Really delete gcc/loop-unswitch.c.
[official-gcc.git] / gcc / ChangeLog
blobe5033a04f268f5867774a4c76c4514f7634b44f4
1 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
3         * loop-unswitch.c: Delete.
5 2014-05-07  Richard Biener  <rguenther@suse.de>
7         * config.gcc: Always set need_64bit_hwint to yes.
9 2014-05-07  Chung-Ju Wu  <jasonwucj@gmail.com>
11         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
12         of using optimize_size.
14 2014-05-06  Mike Stump  <mikestump@comcast.net>
16         * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
18 2014-05-06  Joseph Myers  <joseph@codesourcery.com>
20         * config/i386/sse.md (*mov<mode>_internal)
21         (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
22         (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
23         (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
24         (*<code><mode>3, *andnot<mode>3<mask_name>)
25         (<mask_codefor><code><mode>3<mask_name>): Only consider
26         TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
28 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
30         Revert:
31         2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
33         * lra-constraints.c (valid_address_p): Move earlier in file.
34         Add a constraint argument to the address_info version.
35         (satisfies_memory_constraint_p): New function.
36         (satisfies_address_constraint_p): Likewise.
37         (process_alt_operands, curr_insn_transform): Use them.
38         (process_address): Pass the constraint to valid_address_p when
39         checking address operands.
41 2014-05-06  Richard Sandiford  <r.sandiford@uk.ibm.com>
43         * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
44         to their respective blocks.  Fix inadvertent use of "node".
46 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
48         * emit-rtl.c (init_derived_machine_modes): New functionm, split
49         out from...
50         (init_emit_once): ...here.
51         * rtl.h (init_derived_machine_modes): Declare.
52         * toplev.c (do_compile): Call it even if no_backend.
54 2014-05-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
55             Mike Stump  <mikestump@comcast.net>
56             Richard Sandiford  <rdsandiford@googlemail.com>
57             Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
59         * alias.c (ao_ref_from_mem): Use wide-int interfaces.
60         (rtx_equal_for_memref_p): Update comment.
61         (adjust_offset_for_component_ref): Use wide-int interfaces.
62         * builtins.c (get_object_alignment_2): Likewise.
63         (c_readstr): Likewise.
64         (target_char_cast): Add comment.
65         (determine_block_size): Use wide-int interfaces.
66         (expand_builtin_signbit): Likewise.
67         (fold_builtin_int_roundingfn): Likewise.
68         (fold_builtin_bitop): Likewise.
69         (fold_builtin_bswap): Likewise.
70         (fold_builtin_logarithm): Use signop.
71         (fold_builtin_pow): Likewise.
72         (fold_builtin_memory_op): Use wide-int interfaces.
73         (fold_builtin_object_size): Likewise.
74         * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
75         nb_iterations_estimate.
76         (record_niter_bound): Use wide-int interfaces.
77         (get_estimated_loop_iterations_int): Likewise.
78         (get_estimated_loop_iterations): Likewise.
79         (get_max_loop_iterations): Likewise.
80         * cfgloop.h: Include wide-int.h.
81         (struct nb_iter_bound): Change bound to widest_int.
82         (struct loop): Change nb_iterations_upper_bound and
83         nb_iterations_estimate to widest_int.
84         (record_niter_bound): Switch to use widest_int.
85         (get_estimated_loop_iterations): Likewise.
86         (get_max_loop_iterations): Likewise.
87         (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
88         update for wide-int.
89         * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
90         * combine.c (try_combine): Likewise.
91         (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
92         * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
93         interfaces.
94         (aarch64_float_const_representable_p): Likewise.
95         * config/arc/arc.c: Include wide-int.h.
96         (arc_can_use_doloop_p): Use wide-int interfaces.
97         * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
98         (vfp3_const_double_index): Likewise.
99         * config/avr/avr.c (avr_out_round): Likewise.
100         (avr_fold_builtin): Likewise.
101         * config/bfin/bfin.c (bfin_local_alignment): Likewise.
102         (bfin_can_use_doloop_p): Likewise.
103         * config/darwin.c (darwin_mergeable_constant_section): Likewise.
104         (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
105         * config/i386/i386.c: Include wide-int.h.
106         (ix86_data_alignment): Use wide-int interfaces.
107         (ix86_local_alignment): Likewise.
108         (ix86_emit_swsqrtsf): Update real_from_integer.
109         * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
110         * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
111         * config/rs6000/predicates.md (any_operand): Add const_wide_int.
112         (zero_constant): Likewise.
113         (input_operand): Likewise.
114         (splat_input_operand): Likewise.
115         (non_logical_cint_operand): Change const_double to const_wide_int.
116         * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
117         (easy_altivec_constant): Remove comment.
118         (paired_expand_vector_init): Use CONSTANT_P.
119         (rs6000_legitimize_address): Handle CONST_WIDE_INT.
120         (rs6000_emit_move): Update checks.
121         (rs6000_aggregate_candidate): Use wide-int interfaces.
122         (rs6000_expand_ternop_builtin): Likewise.
123         (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
124         (rs6000_assemble_integer): Likewise.
125         (rs6000_hash_constant): Likewise.
126         (output_toc): Likewise.
127         (rs6000_rtx_costs): Likewise.
128         (rs6000_emit_swrsqrt); Update call to real_from_integer.
129         * config/rs6000/rs6000-c.c: Include wide-int.h.
130         (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
131         * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
132         * config/rs6000/rs6000.md: Use const_scalar_int_operand.
133         Handle CONST_WIDE_INT.
134         * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
135         Use tree_fits_uhwi_p.
136         * config/sparc/sparc.c: Include wide-int.h.
137         (sparc_fold_builtin): Use wide-int interfaces.
138         * config/vax/vax.c: Include wide-int.h.
139         (vax_float_literal): Use real_from_integer.
140         * coretypes.h (struct hwivec_def): New.
141         (hwivec): New.
142         (const_hwivec): New.
143         * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
144         (equiv_constant): Handle CONST_WIDE_INT.
145         * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
146         (cselib_hash_rtx): Handle CONST_WIDE_INT.
147         * dbxout.c (stabstr_U): Use wide-int interfaces.
148         (dbxout_type): Update to use cst_fits_shwi_p.
149         * defaults.h (LOG2_BITS_PER_UNIT): Define.
150         (TARGET_SUPPORTS_WIDE_INT): Add default.
151         * dfp.c: Include wide-int.h.
152         (decimal_real_to_integer2): Use wide-int interfaces and rename to
153         decimal_real_to_integer.
154         * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
155         decimal_real_to_integer.
156         * doc/generic.texi (Constant expressions): Update for wide_int.
157         * doc/rtl.texi (const_double): Likewise.
158         (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
159         (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
160         * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
161         (REAL_VALUE_FROM_INT): Remove.
162         (TARGET_SUPPORTS_WIDE_INT): New.
163         * doc/tm.texi: Regenerate.
164         * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
165         * double-int.h: Include wide-int.h.
166         (struct wi::int_traits): New.
167         * dwarf2out.c (get_full_len): New.
168         (dw_val_equal_p): Add case dw_val_class_wide_int.
169         (size_of_loc_descr): Likewise.
170         (output_loc_operands): Likewise.
171         (insert_double): Remove.
172         (insert_wide_int): New.
173         (add_AT_wide): New.
174         (print_die): Add case dw_val_class_wide_int.
175         (attr_checksum): Likewise.
176         (attr_checksum_ordered): Likewise.
177         (same_dw_val_p): Likewise.
178         (size_of_die): Likewise.
179         (value_format): Likewise.
180         (output_die): Likewise.
181         (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
182         Use wide-int.
183         (clz_loc_descriptor): Use wide-int interfaces.
184         (mem_loc_descriptor): Likewise.  Handle CONST_WIDE_INT.
185         (loc_descriptor): Use wide-int interfaces.  Handle CONST_WIDE_INT.
186         (round_up_to_align): Use wide-int interfaces.
187         (field_byte_offset): Likewise.
188         (insert_double): Rename to insert_wide_int.  Use wide-int interfaces.
189         (add_const_value_attribute): Handle CONST_WIDE_INT.  Update
190         CONST_DOUBLE handling.  Use wide-int interfaces.
191         (add_bound_info): Use tree_fits_uhwi_p.  Use wide-int interfaces.
192         (gen_enumeration_type_die): Use add_AT_wide.
193         (hash_loc_operands): Add case dw_val_class_wide_int.
194         (compare_loc_operands): Likewise.
195         * dwarf2out.h: Include wide-int.h.
196         (wide_int_ptr): New.
197         (enum dw_val_class): Add dw_val_class_wide_int.
198         (struct dw_val_struct): Add val_wide.
199         * emit-rtl.c (const_wide_int_htab): New.
200         (const_wide_int_htab_hash): New.
201         (const_wide_int_htab_eq): New.
202         (lookup_const_wide_int): New.
203         (const_double_htab_hash): Use wide-int interfaces.
204         (const_double_htab_eq): Likewise.
205         (rtx_to_double_int): Conditionally compile for wide-int.
206         (immed_double_int_const): Rename to immed_wide_int_const and
207         update for wide-int.
208         (immed_double_const): Conditionally compile for wide-int.
209         (init_emit_once): Use wide-int interfaces.
210         * explow.c (plus_constant): Likewise.
211         * expmed.c (mask_rtx): Move further up file.  Use wide-int interfaces.
212         (lshift_value): Use wide-int interfaces.
213         (expand_mult): Likewise.
214         (choose_multiplier): Likewise.
215         (expand_smod_pow2): Likewise.
216         (make_tree): Likewise.
217         * expr.c (convert_modes): Consolidate handling of constants.
218         Use wide-int interfaces.
219         (emit_group_load_1): Add note.
220         (store_expr): Update comment.
221         (get_inner_reference): Use wide-int interfaces.
222         (expand_constructor): Update comment.
223         (expand_expr_real_2): Use wide-int interfaces.
224         (expand_expr_real_1): Likewise.
225         (reduce_to_bit_field_precision): Likewise.
226         (const_vector_from_tree): Likewise.
227         * final.c: Include wide-int-print.h.
228         (output_addr_const): Handle CONST_WIDE_INT.  Use CONST_DOUBLE_AS_INT_P.
229         * fixed-value.c: Include wide-int.h.
230         (fixed_from_string): Use wide-int interfaces.
231         (fixed_to_decimal): Likewise.
232         (fixed_convert_from_real): Likewise.
233         (real_convert_from_fixed): Likewise.
234         * fold-const.h (mem_ref_offset): Return an offset_int.
235         (div_if_zero_remainder): Remove code parameter.
236         * fold-const.c (div_if_zero_remainder): Remove code parameter.
237         Use wide-int interfaces.
238         (may_negate_without_overflow_p): Use wide-int interfaces.
239         (negate_expr_p): Likewise.
240         (fold_negate_expr): Likewise.
241         (int_const_binop_1): Likewise.
242         (const_binop): Likewise.
243         (fold_convert_const_int_from_int): Likewise.
244         (fold_convert_const_int_from_real): Likewise.
245         (fold_convert_const_int_from_fixed): Likewise.
246         (fold_convert_const_fixed_from_int): Likewise.
247         (all_ones_mask_p): Take an unsigned size.  Use wide-int interfaces.
248         (sign_bit_p): Use wide-int interfaces.
249         (make_range_step): Likewise.
250         (build_range_check): Likewise.  Pass an integer of the correct type
251         instead of using integer_one_node.
252         (range_predecessor): Pass an integer of the correct type instead
253         of using integer_one_node.
254         (range_successor): Likewise.
255         (merge_ranges): Likewise.
256         (unextend): Use wide-int interfaces.
257         (extract_muldiv_1): Likewise.
258         (fold_div_compare): Likewise.
259         (fold_single_bit_test): Likewise.
260         (fold_sign_changed_comparison): Likewise.
261         (try_move_mult_to_index): Update calls to div_if_zero_remainder.
262         (fold_plusminus_mult_expr): Use wide-int interfaces.
263         (native_encode_int): Likewise.
264         (native_interpret_int): Likewise.
265         (fold_unary_loc): Likewise.
266         (pointer_may_wrap_p): Likewise.
267         (size_low_cst): Likewise.
268         (mask_with_tz): Likewise.
269         (fold_binary_loc): Likewise.
270         (fold_ternary_loc): Likewise.
271         (multiple_of_p): Likewise.
272         (tree_call_nonnegative_warnv_p): Update calls to
273         tree_int_cst_min_precision and real_from_integer.
274         (fold_negate_const): Use wide-int interfaces.
275         (fold_abs_const): Likewise.
276         (fold_relational_const): Use tree_int_cst_lt.
277         (round_up_loc): Use wide-int interfaces.
278         * genemit.c (gen_exp): Add CONST_WIDE_INT case.
279         * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
280         * gengtype.c: Remove include of double-int.h.
281         (do_typedef): Use wide-int interfaces.
282         (open_base_files): Add wide-int.h.
283         (main): Add offset_int and widest_int typedefs.
284         * gengtype-lex.l: Handle "^".
285         (CXX_KEYWORD): Add "static".
286         * gengtype-parse.c (require3): New.
287         (require_template_declaration): Handle constant template arguments
288         and nested templates.
289         * gengtype-state.c: Don't include "double-int.h".
290         * genpreds.c (write_one_predicate_function): Update comment.
291         (write_tm_constrs_h): Add check for hval and lval use in
292         CONST_WIDE_INT.
293         * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
294         (add_to_sequence): Likewise.
295         * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
296         and const_double_operand.
297         * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
298         interfaces.
299         * gimple-fold.c (get_base_constructor): Likewise.
300         (fold_array_ctor_reference): Likewise.
301         (fold_nonarray_ctor_reference): Likewise.
302         (fold_const_aggregate_ref_1): Likewise.
303         (gimple_val_nonnegative_real_p): Likewise.
304         (gimple_fold_indirect_ref): Likewise.
305         * gimple-pretty-print.c (dump_ssaname_info): Likewise.
306         * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
307         (struct slsr_cand_d): Change index to be widest_int.
308         (struct incr_info_d): Change incr to be widest_int.
309         (alloc_cand_and_find_basis): Use wide-int interfaces.
310         (slsr_process_phi): Likewise.
311         (backtrace_base_for_ref): Likewise.  Return a widest_int.
312         (restructure_reference): Take a widest_int instead of a double_int.
313         (slsr_process_ref): Use wide-int interfaces.
314         (create_mul_ssa_cand): Likewise.
315         (create_mul_imm_cand): Likewise.
316         (create_add_ssa_cand): Likewise.
317         (create_add_imm_cand): Take a widest_int instead of a double_int.
318         (slsr_process_add): Use wide-int interfaces.
319         (slsr_process_cast): Likewise.
320         (slsr_process_copy): Likewise.
321         (dump_candidate): Likewise.
322         (dump_incr_vec): Likewise.
323         (replace_ref): Likewise.
324         (cand_increment): Likewise.  Return a widest_int.
325         (cand_abs_increment): Likewise.
326         (replace_mult_candidate): Take a widest_int instead of a double_int.
327         (replace_unconditional_candidate): Use wide-int interfaces.
328         (incr_vec_index): Take a widest_int instead of a double_int.
329         (create_add_on_incoming_edge): Likewise.
330         (create_phi_basis): Use wide-int interfaces.
331         (replace_conditional_candidate): Likewise.
332         (record_increment): Take a widest_int instead of a double_int.
333         (record_phi_increments): Use wide-int interfaces.
334         (phi_incr_cost): Take a widest_int instead of a double_int.
335         (lowest_cost_path): Likewise.
336         (total_savings): Likewise.
337         (analyze_increments): Use wide-int interfaces.
338         (ncd_with_phi): Take a widest_int instead of a double_int.
339         (ncd_of_cand_and_phis): Likewise.
340         (nearest_common_dominator_for_cands): Likewise.
341         (insert_initializers): Use wide-int interfaces.
342         (all_phi_incrs_profitable): Likewise.
343         (replace_one_candidate): Likewise.
344         (replace_profitable_candidates): Likewise.
345         * godump.c: Include wide-int-print.h.
346         (go_output_typedef): Use wide-int interfaces.
347         * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
348         * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
349         (build_loop_iteration_domains): Likewise.
350         * hooks.h: Include wide-int.h rather than double-int.h.
351         (hook_bool_dint_dint_uint_bool_true): Delete.
352         (hook_bool_wint_wint_uint_bool_true): Declare.
353         * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
354         (hook_bool_wint_wint_uint_bool_true): New.
355         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
356         interfaces.
357         (ubsan_expand_si_overflow_mul_check): Likewise.
358         * ipa-devirt.c (get_polymorphic_call_info): Likewise.
359         * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
360         (get_ancestor_addr_info): Likewise.
361         (ipa_modify_call_arguments): Likewise.
362         * loop-doloop.c (doloop_modify): Likewise.
363         (doloop_optimize): Likewise.
364         * loop-iv.c (iv_number_of_iterations): Likewise.
365         * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
366         (unroll_loop_constant_iterations): Likewise.
367         (decide_unroll_runtime_iterations): Likewise.
368         (unroll_loop_runtime_iterations): Likewise.
369         (decide_peel_simple): Likewise.
370         (decide_unroll_stupid): Likewise.
371         * lto-streamer-in.c (streamer_read_wi): Add.
372         (input_cfg): Use wide-int interfaces.
373         (lto_input_tree_1): Likewise.
374         * lto-streamer-out.c (streamer_write_wi): Add.
375         (hash_tree): Use wide-int interfaces.
376         (output_cfg): Likewise.
377         * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
378         (GTFILES): Add wide-int.h and signop.h.
379         (TAGS): Look for .cc files too.
380         * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
381         * optabs.c (expand_subword_shift): Likewise.
382         (expand_doubleword_shift): Likewise.
383         (expand_absneg_bit): Likewise.
384         (expand_copysign_absneg): Likewise.
385         (expand_copysign_bit): Likewise.
386         * postreload.c (reload_cse_simplify_set): Likewise.
387         * predict.c (predict_iv_comparison): Likewise.
388         * pretty-print.h: Include wide-int-print.h.
389         (pp_wide_int) New.
390         * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
391         * print-tree.c: Include wide-int-print.h.
392         (print_node_brief): Use wide-int interfaces.
393         (print_node): Likewise.
394         * read-rtl.c (validate_const_wide_int): New.
395         (read_rtx_code): Add CONST_WIDE_INT case.
396         * real.c: Include wide-int.h.
397         (real_to_integer2): Delete.
398         (real_to_integer): New function, returning a wide_int.
399         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
400         (ten_to_ptwo): Update call to real_from_integer.
401         (real_digit): Likewise.
402         * real.h: Include signop.h, wide-int.h and insn-modes.h.
403         (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
404         (REAL_VALUE_TO_INT): Delete.
405         (real_to_integer): Declare a wide-int form.
406         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
407         * recog.c (const_int_operand): Improve comment.
408         (const_scalar_int_operand): New.
409         (const_double_operand): Add a separate definition for CONST_WIDE_INT.
410         * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
411         (split_double): Likewise.
412         * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
413         (rtx_size): Likewise.
414         (rtx_alloc_stat_v): New.
415         (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
416         (cwi_output_hex): New.
417         (iterative_hash_rtx): Handle CONST_WIDE_INT.
418         (cwi_check_failed_bounds): New.
419         * rtl.def (CONST_WIDE_INT): New.
420         * rtl.h: Include <utility> and wide-int.h.
421         (struct hwivec_def): New.
422         (CWI_GET_NUM_ELEM): New.
423         (CWI_PUT_NUM_ELEM): New.
424         (struct rtx_def): Add num_elem and hwiv.
425         (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
426         (CASE_CONST_UNIQUE): Likewise.
427         (CASE_CONST_ANY): Likewise.
428         (CONST_SCALAR_INT_P): Likewise.
429         (CONST_WIDE_INT_P): New.
430         (CWI_ELT): New.
431         (HWIVEC_CHECK): New.
432         (cwi_check_failed_bounds): New.
433         (CWI_ELT): New.
434         (HWIVEC_CHECK): New.
435         (CONST_WIDE_INT_VEC) New.
436         (CONST_WIDE_INT_NUNITS) New.
437         (CONST_WIDE_INT_ELT) New.
438         (rtx_mode_t): New type.
439         (wi::int_traits <rtx_mode_t>): New.
440         (wi::shwi): New.
441         (wi::min_value): New.
442         (wi::max_value): New.
443         (rtx_alloc_v) New.
444         (const_wide_int_alloc): New.
445         (immed_wide_int_const): New.
446         * sched-vis.c (print_value): Handle CONST_WIDE_INT.
447         * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
448         * signop.h: New file.
449         * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
450         (simplify_const_unary_operation): Use wide-int interfaces.
451         (simplify_binary_operation_1): Likewise.
452         (simplify_const_binary_operation): Likewise.
453         (simplify_const_relational_operation): Likewise.
454         (simplify_immed_subreg): Likewise.
455         * stmt.c (expand_case): Likewise.
456         * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
457         signop rather than a bool.
458         * stor-layout.c (layout_type): Use wide-int interfaces.
459         (initialize_sizetypes): Update calls to
460         set_min_and_max_values_for_integral_type.
461         (set_min_and_max_values_for_integral_type): Take a signop rather
462         than a bool.  Use wide-int interfaces.
463         (fixup_signed_type): Update accordingly.  Remove
464         HOST_BITS_PER_DOUBLE_INT limit.
465         (fixup_unsigned_type): Likewise.
466         * system.h (STATIC_CONSTANT_P): New.
467         (STATIC_ASSERT): New.
468         * target.def (can_use_doloop_p): Take widest_ints rather than
469         double_ints.
470         * target.h: Include wide-int.h rather than double-int.h.
471         * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
472         than double_ints.
473         * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
474         rather than INT_CST_LT_UNSIGNED.
475         (can_use_doloop_if_innermost): Take widest_ints rather than
476         double_ints.
477         * tree-affine.c: Include wide-int-print.h.
478         (double_int_ext_for_comb): Delete.
479         (wide_int_ext_for_comb): New.
480         (aff_combination_zero): Use wide-int interfaces.
481         (aff_combination_const): Take a widest_int instead of a double_int.
482         (aff_combination_elt): Use wide-int interfaces.
483         (aff_combination_scale): Take a widest_int instead of a double_int.
484         (aff_combination_add_elt): Likewise.
485         (aff_combination_add_cst): Likewise.
486         (aff_combination_add): Use wide-int interfaces.
487         (aff_combination_convert): Likewise.
488         (tree_to_aff_combination): Likewise.
489         (add_elt_to_tree): Take a widest_int instead of a double_int.
490         (aff_combination_to_tree): Use wide-int interfaces.
491         (aff_combination_remove_elt): Likewise.
492         (aff_combination_add_product): Take a widest_int instead of
493         a double_int.
494         (aff_combination_mult): Use wide-int interfaces.
495         (aff_combination_expand): Likewise.
496         (double_int_constant_multiple_p): Delete.
497         (wide_int_constant_multiple_p): New.
498         (aff_combination_constant_multiple_p): Take a widest_int pointer
499         instead of a double_int pointer.
500         (print_aff): Use wide-int interfaces.
501         (get_inner_reference_aff): Take a widest_int pointer
502         instead of a double_int pointer.
503         (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
504         * tree-affine.h: Include wide-int.h.
505         (struct aff_comb_elt): Change type of coef to widest_int.
506         (struct affine_tree_combination): Change type of offset to widest_int.
507         (double_int_ext_for_comb): Delete.
508         (wide_int_ext_for_comb): New.
509         (aff_combination_const): Use widest_int instead of double_int.
510         (aff_combination_scale): Likewise.
511         (aff_combination_add_elt): Likewise.
512         (aff_combination_constant_multiple_p): Likewise.
513         (get_inner_reference_aff): Likewise.
514         (aff_comb_cannot_overlap_p): Likewise.
515         (aff_combination_zero_p): Use wide-int interfaces.
516         * tree.c: Include tree.h.
517         (init_ttree): Use make_int_cst.
518         (tree_code_size): Removed code for INTEGER_CST case.
519         (tree_size): Add INTEGER_CST case.
520         (make_node_stat): Update comment.
521         (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
522         (build_int_cst_type): Use wide-int interfaces.
523         (double_int_to_tree): Likewise.
524         (double_int_fits_to_tree_p): Delete.
525         (force_fit_type_double): Delete.
526         (force_fit_type): New.
527         (int_cst_hash_hash): Use wide-int interfaces.
528         (int_cst_hash_eq): Likewise.
529         (build_int_cst_wide): Delete.
530         (wide_int_to_tree): New.
531         (cache_integer_cst): Use wide-int interfaces.
532         (build_low_bits_mask): Likewise.
533         (cst_and_fits_in_hwi): Likewise.
534         (real_value_from_int_cst): Likewise.
535         (make_int_cst_stat): New.
536         (integer_zerop): Use wide_int interfaces.
537         (integer_onep): Likewise.
538         (integer_all_onesp): Likewise.
539         (integer_pow2p): Likewise.
540         (integer_nonzerop): Likewise.
541         (tree_log2): Likewise.
542         (tree_floor_log2): Likewise.
543         (tree_ctz): Likewise.
544         (int_size_in_bytes): Likewise.
545         (mem_ref_offset): Return an offset_int rather than a double_int.
546         (build_type_attribute_qual_variant): Use wide_int interfaces.
547         (type_hash_eq): Likewise
548         (tree_int_cst_equal): Likewise.
549         (tree_int_cst_lt): Delete.
550         (tree_int_cst_compare): Likewise.
551         (tree_fits_shwi_p): Use wide_int interfaces.
552         (tree_fits_uhwi_p): Likewise.
553         (tree_int_cst_sign_bit): Likewise.
554         (tree_int_cst_sgn): Likewise.
555         (tree_int_cst_min_precision): Take a signop rather than a bool.
556         (simple_cst_equal): Use wide_int interfaces.
557         (compare_tree_int): Likewise.
558         (iterative_hash_expr): Likewise.
559         (int_fits_type_p): Likewise.  Use tree_int_cst_lt rather than
560         INT_CST_LT.
561         (get_type_static_bounds): Use wide_int interfaces.
562         (tree_int_cst_elt_check_failed): New.
563         (build_common_tree_nodes): Reordered to set prec before filling in
564         value.
565         (int_cst_value): Check cst_and_fits_in_hwi.
566         (widest_int_cst_value): Use wide_int interfaces.
567         (upper_bound_in_type): Likewise.
568         (lower_bound_in_type): Likewise.
569         (num_ending_zeros): Likewise.
570         (drop_tree_overflow): Likewise.
571         * tree-call-cdce.c (check_pow): Update call to real_from_integer.
572         (gen_conditions_for_pow_cst_base): Likewise.
573         * tree-cfg.c: Include wide-int.h and wide-int-print.h.
574         (group_case_labels_stmt): Use wide-int interfaces.
575         (verify_gimple_assign_binary): Likewise.
576         (print_loop): Likewise.
577         * tree-chrec.c (tree_fold_binomial): Likewise.
578         * tree-core.h (struct tree_base): Add int_length.
579         (struct tree_int_cst): Change rep of value.
580         * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
581         (dr_may_alias_p): Likewise.
582         (max_stmt_executions_tree): Likewise.
583         * tree.def (INTEGER_CST): Update comment.
584         * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
585         * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
586         * tree-dump.c: Include wide-int.h and wide-int-print.h.
587         (dequeue_and_dump): Use wide-int interfaces.
588         * tree.h: Include wide-int.h.
589         (NULL_TREE): Moved to earlier loc in file.
590         (TREE_INT_CST_ELT_CHECK): New.
591         (tree_int_cst_elt_check_failed): New.
592         (TYPE_SIGN): New.
593         (TREE_INT_CST): Delete.
594         (TREE_INT_CST_LOW): Use wide-int interfaces.
595         (TREE_INT_CST_HIGH): Delete.
596         (TREE_INT_CST_NUNITS): New.
597         (TREE_INT_CST_EXT_NUNITS): Likewise.
598         (TREE_INT_CST_OFFSET_NUNITS): Likewise.
599         (TREE_INT_CST_ELT): Likewise.
600         (INT_CST_LT): Delete.
601         (tree_int_cst_elt_check): New (two forms).
602         (type_code_size): Update comment.
603         (make_int_cst_stat, make_int_cst): New.
604         (tree_to_double_int): Delete.
605         (double_int_fits_to_tree_p): Delete.
606         (force_fit_type_double): Delete.
607         (build_int_cstu): Replace with out-of-line function.
608         (build_int_cst_wide): Delete.
609         (tree_int_cst_lt): Define inline.
610         (tree_int_cst_le): New.
611         (tree_int_cst_compare): Define inline.
612         (tree_int_cst_min_precision): Take a signop rather than a bool.
613         (wi::int_traits <const_tree>): New.
614         (wi::int_traits <tree>): New.
615         (wi::extended_tree): New.
616         (wi::int_traits <wi::extended_tree>): New.
617         (wi::to_widest): New.
618         (wi::to_offset): New.
619         (wi::fits_to_tree_p): New.
620         (wi::min_value): New.
621         (wi::max_value): New.
622         * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
623         (copy_tree_body_r): Likewise.
624         * tree-object-size.c (compute_object_offset): Likewise.
625         (addr_object_size): Likewise.
626         * tree-predcom.c: Include wide-int-print.h.
627         (struct dref_d): Change type of offset to widest_int.
628         (dump_dref): Call wide-int printer.
629         (aff_combination_dr_offset): Use wide-int interfaces.
630         (determine_offset): Take a widest_int pointer rather than a
631         double_int pointer.
632         (split_data_refs_to_components): Use wide-int interfaces.
633         (suitable_component_p): Likewise.
634         (order_drefs): Likewise.
635         (add_ref_to_chain): Likewise.
636         (valid_initializer_p): Likewise.
637         (determine_roots_comp): Likewise.
638         * tree-pretty-print.c: Include wide-int-print.h.
639         (dump_generic_node): Use wide-int interfaces.
640         * tree-sra.c (sra_ipa_modify_expr): Likewise.
641         * tree-ssa-address.c (addr_for_mem_ref): Likewise.
642         (move_fixed_address_to_symbol): Likewise.
643         (move_hint_to_base): Likewise.
644         (move_pointer_to_base): Likewise.
645         (move_variant_to_index): Likewise.
646         (most_expensive_mult_to_index): Likewise.
647         (addr_to_parts): Likewise.
648         (copy_ref_info): Likewise.
649         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
650         (indirect_refs_may_alias_p): Likewise.
651         (stmt_kills_ref_p_1): Likewise.
652         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
653         * tree-ssa-ccp.c: Update comment at top of file.  Include
654         wide-int-print.h.
655         (struct prop_value_d): Change type of mask to widest_int.
656         (extend_mask): New function.
657         (dump_lattice_value): Use wide-int interfaces.
658         (get_default_value): Likewise.
659         (set_constant_value): Likewise.
660         (set_value_varying): Likewise.
661         (valid_lattice_transition): Likewise.
662         (set_lattice_value): Likewise.
663         (value_to_double_int): Delete.
664         (value_to_wide_int): New.
665         (get_value_from_alignment): Use wide-int interfaces.
666         (get_value_for_expr): Likewise.
667         (do_dbg_cnt): Likewise.
668         (ccp_finalize): Likewise.
669         (ccp_lattice_meet): Likewise.
670         (bit_value_unop_1): Use widest_ints rather than double_ints.
671         (bit_value_binop_1): Likewise.
672         (bit_value_unop): Use wide-int interfaces.
673         (bit_value_binop): Likewise.
674         (bit_value_assume_aligned): Likewise.
675         (evaluate_stmt): Likewise.
676         (ccp_fold_stmt): Likewise.
677         (visit_cond_stmt): Likewise.
678         (ccp_visit_stmt): Likewise.
679         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
680         (constant_pointer_difference): Likewise.
681         (associate_pointerplus): Likewise.
682         (combine_conversions): Likewise.
683         * tree-ssa-loop.h: Include wide-int.h.
684         (struct tree_niter_desc): Change type of max to widest_int.
685         * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
686         * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
687         (remove_redundant_iv_tests): Likewise.
688         (canonicalize_loop_induction_variables): Likewise.
689         * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
690         (constant_multiple_of): Take a widest_int pointer instead of
691         a double_int pointer.
692         (get_computation_aff): Use wide-int interfaces.
693         (ptr_difference_cost): Likewise.
694         (difference_cost): Likewise.
695         (get_loop_invariant_expr_id): Likewise.
696         (get_computation_cost_at): Likewise.
697         (iv_elimination_compare_lt): Likewise.
698         (may_eliminate_iv): Likewise.
699         * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
700         instead of double_int.
701         (max_loop_iterations): Likewise.
702         (max_stmt_executions): Likewise.
703         (estimated_stmt_executions): Likewise.
704         * tree-ssa-loop-niter.c: Include wide-int-print.h.
705         (split_to_var_and_offset): Use wide-int interfaces.
706         (determine_value_range): Likewise.
707         (bound_difference_of_offsetted_base): Likewise.
708         (bounds_add): Take a widest_int instead of a double_int.
709         (number_of_iterations_ne_max): Use wide-int interfaces.
710         (number_of_iterations_ne): Likewise.
711         (number_of_iterations_lt_to_ne): Likewise.
712         (assert_loop_rolls_lt): Likewise.
713         (number_of_iterations_lt): Likewise.
714         (number_of_iterations_le): Likewise.
715         (number_of_iterations_cond): Likewise.
716         (number_of_iterations_exit): Likewise.
717         (finite_loop_p): Likewise.
718         (derive_constant_upper_bound_assign): Likewise.
719         (derive_constant_upper_bound): Return a widest_int.
720         (derive_constant_upper_bound_ops): Likewise.
721         (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
722         (record_estimate): Take a widest_int rather than a double_int.
723         (record_nonwrapping_iv): Use wide-int interfaces.
724         (double_int_cmp): Delete.
725         (wide_int_cmp): New.
726         (bound_index): Take a widest_int rather than a double_int.
727         (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
728         (maybe_lower_iteration_bound): Likewise.
729         (estimate_numbers_of_iterations_loop): Likewise.
730         (estimated_loop_iterations): Take a widest_int pointer than than
731         a double_int pointer.
732         (estimated_loop_iterations_int): Use wide-int interfaces.
733         (max_loop_iterations): Take a widest_int pointer than than
734         a double_int pointer.
735         (max_loop_iterations_int): Use wide-int interfaces.
736         (max_stmt_executions): Take a widest_int pointer than than
737         a double_int pointer.
738         (estimated_stmt_executions): Likewise.
739         (n_of_executions_at_most): Use wide-int interfaces.
740         (scev_probably_wraps_p): Likewise.
741         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
742         to real_to_integer.
743         * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
744         interfaces.
745         * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
746         double_ints.  Adjust for trailing_wide_ints <3> representation.
747         (set_nonzero_bits): Likewise.
748         (get_range_info): Return wide_ints rather than double_ints.
749         Adjust for trailing_wide_ints <3> representation.
750         (get_nonzero_bits): Likewise.
751         (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
752         representation.
753         * tree-ssanames.h (struct range_info_def): Replace min, max and
754         nonzero_bits with a trailing_wide_ints <3>.
755         (set_range_info): Use wide_int_refs rather than double_ints.
756         (set_nonzero_bits): Likewise.
757         (get_range_info): Return wide_ints rather than double_ints.
758         (get_nonzero_bits): Likewise.
759         * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
760         * tree-ssa-pre.c (phi_translate_1): Likewise.
761         * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
762         (acceptable_pow_call): Likewise.
763         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
764         interfaces.
765         (vn_reference_fold_indirect): Likewise.
766         (vn_reference_maybe_forwprop_address): Likewise.
767         (valueize_refs_1): Likewise.
768         * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
769         * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
770         tree_int_cst_lt and tree_int_cst_le.
771         * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
772         interfaces.
773         (streamer_alloc_tree): Likewise.
774         * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
775         (streamer_write_tree_header): Likewise.
776         (streamer_write_integer_cst): Likewise.
777         * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
778         (build_constructors): Likewise.
779         (array_value_type): Likewise.
780         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
781         (vect_check_gather): Likewise.
782         * tree-vect-generic.c (build_replicated_const): Likewise.
783         (expand_vector_divmod): Likewise.
784         * tree-vect-loop.c (vect_transform_loop): Likewise.
785         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
786         (vect_do_peeling_for_alignment): Likewise.
787         * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
788         * tree-vrp.c: Include wide-int.h.
789         (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
790         (extract_range_from_assert): Use wide-int interfaces.
791         (vrp_int_const_binop): Likewise.
792         (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
793         double_int pointers.
794         (ranges_from_anti_range): Use wide-int interfaces.
795         (quad_int_cmp): Delete.
796         (quad_int_pair_sort): Likewise.
797         (extract_range_from_binary_expr_1): Use wide-int interfaces.
798         (extract_range_from_unary_expr_1): Likewise.
799         (adjust_range_with_scev): Likewise.
800         (masked_increment): Take and return wide_ints rather than double_ints.
801         (register_edge_assert_for_2): Use wide-int interfaces.
802         (check_array_ref): Likewise.
803         (search_for_addr_array): Likewise.
804         (maybe_set_nonzero_bits): Likewise.
805         (union_ranges): Pass an integer of the correct type instead of
806         using integer_one_node.
807         (intersect_ranges): Likewise.
808         (simplify_truth_ops_using_ranges): Likewise.
809         (simplify_bit_ops_using_ranges): Use wide-int interfaces.
810         (range_fits_type_p): Likewise.
811         (simplify_cond_using_ranges): Likewise.  Take a signop rather than
812         a bool.
813         (simplify_conversion_using_ranges): Use wide-int interfaces.
814         (simplify_float_conversion_using_ranges): Likewise.
