree-optimization/79389
19513 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19516 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19518 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19519 function comment to reflect reality.
19520 (loop_exits_before_overflow): Fix typo in function description.
19522 2017-02-24 Richard Biener <rguenther@suse.de>
19524 PR tree-optimization/79389
19525 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19526 properly that a threading opportunity exists. Detect conditional
19527 copy/constant propagation opportunities.
19529 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19531 * config/visium/visium.md (type): Add trap.
19532 (b): New mode attribute.
19533 (*btst): Rename into...
19534 (*btst<mode>): ...this and adjust.
19535 (*cbranchsi4_btst_insn): Rename into...
19536 (*cbranch<mode>4_btst_insn): ...this and adjust.
19537 (trap): New define_insn.
19539 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19541 PR tree-optimization/79389
19542 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19543 (noce_reversed_cond_code): New function.
19544 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19545 reversed_comparison_code. Formatting fix.
19546 (noce_try_store_flag): Test rev_cond != NULL in addition to
19547 reversed_comparison_code.
19548 (noce_try_store_flag_constants): Likewise.
19549 (noce_try_store_flag_mask): Likewise.
19550 (noce_try_addcc): Use rev_cond if non-NULL instead of
19551 reversed_comparison_code.
19552 (noce_try_cmove_arith): Likewise. Formatting fixes.
19553 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19554 (noce_find_if_block): Initialize rev_cond.
19555 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19556 instead of false as last argument never attempt to reverse it
19559 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19561 PR tree-optimization/79663
19562 * tree-predcom.c (combine_chains): Process refs in reverse order
19563 only for ZERO length chains, and add explaining comment.
19565 2017-02-23 Jeff Law <law@redhat.com>
19567 PR tree-optimization/79578
19568 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19569 in call to operand_equal_p.
19571 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19574 * config/i386/cpuid.h: Fix another undefined behavior.
19576 2017-02-23 Richard Biener <rguenther@suse.de>
19578 PR tree-optimization/79683
19579 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19580 vector types for data-refs.
19582 2017-02-23 Martin Liska <mliska@suse.cz>
19584 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19586 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19588 PR middle-end/79665
19589 * internal-fn.c (get_range_pos_neg): Moved to ...
19590 * tree.c (get_range_pos_neg): ... here. No longer static.
19591 * tree.h (get_range_pos_neg): New prototype.
19592 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19593 are known to be in between 0 and signed maximum inclusive, try to
19594 expand both unsigned and signed divmod and use the cheaper one from
19597 2017-02-22 Jeff Law <law@redhat.com>
19599 PR tree-optimization/79578
19600 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19601 to compare base operands.
19603 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19606 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19607 gpc_reg_operand instead of fpr_reg_operand.
19609 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19611 * config/mips/mips.c (mips_return_in_memory): Force FP
19612 vector types to be returned in memory for o32 ABI.
19614 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19616 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19617 instead of DW_TAG_member for static data member declarations and don't
19618 set no_linkage_name for static inline data members.
19619 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19622 2017-02-22 Martin Liska <mliska@suse.cz>
19624 * doc/invoke.texi: Replace inequality signs with square brackets
19627 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19629 PR tree-optimization/68644
19630 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19632 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19635 * lra-constraints.c (simplify_operand_subreg): Handle
19636 WORD_REGISTER_OPERATIONS targets.
19638 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19641 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19642 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19643 elimination by swapping fld*.
19645 2017-02-22 Richard Biener <rguenther@suse.de>
19647 PR tree-optimization/79673
19648 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19649 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19650 irrelevant address-space qualifiers and avoiding a
19651 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19653 2017-02-22 Richard Biener <rguenther@suse.de>
19655 PR tree-optimization/79666
19656 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19657 to not symbolically negate if that may introduce undefined
19660 2017-02-22 Martin Liska <mliska@suse.cz>
19663 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19664 * data-streamer-out.c (streamer_write_gcov_count_stream):
19666 * value-prof.c (stream_out_histogram_value): Make assert more
19667 precise based on type of counter.
19669 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19672 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19673 Use nonimmediate_operand instead of memory_operand for operand 1.
19674 (float-extend standard_x87sse_constant_load splitter): Ditto.
19676 2017-02-21 Jeff Law <law@redhat.com>
19678 PR tree-optimization/79621
19679 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19680 blocks with edges to themselves.
19682 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19685 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19686 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19687 Use gimple_call_builtin_p.
19690 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19691 on temporarily removed DEBUG_INSNs.
19693 PR tree-optimization/79649
19694 * tree-loop-distribution.c (classify_partition): Give up on
19695 non-generic address space loads/stores.
19697 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19699 * doc/loop.texi (Loop manipulation): Remove nonexistent
19700 tree_ssa_loop_version from the documentation.
19701 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19703 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19706 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19707 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19708 * config/rs6000/rs6000.c: Include except.h.
19709 (rs6000_expand_split_stack_prologue): Call
19710 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19712 2017-02-21 Martin Jambor <mjambor@suse.cz>
19715 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19716 have been analyzed.
19718 2017-02-21 Martin Jambor <mjambor@suse.cz>
19720 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19721 for backward compatibility only.
19722 * doc/invoke.texi (Option Summary): Remove all references to
19723 -fipa-cp-alignment.
19725 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19729 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19731 * lra-constraints.c (curr_insn_transform): Handle
19732 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19734 2017-02-21 Martin Liska <mliska@suse.cz>
19736 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19738 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19740 PR translation/79638
19741 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19743 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19746 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19747 (arm_function_ok_for_sibcall): Return false for an indirect call by
19748 descriptor if all the argument registers are used.
19749 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19750 alignment of the function.
19752 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19754 PR tree-optimization/61441
19755 * simplify-rtx.c (simplify_const_unary_operation): For
19756 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19757 the sNaN unmodified.
19759 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19761 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19762 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19763 instead of SYSTEM_HEADER_DIR.
19765 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19766 Martin Liška <mliska@suse.cz>
19768 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19769 Fix typos and grammar, use active voice, and clarify.
19771 2017-02-20 Marek Polacek <polacek@redhat.com>
19773 PR middle-end/79537
19774 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19777 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19779 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19782 * config/i386/i386.c (ix86_expand_builtin): Handle
19783 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19784 ix86_builtins_isa[fcode].isa as a requirement of those
19785 flags and any other flag in the bitmask.
19786 (ix86_init_mmx_sse_builtins): Use 0 instead of
19787 ~OPTION_MASK_ISA_64BIT as mask.
19788 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19789 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19790 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19791 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19793 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19796 * lra-constraints.c (split_reg): Check requested split mode
19797 is supported by the register.
19799 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19801 * lra-constraints.c (simplify_operand_subreg): Remove early
19804 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19807 * lra-constraints.c (curr_insn_transform): Tighten condition
19808 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19810 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19813 * lra-constraints.c (curr_insn_transform): Handle
19814 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19816 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19819 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19821 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19823 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19826 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19829 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19831 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19834 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19836 * final.c (last_columnnum, override_columnnum): New variables.
19837 (final_start_function): Set last_columnnum, pass it to begin_prologue
19838 hook and pass 0 to dwarf2out_begin_prologue.
19839 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19840 to source_line debug hook.
19841 (notice_source_line): Compute last_columnnum and for debug_column_info
19842 return true on column changes.
19843 * debug.h (struct gcc_debug_hooks): Add column argument to
19844 source_line and begin_prologue hooks.
19845 (debug_nothing_int_charstar_int_bool): Remove prototype.
19846 (debug_nothing_int_int_charstar,
19847 debug_nothing_int_int_charstar_int_bool): New prototypes.
19848 (dwarf2out_begin_prologue): Add column argument.
19849 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19850 begin_prologue hooks.
19851 (debug_nothing_int_charstar_int_bool): Remove.
19852 (debug_nothing_int_int_charstar,
19853 debug_nothing_int_int_charstar_int_bool): New functions.
19854 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19855 through to dwarf2out_source_line.
19856 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19857 (dwarf2out_source_line): Add column argument, emit it if requested.
19858 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19860 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19861 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19862 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19863 through to dwarf2out_begin_prologue.
19864 (vmsdbgout_source_line): Add column argument, pass it through to
19865 dwarf2out_source_line.
19866 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19867 dbxout_source_line caller.
19868 (dbxout_source_line): Add column argument.
19870 * common.opt (gno-column-info, gcolumn-info): New options.
19871 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19872 (check_die): Also test for multiple DW_AT_decl_column attributes.
19873 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19874 DW_AT_decl_column if requested.
19875 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19877 (gen_variable_die): Likewise.
19878 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19879 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19882 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19883 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19884 (ix86_handle_option): Handle OPT_m3dnowa.
19885 * doc/invoke.texi (-m3dnowa): Document.
19886 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19887 -m3dnowa instead of -m3dnow -march=athlon.
19890 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19891 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19893 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19896 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19897 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19898 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19899 generator for vsx_xxpermdi_<mode>_be.
19900 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19901 force big-endian semantics.
19902 (vsx_xxpermdi_<mode>_be): New define_expand with same
19903 implementation as previous version of vsx_xxpermdi_<mode>.
19905 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19907 PR tree-optimization/79327
19908 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19909 variable, its initialization and use.
19911 2017-02-17 Julia Koval <julia.koval@intel.com>
19913 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19914 (OPTION_MASK_ISA_PKU_UNSET): New.
19915 (ix86_handle_option): Handle -mrdpid.
19916 * config/i386/cpuid.h (bit_RDPID): New.
19917 * config/i386/driver-i386.c (host_detect_local_cpu):
19918 Detect RDPID feature.
19919 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19920 * config/i386/i386-c.c (ix86_target_macros_internal):
19922 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19923 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19924 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19925 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19926 * config/i386/i386.md (define_insn "rdpid"): New.
19927 * config/i386/i386.opt Add -mrdpid.
19928 * config/i386/immintrin.h (_rdpid_u32): New.
19930 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19932 PR rtl-optimization/79541
19933 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19934 instead of transforming it into USE.
19936 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19938 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19939 If HONOR_SNANS (SFmode) force the input to a register.
19940 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19941 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19942 an frsp or similar insn.
19944 2017-02-17 Martin Liska <mliska@suse.cz>
19946 PR rtl-optimization/79577
19947 * params.def (selsched-max-sched-times): Increase minimum to 1.
19949 2017-02-17 Martin Liska <mliska@suse.cz>
19951 PR rtl-optimization/79574
19952 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19954 2017-02-17 Martin Liska <mliska@suse.cz>
19956 PR tree-optimization/79529
19957 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19958 ssa_defined_default_def_p to handle cases which are implicitly
19960 * tree-ssa.c (ssa_defined_default_def_p): New function.
19961 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19962 which are implicitly defined.
19963 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19965 2017-02-17 Richard Biener <rguenther@suse.de>
19967 PR middle-end/79576
19968 * params.def (max-ssa-name-query-depth): Limit to 10.
19970 2017-02-17 Richard Biener <rguenther@suse.de>
19972 PR tree-optimization/79552
19973 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19976 2017-02-17 Richard Biener <rguenther@suse.de>
19979 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19981 2017-02-17 Marek Polacek <polacek@redhat.com>
19983 PR middle-end/79536
19984 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19985 (fold_negate_expr): New wrapper.
19987 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19989 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19990 Correct terminology and de-emphasize pre-standard behavior.
19992 2017-02-16 Alan Modra <amodra@gmail.com>
19994 PR rtl-optimization/79286
19995 * ira.c (def_dominates_uses): New function.
19996 (update_equiv_regs): Don't create an equivalence for insns that
19997 may trap where the register def does not dominate the use.
19999 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
20001 PR rtl-optimization/78127
20002 * lra.c (lra): Call lra_eliminate before finish the loop after
20005 2017-02-16 Richard Biener <rguenther@suse.de>
20007 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
20009 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
20010 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
20011 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
20012 (isl_val_int_from_wi): New function.
20013 (extract_affine_gmp): Rename to ...
20014 (extract_affine_wi): ... this, take a widest_int.
20015 (extract_affine_int): Just wrap extract_affine_wi.
20016 (add_param_constraints): Use isl_val_int_from_wi.
20017 (add_loop_constraints): Likewise, and extract_affine_wi.
20019 2017-02-15 Jeff Law <law@redhat.com>
20021 PR middle-end/79521
20022 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
20023 ira_init_register_move_cost_if_necessary.
20025 2017-02-15 Martin Sebor <msebor@redhat.com>
20027 PR middle-end/32003
20028 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
20029 removed in a prior commit.
20031 2017-02-15 Bin Cheng <bin.cheng@arm.com>
20033 PR tree-optimization/79347
20034 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
20035 counters during peeling.
20037 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
20039 * Makefile.in (site.exp): Remove "set ISLVER".
20041 2017-02-15 Jakub Jelinek <jakub@redhat.com>
20044 * real.c (real_from_integer): Call real_convert even for decimal.
20046 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
20049 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
20051 2017-02-14 Andrew Pinski <apinski@cavium.com>
20053 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
20054 cores and change the partno/implementer to be correct.
20055 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
20056 the 'B" as the implementer.
20057 * config/aarch64/aarch64-tune.md: Regenerate.
20059 2017-02-14 Carl Love <cel@us.ibm.com>
20061 * config/rs6000/rs6000.c: Add case statement entry to make the
20062 xvcvuxdsp built-in argument unsigned.
20063 * config/rs6000/vsx.md: Fix the source and return operand types so they
20064 match the instruction definitions from the ISA document. Fix typo
20065 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20068 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20071 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20072 member early_clobber_alts.
20073 * lra-lives.c (reg_early_clobber_p): New.
20074 (process_bb_lives): Use it.
20075 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20076 (debug_operand_data): Initialize early_clobber_alts.
20077 (setup_operand_alternative): Set up early_clobber_alts.
20078 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20079 alternatives to new_insn_reg.
20080 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20082 (lra_update_insn_regno_info): Pass the new arg.
20084 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20086 PR middle-end/79505
20087 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20088 (new_oacc_loop_raw): Don't clear already cleared fields.
20091 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20092 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20093 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20095 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20098 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20100 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20103 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20104 the extra instruction to the right place to store 128-bit constant
20107 2017-02-14 Martin Sebor <msebor@redhat.com>
20109 PR middle-end/79448
20110 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20111 warning for strings of unknown length.
20113 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20115 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20117 2017-02-14 Jeff Law <law@redhat.com>
20120 * ira-costs.c (scan_one_insn): Initialize register move costs
20121 for pseudos seen in USE/CLOBBER insns.
20123 PR tree-optimization/79095
20124 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20125 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20126 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20127 if the operands are known to be not equal, then the resulting range
20129 (intersect_ranges): If the new range is ~[0,0] and the old range is
20130 wide, then prefer ~[0,0].
20131 * tree-vrp.c (overflow_comparison_p_1): New function.
20132 (overflow_comparison_p): New function.
20133 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20134 if NAME is used in an overflow test.
20135 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20136 overflow check that can be expressed as an equality test, then adjust
20137 ops to be that equality test.
20139 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20141 * config/s390/s390-builtin-types.def: Remove flags argument.
20142 * config/s390/s390.c (s390_init_builtins): Likewise.
20144 2017-02-14 Martin Liska <mliska@suse.cz>
20146 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20147 vector. Fix trailing white spaces.
20149 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20151 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20154 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20156 PR rtl-optimization/68664
20157 * config/arm/arm.c (arm_sched_can_speculate_insn):
20158 New function. Declare prototype.
20159 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20161 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20163 PR rtl-optimization/68664
20164 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20166 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20168 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20170 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20171 max skip bytes for function, loop and jump.