815         (vrp_finalize): Likewise.
816         * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
817         (gimple_stringops_transform): Likewise.
818         * varasm.c (decode_addr_const): Likewise.
819         (const_hash_1): Likewise.
820         (const_rtx_hash_1): Likewise
821         (output_constant): Likewise.
822         (array_size_for_constructor): Likewise.
823         (output_constructor_regular_field): Likewise.
824         (output_constructor_bitfield): Likewise.
825         * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
826         * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
827         GENERATOR_FILEs.
828         * gencheck.c: Define BITS_PER_UNIT.
829         * wide-int.cc: New.
830         * wide-int.h: New.
831         * wide-int-print.cc: New.
832         * wide-int-print.h: New.
834 2014-05-06  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
836         * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
838 2014-05-06  Richard Biener  <rguenther@suse.de>
840         * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
841         TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
842         (TODO_verify_all): Adjust.
843         * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
844         TODO_verify_stmts and TODO_verify_rtl_sharing.
845         * bb-reorder.c: Likewise.
846         * cfgexpand.c: Likewise.
847         * cprop.c: Likewise.
848         * cse.c: Likewise.
849         * function.c: Likewise.
850         * fwprop.c: Likewise.
851         * gcse.c: Likewise.
852         * gimple-ssa-isolate-paths.c: Likewise.
853         * gimple-ssa-strength-reduction.c: Likewise.
854         * ipa-split.c: Likewise.
855         * loop-init.c: Likewise.
856         * loop-unroll.c: Likewise.
857         * lower-subreg.c: Likewise.
858         * modulo-sched.c: Likewise.
859         * postreload-gcse.c: Likewise.
860         * predict.c: Likewise.
861         * recog.c: Likewise.
862         * sched-rgn.c: Likewise.
863         * store-motion.c: Likewise.
864         * tracer.c: Likewise.
865         * trans-mem.c: Likewise.
866         * tree-call-cdce.c: Likewise.
867         * tree-cfg.c: Likewise.
868         * tree-cfgcleanup.c: Likewise.
869         * tree-complex.c: Likewise.
870         * tree-eh.c: Likewise.
871         * tree-emutls.c: Likewise.
872         * tree-if-conv.c: Likewise.
873         * tree-into-ssa.c: Likewise.
874         * tree-loop-distribution.c: Likewise.
875         * tree-object-size.c: Likewise.
876         * tree-parloops.c: Likewise.
877         * tree-pass.h: Likewise.
878         * tree-sra.c: Likewise.
879         * tree-ssa-ccp.c: Likewise.
880         * tree-ssa-copy.c: Likewise.
881         * tree-ssa-copyrename.c: Likewise.
882         * tree-ssa-dce.c: Likewise.
883         * tree-ssa-dom.c: Likewise.
884         * tree-ssa-dse.c: Likewise.
885         * tree-ssa-forwprop.c: Likewise.
886         * tree-ssa-ifcombine.c: Likewise.
887         * tree-ssa-loop-ch.c: Likewise.
888         * tree-ssa-loop-ivcanon.c: Likewise.
889         * tree-ssa-loop.c: Likewise.
890         * tree-ssa-math-opts.c: Likewise.
891         * tree-ssa-phiopt.c: Likewise.
892         * tree-ssa-phiprop.c: Likewise.
893         * tree-ssa-pre.c: Likewise.
894         * tree-ssa-reassoc.c: Likewise.
895         * tree-ssa-sink.c: Likewise.
896         * tree-ssa-strlen.c: Likewise.
897         * tree-ssa-tail-merge.c: Likewise.
898         * tree-ssa-uncprop.c: Likewise.
899         * tree-switch-conversion.c: Likewise.
900         * tree-tailcall.c: Likewise.
901         * tree-vect-generic.c: Likewise.
902         * tree-vectorizer.c: Likewise.
903         * tree-vrp.c: Likewise.
904         * tsan.c: Likewise.
905         * var-tracking.c: Likewise.
906         * bt-load.c: Likewise.
907         * cfgcleanup.c: Likewise.
908         * combine-stack-adj.c: Likewise.
909         * combine.c: Likewise.
910         * compare-elim.c: Likewise.
911         * config/epiphany/resolve-sw-modes.c: Likewise.
912         * config/i386/i386.c: Likewise.
913         * config/mips/mips.c: Likewise.
914         * config/s390/s390.c: Likewise.
915         * config/sh/sh_treg_combine.cc: Likewise.
916         * config/sparc/sparc.c: Likewise.
917         * dce.c: Likewise.
918         * dse.c: Likewise.
919         * final.c: Likewise.
920         * ifcvt.c: Likewise.
921         * mode-switching.c: Likewise.
922         * passes.c: Likewise.
923         * postreload.c: Likewise.
924         * ree.c: Likewise.
925         * reg-stack.c: Likewise.
926         * regcprop.c: Likewise.
927         * regrename.c: Likewise.
928         * web.c: Likewise.
930 2014-05-06  Richard Biener  <rguenther@suse.de>
932         PR middle-end/61070
933         * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
934         * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
936 2014-05-05  Jan Hubicka  <hubicka@ucw.cz>
938         PR ipa/60965
939         * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
941 2014-05-05  Radovan Obradovic  <robradovic@mips.com>
942             Tom de Vries  <tom@codesourcery.com>
944         * target.def (call_fusage_contains_non_callee_clobbers): New DEFHOOKPOD.
945         * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
946         Hooks to @menu.
947         (@node Miscellaneous Register Hooks): New node.
948         (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
949         * doc/tm.texi: Regenerate.
951 2014-05-05  Marek Polacek  <polacek@redhat.com>
953         PR driver/61065
954         * opts.c (common_handle_option): Call error_at instead of warning_at.
956 2014-05-05  Richard Biener  <rguenther@suse.de>
958         * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
959         from last_verified if update_ssa ran.  Move TODO_verify_rtl_sharing
960         under the TODO_verify_il umbrella.
962 2014-05-05  Richard Biener  <rguenther@suse.de>
964         * passes.c (execute_function_todo): Move TODO_verify_flow under
965         the TODO_verify_ul umbrella.
967 2014-05-05  Richard Biener  <rguenther@suse.de>
969         PR middle-end/61010
970         * fold-const.c (fold_binary_loc): Consistently avoid
971         canonicalizing X & CST away from a CST that is the mask
972         of a mode.
974 2014-05-05  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
976         * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
977         int argument to enum machine_mode.
978         (picochip_class_max_nregs): Ditto.
979         * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
980         (picochip_class_max_nregs): Ditto.
982 2014-05-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
984         * target.def: Add new target hook.
985         * doc/tm.texi: Regenerate.
986         * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
987         * targhooks.c (default_keep_leaf_when_profiled): New function.
989         * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
990         (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
992 2014-05-05  Bin Cheng  <bin.cheng@arm.com>
994         PR tree-optimization/60363
995         * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
996         (copy_phi_args): New parameters.  Call get_value_locus_in_path.
997         (update_destination_phis): New parameter.
998         (create_edge_and_update_destination_phis): Ditto.
999         (ssa_fix_duplicate_block_edges): Pass new arguments.
1000         (thread_single_edge): Ditto.
1002 2014-05-04  Peter Bergner  <bergner@vnet.ibm.com>
1004         * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
1005         (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
1006         (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
1007         * config/rs6000/rs6000-builtin.def (BU_MISC_1):
1008         Use RS6000_BTM_HARD_FLOAT.
1009         (BU_MISC_2): Likewise.
1010         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
1011         RS6000_BTM_HARD_FLOAT.
1012         (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
1013         is explicitly used.
1014         (rs6000_invalid_builtin): Add hard floating builtin support.
1015         (rs6000_expand_builtin): Relax the gcc_assert to allow the new
1016         hard float builtins.
1017         (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
1019 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1021         * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
1022         Add missing function* argument.
1024 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1026         * lra-constraints.c (valid_address_p): Move earlier in file.
1027         Add a constraint argument to the address_info version.
1028         (satisfies_memory_constraint_p): New function.
1029         (satisfies_address_constraint_p): Likewise.
1030         (process_alt_operands, curr_insn_transform): Use them.
1031         (process_address): Pass the constraint to valid_address_p when
1032         checking address operands.
1034 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1036         * config/mips/mips.c (mips_isa_rev): New variable.
1037         (mips_set_architecture): Set it.
1038         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
1039         from mips_isa_rev.
1040         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
1041         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
1042         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
1043         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
1044         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
1045         conditions in terms of mips_isa_rev.
1046         (mips_isa_rev): Declare.
1048 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1050         * config/sh/sh-mem.cc: Use tabs instead of spaces.
1051         (prob_unlikely, prob_likely): Make variables const.
1053 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
1055         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
1057 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1059         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
1061 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1063         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
1064         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
1065         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
1066         functions.
1067         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
1068         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
1069         sh_pass_in_reg_p.
1070         Replace usage of ROUND_REG with sh_round_reg.
1071         Use CEIL instead of ROUND_ADVANCE.
1073 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1075         PR target/61026
1076         * config/sh/sh.c: Include stdlib headers before everything else.
1078 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
1080         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
1081         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
1082         (gimplify_adjust_omp_clauses): Simd region is never
1083         directly nested in combined parallel.  Instead, for linear
1084         with copyin/copyout, if in combined for simd loop, make decl
1085         firstprivate/lastprivate on OMP_FOR.
1086         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
1087         expand_omp_for_static_chunk): When setting endvar, also set
1088         fd->loop.v to the same value.
1090 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1092         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
1094 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
1096         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
1097         expression.
1099 2014-05-02  Marek Polacek  <polacek@redhat.com>
1101         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
1103 2014-05-02  Kito Cheng  <kito@0xlab.org>
1105         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
1106         to a C expression marco.
1107         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
1108         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
1109         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
1110         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
1111         HONOR_REG_ALLOC_ORDER.
1112         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
1114 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1116         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
1118 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1120         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
1122 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
1124         * tree-if-conv.c (is_cond_scalar_reduction): New function.
1125         (convert_scalar_cond_reduction): Likewise.
1126         (predicate_scalar_phi): Add recognition and transformation
1127         of simple conditioanl reduction to be vectorizable.
1129 2014-05-01  Marek Polacek  <polacek@redhat.com>
1131         PR c/43245
1132         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
1134 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
1136         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
1137         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
1138         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
1139         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
1140         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
1141         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
1142         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
1143         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
1145 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
1147         * config/arc/arc.opt (mlra): Move comment above option name
1148         to avoid mis-parsing as language options.
1150 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1152         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
1153         * config/sol2.h: ... here.
1154         * config/sol2-10.h: Remove.
1156         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
1157         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
1158         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
1159         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
1160         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
1161         * config/sol2.h: ... here.
1162         (SECTION_NAME_FORMAT): Don't redefine.
1163         (STARTFILE_ARCH32_SPEC): Rename to ...
1164         (STARTFILE_ARCH_SPEC): ... this.
1165         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
1166         * config/sparc/sol2.h: ... here.
1167         (SECTION_NAME_FORMAT): Don't undef.
1168         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
1169         (SUBTARGET_EXTRA_SPECS): Remove.
1170         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
1172         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
1173         (MD_STARTFILE_PREFIX): Remove.
1174         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
1175         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
1176         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
1177         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
1178         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
1179         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
1180         * config/i386/sol2.h: ... here.
1181         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
1182         * config/i386/sol2-bi.h: Remove.
1183         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
1184         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
1186         * config/i386/t-sol2-64: Rename to ...
1187         * config/i386/t-sol2: ... this.
1188         * config/sparc/t-sol2-64: Rename to ...
1189         * config/sparc/t-sol2: ... this.
1191         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
1192         sol2_tm_file_head, sol2_tm_file_tail.
1193         Include ${cpu_type}/sol2.h before sol2.h.
1194         Remove sol2-10.h.
1195         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
1196         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
1197         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
1198         Reflect i386/t-sol2-64 renaming.
1199         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
1200         Reflect sparc/t-sol2-64 renaming.
1202 2014-04-30  Richard Biener  <rguenther@suse.de>
1204         * passes.c (execute_function_todo): Move TODO_verify_stmts
1205         and TODO_verify_ssa under the TODO_verify_il umbrella.
1206         * tree-ssa.h (verify_ssa): Adjust prototype.
1207         * tree-ssa.c (verify_ssa): Add parameter to tell whether
1208         we should verify SSA operands.
1209         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
1210         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
1211         whether we should verify whether not throwing stmts have EH info.
1212         * graphite-scop-detection.c (create_sese_edges): Adjust.
1213         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
1214         * tree-eh.c (lower_try_finally_switch): Do not add the
1215         default case label twice.
1217 2014-04-30  Marek Polacek  <polacek@redhat.com>
1219         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
1220         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
1221         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
1222         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
1224 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
1226         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
1227         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
1228         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
1229         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
1230         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
1231         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
1232         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
1233         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
1235 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
1237         * tree-cfg.c (dump_function_to_file): Dump the return type of
1238         functions, in a line to itself before the function body, mimicking
1239         the layout of a C function.
1241 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
1243         PR tree-optimization/60971
1244         * tree-tailcall.c (process_assignment): Reject conversions which
1245         reduce precision.
1247 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
1249         * calls.c (initialize_argument_information): Always treat
1250         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
1251         (expand_call): Likewise.
1252         (emit_library_call_calue_1): Likewise.
1253         * expr.c (PUSH_ARGS_REVERSED): Do not define.
1254         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
1255         code accordingly.
1257 2014-04-29  Nick Clifton  <nickc@redhat.com>
1259         * config/msp430/msp430.md (umulsidi): Fix typo.
1260         (mulhisi3): Enable even inside interrupt handlers.
1261         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
1262         bigger return address pushed in large mode.
1264 2014-04-29  Nick Clifton  <nickc@redhat.com>
1266         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
1267         (arc_init_reg_tables): Use a machine_mode enum to iterate over
1268         available modes.
1269         * config/m32r/m32r.c (init_reg_tables): Likewise.
1270         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
1271         enum to hold the modes.
1273 2014-04-29  Richard Biener  <rguenther@suse.de>
1275         * dominance.c (free_dominance_info): Add overload with
1276         function parameter.
1277         (dom_info_state): Likewise.
1278         (dom_info_available_p): Likewise.
1279         * basic-block.h (free_dominance_info, dom_info_state,
1280         dom_info_available_p): Declare overloads.
1281         * passes.c (execute_function_todo): Verify that verifiers
1282         don't change dominator info state.  Drop dominator info
1283         for IPA pass invocations.
1284         * cgraph.c (release_function_body): Restore asserts that
1285         dominator information is released.
1287 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
1289         * doc/invoke.texi: Fix typo.
1290         * tree-vrp.c: Fix typos.
1291         * gimple.c (infer_nonnull_range): Reorder operands of an &&
1292         condition.
1294 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1296         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
1298 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
1300         * config/aarch64/aarch64-builtins.c
1301         (aarch64_types_storestruct_lane_qualifiers): New.
1302         (TYPES_STORESTRUCT_LANE): Likewise.
1303         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
1304         (st3_lane): Likewise.
1305         (st4_lane): Likewise.
1306         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
1307         (vec_store_lanesci_lane<mode>): Likewise.
1308         (vec_store_lanesxi_lane<mode>): Likewise.
1309         (aarch64_st2_lane<VQ:mode>): Likewise.
1310         (aarch64_st3_lane<VQ:mode>): Likewise.
1311         (aarch64_st4_lane<VQ:mode>): Likewise.
1312         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
1313         * config/aarch64/arm_neon.h
1314         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
1315         use new macro arguments.
1316         (__ST3_LANE_FUNC): Likewise.
1317         (__ST4_LANE_FUNC): Likewise.
1318         * config/aarch64/iterators.md (V_TWO_ELEM): New.
1319         (V_THREE_ELEM): Likewise.
1320         (V_FOUR_ELEM): Likewise.
1322 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
1324         * doc/gimple.texi: Replace the description of the now-defunct
1325         union gimple_statement_d with a diagram showing the
1326         gimple_statement_base class hierarchy and its relationships to
1327         the GSS_ and GIMPLE_ enums.
1329 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
1331         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
1332         * config/aarch64/aarch64.c
1333         (aarch64_cannot_change_mode_class): Weaken conditions.
1334         (aarch64_modes_tieable_p): New.
1335         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
1337 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
1339         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1340         (loadsync_<mode>): Change mode.
1341         (load_quadpti, store_quadpti): New.
1342         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1343         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1345 2014-04-28  Martin Jambor  <mjambor@suse.cz>
1347         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
1348         same alias type as the original statement.
1349         (subreplacement_assignment_data): New type.
1350         (handle_unscalarized_data_in_subtree): New type of parameter,
1351         generate new memory accesses with same alias type as the original
1352         statement.
1353         (load_assign_lhs_subreplacements): Likewise.
1354         (sra_modify_constructor_assign): Generate new memory accesses with
1355         same alias type as the original statement.
1357 2014-04-28  Richard Biener  <rguenther@suse.de>
1359         * tree-pass.h (TODO_verify_il): Define.
1360         (TODO_verify_all): Complete properly.
1361         * passes.c (execute_function_todo): Move existing loop-closed
1362         SSA verification under TODO_verify_il.
1363         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
1364         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
1365         Fix tree sharing issue.
1367 2014-04-28  Richard Biener  <rguenther@suse.de>
1369         PR middle-end/60092
1370         * builtins.def (DEF_C11_BUILTIN): Add.
1371         (BUILT_IN_ALIGNED_ALLOC): Likewise.
1372         * coretypes.h (enum function_class): Add function_c11_misc.
1373         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
1374         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
1375         (call_may_clobber_ref_p_1): Likewise.
1376         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1377         (mark_all_reaching_defs_necessary_1): Likewise.
1378         (propagate_necessity): Likewise.
1379         (eliminate_unnecessary_stmts): Likewise.
1380         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
1382 2014-04-28  Richard Biener  <rguenther@suse.de>
1384         * tree-vrp.c (vrp_var_may_overflow): Remove.
1385         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
1386         with overflow immediately bump to one before that value and
1387         let iteration figure out overflow status.
1389 2014-04-28  Richard Biener  <rguenther@suse.de>
1391         * configure.ac: Do valgrind header checks unconditionally.
1392         Add --enable-valgrind-annotations.
1393         * system.h: Guard valgrind header inclusion with
1394         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
1395         * alloc-pool.c (pool_alloc, pool_free): Use
1396         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
1397         to guard possibly dead code.
1398         * config.in: Regenerated.
1399         * configure: Likewise.
1401 2014-04-28  Jeff Law  <law@redhat.com>
1403         PR tree-optimization/60902
1404         * tree-ssa-threadedge.c
1405         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
1406         over real defs when invalidating outputs from statements that do not
1407         produce useful outputs for threading.
1409 2014-04-28  Richard Biener  <rguenther@suse.de>
1411         PR tree-optimization/60979
1412         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
1413         SCOPs that end in a block with a successor with abnormal
1414         predecessors.
1416 2014-04-28  Richard Biener  <rguenther@suse.de>
1418         * tree-pass.h (execute_pass_list): Adjust prototype.
1419         * passes.c (pass_manager::execute_early_local_passes):
1420         Adjust.
1421         (do_per_function): Change callback signature, push all actual
1422         work to the callbals.
1423         (do_per_function_toporder): Likewise.
1424         (execute_function_dump): Adjust.
1425         (execute_function_todo): Likewise.
1426         (clear_last_verified): Likewise.
1427         (verify_curr_properties): Likewise.
1428         (update_properties_after_pass): Likewise.
1429         (execute_pass_list_1): Split out from ...
1430         (execute_pass_list): ... here.  Adjust.
1431         (execute_ipa_pass_list): Likewise.
1432         * cgraphunit.c (cgraph_add_new_function): Adjust.
1433         (analyze_function): Likewise.
1434         (expand_function): Likewise.
1435         * cgraph.c (release_function_body): Free dominance info
1436         here instead of asserting it was magically freed elsewhere.
1438 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
1440         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
1441         * configure: Regenerate.
1442         * config/sparc/sparc.opt (muser-mode): New option.
1443         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
1444         for LEON3.
1445         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
1446         * doc/invoke.texi (SPARC options): Document -muser-mode.
1448 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
1450         * cselib.c (find_slot_memmode): Delete.
1451         (cselib_hasher): Change compare_type to a struct.
1452         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
1453         constants.
1454         (preserve_constants_and_equivs): Adjust for new compare_type.
1455         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
1456         (wrap_constant): Delete.
1457         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
1459 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
1461         * doc/install.texi (Building with profile feedback): Remove
1462         outdated sentence.
1464 2014-04-26  Tom de Vries  <tom@codesourcery.com>
1466         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
1467         array accesses.
1469 2014-04-25  Cary Coutant  <ccoutant@google.com>
1471         PR debug/60929
1472         * dwarf2out.c (should_move_die_to_comdat): A type definition
1473         can contain a subprogram definition, but don't move it to a
1474         comdat unit.
1475         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
1476         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
1477         from original DIE.
1478         (clone_tree_hash): Rename to...
1479         (clone_tree_partial): ...this; change callers.  Copy
1480         DW_TAG_subprogram DIEs as declarations.
1481         (copy_decls_walk): Don't copy children of a declaration into a
1482         type unit.
1484 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
1486         PR target/60969
1487         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
1488         alternative 12.
1490 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
1492         * config/arm/predicates.md (call_insn_operand): Add long_call check.
1493         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
1494         reg for long_call.
1495         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
1496         restriction.
1498 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1500         * config/arm/arm.c (arm_cortex_a8_tune): Initialise
1501         T16-related fields.
1503 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1505         PR tree-optimization/60930
1506         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
1507         creating a multiply candidate by folding two constant
1508         multiplicands when the result overflows.
1510 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
1512         PR tree-optimization/60960
1513         * tree-vect-generic.c (expand_vector_operation): Only call
1514         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
1516 2014-04-25  Tom de Vries  <tom@codesourcery.com>
1518         * expr.c (clobber_reg_mode): New function.
1519         * expr.h (clobber_reg): New function.
1521 2014-04-25  Tom de Vries  <tom@codesourcery.com>
1523         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
1524         clobbers.
1526 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
1527             Tom de Vries  <tom@codesourcery.com>
1529         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
1530         handle.
1531         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
1532         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
1533         new argument to find_all_hard_reg_sets call.
1535 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1537         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
1538         Use HOST_WIDE_INT_C for mask literal.
1539         (aarch_rev16_shleft_mask_imm_p): Likewise.
1541 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
1543         PR target/60941
1544         * config/sparc/sparc.md (ashlsi3_extend): Delete.
1546 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
1548         PR preprocessor/56540
1549         * config/i386/i386-c.c (ix86_target_macros): Define
1550         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
1552 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1554         * configure.ac (tga_func): Remove.
1555         (LIB_TLS_SPEC): Remove.
1556         * configure: Regenerate.
1557         * config.in: Regenerate.
1558         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
1560 2014-04-25  Richard Biener  <rguenther@suse.de>
1562         PR ipa/60912
1563         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
1564         call stmt use/clobber sets during stmt walk instead of
1565         walking the possibly incomplete set of caller edges.
1567 2014-04-25  Richard Biener  <rguenther@suse.de>
1569         PR ipa/60911
1570         * passes.c (apply_ipa_transforms): Inline into only caller ...
1571         (execute_one_pass): ... here.  Properly bring in function
1572         bodies for nodes we want to apply IPA transforms to.
1574 2014-04-24  Cong Hou  <congh@google.com>
1576         PR tree-optimization/60896
1577         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
1578         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
1579         (vect_mark_pattern_stmts): Set the def type of all statements in
1580         PATTERN_DEF_SEQ as vect_internal_def.
1582 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
1584         * doc/extend.texi (PowerPC Built-in Functions): Document new
1585         powerpc extended divide, bcd, pack/unpack 128-bit, builtin
1586         functions.
1587         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
1589         * config/rs6000/predicates.md (const_0_to_3_operand): New
1590         predicate to match 0..3 integer constants.
1592         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
1593         to support adding miscellaneous builtin functions.
1594         (BU_DFP_MISC_2): Likewise.
1595         (BU_P7_MISC_1): Likewise.
1596         (BU_P7_MISC_2): Likewise.
1597         (BU_P8V_MISC_3): Likewise.
1598         (BU_MISC_1): Likewise.
1599         (BU_MISC_2): Likewise.
1600         (DIVWE): Add extended divide builtin functions.
1601         (DIVWEO): Likewise.
1602         (DIVWEU): Likewise.
1603         (DIVWEUO): Likewise.
1604         (DIVDE): Likewise.
1605         (DIVDEO): Likewise.
1606         (DIVDEU): Likewise.
1607         (DIVDEUO): Likewise.
1608         (DXEX): Add decimal floating-point builtin functions.
1609         (DXEXQ): Likewise.
1610         (DDEDPD): Likewise.
1611         (DDEDPDQ): Likewise.
1612         (DENBCD): Likewise.
1613         (DENBCDQ): Likewise.
1614         (DIEX): Likewise.
1615         (DIEXQ): Likewise.
1616         (DSCLI): Likewise.
1617         (DSCLIQ): Likewise.
1618         (DSCRI): Likewise.
1619         (DSCRIQ): Likewise.
1620         (CDTBCD): Add new BCD builtin functions.
1621         (CBCDTD): Likewise.
1622         (ADDG6S): Likewise.
1623         (BCDADD): Likewise.
1624         (BCDADD_LT): Likewise.
1625         (BCDADD_EQ): Likewise.
1626         (BCDADD_GT): Likewise.
1627         (BCDADD_OV): Likewise.
1628         (BCDSUB): Likewise.
1629         (BCDSUB_LT): Likewise.
1630         (BCDSUB_EQ): Likewise.
1631         (BCDSUB_GT): Likewise.
1632         (BCDSUB_OV): Likewise.
1633         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
1634         (UNPACK_TD): Likewise.
1635         (PACK_TF): Likewise.
1636         (UNPACK_TF): Likewise.
1637         (UNPACK_TF_0): Likewise.
1638         (UNPACK_TF_1): Likewise.
1639         (PACK_V1TI): Likewise.
1640         (UNPACK_V1TI): Likewise.
1642         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1643         support for decimal floating point builtin functions.
1644         (rs6000_expand_ternop_builtin): Add checks for the new builtin
1645         functions that take constant arguments.
1646         (rs6000_invalid_builtin): Add decimal floating point builtin
1647         support.
1648         (rs6000_init_builtins): Setup long double, _Decimal64, and
1649         _Decimal128 types for new builtin functions.
1650         (builtin_function_type): Set the unsigned flags appropriately for
1651         the new builtin functions.
1652         (rs6000_opt_masks): Add support for decimal floating point builtin
1653         functions.
1655         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
1656         floating point builtin functions.
1657         (RS6000_BTM_COMMON): Likewise.
1658         (RS6000_BTI_long_double): Likewise.
1659         (RS6000_BTI_dfloat64): Likewise.
1660         (RS6000_BTI_dfloat128): Likewise.
1661         (long_double_type_internal_node): Likewise.
1662         (dfloat64_type_internal_node): Likewise.
1663         (dfloat128_type_internal_node): Likewise.
1665         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
1666         2.07 bcd arithmetic instructions.
1667         (UNSPEC_BCDSUB): Likewise.
1668         (UNSPEC_BCD_OVERFLOW): Likewise.
1669         (UNSPEC_BCD_ADD_SUB): Likewise.
1670         (bcd_add_sub): Likewise.
1671         (BCD_TEST): Likewise.
1672         (bcd<bcd_add_sub>): Likewise.
1673         (bcd<bcd_add_sub>_test): Likewise.
1674         (bcd<bcd_add_sub>_test2): Likewise.
1675         (bcd<bcd_add_sub>_<code>): Likewise.
1676         (peephole2 for combined bcd ops): Likewise.
1678         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
1679         decimal floating point builtin functions.
1680         (UNSPEC_DENBCD): Likewise.
1681         (UNSPEC_DXEX): Likewise.
1682         (UNSPEC_DIEX): Likewise.
1683         (UNSPEC_DSCLI): Likewise.
1684         (UNSPEC_DSCRI): Likewise.
1685         (D64_D128): Likewise.
1686         (dfp_suffix): Likewise.
1687         (dfp_ddedpd_<mode>): Likewise.
1688         (dfp_denbcd_<mode>): Likewise.
1689         (dfp_dxex_<mode>): Likewise.
1690         (dfp_diex_<mode>): Likewise.
1691         (dfp_dscli_<mode>): Likewise.
1692         (dfp_dscri_<mode>): Likewise.
1694         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
1695         builtin functions.
1696         (UNSPEC_CDTBCD): Likewise.
1697         (UNSPEC_CBCDTD): Likewise.
1698         (UNSPEC_DIVE): Add support for new extended divide builtin
1699         functions.
1700         (UNSPEC_DIVEO): Likewise.
1701         (UNSPEC_DIVEU): Likewise.
1702         (UNSPEC_DIVEUO): Likewise.
1703         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
1704         pack/unpack 128-bit types.
1705         (UNSPEC_PACK_128BIT): Likewise.
1706         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
1707         (udiv<mode>3): Use idiv_ldiv mode attribute.
1708         (div<mode>3): Likewise.
1709         (addg6s): Add new BCD builtin functions.
1710         (cdtbcd): Likewise.
1711         (cbcdtd): Likewise.
1712         (UNSPEC_DIV_EXTEND): Add support for new extended divide
1713         instructions.
1714         (div_extend): Likewise.
1715         (div<div_extend>_<mode>"): Likewise.
1716         (FP128_64): Add support for new builtin functions to pack/unpack
1717         128-bit types.
1718         (unpack<mode>): Likewise.
1719         (unpacktf_0): Likewise.
1720         (unpacktf_1): Likewise.
1721         (unpack<mode>_dm): Likewise.
1722         (unpack<mode>_nodm): Likewise.
1723         (pack<mode>): Likewise.
1724         (unpackv1ti): Likewise.
1725         (packv1ti): Likewise.
1727 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
1729         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
1730         is disabled.
1732 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
1734         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
1735         * gimplify.c (omp_is_private): Change last argument's type to int.
1736         Only diagnose lastprivate if the simd argument is 1, only diagnose
1737         linear if the simd argument is 2.
1738         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
1739         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
1740         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
1741         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
1742         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
1743         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
1744         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
1745         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
1746         * tree-nested.c (convert_nonlocal_omp_clauses,
1747         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
1749 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
1751         PR target/60822
1752         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
1753         operand 1.
1755 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
1757         * flag-types.h (enum ivar_visibility): Add.
1759 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
1761         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
1762         function * argument.
1764 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
1766         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
1768 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
1769             Tom de Vries  <tom@codesourcery.com>
1771         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
1772         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
1773         reg-note.
1774         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
1775         * emit-rtl.c (try_split): Same.
1777 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
1778             Tom de Vries  <tom@codesourcery.com>
1780         * common.opt (fuse-caller-save): New option.
1782 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
1784         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of elements
1785         for big-endian.
1787 2014-04-24  Richard Biener  <rguenther@suse.de>
1789         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
1790         during TER and instead use the sepops interface for expanding
1791         non-GIMPLE_SINGLE_RHS.
1793 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1795         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
1796         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
1798 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1800         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
1801         assembler 64-bit option.
1802         * configure: Regenerate.
1804 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1806         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
1807         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
1808         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
1809         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
1810         (TARGET_CRYPTO): Take TARGET_SIMD into account.
1812 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1814         * config/aarch64/aarch64-builtins.c
1815         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
1816         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
1817         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
1818         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
1819         builtins.
1820         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
1821         (Vrevsuff): New mode attribute.
1823 2014-04-24  Terry Guo  <terry.guo@arm.com>
1825         * config/arm/arm.h (machine_function): Define variable
1826         after_arm_reorg here.
1827         * config/arm/arm.c (after_arm_reorg): Remove the definition.
1828         (arm_split_constant): Update the way to access variable
1829         after_arm_reorg.
1830         (arm_reorg): Ditto.
1831         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
1833 2014-04-23  Tom de Vries  <tom@codesourcery.com>
1835         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
1837 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
1839         * is-a.h: Update comments to reflect the following changes to the
1840         "pointerness" of the API, making the template parameter match the
1841         return type, allowing use of is-a.h with typedefs of pointers.
1842         (is_a_helper::cast): Return a T rather then a pointer to a T, so
1843         that the return type matches the parameter to the is_a_helper.
1844         (as_a): Likewise.
1845         (dyn_cast): Likewise.
1847         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
1848         pointer from the is-a.h API.
1850         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
1851         (is_a_helper <cgraph_node *>::test): ...this, matching change to
1852         is-a.h API.
1853         (is_a_helper <varpool_node>::test): Likewise, convert to...
1854         (is_a_helper <varpool_node *>::test): ...this.
1856         (varpool_first_variable): Update for removal of implicit pointer
1857         from the is-a.h API.
1858         (varpool_next_variable): Likewise.
1859         (varpool_first_static_initializer): Likewise.
1860         (varpool_next_static_initializer): Likewise.
1861         (varpool_first_defined_variable): Likewise.
1862         (varpool_next_defined_variable): Likewise.
1863         (cgraph_first_defined_function): Likewise.
1864         (cgraph_next_defined_function): Likewise.
1865         (cgraph_first_function): Likewise.
1866         (cgraph_next_function): Likewise.
1867         (cgraph_first_function_with_gimple_body): Likewise.
1868         (cgraph_next_function_with_gimple_body): Likewise.
1869         (cgraph_alias_target): Likewise.
1870         (varpool_alias_target): Likewise.
1871         (cgraph_function_or_thunk_node): Likewise.
1872         (varpool_variable_node): Likewise.
1873         (symtab_real_symbol_p): Likewise.
1874         * cgraphunit.c (referred_to_p): Likewise.
1875         (analyze_functions): Likewise.
1876         (handle_alias_pairs): Likewise.
1877         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
1878         * gimple-ssa.h (gimple_vuse_op): Likewise.
1879         (gimple_vdef_op): Likewise.
1880         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
1881         * gimple.c (gimple_build_asm_1): Likewise.
1882         (gimple_build_try): Likewise.
1883         (gimple_build_resx): Likewise.
1884         (gimple_build_eh_dispatch): Likewise.
1885         (gimple_build_omp_for): Likewise.
1886         (gimple_omp_for_set_clauses): Likewise.
1888         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
1889         (is_a_helper <gimple_statement_asm *>::test): ...this.
1890         (is_a_helper <gimple_statement_bind>::test): Convert to...
1891         (is_a_helper <gimple_statement_bind *>::test): ...this.
1892         (is_a_helper <gimple_statement_call>::test): Convert to...
1893         (is_a_helper <gimple_statement_call *>::test): ...this.
1894         (is_a_helper <gimple_statement_catch>::test): Convert to...
1895         (is_a_helper <gimple_statement_catch *>::test): ...this.
1896         (is_a_helper <gimple_statement_resx>::test): Convert to...
1897         (is_a_helper <gimple_statement_resx *>::test): ...this.
1898         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
1899         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
1900         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
1901         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
1902         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
1903         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
1904         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
1905         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
1906         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
1907         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
1908         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
1909         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
1910         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
1911         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
1912         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
1913         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
1914         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
1915         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
1916         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
1917         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
1918         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
1919         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
1920         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
1921         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
1922         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
1923         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
1924         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
1925         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
1926         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
1927         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
1928         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
1929         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
1930         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
1931         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
1932         (is_a_helper <gimple_statement_phi>::test): Convert to...
1933         (is_a_helper <gimple_statement_phi *>::test): ...this.
1934         (is_a_helper <gimple_statement_transaction>::test): Convert to...
1935         (is_a_helper <gimple_statement_transaction *>::test): ...this.
1936         (is_a_helper <gimple_statement_try>::test): Convert to...
1937         (is_a_helper <gimple_statement_try *>::test): ...this.
1938         (is_a_helper <gimple_statement_wce>::test): Convert to...
1939         (is_a_helper <gimple_statement_wce *>::test): ...this.
1940         (is_a_helper <const gimple_statement_asm>::test): Convert to...
1941         (is_a_helper <const gimple_statement_asm *>::test): ...this.
1942         (is_a_helper <const gimple_statement_bind>::test): Convert to...
1943         (is_a_helper <const gimple_statement_bind *>::test): ...this.
1944         (is_a_helper <const gimple_statement_call>::test): Convert to...
1945         (is_a_helper <const gimple_statement_call *>::test): ...this.
1946         (is_a_helper <const gimple_statement_catch>::test): Convert to...
1947         (is_a_helper <const gimple_statement_catch *>::test): ...this.
1948         (is_a_helper <const gimple_statement_resx>::test): Convert to...
1949         (is_a_helper <const gimple_statement_resx *>::test): ...this.
1950         (is_a_helper <const gimple_statement_eh_dispatch>::test):
1951         Convert to...
1952         (is_a_helper <const gimple_statement_eh_dispatch *>::test):
1953         ...this.
1954         (is_a_helper <const gimple_statement_eh_filter>::test): Convert
1955         to...
1956         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
1957         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
1958         Convert to...
1959         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
1960         ...this.
1961         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
1962         Convert to...
1963         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
1964         ...this.
1965         (is_a_helper <const gimple_statement_omp_return>::test): Convert
1966         to...
1967         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
1968         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
1969         to...
1970         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
1971         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
1972         to...
1973         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
1974         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
1975         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
1976         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert
1977         to...
1978         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
1979         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
1980         to...
1981         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
1982         (is_a_helper <const gimple_statement_omp_target>::test): Convert
1983         to...
1984         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
1985         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
1986         to...
1987         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
1988         (is_a_helper <const gimple_statement_omp_single>::test): Convert
1989         to...