20173 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20175 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20176 ABS_EXPR for gimple dump.
20178 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20181 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20183 PR tree-optimization/79408
20184 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20185 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20186 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20187 also if rhs1 is INTEGER_CST.
20189 2017-02-14 Richard Biener <rguenther@suse.de>
20191 PR middle-end/79432
20192 * tree-into-ssa.c (insert_phi_nodes): When the function can
20193 have abnormal edges rewrite SSA names with broken use-def
20194 dominance out of SSA and register them for PHI insertion.
20196 2017-02-13 Martin Sebor <msebor@redhat.com>
20198 PR middle-end/79496
20199 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20200 clearing info.nowrite flag when snprintf size argument is a range.
20202 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20204 * cprop.c (cprop_jump): Add missing space in string literal.
20205 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20206 (get_constraint_for_component_ref): Likewise.
20207 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20208 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20209 * lra-constraints.c (process_alt_operands): Likewise.
20210 * ipa-inline.c (inline_small_functions): Likewise.
20211 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20212 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20213 * trans-mem.c (diagnose_tm_1_op): Likewise.
20214 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20215 (grid_parallel_clauses_gridifiable): Likewise.
20217 * config/nvptx/mkoffload.c (process): Add space in between
20220 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20221 "MOD4_SSE_REGS" and "ALL_REGS".
20223 * spellcheck.c (test_data): Add , in between "foo" and "food".
20225 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20228 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20229 boundary crossing check and subsequent code generation agree.
20231 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20233 * config/aarch64/aarch64.c (has_memory_op): Delete.
20234 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20237 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20239 PR rtl-optimization/79388
20240 PR rtl-optimization/79450
20241 * combine.c (distribute_notes): When removing TEM_INSN for which
20242 corresponding dest has last value recorded, invalidate that last
20245 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20247 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20248 of explicit '@'. Add missing assembly comment marker on branch costs
20251 2017-02-13 Nathan Sidwell <nathan@acm.org>
20253 * gengtype-lex.l (<in_struct>): Add '/'.
20255 2017-02-13 Martin Liska <mliska@suse.cz>
20258 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20260 2017-02-13 Richard Biener <rguenther@suse.de>
20262 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20264 * configure: Re-generate.
20265 * config.in: Likewise.
20266 * graphite-dependences.c: Simplify as if
20267 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20268 * graphite-isl-ast-to-gimple.c: Likewise.
20269 * graphite-optimize-isl.c: Likewise.
20270 * graphite-poly.c: Likewise.
20271 * graphite-sese-to-poly.c: Likewise.
20272 * graphite.h: Likewise.
20273 * toplev.c: Include isl/version.h and use isl_version () for
20274 printing the ISL version.
20275 * doc/install.texi: Update ISL requirement.
20277 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20279 * doc/standards.texi (Standards): Update reference to
20282 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20284 * doc/extend.texi (Named Address Spaces): sourceware.org now
20286 * doc/install.texi (Binaries): Ditto.
20289 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20291 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20292 standard terminology "stringize"/"stringizing" throughout.
20293 * doc/cppinternals.texi: Likewise.
20295 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20297 * doc/extend.texi: Fix some spelling mistakes and typos.
20298 * doc/invoke.texi: Likewise.
20300 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20303 * params.def (inline-min-speedup) Change from 10 to 8.
20305 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20307 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20310 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20313 * ipa-inline-analysis.c (get_minimal_bb): New function.
20314 (record_modified): Use it.
20315 (remap_edge_change_prob): Handle also ancestor functions.
20317 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20319 * doc/contrib.texi (Contributors): Remove broken link into
20320 the Mauve CVS repository.
20322 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20324 PR middle-end/79454
20325 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20326 result computation whenever lhs doesn't have vector mode, not
20327 just when it has BLKmode.
20329 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20331 * doc/makefile.texi (profiledbootstrap): Refer to the
20332 installation instructions only in textual form.
20334 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20337 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20339 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20341 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20342 (Specific): Update mingw-w64 reference.
20344 (Specific): Remove broken link to Renesas RX processor.
20346 2017-02-10 Richard Biener <rguenther@suse.de>
20348 * toplev.c (process_options): Do not mention obsolete graphite
20349 options when printing sorry message about missing graphite support.
20350 Mention -floop-nest-optimize.
20352 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20354 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20355 (vtst_p16): Likewise.
20356 (vtstq_p8): Likewise.
20357 (vtstq_p16): Likewise.
20359 (vtstq_p64): Likewise.
20360 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20361 (vset_lane_p64): New.
20362 (vsetq_lane_p64): New.
20364 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20366 PR tree-optimization/79411
20367 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20368 stmt operands are SSA_NAMEs used in abnormal phis.
20369 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20372 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20375 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20378 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20380 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20382 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20384 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20388 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20389 not arbitrary TREE_CONSTANT.
20392 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20393 "omp declare target link" attribute unless is_global_var.
20394 * omp-offload.c (find_link_var_op): Likewise.
20396 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20397 Chung-Lin Tang <cltang@codesourcery.com>
20399 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20401 (gimplify_adjust_omp_clauses): Don't delete TILE.
20402 (gimplify_omp_for): Deal with TILE.
20403 * internal-fn.c (expand_GOACC_TILE): New function.
20404 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20406 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20407 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20409 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20410 avoid DIV for outermost collapse var.
20411 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20412 Remove out of date comments, fix whitespace.
20413 * omp-general.c (omp_extract_for_data): Deal with tiling.
20414 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20415 adjust OLF_DIM_BASE value.
20416 (struct omp_for_data): Add tiling field.
20417 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20418 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20419 for auto loops. Remove default auto determining, moved to
20420 oacc_loop_fixed_partitions.
20421 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20422 stmts, add e_mask field.
20423 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20424 (oacc_thread_numbers): Use oacc_dim_call.
20425 (oacc_xform_tile): New.
20426 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20427 (finish_oacc_loop): Adjust for ifns vector.
20428 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20429 add case for GOACC_TILE fns.
20430 (oacc_loop_xform_loop): Delete.
20431 (oacc_loop_process): Iterate over call list directly, and add
20432 handling for GOACC_TILE fns.
20433 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20435 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20436 vector partitioning to outer loops. Assign 2 partitions to loops
20437 when available. Add TILE handling.
20438 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20439 (execite_oacc_device_lower): Process GOACC_TILE fns,
20440 ignore unknown specs.
20441 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20442 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20443 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20445 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20447 * configure.ac (ACX_BUGURL): Update.
20448 * configure: Regenerate.
20450 2017-02-09 Richard Biener <rguenther@suse.de>
20452 PR tree-optimization/69823
20453 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20454 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20456 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20458 * config/arc/arc-c.def: Add __NPS400__ definition.
20459 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20460 (TARGET_NPS400): Define.
20462 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20464 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20466 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20467 pointer, arch_info.
20468 (arc_cpu_types): Fill the arch_info field with a pointer into the
20469 arc_arch_types table.
20470 (arc_selected_cpu): Declare.
20471 * config/arc/arc.c (arc_selected_cpu): Make global.
20472 (arc_selected_arch): Delete.
20473 (arc_base_cpu): Delete.
20474 (arc_override_options): Remove references to deleted variables,
20475 update access to arch information.
20476 (ARC_OPT): Update access to arch information.
20477 (ARC_OPTX): Likewise.
20478 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20479 (TARGET_ARC600): Update access to arch information.
20480 (TARGET_ARC601): Likewise.
20481 (TARGET_ARC700): Likewise.
20482 (TARGET_EM): Likewise.
20483 (TARGET_HS): Likewise.
20484 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20487 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20490 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20491 condition/operands for integer GE/LE/GEU/LEU operations.
20493 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20495 PR translation/79397
20496 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20499 2017-02-08 Martin Jambor <mjambor@suse.cz>
20502 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20503 whether allocation happened.
20504 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20505 nothing was allocated.
20507 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20509 PR tree-optimization/79408
20510 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20511 constant, but SSA_NAME with a known integer range, use the minimum
20512 of that range instead of op1 to determine if modulo can be replaced
20513 with its first operand.
20515 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20517 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20519 2017-02-08 Richard Biener <rguenther@suse.de>
20521 PR tree-optimization/71824
20522 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20523 Check all loops contained in the merged region.
20525 2017-02-07 Andrew Pinski <apinski@cavium.com>
20527 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20529 2017-02-07 Andrew Pinski <apinski@cavium.com>
20531 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20532 (thunderxt88): Likewise.
20533 (thunderxt81): Disable LSE and change v8.1 to v8.
20534 (thunderxt83): Likewise.
20536 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20537 Richard Biener <rguenther@suse.de>
20539 PR middle-end/79399
20540 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20541 type from int to size_t.
20542 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20544 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20546 PR rtl-optimization/79386
20547 * cprop.c (bypass_conditional_jumps): Initialize
20548 bypass_last_basic_block already before splitting bbs after
20549 unconditional traps...
20550 (bypass_conditional_jumps): ... rather than here.
20553 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20554 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20555 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20556 fix -masm=intel patterns.
20558 2017-02-07 Richard Biener <rguenther@suse.de>
20560 PR tree-optimization/79256
20561 PR middle-end/79278
20562 * builtins.c (get_object_alignment_2): Use min_align_of_type
20563 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20564 and ADJUST_FIELD_ALIGN.
20566 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20568 * doc/tm.texi: Regenerate.
20569 * stor-layout.c (layout_decl): Adjust.
20570 (update_alignment_for_field): Likewise.
20571 (place_field): Likewise.
20572 (min_align_of_type): Likewise.
20573 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20574 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20575 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20576 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20577 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20578 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20579 * config/i386/i386.c (x86_field_alignment): Likewise.
20580 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20581 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20582 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20583 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20584 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20585 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20589 2017-01-30 Richard Biener <rguenther@suse.de>
20591 PR tree-optimization/79256
20592 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20593 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20596 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20598 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20599 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20600 builtins to SImode and emit a zero-extend, if necessary.
20602 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20604 * docs/invoke.texi (RISC-V Options): Alphabetize.
20606 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20608 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20611 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20613 * config/riscv/riscv.c: New file.
20614 * common/config/riscv/riscv-common.c: Likewise.
20615 * config.gcc: Likewise.
20616 * config/riscv/constraints.md: Likewise.
20617 * config/riscv/elf.h: Likewise.
20618 * config/riscv/generic.md: Likewise.
20619 * config/riscv/linux.h: Likewise.
20620 * config/riscv/multilib-generator: Likewise.
20621 * config/riscv/peephole.md: Likewise.
20622 * config/riscv/pic.md: Likewise.
20623 * config/riscv/predicates.md: Likewise.
20624 * config/riscv/riscv-builtins.c: Likewise.
20625 * config/riscv/riscv-c.c: Likewise.
20626 * config/riscv/riscv-ftypes.def: Likewise.
20627 * config/riscv/riscv-modes.def: Likewise.
20628 * config/riscv/riscv-opts.h: Likewise.
20629 * config/riscv/riscv-protos.h: Likewise.
20630 * config/riscv/riscv.h: Likewise.
20631 * config/riscv/riscv.md: Likewise.
20632 * config/riscv/riscv.opt: Likewise.
20633 * config/riscv/sync.md: Likewise.
20634 * config/riscv/t-elf-multilib: Likewise.
20635 * config/riscv/t-linux: Likewise.
20636 * config/riscv/t-linux-multilib: Likewise.
20637 * config/riscv/t-riscv: Likewise.
20638 * configure.ac: Likewise.
20639 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20640 Waterman as RISC-V maintainers.
20641 * doc/install.texi: Add RISC-V entries.
20642 * doc/invoke.texi: Add RISC-V options section.
20643 * doc/md.texi: Add RISC-V constraints section.
20644 * configure: Regenerated.
20646 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20649 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20650 false values to be constant vectors with all 0 or all 1 bits set.
20651 (vcondu<mode><mode>): Likewise.
20652 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20654 (fpmask_comparison_operator): Update comment.
20655 (vecint_comparison_operator): New predicate.
20656 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20657 vector conditionals when the true and false values are constant
20658 vectors with all 0 bits or all 1 bits set.
20660 2017-02-06 Martin Sebor <msebor@redhat.com>
20662 PR tree-optimization/79376
20663 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20665 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20667 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20668 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20669 to simplify split condition.
20671 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20673 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20674 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20677 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20679 PR rtl-optimization/68664
20680 * target.def (can_speculate_insn): New hook.
20681 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20682 * doc/tm.texi: Regenerate.
20683 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20684 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20685 (rs6000_sched_can_speculate_insn): New function.
20687 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20689 PR tree-optimization/79284
20690 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20691 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20692 vectorizable_mask_load_store, vectorizable_operation,
20693 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20694 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20695 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20696 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20697 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20698 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20699 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20700 is_gimple_assign (stmt). Replace another such test with
20701 is_gimple_assign (stmt).
20703 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20706 * config/avr/avr.c (rtl-iter.h): Include it.
20707 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20708 (avr_legitimate_combined_insn): ...and implementation.
20710 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20712 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20713 * config/s390/s390.c (s390_const_operand_ok)
20714 (s390_canonicalize_comparison, s390_extract_part)
20715 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20716 (s390_contiguous_bitmask_p, s390_rtx_costs)
20717 (legitimize_pic_address): Likewise.
20718 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20719 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20720 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20721 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20722 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20724 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20726 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20727 REGNO($0) == REGNO($1).
20729 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20731 * config/s390/linux.h(SIZE_TYPE): Add comment.
20733 2017-02-06 Julian Brown <julian@codesourcery.com>
20734 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20735 Virendra Pathak <virendra.pathak@broadcom.com>
20737 * config/aarch64/aarch64-cores.def: Change the scheduler
20739 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20740 * config/aarch64/thunderx2t99.md: New file.
20742 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20744 * doc/standards.texi (Go Language): Update link to language
20747 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20749 * tree-eh.c (lower_resx): Sanitize profile.
20750 (cleanup_empty_eh_move_lp): Likewise.
20752 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20755 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20757 * cfgloopmanip.h (loop_version): Update prototype.
20758 * modulo-sched.c (sms_schedule): Update call of loop_version.
20759 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20760 * tree-parloops.c (gen_parallel_loop): Likewise.
20761 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20762 * tree-ssa-loop-split.c (split_loop): Likewise.
20763 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20764 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20766 2017-02-05 Martin Liska <mliska@suse.cz>
20769 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20771 (print_operand_address): Initialize a struct to zero.
20773 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20775 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20776 garbage collector only in textual form.
20778 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20780 * doc/extend.texi (x86 specific memory model extensions for
20781 transactional memory): Simplify a phrase.
20783 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20786 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20787 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20788 (atomic_storedi_1): Likewise.
20790 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20792 PR tree-optimization/79338
20793 * tree-parloops.c (gather_scalar_reductions): Don't call
20794 vect_analyze_loop_form for loop->inner before destroying loop's
20797 2017-02-03 Martin Sebor <msebor@redhat.com>
20799 PR tree-optimization/79327
20800 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20801 when precision has resulted in leading zeros.
20802 (format_integer): Adjust the likely counter to assume an unknown
20803 argument that may be zero is non-zero.
20805 2017-02-03 Jason Merrill <jason@redhat.com>
20808 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20809 avoid copying non-taken branch.
20811 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20813 PR tree-optimization/79340
20814 * tree-vect-loop.c (vectorizable_reduction): Release
20815 vec_defs elements after safe_splicing them into other vectors.