1990         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
1991         (is_a_helper <const gimple_statement_omp_teams>::test): Convert
1992         to...
1993         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
1994         (is_a_helper <const gimple_statement_omp_task>::test): Convert
1995         to...
1996         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
1997         (is_a_helper <const gimple_statement_phi>::test): Convert to...
1998         (is_a_helper <const gimple_statement_phi *>::test): ...this.
1999         (is_a_helper <const gimple_statement_transaction>::test): Convert
2000         to...
2001         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
2002         (is_a_helper <const gimple_statement_with_ops>::test): Convert
2003         to...
2004         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
2005         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
2006         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
2007         (is_a_helper <const gimple_statement_with_memory_ops>::test):
2008         Convert to...
2009         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
2010         ...this.
2011         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert
2012         to...
2013         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
2015         (gimple_use_ops): Update for removal of implicit pointer from the
2016         is-a.h API.
2017         (gimple_set_use_ops): Likewise.
2018         (gimple_vuse): Likewise.
2019         (gimple_vdef): Likewise.
2020         (gimple_vuse_ptr): Likewise.
2021         (gimple_vdef_ptr): Likewise.
2022         (gimple_set_vuse): Likewise.
2023         (gimple_set_vdef): Likewise.
2024         (gimple_omp_return_set_lhs): Likewise.
2025         (gimple_omp_return_lhs): Likewise.
2026         (gimple_omp_return_lhs_ptr): Likewise.
2027         (gimple_call_fntype): Likewise.
2028         (gimple_call_set_fntype): Likewise.
2029         (gimple_call_set_internal_fn): Likewise.
2030         (gimple_call_use_set): Likewise.
2031         (gimple_call_clobber_set): Likewise.
2032         (gimple_bind_vars): Likewise.
2033         (gimple_bind_set_vars): Likewise.
2034         (gimple_bind_body_ptr): Likewise.
2035         (gimple_bind_set_body): Likewise.
2036         (gimple_bind_add_stmt): Likewise.
2037         (gimple_bind_block): Likewise.
2038         (gimple_bind_set_block): Likewise.
2039         (gimple_asm_ninputs): Likewise.
2040         (gimple_asm_noutputs): Likewise.
2041         (gimple_asm_nclobbers): Likewise.
2042         (gimple_asm_nlabels): Likewise.
2043         (gimple_asm_input_op): Likewise.
2044         (gimple_asm_input_op_ptr): Likewise.
2045         (gimple_asm_output_op): Likewise.
2046         (gimple_asm_output_op_ptr): Likewise.
2047         (gimple_asm_set_output_op): Likewise.
2048         (gimple_asm_clobber_op): Likewise.
2049         (gimple_asm_set_clobber_op): Likewise.
2050         (gimple_asm_label_op): Likewise.
2051         (gimple_asm_set_label_op): Likewise.
2052         (gimple_asm_string): Likewise.
2053         (gimple_catch_types): Likewise.
2054         (gimple_catch_types_ptr): Likewise.
2055         (gimple_catch_handler_ptr): Likewise.
2056         (gimple_catch_set_types): Likewise.
2057         (gimple_catch_set_handler): Likewise.
2058         (gimple_eh_filter_types): Likewise.
2059         (gimple_eh_filter_types_ptr): Likewise.
2060         (gimple_eh_filter_failure_ptr): Likewise.
2061         (gimple_eh_filter_set_types): Likewise.
2062         (gimple_eh_filter_set_failure): Likewise.
2063         (gimple_eh_must_not_throw_fndecl): Likewise.
2064         (gimple_eh_must_not_throw_set_fndecl): Likewise.
2065         (gimple_eh_else_n_body_ptr): Likewise.
2066         (gimple_eh_else_e_body_ptr): Likewise.
2067         (gimple_eh_else_set_n_body): Likewise.
2068         (gimple_eh_else_set_e_body): Likewise.
2069         (gimple_try_eval_ptr): Likewise.
2070         (gimple_try_cleanup_ptr): Likewise.
2071         (gimple_try_set_eval): Likewise.
2072         (gimple_try_set_cleanup): Likewise.
2073         (gimple_wce_cleanup_ptr): Likewise.
2074         (gimple_wce_set_cleanup): Likewise.
2075         (gimple_phi_capacity): Likewise.
2076         (gimple_phi_num_args): Likewise.
2077         (gimple_phi_result): Likewise.
2078         (gimple_phi_result_ptr): Likewise.
2079         (gimple_phi_set_result): Likewise.
2080         (gimple_phi_arg): Likewise.
2081         (gimple_phi_set_arg): Likewise.
2082         (gimple_resx_region): Likewise.
2083         (gimple_resx_set_region): Likewise.
2084         (gimple_eh_dispatch_region): Likewise.
2085         (gimple_eh_dispatch_set_region): Likewise.
2086         (gimple_omp_critical_name): Likewise.
2087         (gimple_omp_critical_name_ptr): Likewise.
2088         (gimple_omp_critical_set_name): Likewise.
2089         (gimple_omp_for_clauses): Likewise.
2090         (gimple_omp_for_clauses_ptr): Likewise.
2091         (gimple_omp_for_set_clauses): Likewise.
2092         (gimple_omp_for_collapse): Likewise.
2093         (gimple_omp_for_index): Likewise.
2094         (gimple_omp_for_index_ptr): Likewise.
2095         (gimple_omp_for_set_index): Likewise.
2096         (gimple_omp_for_initial): Likewise.
2097         (gimple_omp_for_initial_ptr): Likewise.
2098         (gimple_omp_for_set_initial): Likewise.
2099         (gimple_omp_for_final): Likewise.
2100         (gimple_omp_for_final_ptr): Likewise.
2101         (gimple_omp_for_set_final): Likewise.
2102         (gimple_omp_for_incr): Likewise.
2103         (gimple_omp_for_incr_ptr): Likewise.
2104         (gimple_omp_for_set_incr): Likewise.
2105         (gimple_omp_for_pre_body_ptr): Likewise.
2106         (gimple_omp_for_set_pre_body): Likewise.
2107         (gimple_omp_parallel_clauses): Likewise.
2108         (gimple_omp_parallel_clauses_ptr): Likewise.
2109         (gimple_omp_parallel_set_clauses): Likewise.
2110         (gimple_omp_parallel_child_fn): Likewise.
2111         (gimple_omp_parallel_child_fn_ptr): Likewise.
2112         (gimple_omp_parallel_set_child_fn): Likewise.
2113         (gimple_omp_parallel_data_arg): Likewise.
2114         (gimple_omp_parallel_data_arg_ptr): Likewise.
2115         (gimple_omp_parallel_set_data_arg): Likewise.
2116         (gimple_omp_task_clauses): Likewise.
2117         (gimple_omp_task_clauses_ptr): Likewise.
2118         (gimple_omp_task_set_clauses): Likewise.
2119         (gimple_omp_task_child_fn): Likewise.
2120         (gimple_omp_task_child_fn_ptr): Likewise.
2121         (gimple_omp_task_set_child_fn): Likewise.
2122         (gimple_omp_task_data_arg): Likewise.
2123         (gimple_omp_task_data_arg_ptr): Likewise.
2124         (gimple_omp_task_set_data_arg): Likewise.
2125         (gimple_omp_taskreg_clauses): Likewise.
2126         (gimple_omp_taskreg_clauses_ptr): Likewise.
2127         (gimple_omp_taskreg_set_clauses): Likewise.
2128         (gimple_omp_taskreg_child_fn): Likewise.
2129         (gimple_omp_taskreg_child_fn_ptr): Likewise.
2130         (gimple_omp_taskreg_set_child_fn): Likewise.
2131         (gimple_omp_taskreg_data_arg): Likewise.
2132         (gimple_omp_taskreg_data_arg_ptr): Likewise.
2133         (gimple_omp_taskreg_set_data_arg): Likewise.
2134         (gimple_omp_task_copy_fn): Likewise.
2135         (gimple_omp_task_copy_fn_ptr): Likewise.
2136         (gimple_omp_task_set_copy_fn): Likewise.
2137         (gimple_omp_task_arg_size): Likewise.
2138         (gimple_omp_task_arg_size_ptr): Likewise.
2139         (gimple_omp_task_set_arg_size): Likewise.
2140         (gimple_omp_task_arg_align): Likewise.
2141         (gimple_omp_task_arg_align_ptr): Likewise.
2142         (gimple_omp_task_set_arg_align): Likewise.
2143         (gimple_omp_single_clauses): Likewise.
2144         (gimple_omp_single_clauses_ptr): Likewise.
2145         (gimple_omp_single_set_clauses): Likewise.
2146         (gimple_omp_target_clauses): Likewise.
2147         (gimple_omp_target_clauses_ptr): Likewise.
2148         (gimple_omp_target_set_clauses): Likewise.
2149         (gimple_omp_target_child_fn): Likewise.
2150         (gimple_omp_target_child_fn_ptr): Likewise.
2151         (gimple_omp_target_set_child_fn): Likewise.
2152         (gimple_omp_target_data_arg): Likewise.
2153         (gimple_omp_target_data_arg_ptr): Likewise.
2154         (gimple_omp_target_set_data_arg): Likewise.
2155         (gimple_omp_teams_clauses): Likewise.
2156         (gimple_omp_teams_clauses_ptr): Likewise.
2157         (gimple_omp_teams_set_clauses): Likewise.
2158         (gimple_omp_sections_clauses): Likewise.
2159         (gimple_omp_sections_clauses_ptr): Likewise.
2160         (gimple_omp_sections_set_clauses): Likewise.
2161         (gimple_omp_sections_control): Likewise.
2162         (gimple_omp_sections_control_ptr): Likewise.
2163         (gimple_omp_sections_set_control): Likewise.
2164         (gimple_omp_for_set_cond): Likewise.
2165         (gimple_omp_for_cond): Likewise.
2166         (gimple_omp_atomic_store_set_val): Likewise.
2167         (gimple_omp_atomic_store_val): Likewise.
2168         (gimple_omp_atomic_store_val_ptr): Likewise.
2169         (gimple_omp_atomic_load_set_lhs): Likewise.
2170         (gimple_omp_atomic_load_lhs): Likewise.
2171         (gimple_omp_atomic_load_lhs_ptr): Likewise.
2172         (gimple_omp_atomic_load_set_rhs): Likewise.
2173         (gimple_omp_atomic_load_rhs): Likewise.
2174         (gimple_omp_atomic_load_rhs_ptr): Likewise.
2175         (gimple_omp_continue_control_def): Likewise.
2176         (gimple_omp_continue_control_def_ptr): Likewise.
2177         (gimple_omp_continue_set_control_def): Likewise.
2178         (gimple_omp_continue_control_use): Likewise.
2179         (gimple_omp_continue_control_use_ptr): Likewise.
2180         (gimple_omp_continue_set_control_use): Likewise.
2181         (gimple_transaction_body_ptr): Likewise.
2182         (gimple_transaction_label): Likewise.
2183         (gimple_transaction_label_ptr): Likewise.
2184         (gimple_transaction_set_body): Likewise.
2185         (gimple_transaction_set_label): Likewise.
2187         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
2188         * ipa-inline-analysis.c (inline_write_summary): Likewise.
2189         * ipa-ref.c (ipa_record_reference): Likewise.
2190         * ipa-reference.c (analyze_function): Likewise.
2191         (ipa_reference_write_optimization_summary): Likewise.
2192         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
2193         (address_taken_from_non_vtable_p): Likewise.
2194         (comdat_can_be_unshared_p_1): Likewise.
2195         * lto-cgraph.c (lto_output_ref): Likewise.
2196         (add_references): Likewise.
2197         (compute_ltrans_boundary): Likewise.
2198         (output_symtab): Likewise.
2199         (input_ref): Likewise.
2200         (input_cgraph_1): Likewise.
2201         (output_cgraph_opt_summary): Likewise.
2202         * lto-streamer-out.c (lto_output): Likewise.
2203         (output_symbol_p): Likewise.
2204         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
2205         (lsei_start_function_in_partition): Likewise.
2206         (lsei_next_variable_in_partition): Likewise.
2207         (lsei_start_variable_in_partition): Likewise.
2208         * symtab.c (insert_to_assembler_name_hash): Likewise.
2209         (unlink_from_assembler_name_hash): Likewise.
2210         (symtab_unregister_node): Likewise.
2211         (symtab_remove_node): Likewise.
2212         (dump_symtab_node): Likewise.
2213         (verify_symtab_base): Likewise.
2214         (verify_symtab_node): Likewise.
2215         (symtab_make_decl_local): Likewise.
2216         (symtab_alias_ultimate_target): Likewise.
2217         (symtab_resolve_alias): Likewise.
2218         (symtab_get_symbol_partitioning_class): Likewise.
2219         * tree-phinodes.c (allocate_phi_node): Likewise.
2220         (reserve_phi_args_for_new_edge): Likewise.
2221         (remove_phi_args): Likewise.
2222         * varpool.c (varpool_node_for_asm): Likewise.
2223         (varpool_remove_unreferenced_decls): Likewise.
2225 2014-04-23  Jeff Law  <law@redhat.com>
2227         PR tree-optimization/60902
2228         * tree-ssa-threadedge.c
2229         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
2230         invalidate outputs from statements that do not produce useful
2231         outputs for threading.
2233 2014-04-23 Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
2235         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
2236         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
2237         machine descriptions for Stack Smashing Protector.
2239 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
2241         * aarch64.md (<optab>_rol<mode>3): New pattern.
2242         (<optab>_rolsi3_uxtw): Likewise.
2243         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
2245 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
2247         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
2248         (arm_cortex_a12_tune): Likewise.
2250 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2252         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
2254 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2256         * config/arm/arm.md (arm_rev16si2): New pattern.
2257         (arm_rev16si2_alt): Likewise.
2258         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
2260 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2262        * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
2263        (rev16<mode>2_alt): Likewise.
2264        * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
2265        * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
2266        (aarch_rev16_shleft_mask_imm_p): Likewise.
2267        (aarch_rev16_p_1): Likewise.
2268        (aarch_rev16_p): Likewise.
2269        * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
2270        (aarch_rev16_shright_mask_imm_p): Likewise.
2271        (aarch_rev16_shleft_mask_imm_p): Likewise.
2273 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2275        * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
2276        * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
2277        rev cost.
2278        (cortex_a53_extra_costs): Likewise.
2279        (cortex_a57_extra_costs): Likewise.
2280        * config/arm/arm.c (cortexa9_extra_costs): Likewise.
2281        (cortexa7_extra_costs): Likewise.
2282        (cortexa8_extra_costs): Likewise.
2283        (cortexa12_extra_costs): Likewise.
2284        (cortexa15_extra_costs): Likewise.
2285        (v7m_extra_costs): Likewise.
2286        (arm_new_rtx_costs): Handle BSWAP.
2288 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2290        * config/arm/arm.c (cortexa8_extra_costs): New table.
2291        (arm_cortex_a8_tune): New tuning struct.
2292        * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
2294 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2296        * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
2298 2014-04-23  Richard Biener  <rguenther@suse.de>
2300         * Makefile.in (OBJS): Remove loop-unswitch.o.
2301         * tree-pass.h (make_pass_rtl_unswitch): Remove.
2302         * passes.def (pass_rtl_unswitch): Likewise.
2303         * loop-init.c (gate_rtl_unswitch): Likewise.
2304         (rtl_unswitch): Likewise.
2305         (pass_data_rtl_unswitch): Likewise.
2306         (pass_rtl_unswitch): Likewise.
2307         (make_pass_rtl_unswitch): Likewise.
2308         * rtl.h (reversed_condition): Likewise.
2309         (compare_and_jump_seq): Likewise.
2310         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
2311         and make static.
2312         * loop-unroll.c (compare_and_jump_seq): Likewise.
2314 2014-04-23  Richard Biener  <rguenther@suse.de>
2316         PR tree-optimization/60903
2317         * tree-ssa-loop-im.c (analyze_memory_references): Remove
2318         commented code block.
2319         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
2320         loop flags to newly created BBs and edges.
2322 2014-04-23  Nick Clifton  <nickc@redhat.com>
2324         * config/msp430/msp430.c (msp430_handle_option): Move function
2325         to msp430-common.c
2326         (msp430_option_override): Simplify mcu and mcpu option handling.
2327         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
2328         support for -mhwmult command line option.
2329         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
2330         -mhwmult command line option.
2331         (msp430_hwmult_enabled): Delete.
2332         (msp43o_output_labelref): Add support for -mhwmult command line
2333         option.
2334         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
2335         (umulsidi3): Likewise.
2336         * config/msp430/msp430.opt (mmcu): Add Report attribute.
2337         (mcpu, mlarge, msmall): Likewise.
2338         (mhwmult): New option.
2339         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
2340         prototype.
2341         (msp430_is_f5_mcu): Remove prototype.
2342         (msp430_use_f5_series_hwmult): Add prototype.
2343         * config/msp430/msp430-opts.h: New file.
2344         * common/config/msp430: New directory.
2345         * common/config/msp430/msp430-common.c: New file.
2346         * config.gcc (msp430): Remove target_has_targetm_common.
2347         * doc/invoke.texi: Document -mhwmult command line option.
2349 2014-04-23  Nick Clifton  <nickc@redhat.com>
2351         * config/i386/cygwin.h (ENDFILE_SPEC): Include
2352         default-manifest.o if it can be found in the search path.
2353         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
2355 2014-04-23  Terry Guo  <terry.guo@arm.com>
2357         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
2359 2014-04-23  Richard Biener  <rguenther@suse.de>
2361         PR middle-end/60895
2362         * tree-inline.c (declare_return_variable): Use mark_addressable.
2364 2014-04-23  Richard Biener  <rguenther@suse.de>
2366         PR middle-end/60891
2367         * loop-init.c (loop_optimizer_init): Make sure to apply
2368         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
2370 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
2372         PR sanitizer/60275
2373         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
2374         New options.
2375         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
2376         if flag_sanitize_undefined_trap_on_error.
2377         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
2378         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
2379         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
2380         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2381         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
2382         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
2383         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
2384         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
2385         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
2386         * ubsan.c (ubsan_instrument_unreachable): Return
2387         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
2388         (ubsan_expand_null_ifn): Emit __builtin_trap ()
2389         if flag_sanitize_undefined_trap_on_error and
2390         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
2391         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
2392         instrument_bool_enum_load): Emit __builtin_trap () if
2393         flag_sanitize_undefined_trap_on_error and
2394         __builtin_handle_*_abort () if !flag_sanitize_recover.
2395         * doc/invoke.texi (-fsanitize-recover,
2396         -fsanitize-undefined-trap-on-error): Document.
2398 2014-04-22  Christian Bruel  <christian.bruel@st.com>
2400         * config/sh/sh.md (mov<mode>): Replace movQIHI.
2401          Force immediates to SImode.
2403 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
2405         * config/nios2/nios2.md (UNSPEC_ROUND): New.
2406         (lroundsfsi2): New.
2407         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
2408         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
2409         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
2410         (nios2_fpu_insn): Add entry for round.
2411         (N2FPU_NO_ERRNO_P): Define.
2412         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
2413         flag_errno_math.
2414         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
2416 2014-04-22  Richard Henderson  <rth@redhat.com>
2418         * config/aarch64/aarch64 (addti3, subti3): New expanders.
2419         (add<GPI>3_compare0): Remove leading * from name.
2420         (add<GPI>3_carryin): Likewise.
2421         (sub<GPI>3_compare0): Likewise.
2422         (sub<GPI>3_carryin): Likewise.
2423         (<su_optab>mulditi3): New expander.
2424         (multi3): New expander.
2425         (madd<GPI>): Remove leading * from name.
2427 2014-04-22  Martin Jambor  <mjambor@suse.cz>
2429         * cgraphclones.c (cgraph_function_versioning): Copy
2430         ipa_transforms_to_apply instead of asserting it is empty.
2432 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
2434         PR target/60868
2435         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
2436         on count_exp to get mode.
2438 2014-04-22  Andrew Pinski  <apinski@cavium.com>
2440         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2441         Handle TLS for ILP32.
2442         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
2443         (tlsie_small_<mode>): this and handle PTR.
2444         (tlsie_small_sidi): New pattern.
2445         (tlsle_small): Change to an expand to handle ILP32.
2446         (tlsle_small_<mode>): New pattern.
2447         (tlsdesc_small): Rename to ...
2448         (tlsdesc_small_<mode>): this and handle PTR.
2450 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2452         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
2454 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
2456         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
2457         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
2458         (aarch64_types_signed_poly_qualifiers): Likewise.
2459         (aarch64_types_unsigned_signed_qualifiers): Likewise.
2460         (aarch64_types_poly_signed_qualifiers): Likewise.
2461         (TYPES_REINTERP_SS): Type macro added.
2462         (TYPES_REINTERP_SU): Likewise.
2463         (TYPES_REINTERP_SP): Likewise.
2464         (TYPES_REINTERP_US): Likewise.
2465         (TYPES_REINTERP_PS): Likewise.
2466         (aarch64_fold_builtin): New expression folding added.
2467         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
2468         Declarations removed.
2469         (REINTERP_SS): Declarations added.
2470         (REINTERP_US): Likewise.
2471         (REINTERP_PS): Likewise.
2472         (REINTERP_SU): Likewise.
2473         (REINTERP_SP): Likewise.
2474         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
2475         (vreinterpretq_p8_f64): Likewise.
2476         (vreinterpret_p16_f64): Likewise.
2477         (vreinterpretq_p16_f64): Likewise.
2478         (vreinterpret_f32_f64): Likewise.
2479         (vreinterpretq_f32_f64): Likewise.
2480         (vreinterpret_f64_f32): Likewise.
2481         (vreinterpret_f64_p8): Likewise.
2482         (vreinterpret_f64_p16): Likewise.
2483         (vreinterpret_f64_s8): Likewise.
2484         (vreinterpret_f64_s16): Likewise.
2485         (vreinterpret_f64_s32): Likewise.
2486         (vreinterpret_f64_s64): Likewise.
2487         (vreinterpret_f64_u8): Likewise.
2488         (vreinterpret_f64_u16): Likewise.
2489         (vreinterpret_f64_u32): Likewise.
2490         (vreinterpret_f64_u64): Likewise.
2491         (vreinterpretq_f64_f32): Likewise.
2492         (vreinterpretq_f64_p8): Likewise.
2493         (vreinterpretq_f64_p16): Likewise.
2494         (vreinterpretq_f64_s8): Likewise.
2495         (vreinterpretq_f64_s16): Likewise.
2496         (vreinterpretq_f64_s32): Likewise.
2497         (vreinterpretq_f64_s64): Likewise.
2498         (vreinterpretq_f64_u8): Likewise.
2499         (vreinterpretq_f64_u16): Likewise.
2500         (vreinterpretq_f64_u32): Likewise.
2501         (vreinterpretq_f64_u64): Likewise.
2502         (vreinterpret_s64_f64): Likewise.
2503         (vreinterpretq_s64_f64): Likewise.
2504         (vreinterpret_u64_f64): Likewise.
2505         (vreinterpretq_u64_f64): Likewise.
2506         (vreinterpret_s8_f64): Likewise.
2507         (vreinterpretq_s8_f64): Likewise.
2508         (vreinterpret_s16_f64): Likewise.
2509         (vreinterpretq_s16_f64): Likewise.
2510         (vreinterpret_s32_f64): Likewise.
2511         (vreinterpretq_s32_f64): Likewise.
2512         (vreinterpret_u8_f64): Likewise.
2513         (vreinterpretq_u8_f64): Likewise.
2514         (vreinterpret_u16_f64): Likewise.
2515         (vreinterpretq_u16_f64): Likewise.
2516         (vreinterpret_u32_f64): Likewise.
2517         (vreinterpretq_u32_f64): Likewise.
2519 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
2521         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
2522         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
2523         (vreinterpret_p8_s8): Likewise.
2524         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
2525         (vreinterpret_p8_s16): Likewise.
2526         (vreinterpret_p8_s32): Likewise.
2527         (vreinterpret_p8_s64): Likewise.
2528         (vreinterpret_p8_f32): Likewise.
2529         (vreinterpret_p8_u8): Likewise.
2530         (vreinterpret_p8_u16): Likewise.
2531         (vreinterpret_p8_u32): Likewise.
2532         (vreinterpret_p8_u64): Likewise.
2533         (vreinterpret_p8_p16): Likewise.
2534         (vreinterpretq_p8_s8): Likewise.
2535         (vreinterpretq_p8_s16): Likewise.
2536         (vreinterpretq_p8_s32): Likewise.
2537         (vreinterpretq_p8_s64): Likewise.
2538         (vreinterpretq_p8_f32): Likewise.
2539         (vreinterpretq_p8_u8): Likewise.
2540         (vreinterpretq_p8_u16): Likewise.
2541         (vreinterpretq_p8_u32): Likewise.
2542         (vreinterpretq_p8_u64): Likewise.
2543         (vreinterpretq_p8_p16): Likewise.
2544         (vreinterpret_p16_s8): Likewise.
2545         (vreinterpret_p16_s16): Likewise.
2546         (vreinterpret_p16_s32): Likewise.
2547         (vreinterpret_p16_s64): Likewise.
2548         (vreinterpret_p16_f32): Likewise.
2549         (vreinterpret_p16_u8): Likewise.
2550         (vreinterpret_p16_u16): Likewise.
2551         (vreinterpret_p16_u32): Likewise.
2552         (vreinterpret_p16_u64): Likewise.
2553         (vreinterpret_p16_p8): Likewise.
2554         (vreinterpretq_p16_s8): Likewise.
2555         (vreinterpretq_p16_s16): Likewise.
2556         (vreinterpretq_p16_s32): Likewise.
2557         (vreinterpretq_p16_s64): Likewise.
2558         (vreinterpretq_p16_f32): Likewise.
2559         (vreinterpretq_p16_u8): Likewise.
2560         (vreinterpretq_p16_u16): Likewise.
2561         (vreinterpretq_p16_u32): Likewise.
2562         (vreinterpretq_p16_u64): Likewise.
2563         (vreinterpretq_p16_p8): Likewise.
2564         (vreinterpret_f32_s8): Likewise.
2565         (vreinterpret_f32_s16): Likewise.
2566         (vreinterpret_f32_s32): Likewise.
2567         (vreinterpret_f32_s64): Likewise.
2568         (vreinterpret_f32_u8): Likewise.
2569         (vreinterpret_f32_u16): Likewise.
2570         (vreinterpret_f32_u32): Likewise.
2571         (vreinterpret_f32_u64): Likewise.
2572         (vreinterpret_f32_p8): Likewise.
2573         (vreinterpret_f32_p16): Likewise.
2574         (vreinterpretq_f32_s8): Likewise.
2575         (vreinterpretq_f32_s16): Likewise.
2576         (vreinterpretq_f32_s32): Likewise.
2577         (vreinterpretq_f32_s64): Likewise.
2578         (vreinterpretq_f32_u8): Likewise.
2579         (vreinterpretq_f32_u16): Likewise.
2580         (vreinterpretq_f32_u32): Likewise.
2581         (vreinterpretq_f32_u64): Likewise.
2582         (vreinterpretq_f32_p8): Likewise.
2583         (vreinterpretq_f32_p16): Likewise.
2584         (vreinterpret_s64_s8): Likewise.
2585         (vreinterpret_s64_s16): Likewise.
2586         (vreinterpret_s64_s32): Likewise.
2587         (vreinterpret_s64_f32): Likewise.
2588         (vreinterpret_s64_u8): Likewise.
2589         (vreinterpret_s64_u16): Likewise.
2590         (vreinterpret_s64_u32): Likewise.
2591         (vreinterpret_s64_u64): Likewise.
2592         (vreinterpret_s64_p8): Likewise.
2593         (vreinterpret_s64_p16): Likewise.
2594         (vreinterpretq_s64_s8): Likewise.
2595         (vreinterpretq_s64_s16): Likewise.
2596         (vreinterpretq_s64_s32): Likewise.
2597         (vreinterpretq_s64_f32): Likewise.
2598         (vreinterpretq_s64_u8): Likewise.
2599         (vreinterpretq_s64_u16): Likewise.
2600         (vreinterpretq_s64_u32): Likewise.
2601         (vreinterpretq_s64_u64): Likewise.
2602         (vreinterpretq_s64_p8): Likewise.
2603         (vreinterpretq_s64_p16): Likewise.
2604         (vreinterpret_u64_s8): Likewise.
2605         (vreinterpret_u64_s16): Likewise.
2606         (vreinterpret_u64_s32): Likewise.
2607         (vreinterpret_u64_s64): Likewise.
2608         (vreinterpret_u64_f32): Likewise.
2609         (vreinterpret_u64_u8): Likewise.
2610         (vreinterpret_u64_u16): Likewise.
2611         (vreinterpret_u64_u32): Likewise.
2612         (vreinterpret_u64_p8): Likewise.
2613         (vreinterpret_u64_p16): Likewise.
2614         (vreinterpretq_u64_s8): Likewise.
2615         (vreinterpretq_u64_s16): Likewise.
2616         (vreinterpretq_u64_s32): Likewise.
2617         (vreinterpretq_u64_s64): Likewise.
2618         (vreinterpretq_u64_f32): Likewise.
2619         (vreinterpretq_u64_u8): Likewise.
2620         (vreinterpretq_u64_u16): Likewise.
2621         (vreinterpretq_u64_u32): Likewise.
2622         (vreinterpretq_u64_p8): Likewise.
2623         (vreinterpretq_u64_p16): Likewise.
2624         (vreinterpret_s8_s16): Likewise.
2625         (vreinterpret_s8_s32): Likewise.
2626         (vreinterpret_s8_s64): Likewise.
2627         (vreinterpret_s8_f32): Likewise.
2628         (vreinterpret_s8_u8): Likewise.
2629         (vreinterpret_s8_u16): Likewise.
2630         (vreinterpret_s8_u32): Likewise.
2631         (vreinterpret_s8_u64): Likewise.
2632         (vreinterpret_s8_p8): Likewise.
2633         (vreinterpret_s8_p16): Likewise.
2634         (vreinterpretq_s8_s16): Likewise.
2635         (vreinterpretq_s8_s32): Likewise.
2636         (vreinterpretq_s8_s64): Likewise.
2637         (vreinterpretq_s8_f32): Likewise.
2638         (vreinterpretq_s8_u8): Likewise.
2639         (vreinterpretq_s8_u16): Likewise.
2640         (vreinterpretq_s8_u32): Likewise.
2641         (vreinterpretq_s8_u64): Likewise.
2642         (vreinterpretq_s8_p8): Likewise.
2643         (vreinterpretq_s8_p16): Likewise.
2644         (vreinterpret_s16_s8): Likewise.
2645         (vreinterpret_s16_s32): Likewise.
2646         (vreinterpret_s16_s64): Likewise.
2647         (vreinterpret_s16_f32): Likewise.
2648         (vreinterpret_s16_u8): Likewise.
2649         (vreinterpret_s16_u16): Likewise.
2650         (vreinterpret_s16_u32): Likewise.
2651         (vreinterpret_s16_u64): Likewise.
2652         (vreinterpret_s16_p8): Likewise.
2653         (vreinterpret_s16_p16): Likewise.
2654         (vreinterpretq_s16_s8): Likewise.
2655         (vreinterpretq_s16_s32): Likewise.
2656         (vreinterpretq_s16_s64): Likewise.
2657         (vreinterpretq_s16_f32): Likewise.
2658         (vreinterpretq_s16_u8): Likewise.
2659         (vreinterpretq_s16_u16): Likewise.
2660         (vreinterpretq_s16_u32): Likewise.
2661         (vreinterpretq_s16_u64): Likewise.
2662         (vreinterpretq_s16_p8): Likewise.
2663         (vreinterpretq_s16_p16): Likewise.
2664         (vreinterpret_s32_s8): Likewise.
2665         (vreinterpret_s32_s16): Likewise.
2666         (vreinterpret_s32_s64): Likewise.
2667         (vreinterpret_s32_f32): Likewise.
2668         (vreinterpret_s32_u8): Likewise.
2669         (vreinterpret_s32_u16): Likewise.
2670         (vreinterpret_s32_u32): Likewise.
2671         (vreinterpret_s32_u64): Likewise.
2672         (vreinterpret_s32_p8): Likewise.
2673         (vreinterpret_s32_p16): Likewise.
2674         (vreinterpretq_s32_s8): Likewise.
2675         (vreinterpretq_s32_s16): Likewise.
2676         (vreinterpretq_s32_s64): Likewise.
2677         (vreinterpretq_s32_f32): Likewise.
2678         (vreinterpretq_s32_u8): Likewise.
2679         (vreinterpretq_s32_u16): Likewise.
2680         (vreinterpretq_s32_u32): Likewise.
2681         (vreinterpretq_s32_u64): Likewise.
2682         (vreinterpretq_s32_p8): Likewise.
2683         (vreinterpretq_s32_p16): Likewise.
2684         (vreinterpret_u8_s8): Likewise.
2685         (vreinterpret_u8_s16): Likewise.
2686         (vreinterpret_u8_s32): Likewise.
2687         (vreinterpret_u8_s64): Likewise.
2688         (vreinterpret_u8_f32): Likewise.
2689         (vreinterpret_u8_u16): Likewise.
2690         (vreinterpret_u8_u32): Likewise.
2691         (vreinterpret_u8_u64): Likewise.
2692         (vreinterpret_u8_p8): Likewise.
2693         (vreinterpret_u8_p16): Likewise.
2694         (vreinterpretq_u8_s8): Likewise.
2695         (vreinterpretq_u8_s16): Likewise.
2696         (vreinterpretq_u8_s32): Likewise.
2697         (vreinterpretq_u8_s64): Likewise.
2698         (vreinterpretq_u8_f32): Likewise.
2699         (vreinterpretq_u8_u16): Likewise.
2700         (vreinterpretq_u8_u32): Likewise.
2701         (vreinterpretq_u8_u64): Likewise.
2702         (vreinterpretq_u8_p8): Likewise.
2703         (vreinterpretq_u8_p16): Likewise.
2704         (vreinterpret_u16_s8): Likewise.
2705         (vreinterpret_u16_s16): Likewise.
2706         (vreinterpret_u16_s32): Likewise.
2707         (vreinterpret_u16_s64): Likewise.
2708         (vreinterpret_u16_f32): Likewise.
2709         (vreinterpret_u16_u8): Likewise.
2710         (vreinterpret_u16_u32): Likewise.
2711         (vreinterpret_u16_u64): Likewise.
2712         (vreinterpret_u16_p8): Likewise.
2713         (vreinterpret_u16_p16): Likewise.
2714         (vreinterpretq_u16_s8): Likewise.
2715         (vreinterpretq_u16_s16): Likewise.
2716         (vreinterpretq_u16_s32): Likewise.
2717         (vreinterpretq_u16_s64): Likewise.
2718         (vreinterpretq_u16_f32): Likewise.
2719         (vreinterpretq_u16_u8): Likewise.
2720         (vreinterpretq_u16_u32): Likewise.
2721         (vreinterpretq_u16_u64): Likewise.
2722         (vreinterpretq_u16_p8): Likewise.
2723         (vreinterpretq_u16_p16): Likewise.
2724         (vreinterpret_u32_s8): Likewise.
2725         (vreinterpret_u32_s16): Likewise.
2726         (vreinterpret_u32_s32): Likewise.
2727         (vreinterpret_u32_s64): Likewise.
2728         (vreinterpret_u32_f32): Likewise.
2729         (vreinterpret_u32_u8): Likewise.
2730         (vreinterpret_u32_u16): Likewise.
2731         (vreinterpret_u32_u64): Likewise.
2732         (vreinterpret_u32_p8): Likewise.
2733         (vreinterpret_u32_p16): Likewise.
2734         (vreinterpretq_u32_s8): Likewise.
2735         (vreinterpretq_u32_s16): Likewise.
2736         (vreinterpretq_u32_s32): Likewise.
2737         (vreinterpretq_u32_s64): Likewise.
2738         (vreinterpretq_u32_f32): Likewise.
2739         (vreinterpretq_u32_u8): Likewise.
2740         (vreinterpretq_u32_u16): Likewise.
2741         (vreinterpretq_u32_u64): Likewise.
2742         (vreinterpretq_u32_p8): Likewise.
2743         (vreinterpretq_u32_p16): Likewise.
2745 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
2747         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
2748         Pattern extended.
2749         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
2750         extended.
2751         (sqabs): Likewise.
2752         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
2753         (vqnegd_s64): Likewise.
2754         (vqabs_s64): Likewise.
2755         (vqabsd_s64): Likewise.
2757 2014-04-22  Richard Henderson  <rth@redhat.com>
2759         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
2760         computation to the top of the loop.
2762 2014-04-22  Renlin  <renlin.li@arm.com>
2763             Jiong Wang  <jiong.wang@arm.com>
2765         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
2766         * config/aarch64/aarch64.c (aarch64_layout_frame)
2767         (aarch64_initial_elimination_offset): Likewise.
2769 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2771         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
2772         Fix indentation.
2774 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
2776         * machmode.h (bitwise_mode_for_mode): Declare.
2777         * stor-layout.h (bitwise_type_for_mode): Likewise.
2778         * stor-layout.c (bitwise_mode_for_mode): New function.
2779         (bitwise_type_for_mode): Likewise.
2780         * builtins.c (fold_builtin_memory_op): Use it instead of
2781         int_mode_for_mode and build_nonstandard_integer_type.
2783 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2785         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
2786         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
2787         (*-*-solaris2*): Simplify.
2788         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
2789         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
2790         *-*-solaris2.9* handling.
2792         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
2793         as bug.
2794         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
2795         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
2796         handling, simplify.
2797         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
2798         * configure: Regenerate.
2800         * config/i386/sol2-9.h: Remove.
2802         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
2803         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
2804         Remove Solaris 9 references.
2806 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
2808         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
2809         (floatuns<GPI:mode><GPF:mode>2): Remove.
2810         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
2811         and floatuns conversions.
2812         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
2813         and floatuns conversions.
2814         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
2815         (w1,w2): New mode attributes for inequal width conversions.
2817 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
2819         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
2820         the output asm format.
2822 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
2824         * config/aarch64/aarch64-simd.md
2825         (aarch64_cm<optab>di): Always split.
2826         (*aarch64_cm<optab>di): New.
2827         (aarch64_cmtstdi): Always split.
2828         (*aarch64_cmtstdi): New.
2830 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
2832         PR tree-optimization/60823
2833         * omp-low.c (ipa_simd_modify_function_body): Go through
2834         all SSA_NAMEs and for those refering to vector arguments
2835         which are going to be replaced adjust SSA_NAME_VAR and,
2836         if it is a default definition, change it into a non-default
2837         definition assigned at the beginning of function from new_decl.
2838         (ipa_simd_modify_stmt_ops): Rewritten.
2839         * tree-dfa.c (set_ssa_default_def): When removing default def,
2840         check for NULL loc instead of NULL *loc.
2842 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2844         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
2845         restrictions on core registers for DImode values in Thumb2.
2847 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
2849         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
2850         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
2852 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
2854         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
2855         (*iordi_notzesidi_di): Likewise.
2856         (*iordi_notsesidi_di): Likewise.
2858 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
2860         * config/arm/arm-protos.h (tune_params): New struct members.
2861         * config/arm/arm.c: Initialise tune_params per processor.
2862         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
2863         for speed, based on new tune_params.
2865 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
2867         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
2868         added.
2869         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
2870         macro.
2871         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
2872         corrected.
2873         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
2874         * config/aarch64/arm_neon.h (vrnd_f64): Added.
2875         (vrnda_f64): Likewise.
2876         (vrndi_f64): Likewise.
2877         (vrndm_f64): Likewise.
2878         (vrndn_f64): Likewise.
2879         (vrndp_f64): Likewise.
2880         (vrndx_f64): Likewise.
2882 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2884         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
2885         GET_MODE_SIZE argument is enum machine_mode.
2887 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
2889         PR target/60910
2890         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
2891         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
2893 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
2895         PR middle-end/60281
2896         * asan.c (asan_emit_stack_protection): Force the base to align to
2897         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
2898         appropriate bits if STRICT_ALIGNMENT.
2899         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
2900         when asan is on.
2901         (expand_used_vars): Leave a space in the stack frame for alignment
2902         if STRICT_ALIGNMENT.
2904 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
2906         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
2907         than a gimple.
2908         (gimple_store_p): Likewise.
2909         (gimple_assign_load_p): Likewise.
2910         (gimple_assign_cast_p): Likewise.
2911         (gimple_clobber_p): Likewise.
2913         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
2914         rather than a gimple.
2915         (gimple_assign_cast_p): Likewise.
2917 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
2919         PR target/60735
2920         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
2921         If mode is DDmode and TARGET_E500_DOUBLE allow move.
2923         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
2924         more debug information for E500 if -mdebug=reg.
2926 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
2928         PR target/60909
2929         * config/i386/i386.c (ix86_expand_builtin)
2930         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
2931         register for target RTX.
2932         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
2934 2014-04-18  Cong Hou  <congh@google.com>
2936         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
2937         the widen-mult pattern by handling two operands with different sizes,
2938         and operands whose size is smaller than half of the result type.
2940 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
2942         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
2943         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
2944         (do_estimate_edge_time): Compute it.
2945         * ipa-inline.c (want_inline_small_function_p): Bypass
2946         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
2948 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
2950         * ipa-inline.c (spec_rem): New static variable.
2951         (dump_overall_stats): New function.
2952         (dump_inline_stats): New function.
2954 2014-04-18  Richard Henderson  <rth@redhat.com>
2956         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
2957         to GET_MODE_SIZE, not a reg_class_t.
2959 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2961         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
2962         (vsx_xxmrglw_<mode>): Likewise.
2964 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
2966         PR target/60876
2967         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
2968         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
2969         (rs6000_init_hard_regno_mode_ok): Likewise.
2971 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
2973         * ipa-inline.c (inline_small_functions): Account only non-cold
2974         functions.
2975         * doc/invoke.texi (inline-unit-growth): Update documentation.
2977 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
2979         * config/rs6000/rs6000.md (addti3, subti3): New.
2981 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
2983         PR target/60863
2984         * config/i386/i386.c (ix86_expand_clear): Remove outdated
2985         comment.  Check optimize_insn_for_size_p instead of
2986         optimize_insn_for_speed_p.
2988 2014-04-17  Martin Jambor  <mjambor@suse.cz>
2990         * gimple-iterator.c (gsi_start_edge): New function.
2991         * gimple-iterator.h (gsi_start_edge): Declare.
2992         * tree-sra.c (single_non_eh_succ): New function.
2993         (disqualify_ops_if_throwing_stmt): Renamed to
2994         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
2995         having one non-EH successor BB.
2996         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
2997         generate loads into replacements.
2998         (sra_modify_assign): Likewise and and also use the simple path for
2999         such statements.
3000         (sra_modify_function_body): Commit statements on edges.
3002 2014-04-17  Richard Biener  <rguenther@suse.de>
3004         PR middle-end/60849
3005         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
3006         comparison results and add clarifying comment.
3008 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
3010         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
3011         (blank_mode): Initialize it.
3012         (emit_mode_size_inline, emit_mode_nunits_inline,
3013         emit_mode_inner_inline): New functions.
3014         (emit_insn_modes_h): Call them and surround their output with
3015         #if GCC_VERSION >= 4001 ... #endif.
3016         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
3017         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
3018         mode_* arrays if the argument is __builtin_constant_p.
3019         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
3020         is enum machine_mode.
3022 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3024         * passes.c (opt_pass::execute): Adjust.
3025         (pass_manager::execute_pass_mode_switching): Likewise.
3026         (early_local_passes::execute): Likewise.
3027         (execute_one_pass): Pass cfun to the pass's execute method.
3028         * tree-pass.h (opt_pass::execute): Add function * argument.
3029         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
3030         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
3031         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
3032         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3033         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
3034         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
3035         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
3036         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
3037         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
3038         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
3039         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
3040         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
3041         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
3042         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
3043         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
3044         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
3045         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
3046         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
3047         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
3048         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3049         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3050         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3051         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3052         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3053         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3054         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3055         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
3056         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
3057         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
3058         Adjust.
3060 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3062         * passes.c (opt_pass::gate): Take function * argument.
3063         (gate_all_early_local_passes): Merge into
3064         (early_local_passes::gate): this.
3065         (gate_all_early_optimizations): Merge into
3066         (all_early_optimizations::gate): this.
3067         (gate_all_optimizations): Mege into
3068         (all_optimizations::gate): this.
3069         (gate_all_optimizations_g): Merge into
3070         (all_optimizations_g::gate): this.
3071         (gate_rest_of_compilation): Mege into
3072         (rest_of_compilation::gate): this.
3073         (gate_postreload): Merge into
3074         (postreload::gate): this.
3075         (dump_one_pass): Pass cfun to the pass's gate method.
3076         (execute_ipa_summary_passes): Likewise.
3077         (execute_one_pass): Likewise.
3078         (ipa_write_summaries_2): Likewise.
3079         (ipa_write_optimization_summaries_1): Likewise.
3080         (ipa_read_summaries_1): Likewise.
3081         (ipa_read_optimization_summaries_1): Likewise.
3082         (execute_ipa_stmt_fixups): Likewise.
3083         * tree-pass.h (opt_pass::gate): Add function * argument.
3084         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
3085         combine-stack-adj.c, combine.c, compare-elim.c,
3086         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3087         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
3088         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
3089         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
3090         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
3091         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
3092         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
3093         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
3094         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
3095         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
3096         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
3097         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
3098         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
3099         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
3100         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3101         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3102         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3103         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3104         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3105         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3106         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3107         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3108         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3109         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
3110         var-tracking.c, vtable-verify.c, web.c: Adjust.
3112 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3114         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
3115         * configure: Regenerate.
3117 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3119         * passes.c (dump_one_pass): don't check pass->has_gate.
3120         (execute_ipa_summary_passes): Likewise.
3121         (execute_one_pass): Likewise.
3122         (ipa_write_summaries_2): Likewise.
3123         (ipa_write_optimization_summaries_1): Likewise.
3124         (ipa_read_optimization_summaries_1): Likewise.
3125         (execute_ipa_stmt_fixups): Likewise.
3126         * tree-pass.h (pass_data::has_gate): Remove.
3127         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
3128         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
3129         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
3130         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3131         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
3132         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
3133         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
3134         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
3135         gimple-low.c, gimple-ssa-isolate-paths.c,
3136         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
3137         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3138         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
3139         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
3140         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
3141         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
3142         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
3143         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
3144         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
3145         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
3146         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
3147         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3148         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3149         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3150         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3151         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3152         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3153         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3154         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3155         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
3156         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
3157         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
3158         Adjust.
3160 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3162         * pass_manager.h (pass_manager::register_dump_files_1): Remove
3163         declaration.
3164         * passes.c (pass_manager::register_dump_files_1): Merge into
3165         (pass_manager::register_dump_files): this, and remove its handling of
3166         properties since the pass always has the properties anyway.
3167         (pass_manager::pass_manager): Adjust.
3169 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3171         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
3172         * passes.c (pass_manager::register_dump_files_1): Remove dead code
3173         dealing with properties.
3174         (pass_manager::register_dump_files): Adjust.
3176 2014-03-20  Mark Wielaard  <mjw@redhat.com>
3178         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
3179         then represent the bound as normal constant value.
3181 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
3183         PR target/60847
3184         Forward port from 4.8 branch
3185         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
3187         * config/i386/bmiintrin.h (_blsi_u32): New.
3188         (_blsi_u64): Ditto.
3189         (_blsr_u32): Ditto.
3190         (_blsr_u64): Ditto.
3191         (_blsmsk_u32): Ditto.
3192         (_blsmsk_u64): Ditto.
3193         (_tzcnt_u32): Ditto.
3194         (_tzcnt_u64): Ditto.
3196 2014-04-17  Kito Cheng  <kito@0xlab.org>
3198         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
3200 2014-04-17  Richard Biener  <rguenther@suse.de>
3202         PR middle-end/60849
3203         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
3204         boolean results for comparisons.
3206 2014-04-17  Richard Biener  <rguenther@suse.de>
3208         PR tree-optimization/60836
3209         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
3210         initial PHI args to be gimple values.
3212 2014-04-17   Richard Biener  <rguenther@suse.de>
3214         PR tree-optimization/60841
3215         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
3216         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
3217         of stmts to SLP build.
3218         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3219         (vect_analyze_slp): Likewise.
3220         (vect_analyze_slp_instance): Likewise.
3221         (vect_build_slp_tree): Limit overall SLP tree growth.
3222         * tree-vectorizer.h (vect_analyze_data_refs,
3223         vect_analyze_slp): Adjust prototypes.
3225 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3227         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
3228         Silvermont.
3230 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3232         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
3233         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
3234         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
3235         for TARGET_SLOW_PSHUFB
3237 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3239         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
3240         * config/i386/i386.c (intel_cost): Ditto.
3242 2014-04-17  Joey Ye  <joey.ye@arm.com>
3244         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
3246 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3248         * opts.c (common_handle_option): Disable -fipa-reference coorectly
3249         with -fuse-profile.
3251 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3253         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
3254         (type_all_derivations_known_p): New predicate.
3255         (type_all_ctors_visible_p): New predicate.
3256         (type_possibly_instantiated_p): New predicate.
3257         (get_odr_type): Compute all_derivations_known.
3258         (dump_odr_type): Dump the flag.
3259         (maybe_record_type): Cleanup.
3260         (record_target_from_binfo): Add bases_to_consider array;
3261         record bases for types w/o instances and skip CXX destructor.
3262         (possible_polymorphic_call_targets_1): Add bases_to_consider
3263         and consider_construction parameters; check if type may have instance.
3264         (get_polymorphic_call_info): Set maybe_in_construction to true
3265         when we know nothing.
3266         (record_targets_from_bases): Skip CXX destructors; they are
3267         never called for types in construction.
3268         (possible_polymorphic_call_targets): Do not record target when
3269         type may not have instance.
3271 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3273         PR ipa/60854
3274         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
3275         external aliases alive, too.
3277 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
3279         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
3280         definition.
3282 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
3284         * final.c (compute_alignments): Do not apply loop alignment to a block
3285         falling through to the exit.
3287 2014-04-16  Catherine Moore  <clm@codesourcery.com>
3289         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
3290         Adjust constraints for microMIPS store patterns.
3292 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
3294         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
3296 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
3298         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
3299         (append_use): Run at -O0.
3300         (append_vdef): Likewise.
3301         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
3302         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
3304 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
3306         PR tree-optimization/60844
3307         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
3308         (propagate_op_to_single_use, remove_visited_stmt_chain,
3309         linearize_expr, repropagate_negates, reassociate_bb): Use it
3310         instead of gsi_remove.
3312 2014-04-16  Martin Jambor  <mjambor@suse.cz>
3314         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
3315         ipa_transforms_to_apply.
3316         (cgraph_function_versioning): Assert that old_node has empty
3317         ipa_transforms_to_apply.
3318         * trans-mem.c (ipa_tm_create_version): Likewise.
3319         * tree-inline.c (tree_function_versioning): Do not duplicate
3320         ipa_transforms_to_apply.
3322 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3324         PR target/60817
3325         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
3326         x86_64-*-* cases.
3327         Pass necessary as flags on 64-bit Solaris/x86.
3328         Use lowercase relocs for x86_64-*-*.
3329         * configure: Regenerate.
3331 2014-04-15  Jan Hubicka  <jh@suse.cz>
3333         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
3334         (maybe_record_node, likely_target_p): Use it.
3336 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3338         PR target/60839
3339         Revert following patch
3341         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3343         PR target/60735
3344         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
3345         software floating point or no floating point registers, do not
3346         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
3347         in GPRs that occurs after we tested for GPRs that would never be
3348         true.
3350         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
3351         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
3352         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
3353         specifically allow DDmode, since that does not use the SPE SIMD
3354         instructions.
3356 2014-03-21  Mark Wielaard  <mjw@redhat.com>
3358         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
3359         as unsigned or int depending on type and value used.
3361 2014-04-15  Richard Biener  <rguenther@suse.de>
3363         PR rtl-optimization/56965
3364         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
3365         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
3366         ... here.
3367         * alias.c (true_dependence_1): Do not call
3368         nonoverlapping_component_refs_p.
3369         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
3370         nonoverlapping_component_refs_p.
3371         (indirect_refs_may_alias_p): Likewise.
3373 2014-04-15  Teresa Johnson  <tejohnson@google.com>
3375         * cfg.c (dump_bb_info): Fix flags check.
3376         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
3378 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3380         PR rtl-optimization/60663
3381         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
3382         avoid 0 cost.
3384 2014-04-15  Richard Biener  <rguenther@suse.de>
3386         * lto-streamer.h (LTO_major_version): Bump to 4.
3388 2014-04-15  Richard Biener  <rguenther@suse.de>
3390         * common.opt (lto_partition_model): New enum.
3391         (flto-partition=): Merge separate options with a single with argument,
3392         add -flto-partition=one support.
3393         * flag-types.h (enum lto_partition_model): Declare.
3394         * opts.c (finish_options): Remove duplicate -flto-partition=
3395         option check.
3396         * lto-wrapper.c (run_gcc): Adjust.
3398 2014-04-15  Richard Biener  <rguenther@suse.de>
3400         * alias.c (ncr_compar): New function.
3401         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
3403 2014-04-15  Richard Biener  <rguenther@suse.de>
3405         * alias.c (record_component_aliases): Do not walk BINFOs.
3407 2014-04-15  Richard Biener  <rguenther@suse.de>
3409         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3410         Add struct function argument and adjust.
3411         (find_func_aliases_for_call): Likewise.
3412         (find_func_aliases): Likewise.
3413         (find_func_clobbers): Likewise.
3414         (intra_create_variable_infos): Likewise.
3415         (compute_points_to_sets): Likewise.
3416         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
3418 2014-04-15  Richard Biener  <rguenther@suse.de>
3420         * tree.c (iterative_hash_expr): Use enum tree_code_class
3421         to store TREE_CODE_CLASS.
3422         (tree_block): Likewise.
3423         (tree_set_block): Likewise.
3424         * tree.h (fold_build_pointer_plus_loc): Use
3425         convert_to_ptrofftype_loc.
3427 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
3429         PR plugins/59335
3430         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
3431         added in 4.9.
3433 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
3435         * cfgloop.h (struct loop): Move force_vectorize down.
3436         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
3437         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
3438         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
3439         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
3440         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
3441         * tree-core.h (enum annot_expr_kind): Add new kind values.
3442         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
3443         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
3444         kinds.
3445         * tree.def (ANNOTATE_EXPR): Tweak comment.
3447 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
3449         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
3450         cxa_pure_virtual).
3452 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
3454         * tree.h (TYPE_IDENTIFIER): Declare.
3455         * tree.c (subrange_type_for_debug_p): Use it.
3456         * godump.c (go_format_type): Likewise.
3457         * dwarf2out.c (is_cxx_auto, modified_type_die,
3458         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
3459         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
3461 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
3463         PR lto/60820
3464         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
3466 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
3468         * config/i386/i386.c (examine_argument): Return bool.  Return true if
3469         parameter should be passed in memory.
3470         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
3471         (construct_container): Update calls to examine_argument.
3472         (function_arg_advance_64): Ditto.
3473         (return_in_memory_32): Merge with ix86_return_in_memory.
3474         (return_in_memory_64): Ditto.
3475         (return_in_memory_ms_64): Ditto.
3477 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
3479         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
3480         * coverage.c (coverage_compute_profile_id): Handle externally visible
3481         symbols.
3483 2014-04-14  Martin Jambor  <mjambor@suse.cz>
3485         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
3486         DECL_DISREGARD_INLINE_LIMITS functions.
3488 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
3490         PR target/60827
3491         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
3493 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
3495         PR target/60827
3496         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
3497         optimize_insn_for_speed_p instead of
3498         optimize_function_for_speed_p.
3500 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
3502         * doc/invoke.texi (free): Document AArch64.
3504 2014-04-14  Richard Biener  <rguenther@suse.de>
3506         PR tree-optimization/60042
3507         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
3508         (insert_into_preds_of_block): Do not prevent PHI insertion
3509         for REFERENCE exprs here ...
3510         (eliminate_dom_walker::before_dom_children): ... but prevent
3511         their use here under similar conditions when applied to the
3512         IL after PRE optimizations.
3514 2014-04-14  Richard Biener  <rguenther@suse.de>
3516         * passes.def: Move early points-to after early SRA.
3518 2014-04-14  Richard Biener  <rguenther@suse.de>
3520         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
3521         check for which sign-changes we allow when forwarding
3522         a converted value into a switch.
3524 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
3526         * stor-layout.c (place_field): Finalize non-constant offset for the
3527         field, if any.
3529 2014-04-14  Richard Biener  <rguenther@suse.de>
3531         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
3532         as argument.
3533         (expand_switch_using_bit_tests_p): Likewise.
3534         (process_switch): Compute and pass on speed_p based on the
3535         switch stmt.
3536         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
3537         optimize_bb_for_speed_p.
3539 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
3541         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
3542         * function.h (struct function): Rename has_force_vect_loops into
3543         has_force_vectorize_loops.
3544         * lto-streamer-in.c (input_cfg): Adjust for renaming.
3545         (input_struct_function_base): Likewise.
3546         * lto-streamer-out.c (output_cfg): Likewise.
3547         (output_struct_function_base): Likewise.
3548         * omp-low.c (expand_omp_simd): Likewise.
3549         * tree-cfg.c (move_sese_region_to_fn): Likewise.
3550         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3551         (version_loop_for_if_conversion): Likewise.
3552         (tree_if_conversion): Likewise.
3553         (main_tree_if_conversion): Likewise.
3554         (gate_tree_if_conversion): Likewise.
3555         * tree-inline.c (copy_loops): Likewise.
3556         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
3557         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
3558         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
3559         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
3560         * tree-vectorizer.c (vectorize_loops): Likewise.
3561         * tree-vectorizer.h (unlimited_cost_model): Likewise.
3563 2014-04-14  Richard Biener  <rguenther@suse.de>
3565         PR lto/60720
3566         * lto-streamer-out.c (wrap_refs): New function.
3567         (lto_output): Wrap symbol references in global initializes in
3568         type-preserving MEM_REFs.
3570 2014-04-14  Christian Bruel  <christian.bruel@st.com>
3572         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
3574 2014-04-14  Christian Bruel  <christian.bruel@st.com>
3576         * config/sh/sh.md (setmemqi): New expand pattern.
3577         * config/sh/sh.h (CLEAR_RATIO): Define.
3578         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
3579         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
3581 2014-04-14  Richard Biener  <rguenther@suse.de>
3583         PR middle-end/55022
3584         * fold-const.c (negate_expr_p): Don't negate directional rounding
3585         division.
3586         (fold_negate_expr): Likewise.
3588 2014-04-14  Richard Biener  <rguenther@suse.de>
3590         PR tree-optimization/59817
3591         PR tree-optimization/60453
3592         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
3593         recursion to catch all CHRECs in the scalar evolution and restrict
3594         the predicate for the remains appropriately.
3596 2014-04-12  Catherine Moore  <clm@codesourcery.com>
3598         * config/mips/constraints.md: Add new register constraint "kb".
3599         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
3600         (*movhi_internal): Likewise.
3601         (*movqi_internal): Likewise.
3602         * config/mips/mips.h (M16_STORE_REGS): New register class.
3603         (REG_CLASS_NAMES): Add M16_STORE_REGS.
3604         (REG_CLASS_CONTENTS): Likewise.
3605         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
3607 2014-04-11  Tobias Burnus  <burnus@net-b.de>
3609         PR c/60194
3610         * doc/invoke.texi (-Wformat-signedness): Document it.
3611         (Wformat=2): Mention that this enables -Wformat-signedness.
3613 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3615         * common/config/epiphany/epiphany-common.c
3616         (epiphany_option_optimization_table): Enable section anchors by
3617         default at -O1 or higher.
3618         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
3619         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
3620         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
3621         carries no extra cost.
3622         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
3623         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
3624         * config/epiphany/predicates.md (memclob_operand): New predicate.
3625         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
3626         Use memclob_operand predicate and X constraint for operand 3.
3628 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3630         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
3631         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
3632         its operands.
3634 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3636         PR rtl-optimization/60651
3637         * mode-switching.c (optimize_mode_switching): Make sure to emit
3638         sets of a lower numbered entity before sets of a higher numbered
3639         entity to a mode of the same or lower priority.
3640         When creating a seginfo for a basic block that starts with a code
3641         label, move the insertion point past the code label.
3642         (new_seginfo): Document and enforce requirement that
3643         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
3644         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
3645         * doc/tm.texi: Regenerate.
3647 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3649         PR target/60811
3650         * config/arc/arc.c (arc_save_restore): Fix assert typo.
3652 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
3654         * BASE-VER: Set to 4.10.0.
3656 2014-04-11  Tobias Burnus  <burnus@net-b.de>
3658         PR other/59055
3659         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
3660         * doc/gcc.texi (Service): Update description in the @menu
3661         * doc/invoke.texi (Option Summary): Remove misplaced and
3662         duplicated @menu.
3664 2014-04-11  Steve Ellcey  <sellcey@mips.com>
3665             Jakub Jelinek  <jakub@redhat.com>
3667         PR middle-end/60556
3668         * expr.c (convert_move): Use emit_store_flag_force instead of
3669         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
3670         argument to it.
3672 2014-04-11  Richard Biener  <rguenther@suse.de>
3674         PR middle-end/60797
3675         * varasm.c (assemble_alias): Avoid endless error reporting
3676         recursion by setting TREE_ASM_WRITTEN.
3678 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3680         * config/s390/s390.md: Add a splitter for NOT rtx.
3682 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
3684         PR rtl-optimization/60663
3685         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
3687 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
3688             Jakub Jelinek  <jakub@redhat.com>
3690         PR lto/60567
3691         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
3692         flag from decl_node to node.
3694 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3696         PR debug/60655
3697         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
3698         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
3699         ameliorating the cases where it can be.
3701 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
3703         Revert
3704         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
3706         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
3707         (loadsync_<mode>): Change mode.
3708         (load_quadpti, store_quadpti): New.
3709         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
3710         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
3711         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
3713 2014-04-09  Cong Hou  <congh@google.com>
3715         PR testsuite/60773
3716         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
3717         documentation.
3719 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3721         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
3722         instead of vnor to exploit possible fusion opportunity in the
3723         future.
3724         (altivec_expand_vec_perm_const_le): Likewise.
3726 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
3728         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
3729         (loadsync_<mode>): Change mode.
3730         (load_quadpti, store_quadpti): New.
3731         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
3732         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
3734 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
3736         PR target/60763
3737         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
3738         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
3739         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
3741 2014-04-08  Richard Biener  <rguenther@suse.de>
3743         PR middle-end/60706
3744         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
3745         a 64bit widest int print double-int similar to on HWI64 hosts.
3747 2014-04-08  Richard Biener  <rguenther@suse.de>
3749         PR tree-optimization/60785
3750         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
3751         default defs properly.
3753 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
3755         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
3756         (Weffc++): Likewise.
3758 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
3760         * ipa-devirt.c (maybe_record_node): When node is not recorded,
3761         set completep to false rather than true.
3763 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
3765         PR target/60504
3766         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
3767         ARM_TARGET2_DWARF_FORMAT.
3769 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
3771         PR target/60609
3772         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
3773         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
3774         ADDR_DIFF_VEC.
3776 2014-04-07  Richard Biener  <rguenther@suse.de>
3778         PR tree-optimization/60766
3779         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
3780         (may_eliminate_iv): Convert cand_value_at result to desired type.
3782 2014-04-07  Jason Merrill  <jason@redhat.com>
3784         PR c++/60731
3785         * common.opt (-fno-gnu-unique): Add.
3786         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
3788 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3790         * haifa-sched.c: Fix outdated function reference and minor
3791         grammar errors in introductory comment.
3793 2014-04-07  Richard Biener  <rguenther@suse.de>
3795         PR middle-end/60750
3796         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
3797         for noreturn calls.
3798         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
3800 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
3802         PR debug/55794
3803         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
3804         size accounting for thunks.
3805         (pa_asm_output_mi_thunk): Use final_start_function() and
3806         final_end_function() to output function start and end directives.
3808 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
3810         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
3811         device specific ISA/ feature information. Remove short_sp and
3812         errata_skip ds.  Add avr_device_specific_features enum to have device
3813         specific info.
3814         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
3815         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
3816         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
3817         updated device specific info.
3818         * config/avr/avr-mcus.def: Merge device specific details to
3819         dev_attribute field.
3820         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
3821         errata_skip.
3822         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
3823         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
3824         assembler if RMW isa supported by current device.
3825         * config/avr/genmultilib.awk: Update as device info structure changed.
3826         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
3828 2014-04-04  Cong Hou  <congh@google.com>
3830         PR tree-optimization/60656
3831         * tree-vect-stmts.c (supportable_widening_operation):
3832         Fix a bug that elements in a vector with vect_used_by_reduction
3833         property are incorrectly reordered when the operation on it is not
3834         consistant with the one in reduction operation.
3836 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
3838         PR rtl-optimization/60155
3839         * gcse.c (record_set_data): New function.
3840         (single_set_gcse): New function.
3841         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
3842         (hoist_code): Likewise.
3843         (get_pressure_class_and_nregs): Likewise.
3845 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
3847         * explow.c (probe_stack_range): Emit a final optimization blockage.
3849 2014-04-04  Anthony Green  <green@moxielogic.com>
3851         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
3852         typos.
3854 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
3856         PR ipa/59626
3857         * lto-cgraph.c (input_overwrite_node): Check that partitioning
3858         flags are set only during streaming.
3859         * ipa.c (process_references, walk_polymorphic_call_targets,
3860         symtab_remove_unreachable_nodes): Drop bodies of always inline
3861         after early inlining.
3862         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
3864 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
3865         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3867         PR debug/60655
3868         * dwarf2out.c (const_ok_for_output_1): Reject expressions
3869         containing a NOT.
3871 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3873         PR bootstrap/60743
3874         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
3875         duration.
3876         (cortex_a53_fdivd): Likewise.
3878 2014-04-04  Martin Jambor  <mjambor@suse.cz>
3880         PR ipa/60640
3881         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
3882         Adjust all callers.
3883         * cgraph.c (clone_of_p): Also return true if thunks match.
3884         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
3885         cgraph_function_or_thunk_node and an obsolete comment.
3886         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
3887         file.
3888         (build_function_decl_skip_args): Likewise.
3889         (set_new_clone_decl_and_node_flags): New function.
3890         (duplicate_thunk_for_node): Likewise.
3891         (redirect_edge_duplicating_thunks): Likewise.
3892         (cgraph_clone_node): New parameter args_to_skip, pass it to
3893         redirect_edge_duplicating_thunks which is called instead of
3894         cgraph_redirect_edge_callee.
3895         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
3896         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
3898 2014-04-04  Jeff Law  <law@redhat.com>
3900         PR target/60657
3901         * config/arm/predicates.md (const_int_I_operand): New predicate.
3902         (const_int_M_operand): Similarly.
3903         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
3904         const_int_operand.
3905         (insv_t2, extv_reg, extzv_t2): Likewise.
3906         (load_multiple_with_writeback): Similarly for const_int_I_operand.
3907         (pop_multiple_with_writeback_and_return): Likewise.
3908         (vfp_pop_multiple_with_writeback): Likewise
3910 2014-04-04  Richard Biener  <rguenther@suse.de>
3912         PR ipa/60746
3913         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
3914         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
3915         non-GIMPLE_LABELs.
3916         * gimplify.h (gimple_add_tmp_var_fn): Declare.
3917         * gimplify.c (gimple_add_tmp_var_fn): New function.
3918         * gimple-expr.h (create_tmp_reg_fn): Declare.
3919         * gimple-expr.c (create_tmp_reg_fn): New function.
3920         * gimple-low.c (record_vars_into): Don't change cfun.
3921         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
3922         code generation without cfun.
3924 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
3926         PR bootstrap/60719
3927         * Makefile.in (install-driver): Fix shell scripting.
3929 2014-04-03  Cong Hou  <congh@google.com>
3931         PR tree-optimization/60505
3932         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
3933         threshold of number of iterations below which no vectorization
3934         will be done.
3935         * tree-vect-loop.c (new_loop_vec_info):
3936         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
3937         * tree-vect-loop.c (vect_analyze_loop_operations):
3938         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
3939         * tree-vect-loop.c (vect_transform_loop):
3940         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
3941         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
3942         of iterations of the loop and see if we should build the epilogue.
3944 2014-04-03  Richard Biener  <rguenther@suse.de>
3946         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
3947         (streamer_tree_cache_create): Adjust.
3948         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
3949         to allow optional nodes array.
3950         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
3951         (streamer_tree_cache_append): Likewise.
3952         (streamer_tree_cache_create): Create nodes array optionally
3953         as specified by parameter.
3954         * lto-streamer-out.c (create_output_block): Avoid maintaining
3955         the node array in the writer cache.
3956         (DFS_write_tree): Remove assertion.
3957         (produce_asm_for_decls): Free the out decl state hash table early.
3958         * lto-streamer-in.c (lto_data_in_create): Adjust for
3959         streamer_tree_cache_create prototype change.
3961 2014-04-03  Richard Biener  <rguenther@suse.de>
3963         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
3964         set TREE_CHAIN to NULL_TREE.
3966 2014-04-03  Richard Biener  <rguenther@suse.de>
3968         PR tree-optimization/60740
3969         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
3970         over all GIMPLE_COND operands.
3972 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
3974         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
3975         (Weffc++): Remove Scott's numbering, merge lists and reference
3976         Wnon-virtual-dtor.
3978 2014-04-03  Nick Clifton  <nickc@redhat.com>
3980         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
3981         properly.
3983 2014-04-03  Martin Jambor  <mjambor@suse.cz>
3985         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
3986         mention gcc_unreachable before failing.
3987         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
3988         removed symbols.
3990 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
3992         PR ipa/60659
3993         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
3994         inconsistent code and instead mark the context inconsistent.
3995         (possible_polymorphic_call_targets): For inconsistent contexts
3996         return empty complete list.
3998 2014-04-02  Anthony Green  <green@moxielogic.com>
4000         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
4001         (extendqisi2, extendhisi2): Define.
4002         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
4003         (WCHAR_TYPE): Change to unsigned int.
4005 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4007         PR tree-optimization/60733
4008         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
4009         insertion point for PHI candidates to be the end of the feeding
4010         block for the PHI argument.
4012 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
4014         PR rtl-optimization/60650
4015         * lra-constraints.c (process_alt_operands): Decrease reject for
4016         earlyclobber matching.
4018 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4020         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
4022 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4024         * config/spu/spu.c (pad_bb): Do not crash when the last
4025         insn is CODE_FOR_blockage.
4027 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4029         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
4030         lies outside the target mode.
4032 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4034         PR target/60735
4035         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
4036         software floating point or no floating point registers, do not
4037         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
4038         in GPRs that occurs after we tested for GPRs that would never be
4039         true.
4041         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
4042         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
4043         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
4044         specifically allow DDmode, since that does not use the SPE SIMD
4045         instructions.
4047 2014-04-02  Richard Biener  <rguenther@suse.de>
4049         PR middle-end/60729
4050         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
4051         MODE_INTs.  Properly use negv_optab.
4052         (expand_abs): Likewise.
4054 2014-04-02  Richard Biener  <rguenther@suse.de>
4056         PR bootstrap/60719
4057         * Makefile.in (install-driver): Guard extra installs with special
4058         names properly.
4060 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
4062         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4063         Document vec_vgbbd.
4065 2014-04-01  Richard Henderson  <rth@redhat.com>
4067         PR target/60704
4068         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
4069         alternative enabled before register allocation.
4071 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
4073         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
4074         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
4075         typo.
4076         (nios2_large_got_address): Remove unneeded 'sym' parameter.
4077         (nios2_got_address): Update nios2_large_got_address call site.
4078         (nios2_delegitimize_address): New function.
4079         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
4080         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
4081         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
4083 2014-04-01  Martin Husemann  <martin@duskware.de>
4085         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
4086         for -mabi=32.
4088 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
4090         PR rtl-optimization/60604
4091         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
4092         check from register_operand.
4093         (register_operand): Redefine in terms of general_operand.
4094         (nonmemory_operand): Use register_operand for the non-constant cases.
4096 2014-04-01  Richard Biener  <rguenther@suse.de>
4098         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
4100 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4102         * doc/invoke.texi (mapp-regs): Clarify.
4104 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
4106         * config/i386/avx512fintrin.h (__v32hi): Define type.
4107         (__v64qi): Likewise.
4108         (_mm512_set1_epi8): Define.
4109         (_mm512_set1_epi16): Define.
4110         (_mm512_set4_epi32): Define.
4111         (_mm512_set4_epi64): Define.
4112         (_mm512_set4_pd): Define.
4113         (_mm512_set4_ps): Define.
4114         (_mm512_setr4_epi64): Define.
4115         (_mm512_setr4_epi32): Define.
4116         (_mm512_setr4_pd): Define.
4117         (_mm512_setr4_ps): Define.
4118         (_mm512_setzero_epi32): Define.
4120 2014-03-31  Martin Jambor  <mjambor@suse.cz>
4122         PR middle-end/60647
4123         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
4124         callsite_arguments_match_p.  Updated all callers.  Also check types of
4125         corresponding formal parameters and actual arguments.
4126         (not_all_callers_have_enough_arguments_p) Renamed to
4127         some_callers_have_mismatched_arguments_p.
4129 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
4131         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
4133 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
4135         PR target/60034
4136         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
4137         section anchor.
4139 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
4141         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
4142         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
4143         Split out
4144         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
4145         Use FMAMODE_NOVF512 mode iterator.
4146         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
4147         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
4148         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
4149         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
4150         Split out
4151         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
4152         Use VF_128_256 mode iterator.
4153         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
4154         Ditto.
4156 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4158         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
4159         static chain if needed.
4161 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
4163         PR target/60697
4164         * lra-constraints.c (index_part_to_reg): New.
4165         (process_address): Use it.
4167 2014-03-27  Jeff Law  <law@redhat.com>
4168             Jakub Jelinek  <jakub@redhat.com>
4170         PR target/60648
4171         * expr.c (do_tablejump): Use simplify_gen_binary rather than
4172         gen_rtx_{PLUS,MULT} to build up the address expression.
4174         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
4175         creating non-canonical RTL.
4177 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4179         PR ipa/60243
4180         * ipa-inline.c (want_inline_small_function_p): Short circuit large
4181         functions; reorganize to make cheap checks first.
4182         (inline_small_functions): Do not estimate growth when dumping;
4183         it is expensive.
4184         * ipa-inline.h (inline_summary): Add min_size.
4185         (growth_likely_positive): New function.
4186         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
4187         (set_cond_stmt_execution_predicate): Cleanup.
4188         (estimate_edge_size_and_time): Compute min_size.
4189         (estimate_calls_size_and_time): Likewise.
4190         (estimate_node_size_and_time): Likewise.
4191         (inline_update_overall_summary): Update min_size.
4192         (do_estimate_edge_time): Likewise.
4193         (do_estimate_edge_size): Update.
4194         (do_estimate_edge_hints): Update.