20818 PR tree-optimization/79327
20819 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20820 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20822 (format_integer): Use wide_int_to_tree instead of build_int_cst
20823 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20824 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20825 of shortest and longest sequence.
20827 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20829 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20830 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20832 2017-02-03 Walter Lee <walt@tilera.com>
20835 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20836 after initial stackframe link reg save.
20837 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20839 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20842 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20843 wu for stxssp alternative.
20845 2017-02-03 Martin Sebor <msebor@redhat.com>
20847 PR tree-optimization/79352
20848 * gimple-fold.c (get_range_strlen): Add argument.
20849 (get_range_strlen): Change return type to bool.
20850 (get_maxval_strlen): Pass in a dummy argument.
20851 * gimple-fold.h (get_range_strlen): Change return type to bool.
20852 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20853 * tree.h (array_at_struct_end_p): Add argument.
20854 * tree.c (array_at_struct_end_p): Handle it.
20856 2017-02-03 Martin Liska <mliska@suse.cz>
20859 * multiple_target.c (create_dispatcher_calls): Redirect edge
20860 from a caller of a dispatcher.
20861 (expand_target_clones): Make the clones local.
20862 (ipa_target_clone): Do both target clones and resolvers.
20863 (ipa_dispatcher_calls): Remove the pass.
20864 (pass_dispatcher_calls::gate): Likewise.
20865 (make_pass_dispatcher_calls): Likewise.
20866 * passes.def (pass_target_clone): Put as very first IPA early
20869 2017-02-03 Martin Liska <mliska@suse.cz>
20871 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20872 in case of a function with ifunc attribute.
20874 2017-02-03 Martin Liska <mliska@suse.cz>
20876 * cgraph.c (cgraph_node::dump): Dump function version info.
20877 * symtab.c (symtab_node::dump_base): Add missing new line.
20879 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20881 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20882 (ifcombine_ifandif): Use it.
20884 2017-02-03 Martin Liska <mliska@suse.cz>
20886 * doc/invoke.texi: Document default value for
20887 use-after-scope-direct-emission-threshold.
20889 2017-02-03 Martin Liska <mliska@suse.cz>
20891 PR tree-optimization/79339
20892 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20893 (format_floating): Likewise.
20895 2017-02-03 Martin Liska <mliska@suse.cz>
20898 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20900 (ipa_node_params_t::remove): Likewise.
20901 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20902 initialization from removed ipa_node_params_t::insert.
20903 (ipa_node_params::~ipa_node_params): Move from removed
20904 ipa_node_params_t::release.
20905 * symbol-summary.h (symbol_summary::m_released): New member.
20906 Do not release a summary twice. Do not allow to call finalizer
20907 for types of a summary that live in GGC memory.
20909 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20911 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20914 2017-02-02 Martin Sebor <msebor@redhat.com>
20916 PR middle-end/79275
20917 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20918 (format_string): Tighten up the range of output for non-constant
20919 strings and correct the expected range for wide non-constant strings.
20921 2017-02-02 Martin Sebor <msebor@redhat.com>
20923 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20925 PR middle-end/32003
20926 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20928 (-fdump-tree-@var): Add to index and document how to come up
20929 with pass-specific option and dump file names.
20930 (-fdump-passes): Clarify where to look for output.
20932 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20934 PR middle-end/77445
20935 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20936 statistics of the analyzed path; allow threading for speed when
20937 any of BBs along the path are optimized for speed.
20939 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20941 PR middle-end/78468
20942 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20943 settings of the virtual registers.
20946 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20948 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20949 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20952 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20954 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20955 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20957 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20959 * config/s390/s390.md: Add missing comments with the expanded
20961 * config/s390/vector.md: Likewise.
20962 * config/s390/vx-builtins.md: Likewise.
20964 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20967 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20968 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20969 conditions on a single line.
20971 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20973 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20974 __S390_VX__ to __VX__.
20976 2017-02-01 Andrew Pinski <apinski@cavium.com>
20978 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20979 stmt_info to record_stmt_cost.
20980 (vect_get_known_peeling_cost): Pass stmt_info if known to
20982 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20983 cpu_vector_cost field into
20984 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20985 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20986 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20987 splitting of scalar_stmt_cost and vec_stmt_cost.
20988 (thunderx_vector_cost): Likewise.
20989 (cortexa57_vector_cost): LIkewise.
20990 (exynosm1_vector_cost): Likewise.
20991 (xgene1_vector_cost): Likewise.
20992 (thunderx2t99_vector_cost): Improve after the splitting of the two
20994 (aarch64_builtin_vectorization_cost): Update for the splitting of
20995 scalar_stmt_cost and vec_stmt_cost.
20997 2017-02-01 Torvald Riegel <triegel@redhat.com>
20998 Richard Henderson <rth@redhat.com>
21000 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
21001 conditional on existance of a fast atomic load.
21002 * optabs-query.c (can_atomic_load_p): New function.
21003 * optabs-query.h (can_atomic_load_p): Declare it.
21004 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
21005 no fast atomic load is available for the particular size of access.
21006 (expand_atomic_compare_and_swap): Likewise.
21007 (expand_atomic_load): Likewise.
21008 (expand_atomic_store): Likewise.
21009 (expand_atomic_fetch_op): Likewise.
21010 * testsuite/lib/target-supports.exp
21011 (check_effective_target_sync_int_128): Remove x86 because it provides
21012 no fast atomic load.
21013 (check_effective_target_sync_int_128_runtime): Likewise.
21015 2017-02-01 Richard Biener <rguenther@suse.de>
21017 * graphite.c: Include tree-vectorizer.h for find_loop_location.
21018 (graphite_transform_loops): Provide opt-info for optimized nests.
21019 * tree-parloop.c (parallelize_loops): Provide opt-info for
21020 parallelized loops.
21022 2017-02-01 Richard Biener <rguenther@suse.de>
21024 PR middle-end/79315
21025 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
21026 was not set before.
21028 2017-02-01 Richard Biener <rguenther@suse.de>
21030 PR tree-optimization/71824
21031 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21032 Verify the loops are valid in the merged SESE region.
21033 (scop_detection::can_represent_loop_1): Check analyzing the
21034 evolution of the number of iterations in the region succeeds.
21036 2017-01-31 Ian Lance Taylor <iant@golang.org>
21038 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
21039 REG_ARGS_SIZE note to 32-bit push insns and call insn.
21041 2017-01-31 David Malcolm <dmalcolm@redhat.com>
21043 PR preprocessor/79210
21044 * input.c (get_substring_ranges_for_loc): Replace line_width
21045 assertion with error-handling.
21047 2017-01-31 Richard Biener <rguenther@suse.de>
21049 PR tree-optimization/77318
21050 * graphite-sese-to-poly.c (extract_affine): Fix assert.
21051 (create_pw_aff_from_tree): Take loop parameter.
21052 (add_condition_to_pbb): Pass loop of the condition to
21053 create_pw_aff_from_tree.
21055 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21057 * config/s390/s390.c (s390_asan_shadow_offset): New function.
21058 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
21060 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21064 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21066 (convert_int_to_float128): Likewise.
21067 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21068 (convert_int_to_float128): Likewise.
21069 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21070 (UNSPEC_IEEE128_CONVERT): Likewise.
21071 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21072 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21073 Use local variables for IBM extended format.
21074 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21075 (fix_trunc<mode>si2_fprs): Likewise.
21076 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21077 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21078 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21079 to know that we can now have integers of all sizes in vector
21081 (fix<uns>_<mode>di2_hw): Likewise.
21082 (float<uns>_<mode>si2_hw): Likewise.
21083 (fix_<mode>si2_hw): Likewise.
21084 (fixuns_<mode>si2_hw): Likewise.
21085 (float<uns>_<mode>di2_hw): Likewise.
21086 (float_<mode>di2_hw): Likewise.
21087 (float_<mode>si2_hw): Likewise.
21088 (floatuns_<mode>di2_hw): Likewise.
21089 (floatuns_<mode>si2_hw): Likewise.
21090 (xscvqp<su>wz_<mode>): Delete, no longer used.
21091 (xscvqp<su>dz_<mode>): Likewise.
21092 (xscv<su>dqp_<mode>): Likewise.
21093 (ieee128_mfvsrd_64bit): Likewise.
21094 (ieee128_mfvsrd_32bit): Likewise.
21095 (ieee128_mfvsrwz): Likewise.
21096 (ieee128_mtvsrw): Likewise.
21097 (ieee128_mtvsrd_64bit): Likewise.
21098 (ieee128_mtvsrd_32bit): Likewise.
21100 2017-01-31 Martin Liska <mliska@suse.cz>
21103 * ipa-prop.c (ipa_free_all_node_params): Call release method
21104 instead of ~sumbol_summary to not to trigger double times
21107 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21109 PR tree-optimization/71691
21110 * bitmap.h (class auto_bitmap): New.
21111 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21112 is_maybe_undefined instead of ssa_undefined_value_p.
21114 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21116 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21117 __S390_ARCH_LEVEL__ to __ARCH__.
21119 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21121 PR tree-optimization/79267
21122 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21123 if should_remove_lhs_p is true.
21125 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21128 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21129 (add_alignment_attribute): New.
21130 (base_type_die): Add alignment attribute.
21131 (subrange_type_die): Likewise.
21132 (modified_type_die): Likewise.
21133 (gen_array_type_die): Likewise.
21134 (gen_descr_array_type_die: Likewise.
21135 (gen_enumeration_type_die): Likewise.
21136 (gen_subprogram_die): Likewise.
21137 (gen_variable_die): Likewise.
21138 (gen_field_die): Likewise.
21139 (gen_ptr_to_mbr_type_die): Likewise.
21140 (gen_struct_or_union_type_die): Likewise.
21141 (gen_subroutine_type_die): Likewise.
21142 (gen_typedef_die): Likewise.
21143 (base_type_cmp): Compare alignment attribute.
21145 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21148 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21149 (setb_unsigned) New pattern for setb with CCUNS.
21150 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21151 subfc./subfe sequence to avoid overflow problems. Generate a
21152 shorter sequence with cmpld/setb for power9.
21153 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21154 for generating subfc. instruction.
21155 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21156 now uses this instruction.
21158 2017-01-30 Ian Lance Taylor <iant@google.com>
21161 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21162 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21164 2017-01-30 Martin Sebor <msebor@redhat.com>
21166 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21167 Move constant to the right of a relational operator.
21168 (get_mpfr_format_length, format_character, format_string): Ditto.
21169 (should_warn_p, maybe_warn): Same.
21171 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21173 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21176 * asan.c (get_translation_unit_decl): Remove function.
21177 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21179 2017-01-30 Martin Liska <mliska@suse.cz>
21181 PR gcov-profile/79259
21182 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21183 -fprofile-generate.
21185 2017-01-30 Martin Liska <mliska@suse.cz>
21188 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21189 Initialize variables with NULL value.
21191 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21194 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21196 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21198 2017-01-30 Richard Biener <rguenther@suse.de>
21200 PR tree-optimization/79276
21201 * tree-vrp.c (process_assert_insertions): Properly adjust common
21202 when removing a duplicate.
21204 * gcc.dg/torture/pr79276.c: New testcase.
21206 2017-01-30 Richard Biener <rguenther@suse.de>
21208 PR tree-optimization/79256
21209 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21210 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21212 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21214 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21217 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21218 ("*r<noxa>sbg_<mode>_sll_bitmask")
21219 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21220 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21221 Use contiguous_bitmask_nowrap_operand.
21223 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21226 * config/rs6000/altivec.h (vec_xl): Revise #define.
21227 (vec_xst): Likewise.
21229 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21231 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21233 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21235 PR rtl-optimization/79194
21236 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21237 traps before call to bypass_conditional_jumps.
21239 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21241 PR tree-optimization/71374
21242 * lra-constraints.c (check_conflict_input_operands): New.
21243 (match_reload): Use it.
21245 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21248 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21249 account to calculate conflict_set.
21251 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21253 PR rtl-optimization/78559
21254 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21255 other_insn in combine.
21257 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21259 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21260 uint16_type_node for BT_UINT16.
21262 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21264 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21265 "RTL Tests" to menu.
21266 (GIMPLE Tests): New node.
21267 (RTL Tests): New node.
21269 2017-01-27 Richard Biener <rguenther@suse.de>
21271 PR tree-optimization/79245
21272 * tree-loop-distribution.c (distribute_loop): Apply cost
21273 modeling also to detected patterns.
21275 2017-01-27 Richard Biener <rguenther@suse.de>
21277 PR tree-optimization/71433
21278 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21279 (compare_assert_loc): New function.
21280 (process_assert_insertions): Sort and optimize assert locations
21281 to remove duplicates and push down identical assertions on
21282 edges to their destination block.
21284 2017-01-27 Richard Biener <rguenther@suse.de>
21286 PR tree-optimization/79244
21287 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21288 out SSA names even if abnormal.
21290 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21292 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21293 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21294 instead of MPFR_RNDN.
21296 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21299 * arm.c (arm_option_override): Don't call build_target_option_node
21300 until after doing all option overrides.
21301 (arm_valid_target_attribute_tree): Likewise.
21303 2017-01-27 Martin Liska <mliska@suse.cz>
21305 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21306 for {cd}tors and C++ {cd}tors.
21308 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21310 * config/s390/s390.md ("*setmem_long_and")
21311 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21313 2017-01-26 Martin Sebor <msebor@redhat.com>
21315 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21318 2017-01-26 Martin Sebor <msebor@redhat.com>
21320 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21321 HAVE_DFmode before using XFmode or DFmode.
21322 (parse_directive): Avoid using the z length modifier to avoid
21323 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21325 PR middle-end/78703
21326 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21327 to accept adjustment as an array.
21328 (get_int_range): New function.
21329 (struct directive): Make width and prec arrays.
21330 (directive::set_width, directive::set_precision): Call get_int_range.
21331 (format_integer, format_floating): Handle width and precision ranges.
21332 (format_string, parse_directive): Same.
21334 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21337 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21338 comdat_type_p set, just clone them, but keep the children in the
21342 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21343 which have direct callers with -fvar-tracking-assignments enabled
21345 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21346 inside of type units.
21348 2017-01-26 Martin Sebor <msebor@redhat.com>
21350 PR middle-end/78703
21351 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21353 (struct format_result): Replace number_chars, number_chars_min,
21354 and number_chars_max with a single member of struct result_range.
21356 (format_result::operator+=): Adjust.
21357 (struct fmtresult): Remove bounded. Handle likely and unlikely
21359 (fmtresult::adjust_for_width_or_precision): New function.
21360 (fmtresult:type_max_digits): New function.
21361 (bytes_remaining): Handle likely and unlikely counters.
21362 (min_bytes_remaining): Remove.
21363 (format_percent): Simplify.
21364 (format_integer, format_floating): Set likely and unlikely counters.
21365 (get_string_length, format_character, format_string): Same.
21366 (format_plain, should_warn_p): New function.
21367 (maybe_warn): Call should_warn_p. Update diagnostic messages
21368 and handle those for all directives, including plain strings.
21369 (format_directive): Handle likely and unlikely counters.
21370 Remove unnecessary quoting from diagnostics. Add an informational
21372 (add_bytes): Remove.
21373 (pass_sprintf_length::compute_format_length): Simplify.
21374 (try_substitute_return_value): Handle likely and unlikely counters.
21376 2017-01-26 Carl Love <cel@us.ibm.com>
21378 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21379 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21381 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21384 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21385 endianess for subregs into account.
21386 * lra-constraints.c (lra_constraints): Do risky transformations
21387 always on the first iteration.