4195         (growth_likely_positive): New function.
4197 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
4199         PR target/60693
4200         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
4201         also if addr has VOIDmode.
4203 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4205         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
4206         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
4207         Declare extern.
4208         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
4209         instructions as well as AdvancedSIMD loads.
4211 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4213         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
4214         Use crypto_aese type.
4215         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
4216         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
4217         crypto_aese, crypto_aesmc.  Move to types.md.
4218         * config/arm/types.md (crypto_aes): Split into crypto_aese,
4219         crypto_aesmc.
4220         * config/arm/iterators.md (crypto_type): Likewise.
4222 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4224         * cgraph.c: Include expr.h and tree-dfa.h.
4225         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
4226         remove LHS.
4228 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
4230         PR target/60675
4231         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
4232         regs from checking multi-reg pseudos.
4234 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4236         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
4238 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4240         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
4241         if it would clobber the stack pointer, even temporarily.
4243 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
4245         * mode-switching.c: Make small adjustments to the top comment.
4247 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
4249         * config/rs6000/constraints.md (wD constraint): New constraint to
4250         match the constant integer to get the top DImode/DFmode out of a
4251         vector in a VSX register.
4253         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
4254         match the constant integer to get the top DImode/DFmode out of a
4255         vector in a VSX register.
4257         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
4258         for ISA 2.07.
4260         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4261         vbpermq builtins.
4263         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
4264         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
4266         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
4267         Optimize vec_extract of 64-bit values, where the value being
4268         extracted is in the top word, where we can use scalar
4269         instructions.  Add direct move and store support.  Combine the big
4270         endian/little endian vector select load support into a single insn.
4271         (vsx_extract_<mode>_internal1): Likewise.
4272         (vsx_extract_<mode>_internal2): Likewise.
4273         (vsx_extract_<mode>_load): Likewise.
4274         (vsx_extract_<mode>_store): Likewise.
4275         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
4276         combined into vsx_extract_<mode>_load.
4277         (vsx_extract_<mode>_one_le): Likewise.
4279         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
4280         define the top 64-bit vector element.
4282         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
4283         constraint.
4285         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4286         Document vec_vbpermq builtin.
4288         PR target/60672
4289         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
4290         enable use of xxsldwi and xxpermdi builtin functions.
4291         (vec_xxpermdi): Likewise.
4293         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4294         Document use of vec_xxsldwi and vec_xxpermdi builtins.
4296 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
4298         PR rtl-optimization/60650
4299         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
4300         first_p.  Use it.
4301         (find_spills_for): New.
4302         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
4303         Spill all pseudos on the second iteration.
4305 2014-03-27  Marek Polacek  <polacek@redhat.com>
4307         PR c/50347
4308         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
4309         types.
4311 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4313         * config/s390/s390.c (s390_can_use_return_insn): Check for
4314         call-saved FPRs on 31 bit.
4316 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
4318         PR middle-end/60682
4319         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
4320         if they need regimplification, just drop them instead of
4321         calling gimple_regimplify_operands on them.
4323 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4325         PR target/60580
4326         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
4327         (aarch64_frame_pointer_required): Adjust logic.
4328         (aarch64_can_eliminate): Adjust logic.
4329         (aarch64_override_options_after_change): Adjust logic.
4331 2014-03-27  Dehao Chen  <dehao@google.com>
4333         * ipa-inline.c (early_inliner): Update node's inline info.
4335 2014-03-26  Dehao Chen  <dehao@google.com>
4337         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
4338         compiler inserted conditional jumps for NAN float check.
4340 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
4342         * ubsan.h (ubsan_create_data): Change second argument's type
4343         to const location_t *.
4344         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
4345         _("<unknown>").
4346         (ubsan_create_data): Change second argument to const location_t *PLOC.
4347         Create Loc field whenever PLOC is non-NULL.
4348         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
4349         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
4350         callers.
4352         PR other/59545
4353         * real.c (real_to_integer2): Change type of low to UHWI.
4355 2014-03-26  Tobias Burnus  <burnus@net-b.de>
4357         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
4358         (CILK_SELF_SPECS): New define.
4359         (driver_self_specs): Use it.
4361 2014-03-26  Richard Biener  <rguenther@suse.de>
4363         * tree-pretty-print.c (percent_K_format): Implement special
4364         case for LTO and its stripped down BLOCK tree.
4366 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
4368         PR sanitizer/60636
4369         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
4371         * tree-vrp.c (simplify_internal_call_using_ranges): If only
4372         one range is range_int_cst_p, but not both, at least optimize
4373         addition/subtraction of 0 and multiplication by 0 or 1.
4374         * gimple-fold.c (gimple_fold_call): Fold
4375         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
4376         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
4377         INTEGER_CSTs, try to fold at least x * 0 and y - y.
4379 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
4381         PR rtl-optimization/60452
4382         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
4383         <case REG>: Return 1 for invalid offsets from the frame pointer.
4385 2014-03-26  Marek Polacek  <polacek@redhat.com>
4387         PR c/37428
4388         * doc/extend.texi (C Extensions): Mention variable-length arrays in
4389         a structure/union.
4391 2014-03-26  Marek Polacek  <polacek@redhat.com>
4393         PR c/39525
4394         * doc/extend.texi (Designated Inits): Describe what happens to omitted
4395         field members.
4397 2014-03-26  Marek Polacek  <polacek@redhat.com>
4399         PR other/59545
4400         * ira-color.c (update_conflict_hard_regno_costs): Perform the
4401         multiplication in unsigned type.
4403 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
4405         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
4407 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
4409         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
4411 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
4413         PR ipa/60315
4414         * cif-code.def (UNREACHABLE) New code.
4415         * ipa-inline.c (inline_small_functions): Skip edges to
4416         __builtlin_unreachable.
4417         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
4418         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
4419         predicate to __bulitin_unreachable.
4420         (set_cond_stmt_execution_predicate): Fix issue when
4421         invert_tree_comparison returns ERROR_MARK.
4422         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
4423         propagate to inline clones.
4424         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
4425         to unreachable.
4426         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
4427         * cgraphclones.c (cgraph_clone_node): If call destination is already
4428         ureachable, do not redirect it back.
4429         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
4430         unreachable.
4432 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
4434         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
4435         Do not modify inline clones.
4437 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
4439         * config/i386/i386.md (general_sext_operand): New mode attr.
4440         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
4441         don't generate (sign_extend (const_int)).
4442         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
4443         operands[2].  Use We constraint instead of <i> and
4444         <general_sext_operand> predicate instead of <general_operand>.
4445         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
4446         * config/i386/constraints.md (We): New constraint.
4447         * config/i386/predicates.md (x86_64_sext_operand,
4448         sext_operand): New predicates.
4450 2014-03-25  Martin Jambor  <mjambor@suse.cz>
4452         PR ipa/60600
4453         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
4454         inconsistent devirtualizations to __builtin_unreachable.
4456 2014-03-25  Marek Polacek  <polacek@redhat.com>
4458         PR c/35449
4459         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
4461 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
4463         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
4464         order of elements for big-endian.
4466 2014-03-25  Richard Biener  <rguenther@suse.de>
4468         PR middle-end/60635
4469         * gimplify-me.c (gimple_regimplify_operands): Update the
4470         re-gimplifed stmt.
4472 2014-03-25  Martin Jambor  <mjambor@suse.cz>
4474         PR ipa/59176
4475         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
4476         (lto_output_varpool_node): Likewise.
4477         (input_overwrite_node): Likewise.
4478         (input_varpool_node): Likewise.
4480 2014-03-25  Richard Biener  <rguenther@suse.de>
4482         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
4483         (run_gcc): Likewise.
4485 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
4487         * combine.c (simplify_compare_const): Add MODE argument.
4488         Handle mode_width 0 as very large mode_width.
4489         (try_combine, simplify_comparison): Adjust callers.
4491         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
4492         type to avoid signed integer overflow.
4493         * explow.c (plus_constant): Likewise.
4495 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4497         * doc/generic.texi: Correct typos.
4499 2014-03-24  Tobias Burnus  <burnus@net-b.de>
4501         * doc/invoke.texi (-flto): Expand section about
4502         using static libraries with LTO.
4504 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4506         PR rtl-optimization/60501
4507         * optabs.def (addptr3_optab): New optab.
4508         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
4509         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
4510         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
4512         * lra.c (emit_add3_insn): Use the addptr pattern if available.
4514         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
4516 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
4518         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
4519         _mm512_set1_pd.
4521         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
4522         (_mm256_undefined_ps): Define.
4523         (_mm256_undefined_pd): Define.
4524         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
4525         (_mm_undefined_pd): Define.
4526         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
4527         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
4528         (_mm512_undefined_ps): Define.
4529         (_mm512_undefined_pd): Define.
4530         Use _mm*_undefined_*.
4531         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
4533 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
4535         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
4536         (lshr_simd): DI mode added.
4537         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
4538         (aarch64_ushr_simddi): Likewise.
4539         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
4540         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
4541         (vshrd_n_u64): Likewise.
4543 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4545         * Makefile.in (s-macro_list): Depend on cc1.
4547 2014-03-23  Teresa Johnson  <tejohnson@google.com>
4549         * ipa-utils.c (ipa_print_order): Use specified dump file.
4551 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
4553         PR rtl-optimization/60601
4554         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
4556         * gcc.c (eval_spec_function): Initialize save_growing_value.
4558 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
4560         PR sanitizer/60613
4561         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
4562         code == MINUS_EXPR, never swap op0 with op1.
4564         * toplev.c (init_local_tick): Avoid signed integer multiplication
4565         overflow.
4566         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
4567         shift by first operand's bitsize.
4569 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
4571         PR target/60610
4572         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
4573         redefine to 1 or 0.
4574         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
4575         TARGET_ISA_64BIT_P(x).
4577 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4579         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
4580         pattern for vector nor instead of subtract from splat(-1).
4581         (altivec_expand_vec_perm_const_le): Likewise.
4583 2014-03-21  Richard Henderson  <rth@twiddle.net>
4585         PR target/60598
4586         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
4587         related insns after epilogue_completed.
4589 2014-03-21  Martin Jambor  <mjambor@suse.cz>
4591         PR ipa/59176
4592         * cgraph.h (symtab_node): New flag body_removed.
4593         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
4594         when removing bodies.
4595         * symtab.c (dump_symtab_base): Dump body_removed flag.
4596         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
4597         had their bodies removed.
4599 2014-03-21  Martin Jambor  <mjambor@suse.cz>
4601         PR ipa/60419
4602         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
4603         in the border.
4605 2014-03-21  Richard Biener  <rguenther@suse.de>
4607         PR tree-optimization/60577
4608         * tree-core.h (struct tree_base): Document nothrow_flag use
4609         in VAR_DECL_NONALIASED.
4610         * tree.h (VAR_DECL_NONALIASED): New.
4611         (may_be_aliased): Adjust.
4612         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
4614 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
4616         * expr.c (expand_expr_real_1): Remove outdated comment.
4618 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
4620         PR middle-end/60597
4621         * ira.c (adjust_cleared_regs): Call copy_rtx on
4622         *reg_equiv[REGNO (loc)].src_p before passing it to
4623         simplify_replace_fn_rtx.
4625         PR target/60568
4626         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
4627         into CONST, put pic register as first operand of PLUS.  Use
4628         gen_const_mem for both 32-bit and 64-bit PIC got loads.
4630 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4632         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
4634 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
4636         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
4637         around for store forwarding issue in the FPU on the UT699.
4638         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
4639         loads and operations if -mfix-ut699 is specified.
4640         (divtf3_hq): Tweak attribute.
4641         (sqrttf2_hq): Likewise.
4643 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
4645         * calls.c (store_one_arg): Remove incorrect const qualification on the
4646         type of the temporary.
4647         * cfgexpand.c (expand_return): Likewise.
4648         * expr.c (expand_constructor): Likewise.
4649         (expand_expr_real_1): Likewise.
4651 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4653         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
4654         of parts.
4656 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
4658         PR target/60039
4659         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
4661 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
4663         * config/arm/aarch-common-protos.h
4664         (alu_cost_table): Fix spelling of "extend".
4665         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
4667 2014-03-19  Richard Biener  <rguenther@suse.de>
4669         PR middle-end/60553
4670         * tree-core.h (tree_type_common): Re-order pointer members
4671         to reduce recursion depth during GC walks.
4673 2014-03-19  Marek Polacek  <polacek@redhat.com>
4675         PR sanitizer/60569
4676         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
4677         before accessing it.
4679 2014-03-19  Richard Biener  <rguenther@suse.de>
4681         PR lto/59543
4682         * lto-streamer-in.c (input_function): In WPA stage do not drop
4683         debug stmts.
4685 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
4687         PR tree-optimization/60559
4688         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
4689         with build_zero_cst assignment.
4691 2014-03-18  Kai Tietz  <ktietz@redhat.com>
4693         PR rtl-optimization/56356
4694         * sdbout.c (sdbout_parms): Verify that parms'
4695         incoming argument is valid.
4696         (sdbout_reg_parms): Likewise.
4698 2014-03-18  Richard Henderson  <rth@redhat.com>
4700         PR target/60562
4701         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
4702         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
4703         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
4705 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
4707         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
4708         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
4709         Italicize plugin event names in description.  Explain that
4710         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
4711         Remind that no GCC functions should be called after PLUGIN_FINISH.
4712         Explain what pragmas with expansion are.
4714 2014-03-18  Martin Liska  <mliska@suse.cz>
4716         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
4717         gimple call statement is update.
4718         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
4719         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
4721 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
4723         PR sanitizer/60557
4724         * ubsan.c (ubsan_instrument_unreachable): Call
4725         initialize_sanitizer_builtins.
4726         (ubsan_pass): Likewise.
4728         PR sanitizer/60535
4729         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
4730         varpool_finalize_decl instead of rest_of_decl_compilation.
4732 2014-03-18  Richard Biener  <rguenther@suse.de>
4734         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
4735         by using bitmap_and_compl instead of bitmap_and_compl_into.
4736         (df_rd_transfer_function): Likewise.
4738 2014-03-18  Richard Biener  <rguenther@suse.de>
4740         * doc/lto.texi (fresolution): Fix typo.
4742 2014-03-18  Richard Biener  <rguenther@suse.de>
4744         * doc/invoke.texi (flto): Update for changes in 4.9.
4746 2014-03-18  Richard Biener  <rguenther@suse.de>
4748         * doc/loop.texi: Remove section on the removed lambda framework.
4749         Update loop docs with recent changes in preserving loop structure.
4751 2014-03-18  Richard Biener  <rguenther@suse.de>
4753         * doc/lto.texi (-fresolution): Document.
4755 2014-03-18  Richard Biener  <rguenther@suse.de>
4757         * doc/contrib.texi: Adjust my name.
4759 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
4761         PR ipa/58721
4762         * internal-fn.c: Include diagnostic-core.h.
4763         (expand_BUILTIN_EXPECT): New function.
4764         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
4765         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
4766         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
4767         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
4768         IFN_BUILTIN_EXPECT.
4769         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
4770         Revert 3 argument __builtin_expect code.
4771         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
4772         * gimple-fold.c (gimple_fold_call): Likewise.
4773         * tree.h (fold_builtin_expect): New prototype.
4774         * builtins.c (build_builtin_expect_predicate): Add predictor
4775         argument, if non-NULL, create 3 argument __builtin_expect.
4776         (fold_builtin_expect): No longer static.  Add ARG2 argument,
4777         pass it through to build_builtin_expect_predicate.
4778         (fold_builtin_2): Adjust caller.
4779         (fold_builtin_3): Handle BUILT_IN_EXPECT.
4780         * internal-fn.def (BUILTIN_EXPECT): New.
4782 2014-03-18  Tobias Burnus  <burnus@net-b.de>
4784         PR ipa/58721
4785         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
4786         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
4787         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
4789 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
4791         PR ipa/58721
4792         * predict.c (combine_predictions_for_bb): Fix up formatting.
4793         (expr_expected_value_1, expr_expected_value): Add predictor argument,
4794         fill what it points to if non-NULL.
4795         (tree_predict_by_opcode): Adjust caller, use the predictor.
4796         * predict.def (PRED_COMPARE_AND_SWAP): Add.
4798 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
4800         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
4801         proper constant for the store mode.
4803 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
4805         * symtab.c (change_decl_assembler_name): Fix transparent alias
4806         chain construction.
4808 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
4810         * config/aarch64/aarch64.c: Correct the comments about the
4811         aarch64 stack layout.
4813 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
4815         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
4816         check for GF_OMP_FOR_KIND_FOR.
4818 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
4820         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
4821         ymm and zmm register names.
4823 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
4825         PR target/60516
4826         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
4827         note creation for the 2010-08-31 changes.
4829 2014-03-17  Marek Polacek  <polacek@redhat.com>
4831         PR middle-end/60534
4832         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
4833         as -fno-tree-loop-vectorize.
4834         (expand_omp_simd): Likewise.
4836 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
4838         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
4839         (eligible_for_call_delay): New prototype.
4840         * config/sparc/sparc.c (tls_call_delay): Rename into...
4841         (eligible_for_call_delay): ...this.  Return false if the instruction
4842         cannot be put in the delay slot of a branch.
4843         (eligible_for_restore_insn): Simplify.
4844         (eligible_for_return_delay): Return false if the instruction cannot be
4845         put in the delay slot of a branch and simplify.
4846         (eligible_for_sibcall_delay): Return false if the instruction cannot be
4847         put in the delay slot of a branch.
4848         * config/sparc/sparc.md (fix_ut699): New attribute.
4849         (tls_call_delay): Delete.
4850         (in_call_delay): Reimplement.
4851         (eligible_for_sibcall_delay): Rename into...
4852         (in_sibcall_delay): ...this.
4853         (eligible_for_return_delay): Rename into...
4854         (in_return_delay): ...this.
4855         (in_branch_delay): Reimplement.
4856         (in_uncond_branch_delay): Delete.
4857         (in_annul_branch_delay): Delete.
4859 2014-03-14  Richard Henderson  <rth@redhat.com>
4861         PR target/60525
4862         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
4863         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
4864         (*floathi<X87MODEF>2_i387_with_temp): Remove.
4865         (floathi splitters): Remove.
4866         (float<SWI48x>xf2): New pattern.
4867         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
4868         code that tried to handle DImode for 32-bit, but which was excluded
4869         by the pattern's condition.  Drop allocation of stack temporary.
4870         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
4871         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
4872         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
4873         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
4874         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
4875         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
4876         (*float<SWI48><MODEF>2_sse_interunit): Remove.
4877         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
4878         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
4879         (*float<SWI48x><X87MODEF>2_i387): Remove.
4880         (all float _with_temp splitters): Remove.
4881         (*float<SWI48x><MODEF>2_i387): New pattern.
4882         (*float<SWI48><MODEF>2_sse): New pattern.
4883         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
4884         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
4886 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
4887             Marek Polacek  <polacek@redhat.com>
4889         PR middle-end/60484
4890         * common.opt (dump_base_name_prefixed): New Variable.
4891         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
4892         if x_dump_base_name_prefixed is already set, set it at the end.
4894 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
4896         PR rtl-optimization/60508
4897         * lra-constraints.c (get_reload_reg): Add new parameter
4898         in_subreg_p.
4899         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
4900         Pass the new parameter values.
4902 2014-03-14  Richard Biener  <rguenther@suse.de>
4904         * common.opt: Revert unintented changes from r205065.
4905         * opts.c: Likewise.
4907 2014-03-14  Richard Biener  <rguenther@suse.de>
4909         PR middle-end/60518
4910         * cfghooks.c (split_block): Properly adjust all loops the
4911         block was a latch of.
4913 2014-03-14  Martin Jambor  <mjambor@suse.cz>
4915         PR lto/60461
4916         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
4917         and simplify it.
4919 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
4921         PR target/59396
4922         * config/avr/avr.c (avr_set_current_function): Pass function name
4923         through default_strip_name_encoding before sanity checking instead
4924         of skipping the first char of the assembler name.
4926 2014-03-13  Richard Henderson  <rth@redhat.com>
4928         PR debug/60438
4929         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
4930         (ix86_force_to_memory, ix86_free_from_memory): Remove.
4931         * config/i386/i386-protos.h: Likewise.
4932         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
4933         in the expander instead of a splitter.
4934         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
4935         any possibility of requiring a memory.
4936         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
4937         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
4938         (fp branch splitters): Update for ix86_split_fp_branch.
4939         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
4940         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
4941         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
4942         (*fop_<MODEF>_2_i387): Remove f/r alternative.
4943         (*fop_<MODEF>_3_i387): Likewise.
4944         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
4945         (splitters for the fop_* register patterns): Remove.
4946         (fscalexf4_i387): Rename from *fscalexf4_i387.
4947         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
4949 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
4951         PR tree-optimization/59779
4952         * tree-dfa.c (get_ref_base_and_extent): Use double_int
4953         type for bitsize and maxsize instead of HOST_WIDE_INT.
4955 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
4957         PR rtl-optimization/57320
4958         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
4959         the CFG after thread_prologue_and_epilogue_insns.
4961 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
4963         PR rtl-optimization/57189
4964         * lra-constraints.c (process_alt_operands): Disfavor spilling
4965         vector pseudos.
4967 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
4969         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
4971 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
4973         PR tree-optimization/59025
4974         PR middle-end/60418
4975         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
4976         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
4978 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
4980         PR target/60486
4981         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
4982         calls of avr_out_plus_1.
4984 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
4986         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
4987         BB's single pred and update the father loop's latch info later.
4989 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
4991         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
4992         (VEC_M): Likewise.
4993         (VEC_N): Likewise.
4994         (VEC_R): Likewise.
4995         (VEC_base): Likewise.
4996         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
4997         registers, we need to swap double words in little endian mode.
4999         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
5000         to be a container mode for 128-bit integer operations added in ISA
5001         2.07.  Unlike TImode and PTImode, the preferred register set is
5002         the Altivec/VMX registers for the 128-bit operations.
5004         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
5005         declarations.
5006         (rs6000_split_128bit_ok_p): Likewise.
5008         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
5009         macros for creating ISA 2.07 normal and overloaded builtin
5010         functions with 3 arguments.
5011         (BU_P8V_OVERLOAD_3): Likewise.
5012         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
5013         for use as overloaded functions.
5014         (VPERM_1TI_UNS): Likewise.
5015         (VSEL_1TI): Likewise.
5016         (VSEL_1TI_UNS): Likewise.
5017         (ST_INTERNAL_1ti): Likewise.
5018         (LD_INTERNAL_1ti): Likewise.
5019         (XXSEL_1TI): Likewise.
5020         (XXSEL_1TI_UNS): Likewise.
5021         (VPERM_1TI): Likewise.
5022         (VPERM_1TI_UNS): Likewise.
5023         (XXPERMDI_1TI): Likewise.
5024         (SET_1TI): Likewise.
5025         (LXVD2X_V1TI): Likewise.
5026         (STXVD2X_V1TI): Likewise.
5027         (VEC_INIT_V1TI): Likewise.
5028         (VEC_SET_V1TI): Likewise.
5029         (VEC_EXT_V1TI): Likewise.
5030         (EQV_V1TI): Likewise.
5031         (NAND_V1TI): Likewise.
5032         (ORC_V1TI): Likewise.
5033         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
5034         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
5035         overloaded builtin.
5036         (VADDUQM): Likewise.
5037         (VSUBCUQ): Likewise.
5038         (VADDEUQM): Likewise.
5039         (VADDECUQ): Likewise.
5040         (VSUBEUQM): Likewise.
5041         (VSUBECUQ): Likewise.
5043         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
5044         __int128_t and __uint128_t types.
5045         (__uint128_type): Likewise.
5046         (altivec_categorize_keyword): Add support for vector __int128_t,
5047         vector __uint128_t, vector __int128, and vector unsigned __int128
5048         as a container type for TImode operations that need to be done in
5049         VSX/Altivec registers.
5050         (rs6000_macro_to_expand): Likewise.
5051         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
5052         to support 128-bit integer instructions vaddcuq, vadduqm,
5053         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
5054         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
5056         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
5057         for V1TImode, and set up preferences to use VSX/Altivec registers.
5058         Setup VSX reload handlers.
5059         (rs6000_debug_reg_global): Likewise.
5060         (rs6000_init_hard_regno_mode_ok): Likewise.
5061         (rs6000_preferred_simd_mode): Likewise.
5062         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
5063         (easy_altivec_constant): Likewise.
5064         (output_vec_const_move): Likewise.
5065         (rs6000_expand_vector_set): Convert V1TImode set and extract to
5066         simple move.
5067         (rs6000_expand_vector_extract): Likewise.
5068         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
5069         addressing.
5070         (rs6000_const_vec): Add support for V1TImode.
5071         (rs6000_emit_le_vsx_load): Swap double words when loading or
5072         storing TImode/V1TImode.
5073         (rs6000_emit_le_vsx_store): Likewise.
5074         (rs6000_emit_le_vsx_move): Likewise.
5075         (rs6000_emit_move): Add support for V1TImode.
5076         (altivec_expand_ld_builtin): Likewise.
5077         (altivec_expand_st_builtin): Likewise.
5078         (altivec_expand_vec_init_builtin): Likewise.
5079         (altivec_expand_builtin): Likewise.
5080         (rs6000_init_builtins): Add support for V1TImode type.  Add
5081         support for ISA 2.07 128-bit integer builtins.  Define type names
5082         for the VSX/Altivec vector types.
5083         (altivec_init_builtins): Add support for overloaded vector
5084         functions with V1TImode type.
5085         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
5086         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
5087         external function.
5088         (rs6000_split_128bit_ok_p): Likewise.
5089         (rs6000_handle_altivec_attribute): Create V1TImode from vector
5090         __int128_t and vector __uint128_t.
5092         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
5093         and mode attributes.
5094         (VSX_M): Likewise.
5095         (VSX_M2): Likewise.
5096         (VSm): Likewise.
5097         (VSs): Likewise.
5098         (VSr): Likewise.
5099         (VSv): Likewise.
5100         (VS_scalar): Likewise.
5101         (VS_double): Likewise.
5102         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
5104         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
5105         we support the ISA 2.07 128-bit integer arithmetic instructions.
5106         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
5107         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
5108         and TImode types for use with the builtin functions.
5109         (V1TI_type_node): Likewise.
5110         (unsigned_V1TI_type_node): Likewise.
5111         (intTI_type_internal_node): Likewise.
5112         (uintTI_type_internal_node): Likewise.
5114         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
5115         128-bit builtin functions.
5116         (UNSPEC_VADDEUQM): Likewise.
5117         (UNSPEC_VADDECUQ): Likewise.
5118         (UNSPEC_VSUBCUQ): Likewise.
5119         (UNSPEC_VSUBEUQM): Likewise.
5120         (UNSPEC_VSUBECUQ): Likewise.
5121         (VM): Add V1TImode to vector mode iterators.
5122         (VM2): Likewise.
5123         (VI_unit): Likewise.
5124         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
5125         (altivec_vaddcuq): Likewise.
5126         (altivec_vsubuqm): Likewise.
5127         (altivec_vsubcuq): Likewise.
5128         (altivec_vaddeuqm): Likewise.
5129         (altivec_vaddecuq): Likewise.
5130         (altivec_vsubeuqm): Likewise.
5131         (altivec_vsubecuq): Likewise.
5133         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
5134         mode iterators.
5135         (BOOL_128): Likewise.
5136         (BOOL_REGS_OUTPUT): Likewise.
5137         (BOOL_REGS_OP1): Likewise.
5138         (BOOL_REGS_OP2): Likewise.
5139         (BOOL_REGS_UNARY): Likewise.
5140         (BOOL_REGS_AND_CR0): Likewise.
5142         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
5143         128-bit integer builtin support.
5144         (vec_vadduqm): Likewise.
5145         (vec_vaddecuq): Likewise.
5146         (vec_vaddeuqm): Likewise.
5147         (vec_vsubecuq): Likewise.
5148         (vec_vsubeuqm): Likewise.
5149         (vec_vsubcuq): Likewise.
5150         (vec_vsubuqm): Likewise.
5152         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5153         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
5154         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
5155         128-bit integer add/subtract to ISA 2.07.
5157 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
5159         * config/arc/arc.c (arc_predicate_delay_insns):
5160         Fix third argument passed to conditionalize_nonjump.
5162 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
5164         * config/aarch64/aarch64-builtins.c
5165         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
5166         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
5167         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
5168         instead of __builtin_lfloor.
5169         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
5171 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5173         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
5174         (tree_ssa_ifcombine_bb_1): New function.
5175         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
5176         is an empty forwarder block to then_bb or vice versa and then_bb
5177         and else_bb are effectively swapped.
5179 2014-03-12  Christian Bruel  <christian.bruel@st.com>
5181         PR target/60264
5182         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
5183         REG_CFA_DEF_CFA note.
5184         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
5185         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
5187 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
5189         PR tree-optimization/60454
5190         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
5192 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5194         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
5195         Do not define target_cpu_default2 to generic.
5196         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
5197         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
5198         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
5200 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5201             Marc Glisse  <marc.glisse@inria.fr>
5203         PR tree-optimization/60502
5204         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
5205         instead of build_low_bits_mask.
5207 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5209         PR middle-end/60482
5210         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
5211         if there are multiple uses, but op doesn't live on E edge.
5212         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
5213         clobber stmts before __builtin_unreachable.
5215 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
5217         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
5218         hard_frame_pointer_rtx.
5219         * cse.c (cse_insn): Remove volatile check.
5220         * cselib.c (cselib_process_insn): Likewise.
5221         * dse.c (scan_insn): Likewise.
5223 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
5225         * config/arc/arc.c (conditionalize_nonjump): New function,
5226         broken out of ...
5227         (arc_ifcvt): ... this.
5228         (arc_predicate_delay_insns): Use it.
5230 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
5232         * config/arc/predicates.md (extend_operand): During/after reload,
5233         allow const_int_operand.
5234         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
5235         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
5236         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
5237         to "i".
5238         (umulsi3_highpart_i): Likewise.
5240 2014-03-11  Richard Biener  <rguenther@suse.de>
5242         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
5243         Add asserts to guard possible wrong-code bugs.
5245 2014-03-11  Richard Biener  <rguenther@suse.de>
5247         PR tree-optimization/60429
5248         PR tree-optimization/60485
5249         * tree-ssa-structalias.c (set_union_with_increment): Properly
5250         take into account all fields that overlap the shifted vars.
5251         (do_sd_constraint): Likewise.
5252         (do_ds_constraint): Likewise.
5253         (get_constraint_for_ptr_offset): Likewise.
5255 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
5257         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
5258         (nios2_compute_frame_layout):
5259         Add calculation of cfun->machine->fp_save_offset.
5260         (nios2_expand_prologue): Correct setting of frame pointer register
5261         in prologue.
5262         (nios2_expand_epilogue): Update recovery of stack pointer from
5263         frame pointer accordingly.
5264         (nios2_initial_elimination_offset): Update calculation of offset
5265         for eliminating to HARD_FRAME_POINTER_REGNUM.
5267 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
5269         PR ipa/60457
5270         * ipa.c (symtab_remove_unreachable_nodes): Don't call
5271         cgraph_get_create_node on VAR_DECLs.
5273 2014-03-10  Richard Biener  <rguenther@suse.de>
5275         PR middle-end/60474
5276         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
5278 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
5280         * config/vms/vms.opt (vms_float_format): New variable.
5282 2014-03-08  Tobias Burnus  <burnus@net-b.de>
5284         * doc/invoke.texi (-fcilkplus): Update implementation status.
5286 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
5287             Richard Biener  <rguenther@suse.de>
5289         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
5290         consistently accross all TUs.
5291         (run_gcc): Enable -fshort-double automatically at link at link-time
5292         and disallow override.
5294 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
5296         PR target/58271
5297         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
5298         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
5299         if they can't be used.
5301 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5303         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
5304         for Solaris 11/x86 ld.
5305         * configure: Regenerate.
5307 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5309         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
5310         (LIB_TLS_SPEC): Save as ld_tls_libs.
5311         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
5312         (HAVE_AS_IX86_TLSLDM): New test.
5313         * configure, config.in: Regenerate.
5314         * config/i386/i386.c (legitimize_tls_address): Fall back to
5315         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
5316         cannot support TLS_MODEL_LOCAL_DYNAMIC.
5317         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
5318         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
5320 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
5322         * common.opt (fira-loop-pressure): Mark as optimization.
5324 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
5326         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
5327         an OpenMP mappable type.
5329 2014-03-06  Matthias Klose  <doko@ubuntu.com>
5331         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
5332         MULTILIB_OSDIRNAMES is not defined.
5334 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
5335             Meador Inge  <meadori@codesourcery.com>
5337         PR target/58595
5338         * config/arm/arm.c (arm_tls_symbol_p): Remove.
5339         (arm_legitimize_address): Call legitimize_tls_address for any
5340         arm_tls_referenced_p expression, handle constant addend.  Call it
5341         before testing for !TARGET_ARM.
5342         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
5344 2014-03-06  Richard Biener  <rguenther@suse.de>
5346         PR middle-end/60445
5347         PR lto/60424
5348         PR lto/60427
5349         Revert
5350         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
5352         * tree-streamer.c (record_common_node): Assert we don't record
5353         nodes with type double.
5354         (preload_common_node): Skip type double, complex double and double
5355         pointer since it is now frontend dependent due to fshort-double option.
5357 2014-03-06  Richard Biener  <rguenther@suse.de>
5359         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
5360         or -fno-lto is specified and the linker has full plugin support.
5361         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
5362         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
5363         * lto-wrapper.c (merge_and_complain): Merge compile-time
5364         optimization levels.
5365         (run_gcc): And pass it through to the link options.
5367 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
5369         PR debug/60381
5370         Revert:
5371         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
5372         PR debug/59992
5373         * cselib.c (remove_useless_values): Skip to avoid quadratic
5374         behavior if the condition moved from...
5375         (cselib_process_insn): ... here holds.
5377 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
5379         PR plugins/59335
5380         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
5381         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
5383         PR plugins/59335
5384         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
5385         (TM_H): Add x86-tune.def.
5387 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5389         * config/aarch64/aarch64.c (generic_tunings):
5390         Use cortexa57_extra_costs.
5392 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
5394         PR lto/60404
5395         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
5396         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
5397         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
5398         cost for in_lto_p.
5400 2014-03-04  Heiher  <r@hev.cc>
5402         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
5403         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
5405 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
5407         * config/i386/predicates.md (const2356_operand): Change to ...
5408         (const2367_operand): ... this.
5409         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
5410         const2367_operand.
5411         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5412         (*avx512pf_scatterpf<mode>sf): Ditto.
5413         (avx512pf_scatterpf<mode>df): Ditto.
5414         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5415         (*avx512pf_scatterpf<mode>df): Ditto.
5416         * config/i386/i386.c (ix86_expand_builtin): Update
5417         incorrect hint operand error message.
5419 2014-03-04  Richard Biener  <rguenther@suse.de>
5421         * lto-section-in.c (lto_get_section_data): Fix const cast.
5423 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
5425         * tree-streamer.c (record_common_node): Assert we don't record
5426         nodes with type double.
5427         (preload_common_node): Skip type double, complex double and double
5428         pointer since it is now frontend dependent due to fshort-double option.
5430 2014-03-04  Richard Biener  <rguenther@suse.de>
5432         PR lto/60405
5433         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
5434         (lto_input_toplevel_asms): Likewise.
5435         * lto-section-in.c (lto_get_section_data): Instead do it here
5436         for every section.
5438 2014-03-04  Richard Biener  <rguenther@suse.de>
5440         PR tree-optimization/60382
5441         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
5442         dead PHIs a reduction.
5444 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
5446         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
5447         hint value.
5448         (_mm_prefetch): Move out of GCC target("sse") pragma.
5449         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
5450         GCC target("prfchw") pragma.
5451         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
5452         for locality <= 2.
5453         * config/i386/i386.c (ix86_option_override_internal): Enable
5454         -mprfchw with -mprefetchwt1.
5456 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
5458         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
5459         Mark as varying.
5461 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
5463         * opts.h (CL_PCH_IGNORE): Define.
5464         * targhooks.c (option_affects_pch_p):
5465         Return false for options that have CL_PCH_IGNORE set.
5466         * opt-functions.awk: Process PchIgnore.
5467         * doc/options.texi: Document PchIgnore.
5469         * config/arc/arc.opt (misize): Add PchIgnore property.
5471 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5473         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
5474         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
5475         constraint on constants to permit them being loaded into
5476         GENERAL_REGS or BASE_REGS.
5478 2014-03-03  Nick Clifton  <nickc@redhat.com>
5480         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
5481         anti-cacnonical alternatives.
5482         (negandhi3_real): New pattern.
5483         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
5485 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
5487         * config/avr/avr-mcus.def: Remove atxmega16x1.
5488         * config/avr/avr-tables.opt: Regenerate.
5489         * config/avr/t-multilib: Regenerate.
5490         * doc/avr-mmcu.texi: Regenerate.
5492 2014-03-03  Tobias Grosser  <tobias@grosser.es>
5493             Mircea Namolaru  <mircea.namolaru@inria.fr>
5495         PR tree-optimization/58028
5496         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
5497         scalar dimensions.
5499 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5501         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
5502         not handled by recognizers.
5504 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
5506         PR middle-end/60175
5507         * function.c (expand_function_end): Don't emit
5508         clobber_return_register sequence if clobber_after is a BARRIER.
5509         * cfgexpand.c (construct_exit_block): Append instructions before
5510         return_label to prev_bb.
5512 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5514         * config/rs6000/constraints.md: Document reserved use of "wc".
5516 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
5518         PR ipa/60150
5519         * ipa.c (function_and_variable_visibility): When dissolving comdat
5520         group, also set all symbols to local.
5522 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
5524         PR ipa/60306
5526         Revert:
5527         2013-12-14   Jan Hubicka  <jh@suse.cz>
5528         PR middle-end/58477
5529         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
5531 2014-03-02  Jon Beniston  <jon@beniston.com>
5533         PR bootstrap/48230
5534         PR bootstrap/50927
5535         PR bootstrap/52466
5536         PR target/46898
5537         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
5538           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
5539         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
5540         (simple_return, *simple_return): New patterns
5541         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
5542         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
5544 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
5546         * dwarf2out.c (gen_subprogram_die): Tidy.
5548 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
5550         PR target/60071
5551         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
5552         (*mov_t_msb_neg_negc): ... this new insn.
5554 2014-02-28  Jason Merrill  <jason@redhat.com>
5556         PR c++/58678
5557         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
5558         function.
5560 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
5562         PR c++/60314
5563         * dwarf2out.c (decltype_auto_die): New static.
5564         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
5565         (gen_type_die_with_usage): Handle 'decltype(auto)'.
5566         (is_cxx_auto): Likewise.
5568 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
5570         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
5571         we are not using general regs only.
5573 2014-02-28  Richard Biener  <rguenther@suse.de>
5575         PR target/60280
5576         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
5577         previous fix and only allow to remove trivial pre-headers
5578         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
5579         (remove_forwarder_block): Properly update the latch of a loop.
5581 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
5583         PR debug/59992
5584         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
5585         (cselib_preserved_hash_table): New.
5586         (preserve_constants_and_equivs): Move preserved vals to it.
5587         (cselib_find_slot): Look it up first.
5588         (cselib_init): Initialize it.
5589         (cselib_finish): Release it.
5590         (dump_cselib_table): Dump it.
5592 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
5594         PR debug/59992
5595         * cselib.c (remove_useless_values): Skip to avoid quadratic
5596         behavior if the condition moved from...
5597         (cselib_process_insn): ... here holds.
5599 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
5601         PR debug/57232
5602         * var-tracking.c (vt_initialize): Apply the same condition to
5603         preserve the CFA base value.
5605 2014-02-28  Joey Ye  <joey.ye@arm.com>
5607         PR target/PR60169
5608         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
5609         if reload in progress or completed.
5611 2014-02-28  Tobias Burnus  <burnus@net-b.de>
5613         PR middle-end/60147
5614         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
5615         NAMELIST_DECL.
5617 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
5619         * doc/tm.texi.in (Condition Code Status): Update documention for
5620         relative locations of cc0-setter and cc0-user.
5622 2014-02-27  Jeff Law  <law@redhat.com>
5624         PR rtl-optimization/52714
5625         * combine.c (try_combine): When splitting an unrecognized PARALLEL
5626         into two independent simple sets, if I3 is a jump, ensure the
5627         pattern we place into I3 is a (set (pc) ...).
5629 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
5630             Jeff Law  <law@redhat.com>
5632         PR rtl-optimization/49847
5633         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
5634         are in different blocks.
5635         * doc/tm.texi (Condition Code Status): Update documention for
5636         relative locations of cc0-setter and cc0-user.
5638 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
5640         PR target/59222
5641         * lra.c (lra_emit_add): Check SUBREG too.
5643 2014-02-27  Andreas Schwab  <schwab@suse.de>
5645         * config/m68k/m68k.c (m68k_option_override): Disable
5646         -flive-range-shrinkage for classic m68k.
5647         (m68k_override_options_after_change): Likewise.
5649 2014-02-27  Marek Polacek  <polacek@redhat.com>
5651         PR middle-end/59223
5652         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
5653         -Wmaybe-uninitialized.
5655 2014-02-27  Alan Modra  <amodra@gmail.com>
5657         PR target/57936
5658         * reload1.c (emit_input_reload_insns): When reload_override_in,
5659         set old to rl->in_reg when rl->in_reg is a subreg.
5661 2014-02-26  Richard Biener  <rguenther@suse.de>
5663         PR bootstrap/60343
5664         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
5666 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
5668         * common/config/i386/predicates.md (const1256_operand): Remove.
5669         (const2356_operand): New.
5670         (const_1_to_2_operand): Remove.
5671         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
5672         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5673         (*avx512pf_gatherpf<mode>sf): Ditto.
5674         (avx512pf_gatherpf<mode>df): Ditto.
5675         (*avx512pf_gatherpf<mode>df_mask): Ditto.
5676         (*avx512pf_gatherpf<mode>df): Ditto.
5677         (avx512pf_scatterpf<mode>sf): Ditto.
5678         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5679         (*avx512pf_scatterpf<mode>sf): Ditto.
5680         (avx512pf_scatterpf<mode>df): Ditto.
5681         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5682         (*avx512pf_scatterpf<mode>df): Ditto.
5683         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
5685 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
5687         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
5688         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
5689         (_mm512_mask_testn_epi64_mask): Move to ...
5690         * config/i386/avx512cdintrin.h: Here.
5691         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
5692         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
5693         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
5694         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
5695         TARGET_AVX512F from TARGET_AVX512CD.
5697 2014-02-26  Richard Biener  <rguenther@suse.de>
5699         PR ipa/60327
5700         * ipa.c (walk_polymorphic_call_targets): Properly guard
5701         call to inline_update_overall_summary.
5703 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
5705         PR target/60280
5706         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
5707         and latches only if requested.  Fix latch if it is removed.
5708         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
5709         LOOPS_HAVE_PREHEADERS.
5711 2014-02-25  Andrew Pinski  <apinski@cavium.com>
5713         * builtins.c (expand_builtin_thread_pointer): Create a new target
5714         when the target is NULL.
5716 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
5718         PR rtl-optimization/60317
5719         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
5720         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
5721         * lra-assigns.c: Include params.h.
5722         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
5723         other reload pseudos considerations.
5725 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5727         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
5728         to use canonical form for nor<mode>3.
5730 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5732         PR target/55426
5733         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
5734         conversions.
5736 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
5738         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
5739         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
5740         (ix86_handle_option): Handle OPT_mprefetchwt1.
5741         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
5742         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
5743         PREFETCHWT1 CPUID.
5744         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5745         OPTION_MASK_ISA_PREFETCHWT1.
5746         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
5747         (PTA_PREFETCHWT1): New.
5748         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
5749         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
5750         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
5751         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
5752         (*prefetch_avx512pf_<mode>_: Change into ...
5753         (*prefetch_prefetchwt1_<mode>: This.
5754         * config/i386/i386.opt (mprefetchwt1): New.
5755         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
5756         (_mm_prefetch): Handle intent to write.
5757         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
5759 2014-02-25  Richard Biener  <rguenther@suse.de>
5761         PR middle-end/60291
5762         * emit-rtl.c (mem_attrs_htab): Remove.
5763         (mem_attrs_htab_hash): Likewise.
5764         (mem_attrs_htab_eq): Likewise.
5765         (set_mem_attrs): Always allocate new mem-attrs when something changed.
5766         (init_emit_once): Do not allocate mem_attrs_htab.
5768 2014-02-25  Richard Biener  <rguenther@suse.de>
5770         PR lto/60319
5771         * lto-opts.c (lto_write_options): Output non-explicit conservative
5772         -fwrapv, -fno-trapv and -fno-strict-overflow.
5773         * lto-wrapper.c (merge_and_complain): Handle merging those options.
5774         (run_gcc): And pass them through.
5776 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
5778         * sel-sched.c (calculate_new_fences): New parameter ptime.
5779         Calculate it as a maximum over all fence cycles.
5780         (sel_sched_region_2): Adjust the call to calculate_new_fences.
5781         Print the final schedule timing when sched_verbose.
5783 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
5785         PR rtl-optimization/60292
5786         * sel-sched.c (fill_vec_av_set): Do not reset target availability
5787         bit fot the fence instruction.
5789 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
5791         * calls.h: Fix typo in comment.
5793 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
5795         * config/pa/pa.c (pa_output_move_double): Don't valididate when
5796         adjusting offsetable addresses.
5798 2014-02-24  Guozhi Wei  <carrot@google.com>
5800         * sparseset.h (sparseset_pop): Fix the wrong index.
5802 2014-02-24  Walter Lee  <walt@tilera.com>
5804         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
5805         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
5806         triplet.
5807         * common/config/tilegx/tilegx-common.c
5808         (TARGET_DEFAULT_TARGET_FLAGS): Define.
5809         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
5810         (LINK_SPEC): Ditto.
5811         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
5812         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
5813         (tilegx_gimplify_va_arg_expr): Handle big endian.
5814         (tilegx_expand_unaligned_load): Ditto.
5815         (tilegx_expand_unaligned_store): Ditto.
5816         (TARGET_RETURN_IN_MSB): New.
5817         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
5818         (TARGET_ENDIAN_DEFAULT): New.
5819         (TARGET_BIG_ENDIAN): Handle big endian.
5820         (BYTES_BIG_ENDIAN): Ditto.
5821         (WORDS_BIG_ENDIAN): Ditto.
5822         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
5823         (ENDIAN_SPEC): New.
5824         (EXTRA_SPECS): New.
5825         * config/tilegx/tilegx.md (extv): Handle big endian.
5826         (extzv): Ditto.
5827         (insn_st<n>): Ditto.
5828         (insn_st<n>_add<bitsuffix>): Ditto.
5829         (insn_stnt<n>): Ditto.
5830         (insn_stnt<n>_add<bitsuffix>):Ditto.
5831         (vec_interleave_highv8qi): Handle big endian.
5832         (vec_interleave_highv8qi_be): New.
5833         (vec_interleave_highv8qi_le): New.
5834         (insn_v1int_h): Handle big endian.
5835         (vec_interleave_lowv8qi): Handle big endian.
5836         (vec_interleave_lowv8qi_be): New.
5837         (vec_interleave_lowv8qi_le): New.
5838         (insn_v1int_l): Handle big endian.
5839         (vec_interleave_highv4hi): Handle big endian.
5840         (vec_interleave_highv4hi_be): New.
5841         (vec_interleave_highv4hi_le): New.
5842         (insn_v2int_h): Handle big endian.
5843         (vec_interleave_lowv4hi): Handle big endian.
5844         (vec_interleave_lowv4hi_be): New.
5845         (vec_interleave_lowv4hi_le): New.
5846         (insn_v2int_l): Handle big endian.
5847         (vec_interleave_highv2si): Handle big endian.
5848         (vec_interleave_highv2si_be): New.
5849         (vec_interleave_highv2si_le): New.
5850         (insn_v4int_h): Handle big endian.
5851         (vec_interleave_lowv2si): Handle big endian.
5852         (vec_interleave_lowv2si_be): New.
5853         (vec_interleave_lowv2si_le): New.
5854         (insn_v4int_l): Handle big endian.
5855         * config/tilegx/tilegx.opt (mbig-endian): New option.
5856         (mlittle-endian): New option.
5857         * doc/install.texi: Document tilegxbe-linux.
5858         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
5860 2014-02-24  Martin Jambor  <mjambor@suse.cz>
5862         PR ipa/60266
5863         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
5864         there are no parameter descriptors.
5866 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
5868         PR rtl-optimization/60268
5869         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
5870         initialization to ...
5871         (sched_rgn_init): ... here.
5872         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
5874 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
5876         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
5877         names.
5879 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
5881         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
5882         definition.
5884 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
5886         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
5887         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
5889 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
5891         * config/microblaze/predicates.md: Add cmp_op predicate.
5892         * config/microblaze/microblaze.md: Add branch_compare instruction
5893         which uses cmp_op predicate and emits cmp insn before branch.
5894         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
5895         to microblaze_expand_conditional_branch and consolidate logic.
5896         (microblaze_expand_conditional_branch): emit branch_compare
5897         insn instead of handling cmp op separate from branch insn.
5899 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5901         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
5902         to permit subregs.
5904 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5906         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
5907         define_insn with define_expand and new define_insn
5908         *altivec_lve<VI_char>x_internal.
5909         (altivec_stve<VI_char>x): Replace define_insn with define_expand
5910         and new define_insn *altivec_stve<VI_char>x_internal.
5911         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
5912         prototype.
5913         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
5914         lve*x built-ins.
5915         (altivec_expand_stvex_be): New function.
5917 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
5919         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
5920         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
5921         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
5922         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
5924 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
5926         PR target/60298
5927         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
5928         instead of emit_move_insn.
5930 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5932         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
5933         vspltw with vsldoi.
5934         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
5935         gen_altivec_vsumsws.
5937 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5939         * config/rs6000/altivec.md (altivec_lvxl): Rename as
5940         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
5941         (altivec_lvxl_<mode>): New define_expand incorporating
5942         -maltivec=be semantics where needed.
5943         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
5944         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
5945         semantics where needed.
5946         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
5947         (altivec_stvx_<mode>): New define_expand incorporating
5948         -maltivec=be semantics where needed.
5949         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
5950         VM2 iterator instead of V4SI.
5951         (altivec_stvxl_<mode>): New define_expand incorporating
5952         -maltivec=be semantics where needed.
5953         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
5954         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
5955         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
5956         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
5957         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
5958         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
5959         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
5960         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
5961         ALTIVEC_BUILTIN_STVXL.
5962         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
5963         (altivec_expand_stvx_be): Likewise.
5964         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
5965         (altivec_expand_lvx_be): Likewise.
5966         (altivec_expand_stvx_be): Likewise.
5967         (altivec_expand_builtin): Add cases for
5968         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
5969         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
5970         (altivec_init_builtins): Add definitions for
5971         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
5972         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
5974 2014-02-21  Catherine Moore  <clm@codesourcery.com>
5976         * doc/invoke.texi (mvirt, mno-virt): Document.
5977         * config/mips/mips.opt (mvirt): New option.
5978         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
5980 2014-02-21  Richard Biener  <rguenther@suse.de>
5982         PR tree-optimization/60276
5983         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
5984         (STMT_VINFO_MIN_NEG_DIST): New macro.
5985         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
5986         STMT_VINFO_MIN_NEG_DIST.
5987         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
5988         made for negative dependence distances still hold.
5990 2014-02-21  Richard Biener  <rguenther@suse.de>
5992         PR middle-end/60291
5993         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
5994         DECL_INITIAL for globals not in the current function context.
5996 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
5998         PR tree-optimization/56490
5999         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
6000         * tree-ssa-uninit.c: Include params.h.
6001         (compute_control_dep_chain): Add num_calls argument, return false
6002         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
6003         num_calls to recursive call.
6004         (find_predicates): Change dep_chain into normal array,
6005         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
6006         variable and adjust compute_control_dep_chain caller.
6007         (find_def_preds): Likewise.
6009 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
6011         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
6012         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
6014 2014-02-21  Nick Clifton  <nickc@redhat.com>
6016         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
6017         (pushhi1): Likewise.
6018         (popqi1): Add mode to pre_dec.
6019         (pophi1): Likewise.
6021 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
6023         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
6024         mode for mask of V8SFmode permutation.
6026 2014-02-20  Richard Henderson  <rth@redhat.com>
6028         PR c++/60272
6029         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
6030         a new pseudo for OLDVAL.
6032 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
6034         PR target/57896
6035         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
6036         gen_reg_rtx if d->testing_p.
6037         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
6038         if d->testing_p and we will certainly return true.
6039         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
6040         if d->testing_p.
6042 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
6044         * emit-rtl.c (gen_reg_rtx): Assert that
6045         crtl->emit.regno_pointer_align_length is non-zero.
6047 2014-02-20  Richard Henderson  <rth@redhat.com>
6049         PR c++/60272
6050         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
6051         on failure the store back into EXPECT.
6053 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
6054             Sandra Loosemore  <sandra@codesourcery.com>
6056         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
6057         * config/nios2/nios2.c (nios2_function_profiler): Add
6058         -fPIC (flag_pic == 2) support.
6059         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
6060         (nios2_large_offset_p): New function.
6061         (nios2_unspec_reloc_p): Move up position, update to use
6062         nios2_large_offset_p.
6063         (nios2_unspec_address): Remove function.
6064         (nios2_unspec_offset): New function.
6065         (nios2_large_got_address): New function.
6066         (nios2_got_address): Add large offset support.
6067         (nios2_legitimize_tls_address): Update usage of removed and new
6068         functions.
6069         (nios2_symbol_binds_local_p): New function.
6070         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
6071         (nios2_legitimize_address): Update to use nios2_large_offset_p.
6072         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
6073         (nios2_print_operand): Merge H/L processing, add hiadj/lo
6074         processing for (const (unspec ...)).
6075         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
6077 2014-02-20  Richard Biener  <rguenther@suse.de>
6079         * tree-cfg.c (replace_uses_by): Mark altered BBs before
6080         doing the substitution.
6081         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
6083 2014-02-20  Martin Jambor  <mjambor@suse.cz>
6085         PR ipa/55260
6086         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
6087         info when checking whether lattices are bottom.
6089 2014-02-20  Richard Biener  <rguenther@suse.de>
6091         PR middle-end/60221
6092         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
6093         regions at -O0.
6095 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
6097         PR ipa/58555
6098         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
6099         parameter specifying the scaling.
6100         (inline_call): Update.
6101         (want_inline_recursively): Guard division by zero.
6102         (recursive_inlining): Update.
6103         * ipa-inline.h (clone_inlined_nodes): Update.
6105 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
6107         PR target/60204
6108         * config/i386/i386.c (classify_argument): Pass structures of size
6109         64 bytes or less in register.
6111 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
6112             Kirill Yukhin  <kirill.yukhin@intel.com>
6114         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
6115         (_mm_rcp28_round_ss): Ditto.
6116         (_mm_rsqrt28_round_sd): Ditto.
6117         (_mm_rsqrt28_round_ss): Ditto.
6118         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
6119         (_mm_rcp14_round_ss): Ditto.
6120         (_mm_rsqrt14_round_sd): Ditto.
6121         (_mm_rsqrt14_round_ss): Ditto.
6122         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
6123         the first input operand, get rid of match_dup.
6124         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
6125         attribute to sse.
6126         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6127         Ditto.
6128         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
6129         operand as the first input operand, set type attribute.
6130         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6131         Set type attribute.
6132         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
6133         operand as the first input operand, set type attribute.
6135 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6137         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
6138         bit of zero.
6140 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
6142         PR target/60207
6143         * config/i386/i386.c (construct_container): Remove TFmode check
6144         for X86_64_INTEGER_CLASS.
6146 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
6148         PR target/59794
6149         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
6150         only when -Wpsabi is enabled.
6152 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
6154          PR target/59799
6155         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
6156         passing arrays in registers are the same as for structs, so remove the
6157         special case for them.
6159 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
6161         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
6162         destination type, extract only the valid bits if the source type is not
6163         integral and has a different mode.
6165 2014-02-19  Richard Biener  <rguenther@suse.de>
6167         PR ipa/60243
6168         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
6169         for all calls.
6171 2014-02-19  Richard Biener  <rguenther@suse.de>
6173         PR ipa/60243
6174         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
6175         (ipa_modify_call_arguments): Emit an argument load explicitely and
6176         preserve virtual SSA form there and for the replacement call.
6177         Do not update SSA form nor free dominance info.
6179 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6181         * ipa.c (function_and_variable_visibility): Also clear WEAK
6182         flag when disolving COMDAT_GROUP.
6184 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6186         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
6187         * ipa-prop.c (ipa_set_jf_known_type): Return early when
6188         not devirtualizing.
6189         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
6190         do more sanity checks.
6191         (detect_type_change): Return true when giving up early.
6192         (compute_complex_assign_jump_func): Fix type parameter of
6193         ipa_set_ancestor_jf.
6194         (compute_complex_ancestor_jump_func): Likewise.
6195         (update_jump_functions_after_inlining): Fix updating of
6196         ancestor function.
6197         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
6199 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6201         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
6202         inline clones when edge disappears.
6204 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
6206         PR target/60203
6207         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
6208         Split 64-bit moves into 2 patterns.  Do not allow the use of
6209         direct move for TDmode in little endian, since the decimal value
6210         has little endian bytes within a word, but the 64-bit pieces are
6211         ordered in a big endian fashion, and normal subreg's of TDmode are
6212         not allowed.
6213         (mov<mode>_64bit_dm): Likewise.
6214         (movtd_64bit_nodm): Likewise.
6216 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
6218         PR tree-optimization/60174
6219         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
6220         statement of an SSA_NAME that occurs in an abnormal PHI node.
6222 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
6224         PR sanitizer/60142
6225         * final.c (SEEN_BB): Remove.
6226         (SEEN_NOTE, SEEN_EMITTED): Renumber.
6227         (final_scan_insn): Don't force_source_line on second
6228         NOTE_INSN_BASIC_BLOCK.
6230 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
6232         PR target/60205
6233         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
6234         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
6235         (type_natural_mode): Warn ABI change when %zmm register is not
6236         available for AVX512F vector value passing.
6238 2014-02-18  Kai Tietz  <ktietz@redhat.com>
6240         PR target/60193
6241         * config/i386/i386.c (ix86_expand_prologue): Use value in
6242         rax register as displacement when restoring %r10 or %rax.
6243         Fix wrong offset when restoring both registers.
6245 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
6247         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
6248         assertion with conditional return.
6250 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
6251             Uros Bizjak  <ubizjak@gmail.com>
6253         PR driver/60233
6254         * config/i386/driver-i386.c (host_detect_local_cpu): If
6255         YMM state is not saved by the OS, also clear has_f16c.  Move
6256         CPUID 0x80000001 handling before YMM state saving checking.
6258 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
6260         PR rtl-optimization/58960
6261         * haifa-sched.c (alloc_global_sched_pressure_data): New,
6262         factored out from ...
6263         (sched_init): ... here.
6264         (free_global_sched_pressure_data): New, factored out from ...
6265         (sched_finish): ... here.
6266         * sched-int.h (free_global_sched_pressure_data): Declare.
6267         * sched-rgn.c (nr_regions_initial): New static global.
6268         (haifa_find_rgns): Initialize it.
6269         (schedule_region): Disable sched-pressure for the newly
6270         generated regions.
6272 2014-02-17  Richard Biener  <rguenther@suse.de>
6274         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
6275         release SSA defs of pattern stmts.
6277 2014-02-17  Richard Biener  <rguenther@suse.de>
6279         * tree-inline.c (expand_call_inline): Release the virtual
6280         operand defined by the call we are about to inline.
6282 2014-02-17  Richard Biener  <rguenther@suse.de>
6284         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
6286 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
6287             Ilya Tocar  <ilya.tocar@intel.com>
6289         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
6290         arguments order in builtin.
6291         (_mm512_permutexvar_epi64): Ditto.
6292         (_mm512_mask_permutexvar_epi64): Ditto
6293         (_mm512_maskz_permutexvar_epi32): Ditto
6294         (_mm512_permutexvar_epi32): Ditto
6295         (_mm512_mask_permutexvar_epi32): Ditto
6297 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6299         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
6300         (p8_vmrgow): Likewise.
6302 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6304         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
6305         endian targets.
6307 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
6309         PR target/60203
6310         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
6311         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
6312         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
6313         using direct move instructions on ISA 2.07.  Also adjust
6314         instruction length for 64-bit.
6315         (mov<mode>_64bit, TFmode/TDmode): Likewise.
6316         (mov<mode>_32bit, TFmode/TDmode): Likewise.
6318 2014-02-15  Alan Modra  <amodra@gmail.com>
6320         PR target/58675
6321         PR target/57935
6322         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
6323         find_replacement on parts of insn rtl that might be reloaded.
6325 2014-02-15  Richard Biener  <rguenther@suse.de>
6327         PR tree-optimization/60183
6328         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
6329         (tree_ssa_phiprop): Calculate and free post-dominators.
6331 2014-02-14  Jeff Law  <law@redhat.com>
6333         PR rtl-optimization/60131
6334         * ree.c (get_extended_src_reg): New function.
6335         (combine_reaching_defs): Use it rather than assuming location of REG.
6336         (find_and_remove_re): Verify first operand of extension is
6337         a REG before adding the insns to the copy list.
6339 2014-02-14  Roland McGrath  <mcgrathr@google.com>
6341         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
6342         * configure: Regenerated.
6343         * config.in: Regenerated.
6344         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
6345         instead of ASM_SHORT.
6347 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
6348             Richard Earnshaw  <rearnsha@arm.com>
6350         PR rtl-optimization/59535
6351         * lra-constraints.c (process_alt_operands): Encourage alternative
6352         when unassigned pseudo class is superset of the alternative class.
6353         (inherit_reload_reg): Don't inherit when optimizing for code size.
6354         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
6355         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
6356         modes not less than 4 for Thumb1.
6358 2014-02-14  Kyle McMartin  <kyle@redhat.com>
6360         PR pch/60010
6361         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
6363 2014-02-14  Richard Biener  <rguenther@suse.de>
6365         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
6366         (get_frame_arg): Drop the assert with langhook types_compatible_p.
6367         Do not strip INDIRECT_REFs.
6369 2014-02-14  Richard Biener  <rguenther@suse.de>
6371         PR lto/60179
6372         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
6373         DECL_FUNCTION_SPECIFIC_TARGET.
6374         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
6375         * tree-streamer-out.c (pack_ts_target_option): Remove.
6376         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
6377         (write_ts_function_decl_tree_pointers): Do not stream
6378         DECL_FUNCTION_SPECIFIC_TARGET.
6379         * tree-streamer-in.c (unpack_ts_target_option): Remove.
6380         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
6381         (lto_input_ts_function_decl_tree_pointers): Do not stream
6382         DECL_FUNCTION_SPECIFIC_TARGET.
6384 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
6386         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
6387         (get_initial_def_for_induction, vectorizable_induction): Ignore
6388         debug stmts when looking for exit_phi.
6389         (vectorizable_live_operation): Fix up condition.
6391 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
6393         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
6394         nreverse() because it changes the content of original tree list.
6396 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
6398         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
6399         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
6401 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
6403         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
6404         GNU coding standards.
6406 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
6408         PR debug/60152
6409         * dwarf2out.c (gen_subprogram_die): Don't call
6410         add_calling_convention_attribute if subr_die is old_die.
6412 2014-02-13  Sharad Singhai  <singhai@google.com>
6414         * doc/optinfo.texi: Fix order of nodes.
6416 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
6418         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
6419         operands[2], not operands[3].
6421 2014-02-13  Richard Biener  <rguenther@suse.de>
6423         PR bootstrap/59878
6424         * doc/install.texi (ISL): Update recommended version to 0.12.2,
6425         mention the possibility of an in-tree build.
6426         (CLooG): Update recommended version to 0.18.1, mention the
6427         possibility of an in-tree build and clarify that the ISL
6428         bundled with CLooG does not work.
6430 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
6432         PR target/43546
6433         * expr.c (compress_float_constant): If x is a hard register,
6434         extend into a pseudo and then move to x.
6436 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6438         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
6439         caused by bad second argument to warning_at() with -mhotpatch and
6440         nested functions (e.g. with gfortran).
6442 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
6444         * opts.c (option_name): Remove "enabled by default" rider.
6446 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
6448         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
6450 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
6451             Uros Bizjak  <ubizjak@gmail.com>
6453         PR target/60151
6454         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
6455         * configure: Regenerated.
6457 2014-02-12  Richard Biener  <rguenther@suse.de>
6459         * vec.c (vec_prefix::calculate_allocation): Move as
6460         inline variant to vec.h.
6461         (vec_prefix::calculate_allocation_1): New out-of-line version.
6462         * vec.h (vec_prefix::calculate_allocation_1): Declare.
6463         (vec_prefix::m_has_auto_buf): Rename to ...
6464         (vec_prefix::m_using_auto_storage): ... this.
6465         (vec_prefix::calculate_allocation): Inline the easy cases
6466         and dispatch to calculate_allocation_1 which doesn't need the
6467         prefix address.
6468         (va_heap::reserve): Use gcc_checking_assert.
6469         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
6470         m_using_auto_storage.
6471         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
6472         member and adjust.
6473         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
6474         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
6475         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
6477 2014-02-12  Richard Biener  <rguenther@suse.de>
6479         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
6480         when we found a dependence.
6482 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
6484         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
6485         common code...
6486         (maybe_fold_stmt): ... into this new function.
6487         * omp-low.c (lower_omp): Update comment.
6489         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
6490         last use.
6492         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
6493         dereference.
6495 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
6497         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
6498         identifiers in comments.
6499         (cortexa53_extra_costs): Likewise.
6500         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
6501         (cortexa7_extra_costs): Likewise.
6502         (cortexa12_extra_costs): Likewise.
6503         (cortexa15_extra_costs): Likewise.
6504         (v7m_extra_costs): Likewise.
6506 2014-02-12  Richard Biener  <rguenther@suse.de>
6508         PR middle-end/60092
6509         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
6510         of posix_memalign being successful.
6511         (lower_stmt): Restrict lowering of posix_memalign to when
6512         -ftree-bit-ccp is enabled.
6514 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
6516         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
6517         arg_loc.
6518         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
6520 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
6522         PR rtl-optimization/60116
6523         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
6524         other_insn once the combination has been validated.
6526 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
6528         PR lto/59468
6529         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
6530         and wrapper.
6531         * ipa-devirt.c: Include demangle.h
6532         (odr_violation_reported): New static variable.
6533         (add_type_duplicate): Update odr_violations.
6534         (maybe_record_node): Add completep parameter; update it.
6535         (record_target_from_binfo): Add COMPLETEP parameter;
6536         update it as needed.
6537         (possible_polymorphic_call_targets_1): Likewise.
6538         (struct polymorphic_call_target_d): Add nonconstruction_targets;
6539         rename FINAL to COMPLETE.
6540         (record_targets_from_bases): Sanity check we found the binfo;
6541         fix COMPLETEP updating.
6542         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
6543         parameter, fix computing of COMPLETEP.
6544         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
6545         at LTO time do demangling.
6546         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
6547         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
6548         parameter.
6549         (gimple_get_virt_method_for_binfo): Likewise.
6550         * gimple-fold.h (gimple_get_virt_method_for_binfo,
6551         gimple_get_virt_method_for_vtable): Update prototypes.
6553 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
6555         PR target/49008
6556         * genautomata.c (add_presence_absence): Fix typo with
6557         {final_}presence_list.
6559 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
6561         PR target/60137
6562         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
6563         for VSX/Altivec vectors that land in GPR registers.
6565 2014-02-11  Richard Henderson  <rth@redhat.com>
6566             Jakub Jelinek  <jakub@redhat.com>
6568         PR debug/59776
6569         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
6570         around drhs if type conversion to lacc->type is not useless.
6572 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6574         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
6575         tuning struct.
6576         (cortex-a57.cortex-a53): Likewise.
6577         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
6579 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6581         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
6582         arm_restrict_it.
6584 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
6586         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
6587         add_options_for_arm_vfp3.
6589 2014-02-11  Jeff Law  <law@redhat.com>
6591         PR middle-end/54041
6592         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
6593         object with an undesirable mode.
6595 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6597         PR libgomp/60107
6598         * config/i386/sol2-9.h: New file.
6599         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
6600         *-*-solaris2.9*): Use it.
6602 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
6604         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
6605         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
6607 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
6609         * config/microblaze/microblaze.c: Extend mcpu version format
6611 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
6613         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
6615 2014-02-10  Richard Henderson  <rth@redhat.com>
6617         PR target/59927
6618         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
6619         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
6620         ms-abi vs -mno-accumulate-outgoing-args.
6621         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
6622         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
6623         respect to ms-abi.
6625 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
6627         PR middle-end/60080
6628         * cfgexpand.c (expand_asm_operands): Attach source location to
6629         ASM_INPUT rtx objects.
6630         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
6632 2014-02-10  Nick Clifton  <nickc@redhat.com>
6634         * config/mn10300/mn10300.c (popcount): New function.
6635         (mn10300_expand_prologue): Include saved registers in stack usage
6636         count.
6638 2014-02-10  Jeff Law  <law@redhat.com>
6640         PR middle-end/52306
6641         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
6642         when changing the SET_DEST of a prior insn to avoid an input reload.
6644 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
6646         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
6647         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
6648         -mcall-openbsd, or -mcall-linux.
6649         (CC1_ENDIAN_BIG_SPEC): Remove.
6650         (CC1_ENDIAN_LITTLE_SPEC): Remove.
6651         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
6652         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
6653         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
6654         and %cc1_endian_default.
6655         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
6657 2014-02-10  Richard Biener  <rguenther@suse.de>
6659         PR tree-optimization/60115
6660         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
6661         MEM_REF handling.  Properly verify that the accesses are not
6662         out of the objects bound.
6664 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6666         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
6667         coretex to cortex.
6669 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
6671         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
6672         proper constants and fix formatting.
6673         (possible_polymorphic_call_targets): Fix formatting.
6675 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
6676             Ilya Tocar  <ilya.tocar@intel.com>
6678         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
6679         (_mm512_loadu_epi32): Renamed into...
6680         (_mm512_loadu_si512): This.
6681         (_mm512_storeu_epi32): Renamed into...
6682         (_mm512_storeu_si512): This.
6683         (_mm512_maskz_ceil_ps): Removed.
6684         (_mm512_maskz_ceil_pd): Ditto.
6685         (_mm512_maskz_floor_ps): Ditto.
6686         (_mm512_maskz_floor_pd): Ditto.
6687         (_mm512_floor_round_ps): Ditto.
6688         (_mm512_floor_round_pd): Ditto.
6689         (_mm512_ceil_round_ps): Ditto.
6690         (_mm512_ceil_round_pd): Ditto.
6691         (_mm512_mask_floor_round_ps): Ditto.
6692         (_mm512_mask_floor_round_pd): Ditto.
6693         (_mm512_mask_ceil_round_ps): Ditto.
6694         (_mm512_mask_ceil_round_pd): Ditto.
6695         (_mm512_maskz_floor_round_ps): Ditto.
6696         (_mm512_maskz_floor_round_pd): Ditto.
6697         (_mm512_maskz_ceil_round_ps): Ditto.
6698         (_mm512_maskz_ceil_round_pd): Ditto.
6699         (_mm512_expand_pd): Ditto.
6700         (_mm512_expand_ps): Ditto.
6701         * config/i386/i386.c (ix86_builtins): Remove
6702         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
6703         (bdesc_args): Ditto.
6704         * config/i386/predicates.md (const1256_operand): New.
6705         (const_1_to_2_operand): Ditto.
6706         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
6707         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6708         (*avx512pf_gatherpf<mode>sf): Ditto.
6709         (avx512pf_gatherpf<mode>df): Ditto.
6710         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6711         (*avx512pf_gatherpf<mode>df): Ditto.
6712         (avx512pf_scatterpf<mode>sf): Ditto.
6713         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6714         (*avx512pf_scatterpf<mode>sf): Ditto.
6715         (avx512pf_scatterpf<mode>df): Ditto.
6716         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6717         (*avx512pf_scatterpf<mode>df): Ditto.
6718         (avx512f_expand<mode>): Removed.
6719         (<shift_insn><mode>3<mask_name>): Change predicate type.
6721 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
6723         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
6724         not at the end of datarefs vector use ordered_remove to avoid
6725         reordering datarefs vector.
6727         PR c/59984
6728         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
6729         mark local addressable non-static vars as GOVD_PRIVATE
6730         instead of GOVD_LOCAL.
6731         * omp-low.c (lower_omp_for): Move gimple_bind_vars
6732         and BLOCK_VARS of gimple_bind_block to new_stmt rather
6733         than copying them.
6735         PR middle-end/60092
6736         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
6737         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
6738         assume_aligned or alloc_align attributes.
6739         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
6740         arguments.  Handle also assume_aligned and alloc_align attributes.
6741         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
6742         calls to functions with assume_aligned or alloc_align attributes.
6743         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
6745 2014-02-08  Terry Guo  <terry.guo@arm.com>
6747         * doc/invoke.texi: Document ARM -march=armv7e-m.
6749 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
6751         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
6752         flag on __cilkrts_rethrow builtin.
6754         PR ipa/60026
6755         * ipa-cp.c (determine_versionability): Fail at -O0
6756         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
6757         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
6759         Revert:
6760         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
6762         PR ipa/60026
6763         * tree-inline.c (copy_forbidden): Fail for
6764         __attribute__((optimize (0))) functions.
6766 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
6768         * varpool.c: Include pointer-set.h.
6769         (varpool_remove_unreferenced_decls): Variables in other partitions
6770         will not be output; be however careful to not lose information
6771         about partitioning.
6773 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
6775         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
6776         lookup in the vtable constructor.
6778 2014-02-07  Jeff Law  <law@redhat.com>
6780         PR target/40977
6781         * config/m68k/m68k.md (ashldi_extsi): Turn into a
6782         define_insn_and_split.
6784         * ipa-inline.c (inline_small_functions): Fix typos.
6786 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
6788         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
6789         (s390_can_use_return_insn): Declare.
6790         * config/s390/s390.h (EPILOGUE_USES): Define.
6791         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
6792         instructions.
6793         (s390_chunkify_start): Handle return JUMP_LABELs.
6794         (s390_early_mach): Emit a main_pool instruction on the entry edge.
6795         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
6796         (s390_can_use_return_insn): New functions.
6797         (s390_fix_long_loop_prediction): Handle conditional returns.
6798         (TARGET_SET_UP_BY_PROLOGUE): Define.
6799         * config/s390/s390.md (ANY_RETURN): New code iterator.
6800         (*creturn, *csimple_return, return, simple_return): New patterns.
6802 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
6804         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
6805         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
6806         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
6807         REG_CFA_RESTORE list when deciding not to restore a register.
6809 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
6811         * config/s390/s390.c: Include tree-pass.h and context.h.
6812         (s390_early_mach): New function, split out from...
6813         (s390_emit_prologue): ...here.