21388 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21389 last_call_used_reg_set.
21390 (process_bb_lives): Define and use last_call_used_reg_set.
21391 * lra.c (lra): Always continue after lra_constraints on the first
21394 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21396 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21398 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21400 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21402 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21403 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21404 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21405 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21406 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21407 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21408 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21409 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21410 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21412 2017-01-26 Marek Polacek <polacek@redhat.com>
21415 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21416 for the third operand.
21418 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21420 PR middle-end/79236
21421 * omp-low.c (struct omp_context): Add simt_stmt field.
21422 (scan_omp_for): Return omp_context *.
21423 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21424 context to the _simt_ SIMD stmt.
21425 (lower_omp_for): For combined SIMD with sibling _simt_
21426 SIMD, make sure to use the same decls in _looptemp_
21427 clauses as in the sibling.
21429 2017-01-26 David Sherwood <david.sherwood@arm.com>
21431 PR middle-end/79212
21432 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21435 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21438 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21439 emit fld b; fld a; if possible.
21441 * brig-builtins.def: Update copyright years.
21442 * config/arm/arm_acle_builtins.def: Update copyright years.
21444 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21447 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21448 constraint instead of o for the stxsd instruction.
21450 2017-01-25 Carl Love <cel@us.ibm.com>
21452 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21453 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21455 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21457 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21459 2017-01-25 Richard Biener <rguenther@suse.de>
21461 PR tree-optimization/69264
21462 * target.def (vector_alignment_reachable): Improve documentation.
21463 * doc/tm.texi: Regenerate.
21464 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21466 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21467 earlier changes with respect to TYPE_USER_ALIGN.
21468 (vector_alignment_reachable_p): Likewise. Improve dumping.
21470 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21473 * config/arm/arm.md (xordi3): Force constant operand into a register
21476 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21478 * doc/invoke.texi (-fstore-merging): Correct default optimization
21479 levels at which it is enabled.
21480 (-O): Move -fstore-merging from list to...
21483 2017-01-25 Richard Biener <rguenther@suse.de>
21486 * omp-expand.c: Include debug.h.
21487 (expand_omp_taskreg): Make sure to generate early debug before
21488 outlining anything from a function.
21489 (expand_omp_target): Likewise.
21490 (grid_expand_target_grid_body): Likewise.
21492 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21495 * asan.c (get_translation_unit_decl): New function.
21496 (asan_add_global): Extract modules file name from globals
21497 TRANSLATION_UNIT_DECL name.
21499 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21502 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21503 for long calls with APCS frame and VFP.
21505 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21507 * cfg.c (original_copy_tables_initialized_p): New function.
21508 * cfg.h (original_copy_tables_initialized_p): New decl.
21509 * cfgrtl.c (relink_block_chain): Guard the call to
21510 free_original_copy_tables with a call to
21511 original_copy_tables_initialized_p.
21512 * cgraph.h (symtab_node::native_rtl_p): New decl.
21513 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21514 (symtab_node::needed_p): Don't assert for early assembly output
21515 for __RTL functions.
21516 (cgraph_node::finalize_function): Set "force_output" for __RTL
21518 (cgraph_node::analyze): Bail out early for __RTL functions.
21519 (analyze_functions): Update assertion to support __RTL functions.
21520 (cgraph_node::expand): Bail out early for __RTL functions.
21521 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21523 * function.h (struct function): Update comment for field
21525 * gimple-expr.c: Include "tree-pass.h".
21526 (gimple_has_body_p): Return false for __RTL functions.
21527 * Makefile.in (OBJS): Add run-rtl-passes.o.
21528 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21530 (gcc::pass_manager::get_clean_slate): New accessor.
21531 * passes.c: Include "insn-addr.h".
21532 (should_skip_pass_p): Add logging. Update logic for running
21533 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21534 property-provider override so it is only done for gimple passes.
21536 (skip_pass): New function.
21537 (execute_one_pass): Call skip_pass when skipping passes.
21538 * read-md.c (md_reader::read_char): Support filtering
21539 the input to a subset of line numbers.
21540 (md_reader::md_reader): Initialize fields
21541 m_first_line and m_last_line.
21542 (md_reader::read_file_fragment): New function.
21543 * read-md.h (md_reader::read_file_fragment): New decl.
21544 (md_reader::m_first_line): New field.
21545 (md_reader::m_last_line): New field.
21546 * read-rtl-function.c (function_reader::create_function): Only
21547 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21548 curr_properties. Set DECL_INITIAL to a dummy block.
21549 (read_rtl_function_body_from_file_range): New function.
21550 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21552 * run-rtl-passes.c: New file.
21553 * run-rtl-passes.h: New file.
21555 2017-01-24 Jeff Law <law@redhat.com>
21557 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21560 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21562 PR tree-optimization/79159
21563 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21564 (record_nonwrapping_iv): Improve boundary using above function if no
21565 value range information.
21567 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21568 Martin Jambor <mjambor@suse.cz>
21570 * brig-builtins.def: New file.
21571 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21572 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21573 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21574 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21575 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21576 * builtin-types.def (BT_INT8): New.
21577 (BT_INT16): Likewise.
21578 (BT_UINT8): Likewise.
21579 (BT_UINT16): Likewise.
21580 (BT_FN_ULONG): Likewise.
21581 (BT_FN_UINT_INT): Likewise.
21582 (BT_FN_UINT_ULONG): Likewise.
21583 (BT_FN_UINT_LONG): Likewise.
21584 (BT_FN_UINT_PTR): Likewise.
21585 (BT_FN_ULONG_PTR): Likewise.
21586 (BT_FN_INT8_FLOAT): Likewise.
21587 (BT_FN_INT16_FLOAT): Likewise.
21588 (BT_FN_UINT32_FLOAT): Likewise.
21589 (BT_FN_UINT16_FLOAT): Likewise.
21590 (BT_FN_UINT8_FLOAT): Likewise.
21591 (BT_FN_UINT64_FLOAT): Likewise.
21592 (BT_FN_UINT16_UINT32): Likewise.
21593 (BT_FN_UINT32_UINT16): Likewise.
21594 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21595 (BT_FN_INT_PTR_INT): Likewise.
21596 (BT_FN_UINT_PTR_UINT): Likewise.
21597 (BT_FN_LONG_PTR_LONG): Likewise.
21598 (BT_FN_ULONG_PTR_ULONG): Likewise.
21599 (BT_FN_VOID_UINT64_UINT64): Likewise.
21600 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21601 (BT_FN_INT8_INT8_INT8): Likewise.
21602 (BT_FN_INT16_INT16_INT16): Likewise.
21603 (BT_FN_INT_INT_INT): Likewise.
21604 (BT_FN_UINT_FLOAT_UINT): Likewise.
21605 (BT_FN_FLOAT_UINT_UINT): Likewise.
21606 (BT_FN_ULONG_UINT_UINT): Likewise.
21607 (BT_FN_ULONG_UINT_PTR): Likewise.
21608 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21609 (BT_FN_UINT_UINT_UINT): Likewise.
21610 (BT_FN_VOID_UINT_PTR): Likewise.
21611 (BT_FN_UINT_UINT_PTR: Likewise.
21612 (BT_FN_UINT32_UINT64_PTR): Likewise.
21613 (BT_FN_INT_INT_UINT_UINT): Likewise.
21614 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21615 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21616 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21617 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21618 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21619 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21620 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21621 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21622 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21623 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21624 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21625 * doc/frontends.texi: List BRIG FE.
21626 * doc/install.texi (Testing): Add BRIG tesring requirements.
21627 * doc/invoke.texi (Overall Options): Mention BRIG.
21628 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21630 2017-01-24 Richard Biener <rguenther@suse.de>
21632 PR translation/79208
21633 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21635 2017-01-24 Martin Jambor <mjambor@suse.cz>
21638 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21639 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21640 and known_contexts.
21642 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21644 PR middle-end/79123
21645 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21646 casts from signed to unsigned really don't have a range.
21648 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21650 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21651 GMP_RNDx for compatiblity.
21653 2017-01-24 Martin Liska <mliska@suse.cz>
21656 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21657 that would prevent us to call alloca with -1 as argument.
21659 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21661 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21662 Avoid -Wformat-security warning.
21664 2017-01-23 Andrew Pinski <apinski@cavium.com>
21666 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21669 2017-01-23 Martin Sebor <msebor@redhat.com>
21671 PR middle-end/78703
21672 * gimple-ssa-sprintf.c (warn_level): New global.
21673 (format_integer): Use it here and throughout the rest of the file.
21674 Use the same switch to compute sign as base.
21675 (maybe_warn): New function.
21676 (format_directive): Factor out warnings into maybe_warn.
21677 Add debugging output. Use warn_level.
21678 (add_bytes): Use warn_level.
21679 (pass_sprintf_length::compute_format_length): Add debugging output.
21680 (try_substitute_return_value): Same.
21681 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21683 PR middle-end/78703
21684 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21685 (struct fmtresult, format_integer, format_floating): Adjust.
21686 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21687 (get_string_length, format_string,format_directive): Same.
21688 (pass_sprintf_length::compute_format_length): Same.
21689 (try_substitute_return_value): Simplify slightly.
21691 PR middle-end/78703
21692 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21693 (fmtresult::operator+=): Outlined.
21694 (struct fmtresult): Add ctors.
21695 (struct conversion_spec): Rename...
21696 (struct directive): ...to this. Add and remove data members.
21697 (directive::set_width, directive::set_precision): New functions.
21698 (format_percent): Use fmtresult ctor.
21699 (get_width_and_precision): Remove.
21700 (format_integer): Make naming changes. Avoid computing width and
21702 (format_floating): Same. Adjust indentation.
21703 (format_character, format_none): New functions.
21704 (format_string): Moved character handling to format_character.
21705 (format_directive): Remove arguments, change return type.
21706 (parse_directive): New function.
21707 (pass_sprintf_length::compute_format_length): Move directive
21708 parsing to parse_directive.
21710 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21712 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21713 (assign_assembler_name_if_needed): ... this.
21714 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21715 (assign_assembler_name_if_needed): ... this.
21716 (free_lang_data_in_cgraph): Adjust callers.
21717 * cgraphunit.c (cgraph_node::analyze): Likewise.
21718 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21721 2017-01-23 Richard Biener <rguenther@suse.de>
21723 PR tree-optimization/79088
21724 PR tree-optimization/79188
21725 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21726 resetting loop bounds after last path deletion. Reset loop
21727 bounds of the target loop, make code match the comments.
21728 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21729 Make sure loops need no fixups.
21731 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21733 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21734 exponent support with double type for first argument.
21735 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21736 type returned by __builtin_vec_extract_sig,
21737 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21738 functions from "vector int" to "vector unsigned int" or from
21739 "vector long long int" to "vector unsigned long long int".
21740 Changed type returned by __builtin_vec_extract_exp,
21741 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21742 functions from "vector int" to "vector unsigned int" or from
21743 "vector long long int" to "vector unsigned long long int".
21744 Changed return type of __builtin_vec_test_data_class,
21745 __builtin_vec_test_data_class_sp, and
21746 __builtin_vec_test_data_class_dp from "vector int" to
21747 "vector bool int" or from "vector long long int" to "vector bool
21748 long long int" and changed second argument type from "unsigned
21749 int" to "int". Added new overloaded function forms "vector float
21750 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21751 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21752 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21753 double, vector unsigned long long int)" and "vector double
21754 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21755 long int)". Changed return type of
21756 __builtin_scalar_test_data_class and
21757 __builtin_scalar_test_data_class_sp and
21758 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21759 int" and changed second argument from "unsigned int" to "int".
21760 Changed type returned by __builtin_scalar_test_neg,
21761 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21762 from "int" to "bool int". Added new overloaded function form
21763 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21764 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21765 exponent double-precision with floating point first argument.
21766 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21767 documentation of scalar_test_data_class, scalar_test_neg,
21768 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21769 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21770 vec_test_data_class built-in functions to reflect refinements in
21771 their type signatures.
21773 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21775 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21777 (aarch64_elf_asm_destructor): Likewise.
21779 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21781 PR rtl-optimization/78634
21782 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21783 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21784 * ifcvt.c (noce_try_cmove): Add missing cost check.
21786 PR rtl-optimization/71724
21787 * combine.c (if_then_else_cond): Look for situations where it is
21788 beneficial to undo the work of one of the recursive calls.
21790 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21792 PR tree-optimization/70754
21793 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21794 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21795 combined stmt before it if not NULL.
21796 (combine_chains): Process refs reversely and compute dominance point
21799 2017-01-23 Martin Liska <mliska@suse.cz>
21801 PR tree-optimization/79196
21802 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21803 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21805 (strlen_optimize_stmt): Call the renamed function.
21807 2017-01-23 Michael Matz <matz@suse.de>
21809 PR tree-optimization/78384
21810 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21812 2017-01-23 Richard Biener <rguenther@suse.de>
21814 PR tree-optimization/79186
21815 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21816 both incoming edges before moving an assert.
21818 2017-01-23 Martin Jambor <mjambor@suse.cz>
21820 * ipa-prop.c (load_from_param_1): Removed.
21821 (load_from_unmodified_param): Bits from load_from_param_1 put back
21823 (load_from_param): Removed.
21824 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21825 with stmt. Reverted back to use of load_from_unmodified_param.
21827 2017-01-23 Martin Jambor <mjambor@suse.cz>
21830 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21831 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21832 field a pointer to garbage collected vector, mark lattices and
21833 ipcp_orig_node with GTY((skip)).
21834 (ipa_get_param_count): Adjust to descriptors being a pointer.
21835 (ipa_get_param): Likewise.
21836 (ipa_get_type): Likewise.
21837 (ipa_get_param_move_cost): Likewise.
21838 (ipa_set_param_used): Likewise.
21839 (ipa_get_controlled_uses): Likewise.
21840 (ipa_set_controlled_uses): Likewise.
21841 (ipa_is_param_used): Likewise.
21842 (ipa_node_params_t): Move into garbage collector. New methods insert
21844 (ipa_node_params_sum): Annotate wth GTY(()).
21845 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21847 (ipa_load_from_parm_agg): Adjust declaration.
21848 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21849 * ipa-profile.c (ipa_profile): Likewise.
21850 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21851 (ipa_populate_param_decls): Make descriptors parameter garbage
21853 (ipa_dump_param): Adjust to descriptors being a pointer.
21854 (ipa_alloc_node_params): Likewise.
21855 (ipa_initialize_node_params): Likewise.
21856 (load_from_param_1): Make descriptors parameter garbage collected.
21857 (load_from_unmodified_param): Likewise.
21858 (load_from_param): Likewise.
21859 (ipa_load_from_parm_agg): Likewise.
21860 (ipa_node_params::~ipa_node_params): Removed.
21861 (ipa_free_all_node_params): Remove call to delete operator.
21862 (ipa_node_params_t::insert): New.
21863 (ipa_node_params_t::remove): Likewise.
21864 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21865 copy known_csts and known_contexts vectors.
21866 (ipa_read_node_info): Adjust to descriptors being a pointer.
21867 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21869 (ipcp_transform_function): Make descriptors variable garbage
21872 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21874 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21875 * config/i386/avx512dqintrin.h: Ditto.
21876 * config/i386/avx512fintrin.h: Ditto.
21877 * config/i386/i386.c: Handle new builtins.
21878 * config/i386/i386-builtin.def: Add new builtins.
21879 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21880 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21882 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21883 Martin Liska <mliska@suse.cz>
21885 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21886 * asan.c (asan_expand_poison_ifn): Support stores and use
21887 appropriate ASAN report function.