6814         (pass_data_s390_early_mach): New pass structure.
6815         (pass_s390_early_mach): New class.
6816         (s390_option_override): Create and register early_mach pass.
6817         Move to end of file.
6819 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
6821         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
6822         to match for the exit block.
6824 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6826         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
6827         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
6828         Reject misaligned operands.
6830 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6832         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
6834 2014-02-07  Richard Biener  <rguenther@suse.de>
6836         PR middle-end/60092
6837         * gimple-low.c (lower_builtin_posix_memalign): New function.
6838         (lower_stmt): Call it to lower posix_memalign in a way
6839         to make alignment info accessible.
6841 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
6843         PR c++/60082
6844         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
6845         __builtin_setjmp_receiver.
6847 2014-02-07  Richard Biener  <rguenther@suse.de>
6849         PR middle-end/60092
6850         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
6851         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
6852         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
6853         Handle BUILT_IN_POSIX_MEMALIGN.
6854         (find_func_clobbers): Likewise.
6855         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6856         (call_may_clobber_ref_p_1): Likewise.
6858 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
6860         PR ipa/59918
6861         * ipa-devirt.c (record_target_from_binfo): Remove overactive
6862         sanity check.
6864 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
6866         PR ipa/59469
6867         * lto-cgraph.c (lto_output_node): Use
6868         symtab_get_symbol_partitioning_class.
6869         (lto_output_varpool_node): likewise.
6870         (symtab_get_symbol_partitioning_class): Move here from
6871         lto/lto-partition.c
6872         * cgraph.h (symbol_partitioning_class): Likewise.
6873         (symtab_get_symbol_partitioning_class): Declare.
6875 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
6877         * ggc.h (ggc_internal_cleared_alloc): New macro.
6878         * vec.h (vec_safe_copy): Handle memory stats.
6879         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
6880         * target-globals.c (save_target_globals): Likewise.
6882 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
6884         PR target/60077
6885         * expr.c (emit_move_resolve_push): Export; be bit more selective
6886         on when to clear alias set.
6887         * expr.h (emit_move_resolve_push): Declare.
6888         * function.h (struct function): Add tail_call_marked.
6889         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
6890         * config/i386/i386-protos.h (ix86_expand_push): Remove.
6891         * config/i386/i386.md (TImode move expander): De not call
6892         ix86_expand_push.
6893         (FP push expanders): Preserve memory attributes.
6894         * config/i386/sse.md (push<mode>1): Remove.
6895         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
6896         (ix86_expand_push): Remove.
6897         * config/i386/mmx.md (push<mode>1): Remove.
6899 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
6901         PR rtl-optimization/60030
6902         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
6903         lopart with paradoxical subreg before shifting it up by hprec.
6905 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6907         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
6908         Remove extra newline at end of file.
6909         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
6910         (arm_issue_rate): Handle cortexa57.
6911         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
6912         (cortex-a57.cortex-a53): Likewise.
6914 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
6916         PR target/59575
6917         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
6918         don't record in REG_FRAME_RELATED_EXPR registers not set in that
6919         bitmask.
6920         (arm_expand_prologue): Adjust all callers.
6921         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
6922         info, registers also at the lowest numbered registers side.  Use
6923         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
6924         XEXP.
6926         PR debug/59992
6927         * var-tracking.c (adjust_mems): Before adding a SET to
6928         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
6930 2014-02-06  Alan Modra  <amodra@gmail.com>
6932         PR target/60032
6933         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
6934         change SDmode to DDmode when lra_in_progress.
6936 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
6938         PR middle-end/59150
6939         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
6940         free_data_ref on the dr first, and before goto again also set dr
6941         to the next dr.  For simd_lane_access, free old datarefs[i] before
6942         overwriting it.  For get_vectype_for_scalar_type failure, don't
6943         free_data_ref if simd_lane_access.
6945         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
6947         PR target/60062
6948         * tree.h (opts_for_fn): New inline function.
6949         (opt_for_fn): Define.
6950         * config/i386/i386.c (ix86_function_regparm): Use
6951         opt_for_fn (decl, optimize) instead of optimize.
6953 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6955         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
6956         for SYMBOL_REF in large memory model.
6958 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6960         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
6961         and crypto support.
6962         (cortex-a57): Likewise.
6963         (cortex-a57.cortex-a53): Likewise.
6965 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
6966             Kugan Vivekanandarajah  <kuganv@linaro.org>
6968         * config/arm/arm.c (arm_vector_alignment_reachable): Check
6969         unaligned_access.
6970         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
6972 2014-02-06  Richard Biener  <rguenther@suse.de>
6974         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
6975         set_loop_copy and initialize_original_copy_tables.
6977 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
6979         * config/aarch64/aarch64-simd.md
6980         (aarch64_ashr_simddi): Change QI to SI.
6982 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
6983             Jakub Jelinek  <jakub@redhat.com>
6985         PR middle-end/60013
6986         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
6987         of the dataflow.
6989 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6991         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
6992         CODE_FOR_altivec_vpku[hw]um to
6993         CODE_FOR_altivec_vpku[hw]um_direct.
6994         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
6995         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
6996         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
6997         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
6999 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7001         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
7002         generation for -maltivec=be.
7003         (altivec_vsumsws): Simplify redundant test.
7005 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7007         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
7008         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
7009         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
7010         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
7011         gen_altivec_vpkuwum.
7012         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
7013         BYTES_BIG_ENDIAN.
7014         (altivec_vpks<VI_char>ss): Likewise.
7015         (altivec_vpks<VI_char>us): Likewise.
7016         (altivec_vpku<VI_char>us): Likewise.
7017         (altivec_vpku<VI_char>um): Likewise.
7018         (altivec_vpku<VI_char>um_direct): New (copy of
7019         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
7020         internal use).
7021         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
7022         target is little endian and -maltivec=be is not specified.
7023         (*altivec_vupkhs<VU_char>_direct): New (copy of
7024         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
7025         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
7026         target is little endian and -maltivec=be is not specified.
7027         (*altivec_vupkls<VU_char>_direct): New (copy of
7028         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
7029         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
7030         little endian and -maltivec=be is not specified.
7031         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
7032         little endian and -maltivec=be is not specified.
7034 2014-02-05  Richard Henderson  <rth@redhat.com>
7036         PR debug/52727
7037         * combine-stack-adj.c: Revert r206943.
7038         * sched-int.h (struct deps_desc): Add last_args_size.
7039         * sched-deps.c (init_deps): Initialize it.
7040         (sched_analyze_insn): Add OUTPUT dependencies between insns that
7041         contain REG_ARGS_SIZE notes.
7043 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
7045         * lto-cgraph.c (asm_nodes_output): Make global.
7046         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
7047         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
7048         (driver_handle_option): Handle OPT_fwpa.
7050 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
7052         PR ipa/59947
7053         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
7054         a comment typo and formatting issue.  If odr_hash hasn't been
7055         created, return vNULL and set *completep to false.
7057         PR middle-end/57499
7058         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
7059         bb with no successors.
7061 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
7063         PR target/59718
7064         * doc/invoke.texi (-march): Clarify documentation for ARM.
7065         (-mtune): Likewise.
7066         (-mcpu): Likewise.
7068 2014-02-05  Richard Biener  <rguenther@suse.de>
7070         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
7071         when not vectorizing because of too many alias checks.
7072         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
7073         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
7075 2014-02-05  Nick Clifton  <nickc@redhat.com>
7077         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
7078         accept extended registers in any mode when compiling for the MN10300.
7080 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
7082         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
7083         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
7084         sanitization attributes.
7085         (can_inline_edge_p): Likewise.
7086         (sanitize_attrs_match_for_inline_p): New function.
7088 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7090         * ipa-prop.c (detect_type_change): Shor circuit testing of
7091         type changes on THIS pointer.
7093 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
7095         PR target/59777
7096         * config/pa/pa.c (legitimize_tls_address): Return original address
7097         if not passed a SYMBOL_REF rtx.
7098         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
7099         addresses.
7100         (pa_emit_move_sequence): Simplify TLS source operands.
7101         (pa_legitimate_constant_p): Reject all TLS constants.
7102         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
7103         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
7105 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7107         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
7108         groups when we know they are controlled by LTO.
7109         * varasm.c (default_binds_local_p_1): If object is in other partition,
7110         it will be resolved locally.
7112 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7114         * config/host-linux.c (linux_gt_pch_use_address): Don't
7115         use SSIZE_MAX because it is not always defined.
7117 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
7119         PR bootstrap/59913
7120         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
7121         threshold for pseudo splitting.
7122         (update_ebb_live_info): Process call argument hard registers and
7123         hard registers from insn definition too.
7124         (max_small_class_regs_num): New constant.
7125         (inherit_in_ebb): Update live hard regs through EBBs.  Update
7126         reloads_num only for small register classes.  Don't split for
7127         outputs of jumps.
7129 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
7131         PR ipa/60058
7132         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
7133         is non-null.
7135 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7137         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
7138         visibility is safe.
7140 2014-02-04  Marek Polacek  <polacek@redhat.com>
7142         * gdbinit.in (pel): Define.
7144 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7146         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
7147         behavior.
7149 2014-02-04  Richard Biener  <rguenther@suse.de>
7151         PR lto/59723
7152         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
7153         in function context local.
7154         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
7155         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
7156         similar to LTO_imported_decl_ref.
7158 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7160         PR tree-optimization/60002
7161         * cgraphclones.c (build_function_decl_skip_args): Clear
7162         DECL_LANG_SPECIFIC.
7164         PR tree-optimization/60023
7165         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
7166         false to gsi_replace.
7167         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
7168         has been in some EH region and vec_stmt could throw, add
7169         vec_stmt into the same EH region.
7170         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
7171         has no lhs, ignore it.
7172         * internal-fn.c (expand_MASK_LOAD): Likewise.
7174         PR ipa/60026
7175         * tree-inline.c (copy_forbidden): Fail for
7176         __attribute__((optimize (0))) functions.
7178         PR other/58712
7179         * omp-low.c (simd_clone_struct_copy): If from->inbranch
7180         is set, copy one less argument.
7181         (expand_simd_clones): Don't subtract clone_info->inbranch
7182         from simd_clone_struct_alloc argument.
7184         PR rtl-optimization/57915
7185         * recog.c (simplify_while_replacing): If all unary/binary/relational
7186         operation arguments are constant, attempt to simplify those.
7188         PR middle-end/59261
7189         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
7190         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
7192 2014-02-04  Richard Biener  <rguenther@suse.de>
7194         PR tree-optimization/60012
7195         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
7196         TBAA disambiguation to all DDRs.
7198 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7200         PR target/59788
7201         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
7202         (LINK_SPEC): Use it for -shared, -shared-libgcc.
7204 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7206         PR ipa/59882
7207         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
7209 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7211         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
7212         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
7214 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7216         PR ipa/59831
7217         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
7218         to figure out targets of polymorphic calls with known decl.
7219         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7220         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
7221         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
7222         (get_polymorphic_call_info): ... here.
7223         (get_polymorphic_call_info_from_invariant): New function.
7225 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7227         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
7228         lookup via vtable pointer; check for type consistency
7229         and turn inconsitent facts into UNREACHABLE.
7230         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7231         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
7232         type inconsistent querries; return UNREACHABLE instead.
7234 2014-02-03  Richard Henderson  <rth@twiddle.net>
7236         PR tree-opt/59924
7237         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
7238         already processed this node.
7239         (normalize_one_pred_1): Pass along mark_set.
7240         (normalize_one_pred): Create and destroy a pointer_set_t.
7241         (normalize_one_pred_chain): Likewise.
7243 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
7245         PR gcov-profile/58602
7246         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
7248 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7250         PR ipa/59831
7251         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
7252         -fno-devirtualize; try to devirtualize by the knowledge of
7253         virtual table pointer given by aggregate propagation.
7254         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7255         (ipa_print_node_jump_functions): Dump also offset that
7256         is relevant for polymorphic calls.
7257         (determine_known_aggregate_parts): Add arg_type parameter; use it
7258         instead of determining the type from pointer type.
7259         (ipa_compute_jump_functions_for_edge): Update call of
7260         determine_known_aggregate_parts.
7261         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
7262         (gimple_get_virt_method_for_binfo): ... here; simplify using
7263         vtable_pointer_value_to_vtable.
7264         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
7265         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
7266         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
7267         (vtable_pointer_value_to_vtable): Break out from ...; handle also
7268         POINTER_PLUS_EXPR.
7269         (vtable_pointer_value_to_binfo): ... here.
7270         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
7272 2014-02-03  Teresa Johnson  <tejohnson@google.com>
7274         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
7275         redef of outer loop index variable.
7277 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
7279         PR c++/53017
7280         PR c++/59211
7281         * doc/extend.texi (Function Attributes): Typo.
7283 2014-02-03  Cong Hou  <congh@google.com>
7285         PR tree-optimization/60000
7286         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
7287         if the vectorized statement is a store.  A store statement can only
7288         appear at the end of pattern statements.
7290 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
7292         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
7293         (ix86_option_override_internal): Default long double to 64-bit for
7294         32-bit Bionic and to 128-bit for 64-bit Bionic.
7296         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
7297         TARGET_LONG_DOUBLE_128 is true.
7298         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
7300         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
7301         (mlong-double-64): Negate -mlong-double-128.
7302         (mlong-double-128): New option.
7304         * config/i386/i386-c.c (ix86_target_macros): Define
7305         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
7307         * doc/invoke.texi: Document -mlong-double-128.
7309 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
7311         PR rtl-optimization/60024
7312         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
7314 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
7316         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
7318 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
7320         PR rtl-optimization/57662
7321         * sel-sched.c (code_motion_path_driver): Do not mark already not
7322         existing blocks in the visiting bitmap.
7324 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
7326         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
7327         on the insn being emitted.
7329 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
7330             Will Deacon  <will.deacon@arm.com>
7332         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
7334 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7336         * config/arm/arm-tables.opt: Regenerate.
7338 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7340         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
7341         for vector types other than V16QImode.
7342         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
7343         define_expand, and call altivec_expand_vec_perm_le when producing
7344         code with little endian element order.
7345         (*altivec_vperm_<mode>_internal): New insn having previous
7346         behavior of altivec_vperm_<mode>.
7347         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
7348         altivec_expand_vec_perm_le when producing code with little endian
7349         element order.
7350         (*altivec_vperm_<mode>_uns_internal): New insn having previous
7351         behavior of altivec_vperm_<mode>_uns.
7353 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7355         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
7356         (altivec_vsumsws): Add handling for -maltivec=be with a little
7357         endian target.
7358         (altivec_vsumsws_direct): New.
7359         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
7360         gen_altivec_vsumsws.
7362 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
7364         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
7365         vtable_pointer_value_to_binfo): New functions.
7366         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
7367         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
7369 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
7371         * config/nios2/nios2.md (load_got_register): Initialize GOT
7372         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
7373         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
7375 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
7377         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
7378         preserverd by passthrough, do not propagate the type.
7380 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
7382         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
7383         (mips_atomic_assign_expand_fenv): New function.
7384         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
7386 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
7388         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
7389         (__builtin_mips_set_fcsr): Likewise.
7390         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
7391         MIPS_USI_FTYPE_VOID.
7392         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
7393         (mips16_expand_set_fcsr): Likewise.
7394         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
7395         (mips16_set_fcsr_stub): Likewise.
7396         (mips16_get_fcsr_one_only_stub): New class.
7397         (mips16_set_fcsr_one_only_stub): Likewise.
7398         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
7399         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
7400         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
7401         (hard_float): New availability predicate.
7402         (mips_builtins): Add get_fcsr and set_fcsr.
7403         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
7404         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
7405         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
7406         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
7407         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
7408         patterns.
7410 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
7412         * config/mips/mips.c (mips_one_only_stub): New class.
7413         (mips_need_mips16_rdhwr_p): Replace with...
7414         (mips16_rdhwr_stub): ...this new variable.
7415         (mips16_stub_call_address): New function.
7416         (mips16_rdhwr_one_only_stub): New class.
7417         (mips_expand_thread_pointer): Use mips16_stub_call_address.
7418         (mips_output_mips16_rdhwr): Delete.
7419         (mips_finish_stub): New function.
7420         (mips_code_end): Use it to handle rdhwr stubs.
7422 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
7424         PR target/60017
7425         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
7426         when calculating size of integer atomic types.
7428 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
7430         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
7432 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
7434         PR tree-optimization/60003
7435         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
7436         * profile.c (branch_prob): Use gimple_call_builtin_p
7437         to check for BUILT_IN_SETJMP_RECEIVER.
7438         * tree-inline.c (copy_bb): Call notice_special_calls.
7440 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
7442         PR bootstrap/59985
7443         * lra-constraints.c (process_alt_operands): Update reload_sum only
7444         on the first pass.
7446 2014-01-31  Richard Henderson  <rth@redhat.com>
7448         PR middle-end/60004
7449         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
7450         until after else_eh is processed.
7452 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
7454         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
7455         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
7456         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
7457         in smmintrin.h, remove them.
7458         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
7459         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
7460         * config/i386/i386.md (ROUND_SAE): Fix value.
7461         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
7462         (const48_operand): New.
7463         * config/i386/subst.md (round), (round_expand): Use
7464         const_4_or_8_to_11_operand.
7465         (round_saeonly), (round_saeonly_expand): Use const48_operand.
7467 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
7469         * config/i386/constraints.md (Yk): Swap meaning with k.
7470         * config/i386/i386.md (movhi_internal): Change Yk to k.
7471         (movqi_internal): Ditto.
7472         (*k<logic><mode>): Ditto.
7473         (*andhi_1): Ditto.
7474         (*andqi_1): Ditto.
7475         (kandn<mode>): Ditto.
7476         (*<code>hi_1): Ditto.
7477         (*<code>qi_1): Ditto.
7478         (kxnor<mode>): Ditto.
7479         (kortestzhi): Ditto.
7480         (kortestchi): Ditto.
7481         (kunpckhi): Ditto.
7482         (*one_cmplhi2_1): Ditto.
7483         (*one_cmplqi2_1): Ditto.
7484         * config/i386/sse.md (): Change k to Yk.
7485         (avx512f_load<mode>_mask): Ditto.
7486         (avx512f_blendm<mode>): Ditto.
7487         (avx512f_store<mode>_mask): Ditto.
7488         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
7489         (avx512f_storedqu<mode>_mask): Ditto.
7490         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
7491         Ditto.
7492         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
7493         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
7494         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
7495         (avx512f_maskcmp<mode>3): Ditto.
7496         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
7497         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
7498         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
7499         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
7500         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
7501         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
7502         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
7503         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
7504         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
7505         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
7506         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
7507         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
7508         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
7509         (vec_extract_lo_<mode>_maskm): Ditto.
7510         (vec_extract_hi_<mode>_maskm): Ditto.
7511         (avx512f_vternlog<mode>_mask): Ditto.
7512         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
7513         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
7514         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
7515         (avx512f_<code>v8div16qi2_mask): Ditto.
7516         (avx512f_<code>v8div16qi2_mask_store): Ditto.
7517         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
7518         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
7519         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
7520         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
7521         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
7522         (*avx512pf_gatherpf<mode>df_mask): Ditto.
7523         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
7524         (*avx512pf_scatterpf<mode>df_mask): Ditto.
7525         (avx512cd_maskb_vec_dupv8di): Ditto.
7526         (avx512cd_maskw_vec_dupv16si): Ditto.
7527         (avx512f_vpermi2var<mode>3_maskz): Ditto.
7528         (avx512f_vpermi2var<mode>3_mask): Ditto.
7529         (avx512f_vpermi2var<mode>3_mask): Ditto.
7530         (avx512f_vpermt2var<mode>3_maskz): Ditto.
7531         (*avx512f_gathersi<mode>): Ditto.
7532         (*avx512f_gathersi<mode>_2): Ditto.
7533         (*avx512f_gatherdi<mode>): Ditto.
7534         (*avx512f_gatherdi<mode>_2): Ditto.
7535         (*avx512f_scattersi<mode>): Ditto.
7536         (*avx512f_scatterdi<mode>): Ditto.
7537         (avx512f_compress<mode>_mask): Ditto.
7538         (avx512f_compressstore<mode>_mask): Ditto.
7539         (avx512f_expand<mode>_mask): Ditto.
7540         * config/i386/subst.md (mask): Change k to Yk.
7541         (mask_scalar_merge): Ditto.
7542         (sd): Ditto.
7544 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
7546         * doc/extend.texi (Vector Extensions): Document ?: in C++.
7548 2014-01-31  Richard Biener  <rguenther@suse.de>
7550         PR middle-end/59990
7551         * builtins.c (fold_builtin_memory_op): Make sure to not
7552         use a floating-point mode or a boolean or enumeral type for
7553         the copy operation.
7555 2014-01-30  DJ Delorie  <dj@redhat.com>
7557         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
7558         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
7559         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
7560         whenever main() has an epilogue.
7562 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7564         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
7565         unused variable "field".
7566         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
7567         (vsx_mergeh_<mode>): Likewise.
7568         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
7569         (altivec_vmrghh): Likewise.
7570         (altivec_vmrghw): Likewise.
7571         (altivec_vmrglb): Likewise.
7572         (altivec_vmrglh): Likewise.
7573         (altivec_vmrglw): Likewise.
7574         (altivec_vspltb): Add missing uses.
7575         (altivec_vsplth): Likewise.
7576         (altivec_vspltw): Likewise.
7577         (altivec_vspltsf): Likewise.
7579 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
7581         PR target/59923
7582         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
7583         frame related instructions.
7585 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
7587         PR rtl-optimization/59959
7588         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
7589         any reload of register whose subreg is invalid.
7591 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
7593         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
7594         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
7595         Add missing return type - void.
7597 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7599         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
7600         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
7601         remove element index adjustment for endian (now handled in vsx.md
7602         and altivec.md).
7603         (altivec_expand_vec_perm_const): Use
7604         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
7605         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
7606         (vsx_xxspltw_<mode>): Adjust element index for little endian.
7607         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
7608         define_expand and a new define_insn *altivec_vspltb_internal;
7609         adjust for -maltivec=be on a little endian target.
7610         (altivec_vspltb_direct): New.
7611         (altivec_vsplth): Divide into a define_expand and a new
7612         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
7613         little endian target.
7614         (altivec_vsplth_direct): New.
7615         (altivec_vspltw): Divide into a define_expand and a new
7616         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
7617         little endian target.
7618         (altivec_vspltw_direct): New.
7619         (altivec_vspltsf): Divide into a define_expand and a new
7620         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
7621         a little endian target.
7623 2014-01-30  Richard Biener  <rguenther@suse.de>
7625         PR tree-optimization/59993
7626         * tree-ssa-forwprop.c (associate_pointerplus): Check we
7627         can propagate form the earlier stmt and avoid the transform
7628         when the intermediate result is needed.
7630 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
7632         * README.Portability: Fix typo.
7634 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
7636         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
7637         comparison_operator with ordered_comparison_operator.
7639 2014-01-30  Nick Clifton  <nickc@redhat.com>
7641         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
7642         Rename to mn10300_store_multiple_regs.
7643         * config/mn10300/mn10300.c: Likewise.
7644         * config/mn10300/mn10300.md (store_movm): Fix typo: call
7645         store_multiple_regs.
7646         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
7647         Call mn10300_store_multiple_regs.
7649 2014-01-30  Nick Clifton  <nickc@redhat.com>
7650             DJ Delorie  <dj@redhat.com>
7652         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
7653         %fp 2 to keep registers after it properly word-aligned.
7654         (rl78_alloc_physical_registers_umul): Handle the case where both
7655         input operands are the same.
7657 2014-01-30  Richard Biener  <rguenther@suse.de>
7659         PR tree-optimization/59903
7660         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
7661         check properly.
7663 2014-01-30  Jason Merrill  <jason@redhat.com>
7665         PR c++/59633
7666         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
7668         PR c++/59645
7669         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
7671 2014-01-30  Richard Biener  <rguenther@suse.de>
7673         PR tree-optimization/59951
7674         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
7676 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
7678         PR target/59784
7679         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
7680         SFmode to DFmode case.
7682 2014-01-29  DJ Delorie  <dj@redhat.com>
7684         * config/msp430/msp430.opt (-minrt): New.
7685         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
7686         if -minrt given.
7687         (ENDFILE_SPEC): Likewise.
7689 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
7691         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
7692         (estimate_function_body_sizes): Use it.
7694 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
7696         PR c++/58561
7697         * dwarf2out.c (is_cxx_auto): New.
7698         (is_base_type): Use it.
7699         (gen_type_die_with_usage): Likewise.
7701 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7703         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
7704         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
7705         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
7706         -maltivec=be with LE targets.
7707         (vsx_mergeh_<mode>): Likewise.
7708         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
7709         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
7710         (altivec_vmrghb): Replace with define_expand and new
7711         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
7712         (altivec_vmrghb_direct): New define_insn.
7713         (altivec_vmrghh): Replace with define_expand and new
7714         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
7715         (altivec_vmrghh_direct): New define_insn.
7716         (altivec_vmrghw): Replace with define_expand and new
7717         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
7718         (altivec_vmrghw_direct): New define_insn.
7719         (*altivec_vmrghsf): Adjust for endianness.
7720         (altivec_vmrglb): Replace with define_expand and new
7721         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
7722         (altivec_vmrglb_direct): New define_insn.
7723         (altivec_vmrglh): Replace with define_expand and new
7724         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
7725         (altivec_vmrglh_direct): New define_insn.
7726         (altivec_vmrglw): Replace with define_expand and new
7727         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
7728         (altivec_vmrglw_direct): New define_insn.
7729         (*altivec_vmrglsf): Adjust for endianness.
7730         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
7731         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
7732         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
7733         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
7734         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
7735         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
7736         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
7737         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
7739 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7741         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
7742         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
7743         whitespace.
7745 2014-01-29  Richard Biener  <rguenther@suse.de>
7747         PR tree-optimization/58742
7748         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
7749         associate_pointerplus_align.
7750         (associate_pointerplus_diff): New function.
7751         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
7752         and associate_pointerplus_diff.
7754 2014-01-29  Richard Biener  <rguenther@suse.de>
7756         * lto-streamer.h (LTO_major_version): Bump to 3.
7757         (LTO_minor_version): Reset to 0.
7759 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
7761         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
7762         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
7763         (arm_file_start): Generate correct asm header for armv7ve.
7764         * config/arm/bpabi.h: Add multilib support for armv7ve.
7765         * config/arm/driver-arm.c: Change the architectures of cortex-a7
7766         and cortex-a15 to armv7ve.
7767         * config/arm/t-aprofile: Add multilib support for armv7ve.
7768         * doc/invoke.texi: Document -march=armv7ve.
7770 2014-01-29  Richard Biener  <rguenther@suse.de>
7772         PR tree-optimization/58742
7773         * tree-ssa-forwprop.c (associate_plusminus): Return true
7774         if we changed sth, defer EH cleanup to ...
7775         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
7776         (simplify_mult): New function.
7778 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
7780         PR middle-end/59917
7781         PR tree-optimization/59920
7782         * tree.c (build_common_builtin_nodes): Remove
7783         __builtin_setjmp_dispatcher initialization.
7784         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
7785         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
7786         instead of gsi_after_labels + manually skipping debug stmts.
7787         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
7788         ignore bbs with IFN_ABNORMAL_DISPATCHER.
7789         * tree-inline.c (copy_edges_for_bb): Remove
7790         can_make_abnormal_goto argument, instead add abnormal_goto_dest
7791         argument.  Ignore computed_goto_p stmts.  Don't call
7792         make_abnormal_goto_edges.  If a call might need abnormal edges
7793         for non-local gotos, see if it already has an edge to
7794         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
7795         with true argument, don't do anything then, otherwise add
7796         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
7797         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
7798         caller.
7799         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
7800         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
7801         (lower_stmt): Don't set data->calls_builtin_setjmp.
7802         (lower_builtin_setjmp): Adjust comment.
7803         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
7804         * tree-cfg.c (found_computed_goto): Remove.
7805         (factor_computed_gotos): Remove.
7806         (make_goto_expr_edges): Return bool, true for computed gotos.
7807         Don't call make_abnormal_goto_edges.
7808         (build_gimple_cfg): Don't set found_computed_goto, don't call
7809         factor_computed_gotos.
7810         (computed_goto_p): No longer static.
7811         (make_blocks): Don't set found_computed_goto.
7812         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
7813         (make_edges): If make_goto_expr_edges returns true, push bb
7814         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
7815         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
7816         vector.  Record mapping between bbs and OpenMP regions if there
7817         are any, adjust make_gimple_omp_edges caller.  Call
7818         handle_abnormal_edges.
7819         (make_abnormal_goto_edges): Remove.
7820         * tree-cfg.h (make_abnormal_goto_edges): Remove.
7821         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
7822         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
7823         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
7824         * internal-fn.def (ABNORMAL_DISPATCHER): New.
7825         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
7826         filling *region also set *region_idx to (*region)->entry->index.
7828         PR other/58712
7829         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
7830         For REGs set ORIGINAL_REGNO.
7832 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
7834         * doc/md.texi: Mention that a target shouldn't implement
7835         vec_widen_(s|u)mul_even/odd pair if it is less efficient
7836         than hi/lo pair.
7838 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
7840         PR tree-optimization/59594
7841         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
7842         a copy of the datarefs vector rather than the vector itself.
7844 2014-01-28  Jason Merrill  <jason@redhat.com>
7846         PR c++/53756
7847         * dwarf2out.c (auto_die): New static.
7848         (gen_type_die_with_usage): Handle C++1y 'auto'.
7849         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
7850         on definition.
7852 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
7854         PR target/59672
7855         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
7856         (SPEC_X32): Likewise.
7857         (SPEC_64): Likewise.
7858         * config/i386/i386.c (ix86_option_override_internal): Turn off
7859         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
7860         for TARGET_16BIT.
7861         (x86_file_start): Output .code16gcc for TARGET_16BIT.
7862         * config/i386/i386.h (TARGET_16BIT): New macro.
7863         (TARGET_16BIT_P): Likewise.
7864         * config/i386/i386.opt: Add m16.
7865         * doc/invoke.texi: Document -m16.
7867 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
7869         PR preprocessor/59935
7870         * input.c (location_get_source_line): Bail out on when line number
7871         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
7873 2014-01-28  Richard Biener  <rguenther@suse.de>
7875         PR tree-optimization/58742
7876         * tree-ssa-forwprop.c (associate_plusminus): Handle
7877         pointer subtraction of the form (T)(P + A) - (T)P.
7879 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7881         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
7882         at const_int_cost.
7884 2014-01-28  Richard Biener  <rguenther@suse.de>
7886         Revert
7887         2014-01-28  Richard Biener  <rguenther@suse.de>
7889         PR rtl-optimization/45364
7890         PR rtl-optimization/59890
7891         * var-tracking.c (local_get_addr_clear_given_value): Handle
7892         already cleared slot.
7893         (val_reset): Handle not allocated local_get_addr_cache.
7894         (vt_find_locations): Use post-order on the inverted CFG.
7896 2014-01-28  Richard Biener  <rguenther@suse.de>
7898         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
7900 2014-01-28  Richard Biener  <rguenther@suse.de>
7902         PR rtl-optimization/45364
7903         PR rtl-optimization/59890
7904         * var-tracking.c (local_get_addr_clear_given_value): Handle
7905         already cleared slot.
7906         (val_reset): Handle not allocated local_get_addr_cache.
7907         (vt_find_locations): Use post-order on the inverted CFG.
7909 2014-01-28  Alan Modra  <amodra@gmail.com>
7911         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
7912         * configure.ac <recursive call for build != host>: Define
7913         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
7914         and LD_FOR_BUILD too.
7915         * configure: Regenerate.
7917 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
7919         * config/i386/i386.c (get_builtin_code_for_version): Separate
7920         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
7921         Broadwell from Haswell.
7923 2014-01-27  Steve Ellcey  <sellcey@mips.com>
7925         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
7926         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
7927         * config/mips/mips.c (mips_option_override): Change setting
7928         of TARGET_DSP.
7929         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
7930         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
7931         Change from Mask to Var.
7933 2014-01-27  Jeff Law  <law@redhat.com>
7935         * ipa-inline.c (inline_small_functions): Fix typo.
7937 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
7939         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
7940         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
7941         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
7942         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
7943         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
7944         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
7945         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
7946         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
7947         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
7948         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
7949         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
7950         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
7951         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
7952         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
7953         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
7954         (_mm512_storeu_epi64): Ditto.
7955         (_mm512_cmpge_epi32_mask): Ditto.
7956         (_mm512_cmpge_epu32_mask): Ditto.
7957         (_mm512_cmpge_epi64_mask): Ditto.
7958         (_mm512_cmpge_epu64_mask): Ditto.
7959         (_mm512_cmple_epi32_mask): Ditto.
7960         (_mm512_cmple_epu32_mask): Ditto.
7961         (_mm512_cmple_epi64_mask): Ditto.
7962         (_mm512_cmple_epu64_mask): Ditto.
7963         (_mm512_cmplt_epi32_mask): Ditto.
7964         (_mm512_cmplt_epu32_mask): Ditto.
7965         (_mm512_cmplt_epi64_mask): Ditto.
7966         (_mm512_cmplt_epu64_mask): Ditto.
7967         (_mm512_cmpneq_epi32_mask): Ditto.
7968         (_mm512_cmpneq_epu32_mask): Ditto.
7969         (_mm512_cmpneq_epi64_mask): Ditto.
7970         (_mm512_cmpneq_epu64_mask): Ditto.
7971         (_mm512_expand_pd): Ditto.
7972         (_mm512_expand_ps): Ditto.
7973         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
7974         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
7975         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
7976         * config/i386/i386.c (ix86_builtins): Add
7977         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
7978         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
7979         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
7980         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
7981         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
7982         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
7983         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
7984         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
7985         IX86_BUILTIN_PMOVUSQW512_MEM.
7986         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
7987         __builtin_ia32_pmovsqd512mem_mask,
7988         __builtin_ia32_pmovqd512mem_mask,
7989         __builtin_ia32_pmovusqw512mem_mask,
7990         __builtin_ia32_pmovsqw512mem_mask,
7991         __builtin_ia32_pmovqw512mem_mask,
7992         __builtin_ia32_pmovusdw512mem_mask,
7993         __builtin_ia32_pmovsdw512mem_mask,
7994         __builtin_ia32_pmovdw512mem_mask,
7995         __builtin_ia32_pmovqb512mem_mask,
7996         __builtin_ia32_pmovusqb512mem_mask,
7997         __builtin_ia32_pmovsqb512mem_mask,
7998         __builtin_ia32_pmovusdb512mem_mask,
7999         __builtin_ia32_pmovsdb512mem_mask,
8000         __builtin_ia32_pmovdb512mem_mask.
8001         (bdesc_args): Add __builtin_ia32_expanddf512,
8002         __builtin_ia32_expandsf512.
8003         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
8004         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
8005         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
8006         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
8007         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
8008         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
8009         (avx512f_<code>v8div16qi2_mask_store): This.
8010         (avx512f_expand<mode>): New.
8012 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
8014         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
8015         New.
8016         (_mm512_mask_prefetch_i64gather_pd): Ditto.
8017         (_mm512_prefetch_i32scatter_pd): Ditto.
8018         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
8019         (_mm512_prefetch_i64scatter_pd): Ditto.
8020         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
8021         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
8022         (_mm512_mask_prefetch_i64gather_ps): Ditto.
8023         (_mm512_prefetch_i32scatter_ps): Ditto.
8024         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
8025         (_mm512_prefetch_i64scatter_ps): Ditto.
8026         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
8027         * config/i386/i386-builtin-types.def: Define
8028         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
8029         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
8030         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
8031         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
8032         IX86_BUILTIN_SCATTERPFQPD.
8033         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
8034         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
8035         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
8036         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
8037         __builtin_ia32_scatterpfqps.
8038         (ix86_expand_builtin): Expand new built-ins.
8039         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
8040         fix memory access data type.
8041         (*avx512pf_gatherpf<mode>_mask): Ditto.
8042         (*avx512pf_gatherpf<mode>): Ditto.
8043         (avx512pf_scatterpf<mode>): Ditto.
8044         (*avx512pf_scatterpf<mode>_mask): Ditto.
8045         (*avx512pf_scatterpf<mode>): Ditto.
8046         (GATHER_SCATTER_SF_MEM_MODE): New.
8047         (avx512pf_gatherpf<mode>df): Ditto.
8048         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8049         (*avx512pf_scatterpf<mode>df): Ditto.
8051 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
8053         PR bootstrap/59934
8054         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
8055         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
8056         reached.
8058 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
8060         * common/config/arm/arm-common.c
8061         (arm_rewrite_mcpu): Handle multiple names.
8062         * config/arm/arm.h
8063         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8065 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
8067         * gimple-builder.h (create_gimple_tmp): Delete.
8069 2014-01-27  Christian Bruel  <christian.bruel@st.com>
8071         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
8072         words comparisons.
8074 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
8076         * config/pa/pa.md (call): Generate indirect long calls to non-local
8077         functions when outputing 32-bit code.
8078         (call_value): Likewise except for special call to buggy powf function.
8080         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
8081         portable runtime and PIC indirect calls.
8082         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
8083         and PIC call sequences.  Use ldo instead of blr to set return register
8084         in PIC call sequence.
8086 2014-01-25  Walter Lee  <walt@tilera.com>
8088         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
8089         avoid clobbering a live register.
8091 2014-01-25  Walter Lee  <walt@tilera.com>
8093         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
8094         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
8095         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
8096         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
8098 2014-01-25  Walter Lee  <walt@tilera.com>
8100         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
8101         arguments on even registers.