21888 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21889 * internal-fn.def (ASAN_POISON_USE): Declare.
21890 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21891 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21892 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21893 ASAN_POISON calls w/o LHS.
21894 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21895 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21896 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21897 * gimplify.c (asan_poison_variables): Add attribute
21898 use_after_scope_memory to variables that really needs to live
21900 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21901 having the attribute.
21903 2017-01-23 Martin Liska <mliska@suse.cz>
21905 * asan.c (create_asan_shadow_var): New function.
21906 (asan_expand_poison_ifn): Likewise.
21907 * asan.h (asan_expand_poison_ifn): New declaration.
21908 * internal-fn.c (expand_ASAN_POISON): Likewise.
21909 * internal-fn.def (ASAN_POISON): New builtin.
21910 * sanopt.c (pass_sanopt::execute): Expand
21911 asan_expand_poison_ifn.
21912 * tree-inline.c (copy_decl_for_dup_finish): Make function
21914 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21915 * tree-ssa.c (is_asan_mark_p): New function.
21916 (execute_update_addresses_taken): Rewrite local variables
21917 (identified just by use-after-scope as addressable) into SSA.
21919 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21921 * doc/install.texi (Specific): opensource.apple.com uses https
21922 now. Remove trailing slash.
21924 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21926 * README.Portability: Remove note on an Irix compatibility issue.
21928 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21930 * gcov.c (INCLUDE_ALGORITHM): Define.
21931 (INCLUDE_VECTOR): Define.
21932 No longer include <vector> and <algorithm> directly.
21934 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21936 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21938 * doc/invoke.texi (Code Gen Options): Ditto.
21940 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21943 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21945 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21947 rtl-optimization/79125
21948 * cprop.c (local_cprop_pass): Handle cases where we make an
21949 unconditional trap.
21951 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21955 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21956 read from, for big endian.
21958 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21960 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21961 register pauth builtins for LP64 only.
21963 2017-01-20 Marek Polacek <polacek@redhat.com>
21966 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21969 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21971 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21973 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21974 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21975 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21977 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21980 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21981 in big-endian mode when they are not a single duplicated value.
21983 2017-01-20 Richard Biener <rguenther@suse.de>
21985 * BASE-VER: Bump to 7.0.1.
21987 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21989 * omp-low.c (omplow_simd_context): New struct. Use it...
21990 (lower_rec_simd_input_clauses): ...here and...
21991 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21992 references to idx, lane, max_vf, is_simt.
21994 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21996 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21999 2017-01-20 Martin Jambor <mjambor@suse.cz>
22001 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
22002 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
22004 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
22005 (GTFILES): Rename hsa.c to hsa-common.c.
22006 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
22007 * hsa-dump.c: Likewise.
22008 * hsa-gen.c: Likewise.
22009 * hsa-regalloc.c: Likewise.
22010 * ipa-hsa.c: Likewise.
22011 * omp-expand.c: Likewise.
22012 * omp-low.c: Likewise.
22013 * toplev.c: Likewise.
22015 2017-01-20 Marek Polacek <polacek@redhat.com>
22018 * doc/invoke.texi: Document -Wduplicated-branches.
22019 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
22020 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
22021 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
22022 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
22023 return 0 only when not OEP_LEXICOGRAPHIC.
22024 (fold_build_cleanup_point_expr): Use the expression
22025 location when building CLEANUP_POINT_EXPR.
22026 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
22027 * tree.c (add_expr): Handle error_mark_node.
22029 2017-01-20 Martin Liska <mliska@suse.cz>
22032 * tree-profile.c (init_ic_make_global_vars): Do not call
22034 (gimple_init_gcov_profiler): Likewise.
22036 2017-01-20 Martin Liska <mliska@suse.cz>
22039 * cgraph.h (maybe_create_reference): Remove argument and
22041 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
22043 * ipa-cp.c (create_specialized_node): Likewise.
22044 * symtab.c (symtab_node::maybe_create_reference): Handle
22045 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
22047 2017-01-20 Martin Liska <mliska@suse.cz>
22049 * read-rtl-function.c (function_reader::create_function): Use
22050 build_decl instread of build_decl_stat.
22052 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
22054 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
22055 * config/i386/avx512dqintrin.h: Ditto.
22056 * config/i386/avx512fintrin.h: Ditto.
22057 * config/i386/i386-builtin-types.def: Add new types.
22058 * config/i386/i386.c: Handle new types.
22059 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
22060 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
22061 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
22062 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
22063 (__builtin_ia32_kshiftridi): New.
22064 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22066 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22070 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22071 define to rs6000_init_stack_protect_guard.
22072 (rs6000_init_stack_protect_guard): New function.
22074 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22075 Yunqiang Su <yunqiang.su@imgtec.com>
22077 * config.gcc (supported_defaults): Add madd4.
22078 (with_madd4): Add validation.
22079 (all_defaults): Add madd4.
22080 * config/mips/mips.opt (mmadd4): New option.
22081 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22083 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22085 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22086 (ISA_HAS_FUSED_MADD4): Likewise.
22087 * doc/invoke.texi (-mmadd4): Document the new option.
22088 * doc/install.texi (--with-madd4): Document the new option.
22090 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22092 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22093 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22094 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22095 (aarch64_init_pauth_hint_builtins): New.
22096 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22097 (aarch64_expand_builtin): Expand new builtins.
22099 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22101 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22102 * combine-stack-adj.c (no_unhandled_cfa): Handle
22103 REG_CFA_TOGGLE_RA_MANGLE.
22104 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22105 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22106 info for return address signing.
22107 (aarch64_expand_epilogue): Likewise.
22109 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22111 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22112 * config/aarch64/aarch64-protos.h
22113 (aarch64_return_address_signing_enabled): New declaration.
22114 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22116 (aarch64_expand_prologue): Sign return address before it's pushed onto
22118 (aarch64_expand_epilogue): Authenticate return address fetched from
22120 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22121 (aarch64_attributes): New function attributes for "sign-return-address".
22122 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22123 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22124 ("*do_return"): Generate combined instructions according to key index.
22125 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22126 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22128 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22129 * config/aarch64/aarch64.opt (msign-return-address=): New.
22130 * doc/extend.texi (AArch64 Function Attributes): Documents
22131 "sign-return-address=".
22132 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22134 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22136 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22137 overall option summary.
22139 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22141 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22142 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22143 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22144 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22146 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22148 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22149 -mpower9-minmax by default for -mcpu=power9.
22150 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22151 128-bit floating point.
22153 2017-01-20 Alan Modra <amodra@gmail.com>
22155 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22156 optimizing for size.
22158 2017-01-20 Alan Modra <amodra@gmail.com>
22161 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22162 for strcmp and strncmp from corresponding builtin decl.
22164 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22166 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22167 instead of i386/rtems-64.h.
22168 * config/i386/rtems-64.h: Remove.
22170 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22174 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22176 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22178 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22180 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22181 Change int to HOST_WIDE_INT.
22182 * config/aarch64/aarch64-protos.h
22183 (aarch64_simd_gen_const_vector_dup): Likewise.
22184 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22186 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22188 * langhooks-def.h (lhd_type_for_size): New decl.
22189 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22190 * langhooks.c (lhd_type_for_size): New function, taken from
22193 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22195 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22196 define_bypass for CR latency.
22197 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22198 (power9-alu2): Add define_bypass for CR latency.
22200 (power9-mul): Update insn latency.
22201 (power9-mul-compare): Update insn latency, bypass latency and remove
22204 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22206 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22208 * config/aarch64/aarch64.md
22209 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22210 aarch64_nopcrelative_literal_loads.
22211 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22213 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22215 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22216 TARGET_LOONGSON_3A.
22217 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22219 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22222 * config.gcc (supported_defaults): Add lxc1-sxc1.
22223 (with_lxc1_sxc1): Add validation.
22224 (all_defaults): Add lxc1-sxc1.
22225 * config/mips/mips.opt (mlxc1-sxc1): New option.
22226 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22228 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22229 __mips_no_lxc1_sxc1.
22230 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22231 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22232 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22234 2017-01-19 Richard Biener <rguenther@suse.de>
22236 PR tree-optimization/72488
22237 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22238 sure to restore SSA info.
22239 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22241 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22243 PR rtl-optimization/79121
22244 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22245 of the inner type when shifting an extended value.
22247 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22250 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22251 interposable aliases.
22253 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22256 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22257 Use the evmergelohi instruction.
22258 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22259 (mov_sitf_e500_subreg8_2_be): Likewise.
22260 (mov_sitf_e500_subreg12_2_le): Likewise.
22261 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22262 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22263 (mov_sitf_e500_subreg8_2_le): Likewise.
22264 (mov_sitf_e500_subreg12_2_be): Likewise.
22266 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22268 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22269 attribute from vecsimple to vecperm.
22270 (altivec_vbpermq2): Likewise.
22272 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22275 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22277 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22278 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22279 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22280 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22281 case where N arg is SIZE_MAX.
22282 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22283 (cmpstrsi): Add pattern.
22285 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22287 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22288 __builtin_vec_revb builtins.
22289 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22290 built-in functions to support generation of the ISA 3.0 XXBR<x>
22291 vector byte reverse instructions.
22292 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22293 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22294 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22295 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22296 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22297 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22298 (P9V_BUILTIN_VEC_REVB): Likewise.
22299 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22300 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22301 (p9_xxbrq_v16qi): Likewise.
22302 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22303 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22304 (p9_xxbrh_v8hi): Likewise.
22305 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22306 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22307 vec_revb built-in functions.
22309 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22311 PR rtl-optimization/78952
22312 * config/i386/i386.md (any_extract): New code iterator.
22313 (*insvqi_2): Use any_extract for source operand.
22314 (*insvqi_3): Use any_shiftrt for source operand.
22316 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22318 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22320 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22322 2017-01-18 Matthias Klose <doko@ubuntu.com>
22324 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22326 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22328 * config/rs6000/altivec.h (vec_bperm): Change #define.
22329 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22330 (altivec_vbpermq2): New define_insn.
22331 (altivec_vbpermd): Likewise.
22332 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22333 function interface.
22334 (VBPERMD): Likewise.
22335 (VBPERM): New polymorphic function interface.
22336 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22337 Add entries for P9V_BUILTIN_VEC_VBPERM.
22338 * doc/extend.texi: Add interfaces for vec_bperm.
22340 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22342 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22343 first letter of error messages.
22344 (s390_resolve_overloaded_builtin): Likewise.
22345 * config/s390/s390.c (s390_expand_builtin): Likewise.
22346 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22347 (s390_valid_target_attribute_inner_p): Likewise.
22348 * config/s390/s390.md ("tabort"): Likewise.
22350 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22352 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22353 (ISA_AVOID_DIV_HILO): New macro.
22354 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22355 (ISA_HAS_DDIV): Likewise.
22357 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22359 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22361 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22363 * doc/invoke.texi (fabi-version): Spelling fix.
22365 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22368 * doc/invoke.texi (fabi-version): Mention mangling fix for
22371 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22374 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22376 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22379 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22380 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22382 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22384 (stack_protect_test): Ditto.
22385 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22386 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22388 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22389 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22390 -mstack-protector-guard-offset=.
22391 (RS/6000 and PowerPC Options): Ditto.
22393 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22395 * config/i386/i386.h (MASK_CLASS_P): New define.
22396 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22397 there are no registers from different register sets also when
22398 mask registers are used. Update function comment.
22399 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22400 to (*k/*r) and (*k/*km) alternatives.
22402 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22404 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22405 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22406 (EH_RETURN_HANDLER_RTX): New define.
22407 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22408 Force frame pointer in EH return functions.
22409 (aarch64_expand_epilogue): Add barrier for eh_return.
22410 (aarch64_final_eh_return_addr): Remove.
22411 (aarch64_eh_return_handler_rtx): New function.
22412 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22414 (aarch64_eh_return_handler_rtx): New prototype.
22416 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22418 * config/rs6000/altivec.h (vec_rlmi): New #define.
22419 (vec_vrlnm): Likewise.
22420 (vec_rlnm): Likewise.
22421 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22422 (UNSPEC_VRLNM): Likewise.
22423 (VIlong): New mode iterator.
22424 (altivec_vrl<VI_char>mi): New define_insn.
22425 (altivec_vrl<VI_char>nm): Likewise.
22426 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22428 (VRLDNM): Likewise.
22429 (RLNM): New polymorphic function entry.
22430 (VRLWMI): New monomorphic function entry.
22431 (VRLDMI): Likewise.
22432 (RLMI): New polymorphic function entry.
22433 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22434 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22435 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22438 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22441 * dwarf2out.c (field_byte_offset): Restore the
22442 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22443 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22444 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22447 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22450 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22452 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22455 * dwarf2out.c (add_data_member_location_attribute): For constant
22456 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22457 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22458 DW_AT_byte_size attributes.
22460 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22462 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22463 after forcing to constant memory when the code model is medium.
22465 2017-01-17 Julia Koval <julia.koval@intel.com>
22468 * config/i386/avx512fintrin.h
22469 (_mm512_i32gather_ps): Change __addr type to void const*.
22470 (_mm512_mask_i32gather_ps): Ditto.
22471 (_mm512_i32gather_pd): Ditto.
22472 (_mm512_mask_i32gather_pd): Ditto.
22473 (_mm512_i64gather_ps): Ditto.
22474 (_mm512_mask_i64gather_ps): Ditto.
22475 (_mm512_i64gather_pd): Ditto.
22476 (_mm512_mask_i64gather_pd): Ditto.
22477 (_mm512_i32gather_epi32): Ditto.
22478 (_mm512_mask_i32gather_epi32): Ditto.
22479 (_mm512_i32gather_epi64): Ditto.
22480 (_mm512_mask_i32gather_epi64): Ditto.
22481 (_mm512_i64gather_epi32): Ditto.
22482 (_mm512_mask_i64gather_epi32): Ditto.
22483 (_mm512_i64gather_epi64): Ditto.
22484 (_mm512_mask_i64gather_epi64): Ditto.
22485 (_mm512_i32scatter_ps): Change __addr type to void*.
22486 (_mm512_mask_i32scatter_ps): Ditto.
22487 (_mm512_i32scatter_pd): Ditto.
22488 (_mm512_mask_i32scatter_pd): Ditto.
22489 (_mm512_i64scatter_ps): Ditto.
22490 (_mm512_mask_i64scatter_ps): Ditto.
22491 (_mm512_i64scatter_pd): Ditto.
22492 (_mm512_mask_i64scatter_pd): Ditto.
22493 (_mm512_i32scatter_epi32): Ditto.
22494 (_mm512_mask_i32scatter_epi32): Ditto.
22495 (_mm512_i32scatter_epi64): Ditto.
22496 (_mm512_mask_i32scatter_epi64): Ditto.
22497 (_mm512_i64scatter_epi32): Ditto.
22498 (_mm512_mask_i64scatter_epi32): Ditto.
22499 (_mm512_i64scatter_epi64): Ditto.
22500 (_mm512_mask_i64scatter_epi64): Ditto.
22501 * config/i386/avx512pfintrin.h
22502 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22503 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22504 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22505 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22506 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22507 (_mm512_prefetch_i32scatter_ps): Ditto.
22508 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22509 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22510 (_mm512_prefetch_i64scatter_pd): Ditto.
22511 (_mm512_prefetch_i64scatter_ps): Ditto.
22512 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22513 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22514 * config/i386/avx512vlintrin.h
22515 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22516 (_mm_mmask_i32gather_ps): Ditto.
22517 (_mm256_mmask_i32gather_pd): Ditto.
22518 (_mm_mmask_i32gather_pd): Ditto.
22519 (_mm256_mmask_i64gather_ps): Ditto.
22520 (_mm_mmask_i64gather_ps): Ditto.
22521 (_mm256_mmask_i64gather_pd): Ditto.
22522 (_mm_mmask_i64gather_pd): Ditto.
22523 (_mm256_mmask_i32gather_epi32): Ditto.
22524 (_mm_mmask_i32gather_epi32): Ditto.
22525 (_mm256_mmask_i32gather_epi64): Ditto.
22526 (_mm_mmask_i32gather_epi64): Ditto.
22527 (_mm256_mmask_i64gather_epi32): Ditto.
22528 (_mm_mmask_i64gather_epi32): Ditto.
22529 (_mm256_mmask_i64gather_epi64): Ditto.
22530 (_mm_mmask_i64gather_epi64): Ditto.
22531 (_mm256_i32scatter_ps): Change __addr type to void*.
22532 (_mm256_mask_i32scatter_ps): Ditto.
22533 (_mm_i32scatter_ps): Ditto.
22534 (_mm_mask_i32scatter_ps): Ditto.
22535 (_mm256_i32scatter_pd): Ditto.
22536 (_mm256_mask_i32scatter_pd): Ditto.
22537 (_mm_i32scatter_pd): Ditto.
22538 (_mm_mask_i32scatter_pd): Ditto.
22539 (_mm256_i64scatter_ps): Ditto.
22540 (_mm256_mask_i64scatter_ps): Ditto.
22541 (_mm_i64scatter_ps): Ditto.
22542 (_mm_mask_i64scatter_ps): Ditto.
22543 (_mm256_i64scatter_pd): Ditto.
22544 (_mm256_mask_i64scatter_pd): Ditto.
22545 (_mm_i64scatter_pd): Ditto.
22546 (_mm_mask_i64scatter_pd): Ditto.
22547 (_mm256_i32scatter_epi32): Ditto.
22548 (_mm256_mask_i32scatter_epi32): Ditto.
22549 (_mm_i32scatter_epi32): Ditto.
22550 (_mm_mask_i32scatter_epi32): Ditto.
22551 (_mm256_i32scatter_epi64): Ditto.
22552 (_mm256_mask_i32scatter_epi64): Ditto.
22553 (_mm_i32scatter_epi64): Ditto.
22554 (_mm_mask_i32scatter_epi64): Ditto.
22555 (_mm256_i64scatter_epi32): Ditto.
22556 (_mm256_mask_i64scatter_epi32): Ditto.
22557 (_mm_i64scatter_epi32): Ditto.
22558 (_mm_mask_i64scatter_epi32): Ditto.
22559 (_mm256_i64scatter_epi64): Ditto.
22560 (_mm256_mask_i64scatter_epi64): Ditto.
22561 (_mm_i64scatter_epi64): Ditto.
22562 (_mm_mask_i64scatter_epi64): Ditto.
22563 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22564 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22565 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22566 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22567 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22568 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22569 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22570 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22571 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22572 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22573 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22574 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22575 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22576 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22577 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22578 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22579 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22580 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22581 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22582 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22583 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22584 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22585 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22586 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22587 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22588 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22589 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22590 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22591 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22592 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22593 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22594 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22595 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22596 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22597 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22598 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22599 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22600 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22601 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22602 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22603 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22604 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22605 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22606 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22607 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22608 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22609 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22610 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22611 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22612 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22613 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22614 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22615 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22616 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22617 definitions accordingly.
22619 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22620 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22623 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22626 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22629 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22630 hard regs for inner regno.
22632 2017-01-17 Martin Liska <mliska@suse.cz>
22635 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22636 assumption and add comment.
22638 2017-01-17 Nathan Sidwell <nathan@acm.org>
22640 * ipa-visibility.c (localize_node): New function, broken out of ...
22641 (function_and_variable_visibility): ... here. Call it.
22643 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22645 PR middle-end/77445
22646 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22647 correctly set frequency of oudgoing edge.
22648 (duplicate_thread_path): Fix profile updating.
22650 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22653 * configure.ac: Add GCC_BASE_VER.
22654 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22655 version from BASE-VER file.
22656 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22657 (gcc.o): Depend on $(BASEVER).
22658 * common.opt (dumpfullversion): New option.
22659 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22660 * doc/invoke.texi: Document -dumpfullversion.
22661 * doc/install.texi: Document --with-gcc-major-version-only.
22662 * configure: Regenerated.
22664 2017-01-17 Richard Biener <rguenther@suse.de>
22666 PR tree-optimization/71433
22667 * tree-vrp.c (register_new_assert_for): Merge same asserts
22668 on all incoming edges.
22669 (process_assert_insertions_for): Handle insertions at the
22672 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22674 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22675 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22677 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22680 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22683 2017-01-17 Alan Modra <amodra@gmail.com>
22686 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22687 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22688 symbolic stack limit when pic.
22690 2017-01-16 Martin Sebor <msebor@redhat.com>
22692 PR tree-optimization/78608
22693 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22695 2017-01-16 Jeff Law <law@redhat.com>
22698 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22699 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22700 for several include directories that may be relative to sysroot.
22701 * config/i386/x-mingw32 (gplus_includedir): Define.
22702 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22703 (native_system_includedir): Likewise.
22704 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22705 override if TARGET_SYSTEM_ROOT is defined.
22706 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22708 PR tree-optimization/79090
22709 PR tree-optimization/33562
22710 PR tree-optimization/61912
22711 PR tree-optimization/77485
22712 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22713 and computed trims into the dump file.
22715 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22717 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22719 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22722 * gimplify.c (gimplify_init_constructor): If want_value and
22723 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22727 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22728 sequence. Formatting fixes.
22729 (doloop_optimize): Formatting fixes.
22732 * gcc.c (debug_level_greater_than_spec_func): New function.
22733 (static_spec_functions): Add debug-level-gt spec function.
22734 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22736 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22737 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22738 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22739 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22740 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22741 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22743 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22745 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22746 QImode fixups to general and mask registers only.
22748 2017-01-16 Carl Love <cel@us.ibm.com>
22750 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22751 for built-in functions
22752 vector signed char vec_nabs (vector signed char)
22753 vector signed short vec_nabs (vector signed short)
22754 vector signed int vec_nabs (vector signed int)
22755 vector signed long long vec_nabs (vector signed long long)
22756 vector float vec_nabs (vector float)
22757 vector double vec_nabs (vector double)
22758 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22760 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22761 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22762 * doc/extend.texi: Update the documentation file for the new built-in
22765 2017-01-16 Martin Sebor <msebor@redhat.com>
22767 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22770 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22772 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22773 UNSPEC_VSX__XXSPLTD to require special splat handling.
22775 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22778 * system.h: Poison strndup.
22780 2017-01-16 Alan Modra <amodra@gmail.com>
22783 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22786 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22788 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22790 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22792 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22793 call recog here. Assert that INSN_CODE (insn) is non-negative.
22795 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22798 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22800 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22801 in the currently scheduled RTL fragment.
22803 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22805 PR rtl-optimization/78751
22806 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22809 2017-01-14 Jeff Law <law@redhat.com>
22811 PR tree-optimization/79090
22812 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22813 variable length stores.
22814 (compute_trims): Delete dead assignment to *trim_tail.
22815 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22818 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22820 PR rtl-optimization/78626
22821 PR rtl-optimization/78727
22822 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22823 of a block, and split such blocks after everything else is finished.
22825 2017-01-14 Alan Modra <amodra@gmail.com>
22828 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22829 target legitimate_combined_insn.
22830 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22831 (rs6000_legitimate_combined_insn): New function.
22832 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22834 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22835 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22836 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22838 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22840 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22842 2017-01-13 Jeff Law <law@redhat.com>
22844 PR tree-optimization/33562
22845 PR tree-optimization/61912
22846 PR tree-optimization/77485
22847 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22849 (delete_dead_assignment): Likewise.
22850 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22851 statement to delete_dead_call and delete_dead_assignment.
22853 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22856 * substring-locations.c (format_warning_va): Strengthen case 1 so
22857 that both endpoints of the substring must be within the format
22858 range for just the substring to be printed.
22860 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22862 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22863 * config/i386/i386.c (ix86_target_string): Add missing options
22864 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22865 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22866 flags_other and ix86_target_other to flags2_other. Display unknown
22868 (ix86_valid_target_attribute_inner_p): Add missing options and
22869 reorder options by implied ISAs, as in ix86_target_string.
22871 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22873 * hash-table.h (hash_table::too_empty_p): New function.
22874 (hash_table::expand): Use it.
22875 (hash_table::traverse): Likewise.
22876 (hash_table::empty_slot): Use sizeof (value_type) instead of
22877 sizeof (PTR) to convert bytes to elements. Shrink the table
22878 if the current size is excessive for the current number of
22881 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22883 * ira-costs.c (record_reg_classes): Break from the inner loop
22884 early once alt_fail is known to be true. Update outer loop
22885 handling accordingly.
22887 2017-01-13 Jeff Law <law@redhat.com>
22889 * tree-ssa-dse.c (decrement_count): New function.
22890 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22891 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22892 when we know the partially dead statement is a mem* function.
22894 PR tree-optimization/61912
22895 PR tree-optimization/77485
22896 * tree-ssa-dse.c: Include expr.h.
22897 (maybe_trim_constructor_store): New function.
22898 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22900 PR tree-optimization/33562
22901 PR tree-optimization/61912
22902 PR tree-optimization/77485
22903 * doc/invoke.texi: Document new dse-max-object-size param.
22904 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22905 * tree-ssa-dse.c: Include params.h.
22906 (dse_store_status): New enum.
22907 (initialize_ao_ref_for_dse): New, partially extracted from
22909 (valid_ao_ref_for_dse, normalize_ref): New.
22910 (setup_live_bytes_from_ref, compute_trims): Likewise.
22911 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22912 (maybe_trim_partially_dead_store): Likewise.
22913 (maybe_trim_complex_store): Likewise.
22914 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22915 Track what bytes live from the original store. Return tri-state
22916 for dead, partially dead or live.
22917 (dse_dom_walker): Add constructor, destructor and new private members.
22918 (delete_dead_call, delete_dead_assignment): New extracted from
22920 (dse_optimize_stmt): Make a member of dse_dom_walker.
22921 Use initialize_ao_ref_for_dse.
22923 PR tree-optimization/33562
22924 PR tree-optimization/61912
22925 PR tree-optimization/77485
22926 * sbitmap.h (bitmap_count_bits): Prototype.
22927 (bitmap_clear_range, bitmap_set_range): Likewise.
22928 * sbitmap.c (bitmap_clear_range): New function.
22929 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22931 2017-01-13 Martin Liska <mliska@suse.cz>
22934 * function.c (set_cfun): Add new argument force.
22935 * function.h (set_cfun): Likewise.
22936 * ipa-inline-transform.c (inline_call): Use the function when
22937 strict alising from is dropped for function we inline to.
22939 2017-01-13 Richard Biener <rguenther@suse.de>
22941 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22942 for dumping GIMPLE INTEGER_CSTs.
22944 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22946 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22947 to 201112L since C++17.
22949 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22952 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22953 if -fsanitize=kernel-address is present.
22955 2017-01-13 Richard Biener <rguenther@suse.de>
22957 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22958 as _Literal ( type ) number in case usual suffixes do not
22959 preserve all information.
22961 2017-01-13 Richard Biener <rguenther@suse.de>
22963 PR tree-optimization/77283
22964 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22965 and ssa-iterators.h.
22966 (is_feasible_trace): Implement a cost model based on joiner
22969 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22972 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22973 char or short to __float128/_Float128 directly.
22975 2017-01-12 Martin Sebor <msebor@redhat.com>
22977 to -Wformat-overflow.
22978 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22979 (min_bytes_remaining): Same.
22980 (get_string_length): Same.
22981 (format_string): Same.
22982 (format_directive): Same.
22984 (pass_sprintf_length::handle_gimple_call): Same.
22986 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22988 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22989 info.nowrite calls with no lhs that can't throw. Return bool
22990 whether gsi_remove has been called or not.
22991 (pass_sprintf_length::handle_gimple_call): Return bool whether
22992 try_substitute_return_value called gsi_remove. Formatting fix.
22993 (pass_sprintf_length::execute): Don't use gsi_remove if
22994 handle_gimple_call returned true.
22997 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22998 be removed due to side-effects, don't remove following barrier nor
22999 turn the successor edge into fallthru edge.
23001 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23004 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
23005 element-reversing loads and stores as not swappable.
23007 2017-01-12 Nathan Sidwell <nathan@acm.org>
23008 Nicolai Stange <nicstange@gmail.com>
23010 * combine.c (try_combine): Don't ignore result of overlap checking
23011 loop. Combine overlap & asm check into single loop.
23013 2017-01-12 Richard Biener <rguenther@suse.de>
23015 * tree-pretty-print.c (dump_generic_node): Provide -gimple
23016 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
23018 2017-01-12 Richard Biener <rguenther@suse.de>
23020 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
23021 and TS_TARGET_OPTION directly derive from TS_BASE.
23022 * tree-core.h (tree_optimization_option): Derive from tree_base.
23023 (tree_target_option): Likewise.
23025 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
23027 * config/i386/i386.c (memory_address_length): Increase len
23028 only when rip_relative_addr_p returns false.
23030 2017-01-11 Julia Koval <julia.koval@intel.com>
23032 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
23033 (OPTION_MASK_ISA_SGX_SET): New.
23034 (ix86_handle_option): Handle OPT_msgx.
23035 * config.gcc: Added sgxintrin.h.
23036 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
23037 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
23038 * config/i386/i386.c (ix86_target_string): Add -msgx.
23040 (ix86_option_override_internal): Handle new options.
23041 (ix86_valid_target_attribute_inner_p): Add sgx.
23042 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
23043 * config/i386/i386.opt: Add msgx.
23044 * config/i386/sgxintrin.h: New file.
23045 * config/i386/x86intrin.h: Add sgxintrin.h.
23047 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23050 * fold-const.c (maybe_nonzero_address): Return 1 for function
23052 (tree_single_nonzero_warnv_p): Don't handle function local objects
23056 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
23059 2017-01-11 David Malcolm <dmalcolm@redhat.com>
23062 * opts.c: Include "spellcheck.h"
23063 (struct string_fragment): New struct.
23064 (struct edit_distance_traits<const string_fragment &>): New
23066 (get_closest_sanitizer_option): New function.
23067 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23069 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23071 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23073 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23074 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23075 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23076 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23077 for initial die_offset if dwarf_split_debug_info.
23078 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23079 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23081 (output_skeleton_debug_sections): Formatting fix. Use
23082 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23083 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23085 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23087 * config/arm/cortex-a53.md: Add bypasses for
23088 cortex_a53_r2f_cvt.
23089 (cortex_a53_r2f): Only use for transfers.
23090 (cortex_a53_f2r): Likewise.
23091 (cortex_a53_r2f_cvt): Add reservation for conversions.
23092 (cortex_a53_f2r_cvt): Likewise.
23094 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23096 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23097 to all inlined functions, change static to extern.
23099 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23102 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23104 (arm_assemble_integer): Likewise.
23106 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23108 * config.gcc: Use new awk script to check CPU, FPU and architecture
23109 parameters for --with-... options.
23110 * config/arm/parsecpu.awk: New file
23111 * config/arm/arm-cpus.in: New file.
23112 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23114 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23116 * config/arm/t-arm: Update dependency rules.
23117 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23118 of processing .def files.
23119 * config/arm/genopt.sh: Deleted.
23120 * config/arm/gentune.sh: Deleted.
23121 * config/arm/arm-cores.def: Deleted.
23122 * config/arm/arm-arches.def: Deleted.
23123 * config/arm/arm-fpus.def: Deleted.
23124 * config/arm/arm-tune.md: Regenerated.
23125 * config/arm/arm-tables.opt: Regenerated.
23126 * config/arm/arm-cpu.h: New generated file.
23127 * config/arm/arm-cpu-data.h: New generated file.
23128 * config/arm/arm-cpu-cdata.h: New generated file.
23130 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23133 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23135 (input_varpool_node): Unpack dynamically_initialized bit.
23137 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23139 PR rtl-optimization/79032
23140 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23141 the alignment of the adjusted memory reference against that of MODE,
23142 instead of the alignment of the original memory reference.
23144 2017-01-11 Martin Jambor <mjambor@suse.cz>
23146 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23148 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23149 decorated functions.
23151 2017-01-11 Richard Biener <rguenther@suse.de>
23153 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23154 set range/nonnull info for PHI results. Do not set it on
23155 stmts marked for removal.
23157 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23159 * expr.c (store_field): In the bitfield case, fetch the return value
23160 from the registers before applying a single big-endian adjustment.
23161 Always do a final load for a BLKmode value not larger than a word.
23163 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23166 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23167 that we correctly handle column numbers greater than
23168 LINE_MAP_MAX_COLUMN_NUMBER.
23170 2017-01-10 Martin Sebor <msebor@redhat.com>
23172 PR middle-end/78245
23173 * gimple-ssa-sprintf.c (get_destination_size): Call
23174 {init,fini}object_sizes.
23175 * tree-object-size.c (addr_object_size): Adjust.
23176 (pass_through_call): Adjust.
23177 (pass_object_sizes::execute): Adjust.
23178 * tree-object-size.h (fini_object_sizes): Declare.
23180 2017-01-10 Martin Sebor <msebor@redhat.com>
23182 PR tree-optimization/78775
23183 * builtins.c (get_size_range): Move...
23184 * calls.c: ...to here.
23185 (alloc_max_size): Accept zero argument.
23186 (operand_signed_p): Remove.
23187 (maybe_warn_alloc_args_overflow): Call get_size_range.
23188 * calls.h (get_size_range): Declare.
23190 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23192 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23193 from TI's devices.csv file as of September 2016.
23194 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23196 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23198 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23199 * doc/invoke.texi: Likewise.
23200 * doc/md.texi: Likewise.
23201 * doc/objc.texi: Likewise.
23203 2017-01-10 Joshua Conner <joshconner@google.com>
23205 * config/arm/fuchsia-elf.h: New file.
23206 * config/fuchsia.h: New file.
23207 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23208 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23210 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23212 2016-01-10 Richard Biener <rguenther@suse.de>
23214 PR tree-optimization/79034
23215 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23216 Propagate out degenerate PHIs in the joiner.
23218 2017-01-10 Martin Liska <mliska@suse.cz>
23220 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23221 (sort_congruence_classes_by_decl_uid): Likewise.
23222 (sort_congruence_class_groups_by_decl_uid): Likewise.
23223 (sem_item_optimizer::merge_classes): Sort class, groups in these
23224 classes and members in the groups by DECL_UID of declarations.
23225 This would make merge operations stable.
23227 2017-01-10 Martin Liska <mliska@suse.cz>
23229 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23230 usage of m_classes_vec.
23231 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23232 (sem_item_optimizer::get_group_by_hash): Likewise.
23233 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23234 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23235 (sem_item_optimizer::verify_classes): Likewise.
23236 (sem_item_optimizer::process_cong_reduction): Likewise.
23237 (sem_item_optimizer::dump_cong_classes): Likewise.
23238 (sem_item_optimizer::merge_classes): Likewise.
23239 * ipa-icf.h (congruence_class_hash): Rename from
23240 congruence_class_group_hash. Remove declaration of m_classes_vec.
23242 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23244 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23245 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23246 * config.gcc: Add avx512vpopcntdqintrin.h.
23247 * config/i386/avx512vpopcntdqintrin.h: New.
23248 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23249 * config/i386/i386-builtin-types.def: Add new types.
23250 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23251 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23252 __builtin_ia32_vpopcountq_v8di_mask): New.
23253 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23254 __AVX512VPOPCNTDQ__.
23255 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23256 (PTA_AVX512VPOPCNTDQ): Define.
23257 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23258 TARGET_AVX512VPOPCNTDQ_P): Define.
23259 * config/i386/i386.opt: Add mavx512vpopcntdq.
23260 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23261 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23263 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23265 PR middle-end/77484
23266 * predict.def (PRED_CALL): Set to 67.
23268 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23270 * expr.c (store_field): In the bitfield case, if the value comes from
23271 a function call and is of an aggregate type returned in registers, do
23272 not modify the field mode; extract the value in all cases if the mode
23273 is BLKmode and the size is not larger than a word.
23275 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23278 * config/i386/cpuid.h: Fix undefined behavior.
23280 2017-01-04 Jeff Law <law@redhat.com>
23282 PR tree-optimization/79007
23283 PR tree-optimization/67955
23284 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23285 conservative for pt.null when flag_non_call_exceptions is on.
23287 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23289 PR translation/79019
23290 PR translation/79020
23291 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23292 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23293 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23295 * config/avr/avr.opt (maccumulate-args): Likewise.
23296 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23297 * common.opt (freport-bug): Likewise.
23298 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23299 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23300 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23301 translatable string.
23302 * config/i386/i386.c (function_value_32): Likewise.
23303 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23304 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23306 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23307 * common/config/msp430/msp430-common.c (msp430_handle_option):
23309 * symtab.c (symtab_node::verify_base): Likewise.
23310 * opts.c (set_debug_level): Likewise.
23311 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23312 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23313 missing whitespace to translatable strings.
23314 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23315 * config/sh/superh.h: Likewise.
23316 * config/i386/xopintrin.h: Likewise.
23317 * config/i386/znver1.md: Likewise.
23318 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23319 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23320 * double-int.h (struct double_int): Likewise.
23321 * double-int.c (div_and_round_double): Likewise.
23322 * wide-int.cc: Likewise.
23323 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23324 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23325 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23326 (crossjumps_occurred): ... this.
23327 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23330 PR tree-optimization/78899
23331 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23332 returning bool return struct loop *, NULL for failure and the new
23334 (versionable_outer_loop_p): Don't version outer loop if it has
23335 dont_vectorized bit set.
23336 (tree_if_conversion): When versioning outer loop, ensure
23337 tree_if_conversion is performed also on the inner loop of the
23338 non-vectorizable outer loop copy.
23339 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23340 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23341 prevent vectorization of it.
23342 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23343 the outer loop vectorization of the non-scalar version is attempted
23344 before vectorization of the inner loop in scalar version. If
23345 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23346 vectorization of its inner loop.
23347 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23348 has 2 inner loops, rename also on edges from bb whose single pred
23349 is outer_loop->header. Fix typo in function comment.
23351 2017-01-09 Martin Sebor <msebor@redhat.com>
23354 * asan.c (asan_emit_stack_protection): Increase local buffer size
23355 to avoid snprintf truncation warning.
23357 2017-01-09 Andrew Pinski <apinski@cavium.com>
23359 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23360 to reference thunderx2t99 for the tuning structure
23361 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23363 (thunderx2t99_extra_costs): This.
23364 * config/aarch64/aarch64-tune.md: Regenerate.
23365 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23366 (vulcan_addrcost_table): This.
23367 (vulcan_regmove_cost): Rename to ...
23368 (thunderx2t99_regmove_cost): This.
23369 (vulcan_vector_cost): Rename to ...
23370 (thunderx2t99_vector_cost): this.
23371 (vulcan_branch_cost): Rename to ...
23372 (thunderx2t99_branch_cost): This.
23373 (vulcan_tunings): Rename to ...
23374 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23375 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23377 2017-01-09 Martin Jambor <mjambor@suse.cz>
23381 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23382 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23383 (propagate_vr_accross_jump_function): Use the above function for all
23384 value range computations for pass-through jump functions and type
23385 converasion from explicit value range values.
23386 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23387 parameters from TYPE_ARG_TYPES.
23388 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23389 (ipa_write_node_info): Stream type of the actual argument.
23390 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23392 2017-01-09 Martin Liska <mliska@suse.cz>
23395 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23396 (lookup_compiler): Do not show error message with have_E.
23398 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23400 PR tree-optimization/78938
23401 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23402 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23403 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23404 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23407 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23409 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23412 2017-01-09 Richard Biener <rguenther@suse.de>
23414 PR tree-optimization/78997
23415 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23416 name condition properly.
23418 2017-01-09 Richard Biener <rguenther@suse.de>
23421 * dwarf2out.c (is_cxx): New overload with context.
23422 (is_naming_typedef_decl): Use it.
23424 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23426 * invoke.texi (Option Summary): Correct spacing in option lists
23427 and add line breaks to fix over-long lines.
23429 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23431 PR middle-end/17660
23433 * extend.texi (Common Variable Attributes): Add xref to GCC
23434 Internals manual to explain mode attribute keywords.
23436 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23439 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23440 and Preprocessor Options.
23441 (Options for Linking): Document -pthread here....
23442 (RS/6000 and PowerPC Options): ...not here.
23443 (Solaris 2 Options): ...or here.
23444 * doc/cppopts.texi: Document -pthread.
23446 2017-01-08 Martin Sebor <msebor@redhat.com>
23448 PR middle-end/77708
23449 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23450 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23451 New member functions.
23452 (format_directive): Used them.
23454 (pass_sprintf_length::handle_gimple_call): Same.
23455 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23456 to avoid truncation for any argument.
23457 (extract_affine_mul): Same.
23458 * tree.c (get_file_function_name): Same.
23460 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23462 PR middle-end/77484
23463 * predict.def (PRED_INDIR_CALL): Set to 86.
23465 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23467 PR preprocessor/54124
23468 * doc/cppopts.texi: Reformat -d subtable to list the full name
23469 of the options. Add cross-reference to the docs for the general
23470 compiler -d options.
23471 * doc/invoke.texi (Developer Options): Add cross-reference to the
23472 preprocessor-specific -d option documentation.
23474 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23476 PR preprocessor/13498
23477 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23478 redudant material, and reflect new command-line options.
23479 (System Headers): Likewise.
23481 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23483 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23484 -isystem, and -idirafter. Copy-edit.
23485 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23486 default for -ftrack-macro-expansion. Delete obsolete and
23487 badly-formatted implementation details about -fdebug-cpp output.
23488 * doc/cppwarnopts.texi: Copy-edit.
23490 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23493 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23494 that the transition from a max line width >= 1<<10 to narrower
23495 lines works correctly.
23497 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23499 * doc/options.texi (PerFunction): New.
23500 * opt-functions.awk (switch_flags): Map both Optimization and
23501 PerFunction to CL_OPTIMIZATION.
23502 * opth-gen.awk: Test for PerFunction flag along with
23504 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23505 it only when the latter is present. Skip those that don't in
23506 the hash function generator.
23507 * common.opt (fvar-tracking): Mark as PerFunction instead of
23509 (fvar-tracking-assignments): Likewise.
23510 (fvar-tracking-assignments-toggle): Likewise.
23511 (fvar-tracking-uninit): Likewise.
23513 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23515 PR translation/79018
23516 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23519 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23522 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23523 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23524 TARGET_LONG_JUMP_TABLE_OFFSETS.
23525 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23526 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23527 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23528 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23529 * config/m68k/m68k.md (tablejump expander): Likewise.
23530 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23531 TARGET_LONG_JUMP_TABLE_OFFSETS.
23532 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23533 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23535 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23536 David Holsgrove <david.holsgrove@xilinx.com>
23538 * common/config/microblaze/microblaze-common.c
23539 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23540 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23542 * config/microblaze/microblaze.c (microblaze_must_save_register)
23543 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23545 (microblaze_eh_return): New function.
23546 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23547 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23548 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23549 * config/microblaze/microblaze.md (eh_return): New pattern.
23551 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23553 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23554 GCC_DIAGNOSTIC_STRINGIFY): Define.
23556 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23558 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23560 * config/arm/arm.md (<mcrr>): New.
23562 * config/arm/arm.c (arm_arch5te): New.
23563 (arm_option_override): Set arm_arch5te.
23564 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23566 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23567 (arm_mcrr_qualifiers): ... this. New.
23568 (MRRC_QUALIFIERS): Define to...
23569 (arm_mrrc_qualifiers): ... this. New.
23570 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23572 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23573 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23574 (MRRCI, mrrc, MRRC): New.
23575 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23576 VUNSPEC_MRRC2): New.
23578 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23580 * config/arm/arm.md (<mcr>): New.
23582 * config/arm/arm.c (arm_coproc_builtin_available): Add
23583 support for mcr, mrc, mcr2 and mrc2.
23584 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23585 (arm_mcr_qualifiers): ... this. New.
23586 (MRC_QUALIFIERS): Define to ...
23587 (arm_mrc_qualifiers): ... this. New.
23588 (MCR_QUALIFIERS): Define to ...
23589 (arm_mcr_qualifiers): ... this. New.
23590 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23592 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23593 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23594 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23595 VUNSPEC_MRC2): New.
23597 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23599 * config/arm/arm.md (*ldc): New.
23603 * config/arm/arm.c (arm_coproc_builtin_available): Add
23604 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23605 (arm_coproc_ldc_stc_legitimate_address): New.
23606 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23607 'qualifier_const_pointer'.
23608 (LDC_QUALIFIERS): Define to...
23609 (arm_ldc_qualifiers): ... this. New.
23610 (STC_QUALIFIERS): Define to...
23611 (arm_stc_qualifiers): ... this. New.
23612 * config/arm/arm-protos.h
23613 (arm_coproc_ldc_stc_legitimate_address): New.
23614 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23615 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23616 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23617 stc2, stcl, stc2l): New.
23618 * config/arm/constraints.md (Uz): New.
23619 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23620 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23621 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23622 VUNSPEC_STC2L): New.
23624 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23626 * config/arm/arm.md (<cdp>): New.
23627 * config/arm/arm.c (neon_const_bounds): Rename this ...
23628 (arm_const_bounds): ... this.
23629 (arm_coproc_builtin_available): New.
23630 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23631 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23632 (CDP_QUALIFIERS): Define to...
23633 (arm_cdp_qualifiers): ... this. New.
23635 (arm_expand_builtin_args): Add case for 6 arguments.
23636 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23637 (arm_const_bounds): ... this.
23638 (arm_coproc_builtin_available): New.
23639 * config/arm/arm_acle.h (__arm_cdp): New.
23641 * config/arm/arm_acle_builtins.def (cdp): New.
23643 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23644 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23645 'arm_const_bounds'.
23646 * config/arm/types.md (coproc): New.
23647 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23648 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23649 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23650 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23652 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23654 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23655 (UBINOP_QUALIFIERS): New.
23657 (acle_builtin_data): New. Change comment.
23658 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23659 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23660 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23661 arm_acle_builtins.def.
23662 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23663 (arm_init_acle_builtins): New.
23664 (CRC32_BUILTIN): Remove.
23665 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23666 crc32cb, crc32ch and crc32cw.
23667 (arm_init_crc32_builtins): Remove.
23668 (arm_init_builtins): Use arm_init_acle_builtins rather
23669 than arm_init_crc32_builtins.
23670 (arm_expand_acle_builtin): New.
23671 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23672 * config/arm/arm_acle_builtins.def: New.
23674 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23676 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23677 (arm_builtin_datum): ... this.
23678 (arm_init_neon_builtin): Rename to ...
23679 (arm_init_builtin): ... this. Add a new parameters PREFIX
23680 and USE_SIG_IN_NAME.
23681 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23682 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23683 'arm_builtin_datum'.
23684 (arm_init_vfp_builtins): Likewise.
23685 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23686 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23687 (arm_expand_neon_args): Rename to ...
23688 (arm_expand_builtin_args): ... this. Rename builtin_arg
23689 enum values and differentiate between ARG_BUILTIN_MEMORY
23690 and ARG_BUILTIN_NEON_MEMORY.
23691 (arm_expand_neon_builtin_1): Rename to ...
23692 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23693 values, arm_expand_builtin_args and add bool parameter NEON.
23694 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23695 (arm_expand_vfp_builtin): Likewise.
23696 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23698 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23700 PR middle-end/77484
23701 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23702 * predict.c (tree_estimate_probability_bb): Reverse direction of
23703 polymorphic call predictor.
23705 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23707 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23708 (determine_pass_name_match): ...this new function and...
23709 (should_skip_pass_p): ...this new function.
23711 2017-01-06 Nathan Sidwell <nathan@acm.org>
23713 * ipa-visibility.c (function_and_variable_visibility): Reformat
23714 comments and long lines. Remove extrneous if.
23715 * symtab.c (symtab_node::make_decl_local): Fix code format.
23716 (symtab_node::set_section_for_node): Fix comment typo.
23718 2017-01-06 Martin Liska <mliska@suse.cz>
23721 * lra-constraints.c: Rename invariant to lra_invariant.
23722 * predict.c (set_even_probabilities): Initialize e to NULL.
23724 2017-01-05 Martin Sebor <msebor@redhat.com>
23726 PR tree-optimization/78910
23727 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23728 (format_integer): Correct off-by-one error in the handling
23729 of precision with negative numbers in signed conversions..
23731 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23733 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23735 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23737 PR tree-optimization/71016
23738 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23739 factor_out_conditional_conversion. Formatting fix.
23740 (factor_out_conditional_conversion): Add cond_stmt argument.
23741 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23742 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23745 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23747 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23748 read-rtl-function.o, and selftest-rtl.o.
23749 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23750 (selftest::aarch64_test_loading_full_dump): New function.
23751 (selftest::aarch64_run_selftests): New function.
23752 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23753 selftest::aarch64_run_selftests.
23754 * config/i386/i386.c
23755 (selftest::ix86_test_loading_dump_fragment_1): New function.
23756 (selftest::ix86_test_loading_call_insn): New function.
23757 (selftest::ix86_test_loading_full_dump): New function.
23758 (selftest::ix86_test_loading_unspec): New function.
23759 (selftest::ix86_run_selftests): Call the new functions.
23760 * emit-rtl.c (maybe_set_max_label_num): New function.
23761 * emit-rtl.h (maybe_set_max_label_num): New decl.
23762 * function.c (instantiate_decls): Guard call to
23763 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23764 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23766 * gensupport.c (gen_reader::gen_reader): Pass "false"
23767 for new "compact" param of rtx_reader.
23768 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23769 rather than an empty string for NULL strings.
23770 * read-md.c: Potentially include config.h rather than bconfig.h.
23771 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23772 (have_error): New global, copied from errors.c.
23773 (md_reader::read_name): Rename to...
23774 (md_reader::read_name_1): ...this, adding "out_loc" param,
23775 and converting "missing name or number" to returning false, rather
23777 (md_reader::read_name): Reimplement in terms of read_name_1.
23778 (md_reader::read_name_or_nil): New function.
23779 (md_reader::read_string): Handle "(nil)" by returning NULL.
23780 (md_reader::md_reader): Add new param "compact".
23781 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23782 (md_reader::read_file): New method.
23783 * read-md.h (md_reader::md_reader): Add new param "compact".
23784 (md_reader::read_file): New method.
23785 (md_reader::is_compact): New accessor.
23786 (md_reader::read_name): Convert return type from void to file_location.
23787 (md_reader::read_name_or_nil): New decl.
23788 (md_reader::read_name_1): New decl.
23789 (md_reader::m_compact): New field.
23790 (noop_reader::noop_reader): Pass "false" for new "compact" param
23792 (rtx_reader::rtx_reader): Add new "compact" param.
23793 (rtx_reader::read_rtx_operand): Make virtual and convert return
23794 type from void to rtx.
23795 (rtx_reader::read_until): New decl.
23796 (rtx_reader::handle_any_trailing_information): New virtual function.
23797 (rtx_reader::postprocess): New virtual function.
23798 (rtx_reader::finalize_string): New virtual function.
23799 (rtx_reader::m_in_call_function_usage): New field.
23800 (rtx_reader::m_reuse_rtx_by_id): New field.
23801 * read-rtl-function.c: New file.
23802 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23803 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23804 (selftest::verify_three_block_rtl_cfg): New decl.
23805 * read-rtl-function.h: New file.
23806 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23807 For host, include function.h, memmodel.h, and emit-rtl.h.
23808 (one_time_initialization): New function.
23809 (struct compact_insn_name): New struct.
23810 (compact_insn_names): New array.
23811 (find_code): Handle insn codes in compact dumps.
23812 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23813 (bind_subst_iter_and_attr): Likewise.
23814 (add_condition_to_string): Likewise.
23815 (add_condition_to_rtx): Likewise.
23816 (apply_attribute_uses): Likewise.
23817 (add_current_iterators): Likewise.
23818 (apply_iterators): Likewise.
23819 (initialize_iterators): Guard usage of apply_subst_iterator with
23820 #ifdef GENERATOR_FILE.
23821 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23822 (md_reader::read_mapping): Likewise.
23823 (add_define_attr_for_define_subst): Likewise.
23824 (add_define_subst_attr): Likewise.
23825 (read_subst_mapping): Likewise.
23826 (check_code_iterator): Likewise.
23827 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23829 (one_time_initialization): New function.
23830 (rtx_reader::read_until): New method.
23831 (read_flags): New function.
23832 (parse_reg_note_name): New function.
23833 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23834 Handle reuse_rtx ids.
23835 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23836 Add parsing support for RTL dumps, mirroring the special-cases in
23837 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23838 values, and calling handle_any_trailing_information.
23839 (rtx_reader::read_rtx_operand): Convert return type from void
23840 to rtx, returning return_rtx. Handle case 'e'. Call
23841 finalize_string on XSTR and XTMPL fields.
23842 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23843 "(nil)" values were omitted. Call the postprocess vfunc on the
23845 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23846 class ctor. Initialize m_in_call_function_usage. Call
23847 one_time_initialization.
23848 * rtl-tests.c (selftest::test_uncond_jump): Call
23849 set_new_first_and_last_insn.
23850 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23851 * selftest-rtl.c: New file.
23852 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23853 (selftest::get_insn_by_uid): New decl.
23854 * selftest-run-tests.c (selftest::run_tests): Call
23855 read_rtl_function_c_tests.
23856 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23857 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23860 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23862 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23863 operands in a special way. Assert that pos+len <= mode precision.
23865 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23867 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23868 3 argument Alias with unlimited for the negative form.
23869 (fno-vect-cost-model): Removed.
23871 2017-01-05 Martin Liska <mliska@suse.cz>
23873 * hsa-gen.c (gen_hsa_divmod): New function.
23874 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23876 2017-01-05 Martin Liska <mliska@suse.cz>
23879 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23882 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23884 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23885 small constant length operands.
23887 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23889 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23890 between loop iterations.
23892 2017-01-05 Martin Liska <mliska@suse.cz>
23895 * gimplify.c (gimplify_decl_expr): Compare to
23896 asan_poisoned_variables instread of checking flags.
23897 (gimplify_target_expr): Likewise.
23898 (gimplify_expr): Likewise.
23899 (gimplify_function_tree): Conditionally initialize
23900 asan_poisoned_variables.
23902 2017-01-04 Jeff Law <law@redhat.com>
23904 PR tree-optimizatin/78812
23905 * rtl.h (contains_mem_rtx_p): Prototype.
23906 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23907 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23908 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23909 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23910 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23912 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23914 * input.c (assert_char_at_range): Default-initialize actual_range.
23916 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23918 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23919 to match the caller.
23921 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23923 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23924 insns after final jump in test to emit dummy move.
23926 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23928 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23929 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23931 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23933 * multiple_target.c (create_dispatcher_calls): Init e_next.
23934 * tree-ssa-loop-split.c (split_loop): Init border.
23935 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23938 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23943 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23944 (altivec_register_operand): Do not return true if the operand
23945 contains a SUBREG mixing SImode and SFmode.
23946 (vsx_register_operand): Likewise.
23947 (vsx_reg_sfsubreg_ok): New predicate.
23948 (vfloat_operand): Do not return true if the operand contains a
23949 SUBREG mixing SImode and SFmode.
23950 (vint_operand): Likewise.
23951 (vlogical_operand): Likewise.
23952 (gpc_reg_operand): Likewise.
23953 (int_reg_operand): Likewise.
23954 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23955 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23956 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23958 (rs6000_emit_move_si_sf_subreg): New helper function.
23959 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23960 fixup SUBREGs involving SImode and SFmode.
23961 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23962 numbers for the new peephole2 optimization.
23963 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23964 the GLIBC math library that do AND/IOR/XOR operations on single
23965 precision floating point.
23966 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23967 target macros to say whether we need to avoid SUBREGs mixing
23969 (TARGET_ALLOW_SF_SUBREG): Likewise.
23970 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23971 (UNSPEC_SI_FROM_SF): Likewise.
23972 (iorxor): Change spacing.
23973 (and_ior_xor): New iterator for AND, IOR, and XOR.
23974 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23975 (movdi_from_sf_zero_ext): Likewise.
23976 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23977 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23978 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23979 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23980 (fms<mode>4): Likewise.
23981 (fnma<mode>4): Likewise.
23982 (fnms<mode>4): Likewise.
23983 (nfma<mode>4): Likewise.
23984 (nfms<mode>4): Likewise.
23986 2017-01-04 Marek Polacek <polacek@redhat.com>
23989 * doc/invoke.texi: Document -Wpointer-compare.
23991 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23993 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23996 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23997 descriptions for -gdwarf-5 and emit them as uleb128 instead of
24000 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
24003 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
24004 documentation of the powerpc_popcntb_ok attribute.
24005 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24006 code to issue warning messages if a requested CPU configuration is
24007 not supported by the binary (assembler and loader) toolchain.
24008 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
24009 made to define a built-in function that has been disabled.
24010 (paired_init_builtins): Add assertion to prevent ICE if attempt is
24011 made to define a built-in function that has been disabled.
24012 (altivec_init_builtins): Add comment explaining why definition
24013 of the DST built-in functions is not preceded by an assertion
24014 check. Add assertions to prevent ICE if attempts are made to
24015 define an altivec predicate or an abs* built-in function that has
24017 (htm_init_builtins): Add comment explaining why definition of the
24018 htm built-in functions is not preceded by an assertion check.
24020 2017-01-04 Jeff Law <law@redhat.com>
24022 PR tree-optimizatin/67955
24023 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
24024 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
24025 the points-to solution does not include pt_null. Use DECL_PT_UID
24028 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
24030 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
24031 Use gen_int_mode instead of gen_lopwart for const_int operands.
24033 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24035 PR tree-optimization/71563
24036 * match.pd: Simplify X << Y into X if Y is known to be 0 or
24037 out of range value - has low bits known to be zero.
24039 2017-01-04 Alan Modra <amodra@gmail.com>
24041 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
24042 * configure: Regenerate.
24043 * config.in: Regenerate.
24045 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24048 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
24049 a substring of the message, but strcmp with the whole message. Ifdef
24050 ENABLE_NLS, translate the message first using dgettext.
24052 2017-01-03 Jeff Law <law@redhat.com>
24054 PR tree-optimizatin/78856
24055 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
24056 (mark_threaded_blocks): Remove code to truncate thread paths that
24057 cross multiple loop headers. Instead invalidate the cached loop
24058 iteration information and handle case of a thread path walking
24059 into an irreducible region.
24061 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24064 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24065 assertions. Add support for doing the signbit if the IEEE 128-bit
24066 floating point value is in a GPR.
24067 * config/rs6000/rs6000.md (Fsignbit): Delete.
24068 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24069 Update the length attribute if the value is in a GPR.
24070 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24071 the sign or zero extension instruction, since the value is always 0/1.
24072 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24075 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24076 extracting SImode to a GPR register so that we can generate a
24077 store, limit the vector to be in a traditional Altivec register
24078 for the vextuwrx instruction.
24080 2017-01-03 Ian Lance Taylor <iant@google.com>
24082 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24084 2017-01-03 Martin Sebor <msebor@redhat.com>
24086 PR tree-optimization/78696
24087 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24088 precision. Use MPFR for %f for greater fidelity. Correct handling
24090 (pass_sprintf_length::compute_format_length): Set width and precision
24091 specified by asrerisk to void_node for vararg functions.
24092 (try_substitute_return_value): Adjust dump output.
24094 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24096 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24098 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24100 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24101 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24102 -mlra/-mno-lra was passed to the compiler.
24104 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24106 PR rtl-optimization/65618
24107 * emit-rtl.c (try_split): Move initialization of "before" and
24108 "after" to just before the call to emit_insn_after_setloc.
24110 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24112 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24114 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24116 * dwarf2out.c (gen_enumeration_type_die): When
24117 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24119 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24121 PR tree-optimization/78965
24122 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24123 Change first argument from const call_info & to call_info &. For %n
24124 set info.nowrite to false.
24126 PR middle-end/78901
24127 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24128 possibly throwing calls.
24130 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24131 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24132 and fns handling, rather than in a separate case SSA_NAME.
24134 2017-01-02 Jeff Law <law@redhat.com>
24136 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24137 fixes for first_period and second_period variables.
24139 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24142 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24143 (*insvqi_1): New insn pattern.
24144 (*insvqi_1_mem_rex64): Ditto.
24145 (*insvqi_2): Ditto.
24146 (*insvqi_3): Rename from *insvqi.
24148 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24150 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24152 * doc/cfg.texi (Edges): Remove reference to Java.
24153 (Maintaining the CFG): Ditto.
24155 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24157 PR middle-end/77674
24158 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24159 transparent aliases.
24161 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24163 PR middle-end/77484
24164 * predict.def (PRED_CALL): Update hitrate.
24165 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24166 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24167 into direct/indirect/polymorphic variants.
24169 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24171 Update copyright years.
24173 * gcc.c (process_command): Update copyright notice dates.
24174 * gcov-dump.c (print_version): Ditto.
24175 * gcov.c (print_version): Ditto.
24176 * gcov-tool.c (print_version): Ditto.
24177 * gengtype.c (create_file): Ditto.
24178 * doc/cpp.texi: Bump @copying's copyright year.
24179 * doc/cppinternals.texi: Ditto.
24180 * doc/gcc.texi: Ditto.
24181 * doc/gccint.texi: Ditto.
24182 * doc/gcov.texi: Ditto.
24183 * doc/install.texi: Ditto.
24184 * doc/invoke.texi: Ditto.
24186 Copyright (C) 2017 Free Software Foundation, Inc.
24188 Copying and distribution of this file, with or without modification,
24189 are permitted in any medium without royalty provided the copyright
24190 notice and this notice are preserved.