8102         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
8103         STACK_BOUNDARY.
8104         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
8105         (BIGGEST_ALIGNMENT): Ditto.
8106         (BIGGEST_FIELD_ALIGNMENT): Ditto.
8108 2014-01-25  Walter Lee  <walt@tilera.com>
8110         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
8111         insns before bundling.
8112         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
8114 2014-01-25  Walter Lee  <walt@tilera.com>
8116         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
8117         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
8118         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
8120 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
8122         * config/mips/constraints.md (kl): Delete.
8123         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
8124         define expands, using...
8125         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
8126         instructions for MIPS16.
8127         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
8128         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
8130 2014-01-25  Walter Lee  <walt@tilera.com>
8132         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
8133         (clzdi2): Ditto.
8134         (ffsdi2): Ditto.
8136 2014-01-25  Walter Lee  <walt@tilera.com>
8138         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
8139         (TARGET_EXPAND_TO_RTL_HOOK): Define.
8141 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
8143         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
8144         Handle XOR.
8146 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
8148         * print-rtl.c (in_call_function_usage): New var.
8149         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
8150         EXPR_LIST mode as mode and not as reg note name.
8152         PR middle-end/59561
8153         * cfgloopmanip.c (copy_loop_info): If
8154         loop->warned_aggressive_loop_optimizations, make sure
8155         the flag is set in target loop too.
8157 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8159         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
8160         flag_cilkplus.
8161         * builtins.def: Likewise.
8162         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
8163         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
8164         * ira.c (ira_setup_eliminable_regset): Likewise.
8165         * omp-low.c (gate_expand_omp): Likewise.
8166         (execute_lower_omp): Likewise.
8167         (diagnose_sb_0): Likewise.
8168         (gate_diagnose_omp_blocks): Likewise.
8169         (simd_clone_clauses_extract): Likewise.
8170         (gate): Likewise.
8172 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8174         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
8175         correction for little endian...
8176         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
8177         here.
8179 2014-01-24  Jeff Law  <law@redhat.com>
8181         PR tree-optimization/59919
8182         * tree-vrp.c (find_assert_locations_1): Do not register asserts
8183         for non-returning calls.
8185 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
8187         * common/config/aarch64/aarch64-common.c
8188         (aarch64_rewrite_mcpu): Handle multiple names.
8189         * config/aarch64/aarch64.h
8190         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8192 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
8194         * input.c (add_file_to_cache_tab): Handle the case where fopen
8195         returns NULL.
8197 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
8199         PR target/59929
8200         * config/i386/i386.md (pushsf splitter): Get stack adjustment
8201         from push operand if code of push isn't PRE_DEC.
8203 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
8205         PR target/59909
8206         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
8207         -mquad-memory-atomic.  Update -mquad-memory documentation to say
8208         it is only used for non-atomic loads/stores.
8210         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
8211         -mquad-memory or -mquad-memory-atomic switches.
8213         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
8214         -mquad-memory-atomic to ISA 2.07 support.
8216         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
8217         to separate support of normal quad word memory operations (ldq, stq)
8218         from the atomic quad word memory operations.
8220         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
8221         support to separate non-atomic quad word operations from atomic
8222         quad word operations.  Disable non-atomic quad word operations in
8223         little endian mode so that we don't have to swap words after the
8224         load and before the store.
8225         (quad_load_store_p): Add comment about atomic quad word support.
8226         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
8227         options printed with -mdebug=reg.
8229         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
8230         -mquad-memory-atomic as the test for whether we have quad word
8231         atomic instructions.
8232         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
8233         or -mp8-vector are used, allow byte/half-word atomic operations.
8235         * config/rs6000/sync.md (load_lockedti): Insure that the address
8236         is a proper indexed or indirect address for the lqarx instruction.
8237         On little endian systems, swap the hi/lo registers after the lqarx
8238         instruction.
8239         (load_lockedpti): Use indexed_or_indirect_operand predicate to
8240         insure the address is valid for the lqarx instruction.
8241         (store_conditionalti): Insure that the address is a proper indexed
8242         or indirect address for the stqcrx. instruction.  On little endian
8243         systems, swap the hi/lo registers before doing the stqcrx.
8244         instruction.
8245         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
8246         insure the address is valid for the stqcrx. instruction.
8248         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8249         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
8250         type of quad memory support is available.
8252 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
8254         PR regression/59915
8255         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
8256         there is a danger of looping.
8258 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
8260         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8261         force flag_ira_loop_pressure if set via command line.
8263 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8265         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
8266         (ashr_simd): New builtin handling DI mode.
8267         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
8268         (aarch64_sshr_simddi): New match pattern.
8269         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
8270         (vshrd_n_s64): Likewise.
8271         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
8273 2014-01-23  Nick Clifton  <nickc@redhat.com>
8275         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
8276         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
8277         favour of mcu specific scripts.
8278         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
8279         430x multilibs.
8281 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
8282             Alex Velenko  <Alex.Velenko@arm.com>
8284         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
8285         (vaddv_s16): Likewise.
8286         (vaddv_s32): Likewise.
8287         (vaddv_u8): Likewise.
8288         (vaddv_u16): Likewise.
8289         (vaddv_u32): Likewise.
8290         (vaddvq_s8): Likewise.
8291         (vaddvq_s16): Likewise.
8292         (vaddvq_s32): Likewise.
8293         (vaddvq_s64): Likewise.
8294         (vaddvq_u8): Likewise.
8295         (vaddvq_u16): Likewise.
8296         (vaddvq_u32): Likewise.
8297         (vaddvq_u64): Likewise.
8298         (vaddv_f32): Likewise.
8299         (vaddvq_f32): Likewise.
8300         (vaddvq_f64): Likewise.
8301         (vmaxv_f32): Likewise.
8302         (vmaxv_s8): Likewise.
8303         (vmaxv_s16): Likewise.
8304         (vmaxv_s32): Likewise.
8305         (vmaxv_u8): Likewise.
8306         (vmaxv_u16): Likewise.
8307         (vmaxv_u32): Likewise.
8308         (vmaxvq_f32): Likewise.
8309         (vmaxvq_f64): Likewise.
8310         (vmaxvq_s8): Likewise.
8311         (vmaxvq_s16): Likewise.
8312         (vmaxvq_s32): Likewise.
8313         (vmaxvq_u8): Likewise.
8314         (vmaxvq_u16): Likewise.
8315         (vmaxvq_u32): Likewise.
8316         (vmaxnmv_f32): Likewise.
8317         (vmaxnmvq_f32): Likewise.
8318         (vmaxnmvq_f64): Likewise.
8319         (vminv_f32): Likewise.
8320         (vminv_s8): Likewise.
8321         (vminv_s16): Likewise.
8322         (vminv_s32): Likewise.
8323         (vminv_u8): Likewise.
8324         (vminv_u16): Likewise.
8325         (vminv_u32): Likewise.
8326         (vminvq_f32): Likewise.
8327         (vminvq_f64): Likewise.
8328         (vminvq_s8): Likewise.
8329         (vminvq_s16): Likewise.
8330         (vminvq_s32): Likewise.
8331         (vminvq_u8): Likewise.
8332         (vminvq_u16): Likewise.
8333         (vminvq_u32): Likewise.
8334         (vminnmv_f32): Likewise.
8335         (vminnmvq_f32): Likewise.
8336         (vminnmvq_f64): Likewise.
8338 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
8340         * config/aarch64/aarch64-simd.md
8341         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
8342         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
8343         (*aarch64_mul3_elt<mode>): Likewise.
8344         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
8345         (*aarch64_mul3_elt_to_64v2df): Likewise.
8346         (*aarch64_mla_elt<mode>): Likewise.
8347         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
8348         (*aarch64_mls_elt<mode>): Likewise.
8349         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
8350         (*aarch64_fma4_elt<mode>): Likewise.
8351         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
8352         (*aarch64_fma4_elt_to_64v2df): Likewise.
8353         (*aarch64_fnma4_elt<mode>): Likewise.
8354         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8355         (*aarch64_fnma4_elt_to_64v2df): Likewise.
8356         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
8357         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
8358         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
8359         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
8360         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8361         (aarch64_sqdmull_lane<mode>_internal): Likewise.
8362         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8364 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8366         * config/aarch64/aarch64-simd.md
8367         (aarch64_be_checked_get_lane<mode>): New define_expand.
8368         * config/aarch64/aarch64-simd-builtins.def
8369         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
8370         New builtin definition.
8371         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
8372         Use new safe be builtin.
8374 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8376         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
8377         New define_insn.
8378         (aarch64_be_st1<mode>): Likewise.
8379         (aarch_ld1<VALL:mode>): Define_expand modified.
8380         (aarch_st1<VALL:mode>): Likewise.
8381         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
8382         (UNSPEC_ST1): Likewise.
8384 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
8386         * config/microblaze/microblaze.md: Add trap insn and attribute
8388 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
8390         PR preprocessor/58580
8391         * input.h (location_get_source_line): Take an additional line_size
8392         parameter.
8393         (void diagnostics_file_cache_fini): Declare new function.
8394         * input.c (struct fcache): New type.
8395         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
8396         New static constants.
8397         (diagnostic_file_cache_init, total_lines_num)
8398         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
8399         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
8400         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
8401         (get_next_line, read_next_line, goto_next_line, read_line_num):
8402         New static function definitions.
8403         (diagnostic_file_cache_fini): New function.
8404         (location_get_source_line): Take an additional output line_len
8405         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
8406         read_line_num.
8407         * diagnostic.c (diagnostic_finish): Call
8408         diagnostic_file_cache_fini.
8409         (adjust_line): Take an additional input parameter for the length
8410         of the line, rather than calculating it with strlen.
8411         (diagnostic_show_locus): Adjust the use of
8412         location_get_source_line and adjust_line with respect to their new
8413         signature.  While displaying a line now, do not stop at the first
8414         null byte.  Rather, display the zero byte as a space and keep
8415         going until we reach the size of the line.
8416         * Makefile.in: Add vec.o to OBJS-libcommon
8418 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8419             Ilya Tocar     <ilya.tocar@intel.com>
8421         * config/i386/avx512fintrin.h (_mm512_kmov): New.
8422         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
8423         (__builtin_ia32_kmov16): Ditto.
8424         * config/i386/i386.md (UNSPEC_KMOV): New.
8425         (kmovw): Ditto.
8427 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8429         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
8430         (_mm512_storeu_si512): Ditto.
8432 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
8434         PR target/52125
8435         * rtl.h (get_referenced_operands): Declare.
8436         * recog.c (get_referenced_operands): New function.
8437         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
8438         operands have been referenced when recording LO_SUM references.
8440 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
8442         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
8444 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
8446         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
8447         Enable for generic and recent AMD targets.
8449 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
8451         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
8452         ARG_SIZE note when adjustment was eliminated.
8454 2014-01-22  Jeff Law  <law@redhat.com>
8456         PR tree-optimization/59597
8457         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
8458         in file.  Accept new argument REGISTERING and use it to modify
8459         dump output appropriately.
8460         (register_jump_thread): Corresponding changes.
8461         (mark_threaded_blocks): Reinstate code to cancel unprofitable
8462         thread paths involving joiner blocks.  Add code to dump cancelled
8463         jump threading paths.
8465 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
8467         PR rtl-optimization/59477
8468         * lra-constraints.c (inherit_in_ebb): Process call for living hard
8469         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
8471 2014-01-22  Tom Tromey  <tromey@redhat.com>
8473         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
8474         PARAMS.
8475         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
8477 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
8479         PR rtl-optimization/59896
8480         * lra-constraints.c (process_alt_operands): Check unused note for
8481         matched operands of insn with no output reloads.
8483 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
8485         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
8486         (mips_move_from_gpr_cost): Likewise.
8488 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
8490         PR rtl-optimization/59858
8491         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
8492         ira_class_hard_regs_num.
8493         (process_alt_operands): Increase reject for dying matched operand.
8495 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
8497         PR target/59003
8498         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
8499         smaller than size, perform several stores or loads and stores
8500         at dst + count - size to store or copy all of size bytes, rather
8501         than just last modesize bytes.
8503 2014-01-20  DJ Delorie  <dj@redhat.com>
8505         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
8506         that CLOBBERs are REGs before propogating their values.
8508 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
8510         PR middle-end/59789
8511         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
8512         (cgraph_inline_failed_type): New function.
8513         * cgraph.h (DEFCIFCODE): Add type.
8514         (cgraph_inline_failed_type_t): New enum.
8515         (cgraph_inline_failed_type): New prototype.
8516         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
8517         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
8518         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
8519         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
8520         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
8521         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
8522         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
8523         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
8524         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
8525         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
8526         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
8527         OPTIMIZATION_MISMATCH.
8528         * tree-inline.c (expand_call_inline): Emit errors during
8529         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
8531 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
8533         PR target/59685
8534         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
8535         mode attribute in insn output.
8537 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
8539         * output.h (output_constant): Delete.
8540         * varasm.c (output_constant): Make private.
8542 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
8544         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
8546 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
8548         PR middle-end/59860
8549         * tree.h (fold_builtin_strcat): New prototype.
8550         * builtins.c (fold_builtin_strcat): No longer static.  Add len
8551         argument, if non-NULL, don't call c_strlen.  Optimize
8552         directly into __builtin_memcpy instead of __builtin_strcpy.
8553         (fold_builtin_2): Adjust fold_builtin_strcat caller.
8554         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
8556 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
8558         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
8559         for SImode_address_operand operands, having only a REG argument.
8561 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8563         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
8564         loader name using mbig-endian.
8565         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
8567 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
8569         * doc/invoke.texi (-march): Clarify documentation for AArch64.
8570         (-mtune): Likewise.
8571         (-mcpu): Likewise.
8573 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
8575         * config/aarch64/aarch64-protos.h
8576         (aarch64_cannot_change_mode_class_ptr): Declare.
8577         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
8578         aarch64_cannot_change_mode_class_ptr): New.
8579         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
8580         backend hook aarch64_cannot_change_mode_class.
8582 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
8584         * common/config/aarch64/aarch64-common.c
8585         (aarch64_handle_option): Don't handle any option order logic here.
8586         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
8587         selected_cpu, warn on architecture version mismatch.
8588         (aarch64_override_options): Fix parsing order for option strings.
8590 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8591             Iain Sandoe  <iain@codesourcery.com>
8593         PR bootstrap/59496
8594         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
8595         warning.  Amend comment to reflect current functionality.
8597 2014-01-20  Richard Biener  <rguenther@suse.de>
8599         PR middle-end/59860
8600         * builtins.c (fold_builtin_strcat): Remove case better handled
8601         by tree-ssa-strlen.c.
8603 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
8605         * config/aarch64/aarch64.opt
8606         (mcpu, march, mtune): Make case-insensitive.
8608 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
8610         PR target/59880
8611         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
8612         if operands[1] is a REG or ZERO_EXTEND of a REG.
8614 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
8616         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
8618 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
8620         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
8621         long non-pic millicode calls.
8623 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8625         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
8627 2014-01-19  Kito Cheng  <kito@0xlab.org>
8629         * builtins.c (expand_movstr): Check movstr expand done or fail.
8631 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
8632             H.J. Lu  <hongjiu.lu@intel.com>
8634         PR target/59379
8635         * config/i386/i386.md (*lea<mode>): Zero-extend return register
8636         to DImode for zero-extended addresses.
8638 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
8640         PR rtl-optimization/57763
8641         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
8642         on the new indirect jump_insn and increment LABEL_NUSES (label).
8644 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
8646         PR bootstrap/59580
8647         PR bootstrap/59583
8648         * config.gcc (x86_archs): New variable.
8649         (x86_64_archs): Likewise.
8650         (x86_cpus): Likewise.
8651         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
8652         --with-arch/--with-cpu= options.
8653         Support --with-arch=/--with-cpu={nehalem,westmere,
8654         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
8656 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
8658         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
8659         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
8661 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
8663         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
8665 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
8667         PR target/58944
8668         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
8669         clear cpp_get_options (parse_in)->warn_unused_macros for
8670         ix86_target_macros_internal with cpp_define.
8672 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
8674         * jump.c (delete_related_insns): Keep (use (insn))s.
8675         * reorg.c (redundant_insn): Check for barriers too.
8677 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
8679         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
8681 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
8683         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
8684         call to $$dyncall when TARGET_LONG_CALLS is true.
8686 2014-01-17  Jeff Law  <law@redhat.com>
8688         * ree.c (combine_set_extension): Temporarily disable test for
8689         changing number of hard registers.
8691 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
8693         PR middle-end/58125
8694         * ipa-inline-analysis.c (inline_free_summary):
8695         Do not free summary of aliases.
8697 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
8699         PR middle-end/59706
8700         * gimplify.c (gimplify_expr): Use create_tmp_var
8701         instead of create_tmp_var_raw.  If cond doesn't have
8702         integral type, don't add the IFN_ANNOTATE builtin at all.
8704 2014-01-17  Martin Jambor  <mjambor@suse.cz>
8706         PR ipa/59736
8707         * ipa-cp.c (prev_edge_clone): New variable.
8708         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
8709         Also resize prev_edge_clone vector.
8710         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
8711         (ipcp_edge_removal_hook): New function.
8712         (ipcp_driver): Register ipcp_edge_removal_hook.
8714 2014-01-17  Andrew Pinski  <apinski@cavium.com>
8715             Steve Ellcey  <sellcey@mips.com>
8717         PR target/59462
8718         * config/mips/mips.c (mips_print_operand): Check operand mode instead
8719         of operator mode.
8721 2014-01-17  Jeff Law  <law@redhat.com>
8723         PR middle-end/57904
8724         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
8725         so that pass_ccp runs first.
8727 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
8729         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
8730         (ix86_adjust_cost): Use !TARGET_XXX.
8731         (do_reorder_for_imul): Likewise.
8732         (swap_top_of_ready_list): Likewise.
8733         (ix86_sched_reorder): Likewise.
8735 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
8737         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8738         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
8739         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
8740         (intel_memset): New.  Duplicate slm_memset.
8741         (intel_cost): New.  Duplicate slm_cost.
8742         (m_INTEL): New macro.
8743         (processor_target_table): Add "intel".
8744         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
8745         with PROCESSOR_INTEL for "intel".
8746         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
8747         PROCESSOR_SILVERMONT.
8748         (ix86_issue_rate): Likewise.
8749         (ix86_adjust_cost): Likewise.
8750         (ia32_multipass_dfa_lookahead): Likewise.
8751         (swap_top_of_ready_list): Likewise.
8752         (ix86_sched_reorder): Likewise.
8753         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
8754         instead of TARGET_OPT_AGU.
8755         * config/i386/i386.h (TARGET_INTEL): New.
8756         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
8757         (processor_type): Add PROCESSOR_INTEL.
8758         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
8759         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
8761 2014-01-17  Marek Polacek  <polacek@redhat.com>
8763         PR c/58346
8764         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
8765         size is zero.
8767 2014-01-17  Richard Biener  <rguenther@suse.de>
8769         PR tree-optimization/46590
8770         * opts.c (default_options_table): Add entries for
8771         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
8772         all enabled at -O1 but not for -Og.
8773         * common.opt (fbranch-count-reg): Remove Init(1).
8774         (fmove-loop-invariants): Likewise.
8775         (ftree-pta): Likewise.
8777 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
8779         * config/i386/i386.c (ix86_data_alignment): For compatibility with
8780         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
8781         decls to at least the GCC 4.8 used alignments.
8783         PR fortran/59440
8784         * tree-nested.c (convert_nonlocal_reference_stmt,
8785         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
8786         of GIMPLE_BIND stmts, adjust associated decls.
8788 2014-01-17  Richard Biener  <rguenther@suse.de>
8790         PR tree-optimization/46590
8791         * vec.h (vec<>::bseach): New member function implementing
8792         binary search according to C89 bsearch.
8793         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
8794         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
8795         bitmap pointer again.  Make accesses_in_loop a flat array.
8796         (mem_ref_obstack): New global.
8797         (outermost_indep_loop): Adjust for mem_ref->stored changes.
8798         (mark_ref_stored): Likewise.
8799         (ref_indep_loop_p_2): Likewise.
8800         (set_ref_stored_in_loop): New helper function.
8801         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
8802         (memref_free): Adjust.
8803         (record_mem_ref_loc): Simplify.
8804         (gather_mem_refs_stmt): Adjust.
8805         (sort_locs_in_loop_postorder_cmp): New function.
8806         (analyze_memory_references): Sort accesses_in_loop after
8807         loop postorder number.
8808         (find_ref_loc_in_loop_cmp): New function.
8809         (for_all_locs_in_loop): Find relevant cluster of locs in
8810         accesses_in_loop and iterate without recursion.
8811         (execute_sm): Avoid uninit warning.
8812         (struct ref_always_accessed): Simplify.
8813         (ref_always_accessed::operator ()): Likewise.
8814         (ref_always_accessed_p): Likewise.
8815         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
8816         loop postorder numbers here.
8817         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
8818         numbers.
8820 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
8822         PR c++/57945
8823         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
8824         on decls for which assemble_alias has been called.
8826 2014-01-17  Nick Clifton  <nickc@redhat.com>
8828         * config/msp430/msp430.opt: (mcpu): New option.
8829         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
8830         (msp430_option_override): Parse target_cpu.  If the MCU name
8831         matches a generic string, clear target_mcu.
8832         (msp430_attr): Allow numeric interrupt values up to 63.
8833         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
8834         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
8835         option.
8836         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
8837         Add mcpu matches.
8838         * config/msp430/msp430.md (popm): Use %J rather than %I.
8839         (addsi3): Use msp430_nonimmediate_operand for operand 2.
8840         (addhi_cy_i): Use immediate_operand for operand 2.
8841         * doc/invoke.texi: Document -mcpu option.
8843 2014-01-17  Richard Biener  <rguenther@suse.de>
8845         PR rtl-optimization/38518
8846         * df.h (df_analyze_loop): Declare.
8847         * df-core.c: Include cfgloop.h.
8848         (df_analyze_1): Split out main part of df_analyze.
8849         (df_analyze): Adjust.
8850         (loop_inverted_post_order_compute): New function.
8851         (loop_post_order_compute): Likewise.
8852         (df_analyze_loop): New function avoiding whole-function
8853         postorder computes.
8854         * loop-invariant.c (find_defs): Use df_analyze_loop.
8855         (find_invariants): Adjust.
8856         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
8858 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
8860         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
8861         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
8863 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
8865         * ipa-ref.c (ipa_remove_stmt_references): Fix references
8866         traversal when removing references.
8868 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
8870         PR ipa/59775
8871         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
8873 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
8875         PR middle-end/56791
8876         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
8877         pushing a reload for an autoinc when we had previously reloaded an
8878         inner part of the address.
8880 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
8882         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
8883         field.
8884         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
8885         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
8886         when not giving up or versioning for alias only because of
8887         loop->safelen.
8888         (vect_analyze_data_ref_dependences): Set to true.
8889         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
8890         is a GIMPLE_PHI.
8891         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
8892         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
8893         to the condition.
8895         PR middle-end/58344
8896         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
8898         PR target/59839
8899         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
8900         operand 0 predicate for gathers, use a new pseudo as subtarget.
8902 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
8904         PR middle-end/59609
8905         * lra-constraints.c (process_alt_operands): Add printing debug info.
8906         Check absence of input/output reloads for matched operands too.
8908 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
8910         PR rtl-optimization/59835
8911         * ira.c (ira_init_register_move_cost): Increase cost for
8912         impossible modes.
8914 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
8916         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
8918 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
8920         PR target/59780
8921         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
8922         non-register objects.  Use gen_(high/low)part more consistently.
8923         Fix assertions.
8925 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
8927         PR target/59844
8928         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
8929         endian support, remove tests for WORDS_BIG_ENDIAN.
8930         (p8_mfvsrd_3_<mode>): Likewise.
8931         (reload_gpr_from_vsx<mode>): Likewise.
8932         (reload_gpr_from_vsxsf): Likewise.
8933         (p8_mfvsrd_4_disf): Likewise.
8935 2014-01-16  Richard Biener  <rguenther@suse.de>
8937         PR rtl-optimization/46590
8938         * lcm.c (compute_antinout_edge): Use postorder iteration.
8939         (compute_laterin): Use inverted postorder iteration.
8941 2014-01-16  Nick Clifton  <nickc@redhat.com>
8943         PR middle-end/28865
8944         * varasm.c (output_constant): Return the number of bytes actually
8945         emitted.
8946         (output_constructor_array_range): Update the field size with the
8947         number of bytes emitted by output_constant.
8948         (output_constructor_regular_field): Likewise.  Also do not
8949         complain if the total number of bytes emitted is now greater
8950         than the expected fieldpos.
8951         * output.h (output_constant): Update prototype and descriptive comment.
8953 2014-01-16  Marek Polacek  <polacek@redhat.com>
8955         PR middle-end/59827
8956         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
8957         it is error_mark_node.
8959 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
8961         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
8962         VALID_AVX256_REG_OR_OI_MODE.
8964 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
8966         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
8967         current procedure should be profiled.
8969 2014-01-15  Andrew Pinski  <apinski@cavium.com>
8971         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
8972         of moving from/to the STACK_REG register class.
8974 2014-01-15  Richard Henderson  <rth@redhat.com>
8976         PR debug/54694
8977         * reginfo.c (global_regs_decl): Globalize.
8978         * rtl.h (global_regs_decl): Declare.
8979         * ira.c (do_reload): Diagnose frame_pointer_needed and it
8980         reserved via global_regs.
8982 2014-01-15  Teresa Johnson  <tejohnson@google.com>
8984         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
8986 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
8988         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
8989         and vmulosh rather than call gen_vec_widen_smult_*.
8990         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
8991         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
8992         (vec_widen_smult_even_v16qi): Likewise.
8993         (vec_widen_umult_even_v8hi): Likewise.
8994         (vec_widen_smult_even_v8hi): Likewise.
8995         (vec_widen_umult_odd_v16qi): Likewise.
8996         (vec_widen_smult_odd_v16qi): Likewise.
8997         (vec_widen_umult_odd_v8hi): Likewise.
8998         (vec_widen_smult_odd_v8hi): Likewise.
8999         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
9000         vmuloub rather than call gen_vec_widen_umult_*.
9001         (vec_widen_umult_lo_v16qi): Likewise.
9002         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
9003         vmulosb rather than call gen_vec_widen_smult_*.
9004         (vec_widen_smult_lo_v16qi): Likewise.
9005         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
9006         rather than call gen_vec_widen_umult_*.
9007         (vec_widen_umult_lo_v8hi): Likewise.
9008         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
9009         rather than call gen_vec_widen_smult_*.
9010         (vec_widen_smult_lo_v8hi): Likewise.
9012 2014-01-15  Jeff Law  <law@redhat.com>
9014         PR tree-optimization/59747
9015         * ree.c (find_and_remove_re): Properly handle case where a second
9016         eliminated extension requires widening a copy created for elimination
9017         of a prior extension.
9018         (combine_set_extension): Ensure that the number of hard regs needed
9019         for a destination register does not change when we widen it.
9021 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9023         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
9024         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
9025         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
9026         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
9027         (avr-*-rtems*): Likewise.
9028         (bfin*-rtems*): Likewise.
9029         (moxie-*-rtems*): Likewise.
9030         (h8300-*-rtems*): Likewise.
9031         (i[34567]86-*-rtems*): Likewise.
9032         (lm32-*-rtems*): Likewise.
9033         (m32r-*-rtems*): Likewise.
9034         (m68k-*-rtems*): Likewise.
9035         (microblaze*-*-rtems*): Likewise.
9036         (mips*-*-rtems*): Likewise.
9037         (powerpc-*-rtems*): Likewise.
9038         (sh-*-rtems*): Likewise.
9039         (sparc-*-rtems*): Likewise.
9040         (sparc64-*-rtems*): Likewise.
9041         (v850-*-rtems*): Likewise.
9042         (m32c-*-rtems*): Likewise.
9044 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
9046         PR rtl-optimization/59511
9047         * ira.c (ira_init_register_move_cost): Use memory costs for some
9048         cases of register move cost calculations.
9049         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
9050         instead of BB frequency.
9051         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
9052         * lra-assigns.c (find_hard_regno_for): Ditto.
9054 2014-01-15  Richard Biener  <rguenther@suse.de>
9056         PR tree-optimization/59822
9057         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
9058         (vectorizable_load): Use it to hoist defs of uses of invariant
9059         loads out of the loop.
9061 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
9062             Kugan Vivekanandarajah  <kuganv@linaro.org>
9064         PR target/59695
9065         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
9066         truncation.
9068 2014-01-15  Richard Biener  <rguenther@suse.de>
9070         PR rtl-optimization/59802
9071         * lcm.c (compute_available): Use inverted postorder to seed
9072         the initial worklist.
9074 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9076         PR target/59803
9077         * config/s390/s390.c (s390_preferred_reload_class): Don't return
9078         ADDR_REGS for invalid symrefs in non-PIC code.
9080 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
9082         PR other/58712
9083         * builtins.c (determine_block_size): Initialize *probable_max_size
9084         even if len_rtx is CONST_INT.
9086 2014-01-14  Andrew Pinski  <apinski@cavium.com>
9088         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
9089         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
9090         (cortexa53_tunings): Likewise.
9091         (aarch64_sched_issue_rate): New function.
9092         (TARGET_SCHED_ISSUE_RATE): Define.
9094 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
9096         * ira-costs.c (find_costs_and_classes): Add missed
9097         ira_init_register_move_cost_if_necessary.
9099 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
9101         PR target/59787
9102         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
9104 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
9106         PR target/59794
9107         * config/i386/i386.c (type_natural_mode): Add a bool parameter
9108         to indicate if type is used for function return value.  Warn ABI
9109         change if the vector mode isn't available for function return value.
9110         (ix86_function_arg_advance): Pass false to type_natural_mode.
9111         (ix86_function_arg): Likewise.
9112         (ix86_gimplify_va_arg): Likewise.
9113         (function_arg_32): Don't warn ABI change.
9114         (ix86_function_value): Pass true to type_natural_mode.
9115         (ix86_return_in_memory): Likewise.
9116         (ix86_struct_value_rtx): Removed.
9117         (TARGET_STRUCT_VALUE_RTX): Likewise.
9119 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9121         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
9122         converting a conditional jump into a conditional return.
9124 2014-01-14  Richard Biener  <rguenther@suse.de>
9126         PR tree-optimization/58921
9127         PR tree-optimization/59006
9128         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
9129         hoisting invariant stmts.
9130         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
9131         invariant loads on the preheader edge if possible.
9133 2014-01-14  Joey Ye  <joey.ye@arm.com>
9135         * doc/plugin.texi (Building GCC plugins): Update to C++.
9137 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
9139         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
9140         (_mm_rcp28_round_ss): Ditto.
9141         (_mm_rsqrt28_round_sd): Ditto.
9142         (_mm_rsqrt28_round_ss): Ditto.
9143         (_mm_rcp28_sd): Ditto.
9144         (_mm_rcp28_ss): Ditto.
9145         (_mm_rsqrt28_sd): Ditto.
9146         (_mm_rsqrt28_ss): Ditto.
9147         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
9148         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
9149         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
9150         (IX86_BUILTIN_RCP28SD): Ditto.
9151         (IX86_BUILTIN_RCP28SS): Ditto.
9152         (IX86_BUILTIN_RSQRT28SD): Ditto.
9153         (IX86_BUILTIN_RSQRT28SS): Ditto.
9154         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
9155         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
9156         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
9157         (ix86_expand_special_args_builtin): Expand new FTYPE.
9158         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
9159         (srcp14<mode>): Make insn unary.
9160         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
9161         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
9162         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
9163         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
9164         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
9165         Fix rounding: make it SAE only.
9166         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
9167         Ditto.
9168         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
9169         Ditto.
9170         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
9171         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
9172         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
9173         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
9174         (round_saeonly_mask_scalar_operand4): Ditto.
9175         (round_saeonly_mask_scalar_op3): Ditto.
9176         (round_saeonly_mask_scalar_op4): Ditto.
9178 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9180         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9181         Implement -maltivec=be for vec_insert and vec_extract.
9183 2014-01-10  DJ Delorie  <dj@redhat.com>
9185         * config/msp430/msp430.md (call_internal): Don't allow memory
9186         references with SP as the base register.
9187         (call_value_internal): Likewise.
9188         * config/msp430/constraints.md (Yc): New.  For memory references
9189         that don't use SP as a base register.
9191         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
9192         "an integer without a # prefix"
9193         * config/msp430/msp430.md (epilogue_helper): Use it.
9195 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
9197         PR target/59617
9198         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
9199         AVX512F gather builtins.
9200         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
9201         on gather decls with INTEGER_TYPE masktype.
9202         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
9203         directly into the builtin rather than hoisting it before loop.
9205         PR tree-optimization/59387
9206         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
9207         (scev_const_prop): If folded_casts and type has undefined overflow,
9208         use force_gimple_operand instead of force_gimple_operand_gsi and
9209         for each added stmt if it is assign with
9210         arith_code_with_undefined_signed_overflow, call
9211         rewrite_to_defined_overflow.
9212         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
9213         gimple-fold.h instead.
9214         (arith_code_with_undefined_signed_overflow,
9215         rewrite_to_defined_overflow): Moved to ...
9216         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
9217         rewrite_to_defined_overflow): ... here.  No longer static.
9218         Include gimplify-me.h.
9219         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
9220         rewrite_to_defined_overflow): New prototypes.
9222 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9224         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
9226 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
9228         * builtins.c (get_object_alignment_2): Minor tweak.
9229         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
9231 2014-01-13  Christian Bruel  <christian.bruel@st.com>
9233         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
9234         optimized non constant lengths.
9236 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
9238         PR libgomp/59194
9239         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
9240         load as __atomic_load_N if possible.
9242 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
9244         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
9245         target parameter.
9246         (rs6000_expand_builtin): Adjust call.
9248 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
9250         PR target/58115
9251         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
9252         * config/rs6000/rs6000.c: Include target-globals.h.
9253         (rs6000_set_current_function): Instead of doing target_reinit
9254         unconditionally, use save_target_globals_default_opts and
9255         restore_target_globals.
9257         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
9258         FPSCR.
9259         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
9260         (rs6000_expand_builtin): Handle mffs and mtfsf.
9261         (rs6000_init_builtins): Define mffs and mtfsf.
9262         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
9263         (rs6000_mffs): New pattern.
9264         (rs6000_mtfsf): New pattern.
9266 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
9268         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
9269         Start narrowing with START.  Apply candidate-use pair
9270         and check overall cost in narrowing.
9271         (iv_ca_prune): Pass new argument.
9273 2014-01-10  Jeff Law  <law@redhat.com>
9275         PR middle-end/59743
9276         * ree.c (combine_reaching_defs): Ensure the defining statement
9277         occurs before the extension when optimizing extensions with
9278         different source and destination hard registers.
9280 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
9282         PR ipa/58585
9283         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
9284         vtables into the type inheritance graph.
9286 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9288         PR rtl-optimization/59754
9289         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
9290         modes in the REGNO != REGNO case.
9292 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9294         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
9296 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9298         PR tree-optimization/59745
9299         * tree-predcom.c (tree_predictive_commoning_loop): Call
9300         free_affine_expand_cache if giving up because components is NULL.
9302         * target-globals.c (save_target_globals): Allocate < 4KB structs using
9303         GC in payload of target_globals struct instead of allocating them on
9304         the heap and the larger structs separately using GC.
9305         * target-globals.h (struct target_globals): Make regs, hard_regs,
9306         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
9307         of GTY((skip)) and change type to void *.
9308         (reset_target_globals): Cast loads from those fields to corresponding
9309         types.
9311 2014-01-10  Steve Ellcey  <sellcey@mips.com>
9313         PR plugins/59335
9314         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
9315         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
9316         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
9318 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
9320         PR target/59744
9321         * aarch64-modes.def (CC_Zmode): New flags mode.
9322         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
9323         represents an equality.
9324         (aarch64_get_condition_code): Handle CC_Zmode.
9325         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
9327 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9329         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
9330         extraction in good case.
9332 2014-01-10  Richard Biener  <rguenther@suse.de>
9334         PR tree-optimization/59374
9335         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
9336         checking after SLP discovery.  Mark stmts not participating
9337         in any SLP instance properly.
9339 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9341         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
9342         when handling a SET rtx.
9344 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9346         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
9347         (cortex-a57): Likewise.
9348         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
9350 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9352         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
9353         non-iwmmxt builtins.
9355 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
9357         PR ipa/58252
9358         PR ipa/59226
9359         * ipa-devirt.c record_target_from_binfo): Take as argument
9360         stack of binfos and lookup matching one for virtual inheritance.
9361         (possible_polymorphic_call_targets_1): Update.
9363 2014-01-10  Huacai Chen  <chenhc@lemote.com>
9365         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
9366         kernel strings for Loongson-2E/2F/3A.
9368 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9370         PR middle-end/59670
9371         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
9372         is_gimple_call before calling gimple_call_internal_p.
9374 2014-01-09  Steve Ellcey  <sellcey@mips.com>
9376         * Makefile.in (TREE_FLOW_H): Remove.
9377         (TREE_SSA_H): Add file names from tree-flow.h.
9378         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
9379         * tree.h: Remove tree-flow.h reference.
9380         * hash-table.h: Remove tree-flow.h reference.
9381         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
9382         reference with tree-ssa-loop.h.
9384 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9386         * doc/invoke.texi: Add -maltivec={be,le} options, and document
9387         default element-order behavior for -maltivec.
9388         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
9389         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
9390         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
9391         when targeting big endian, at least for now.
9392         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
9394 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
9396         PR middle-end/47735
9397         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
9398         var satisfies use_register_for_decl, just take into account type
9399         alignment, rather than decl alignment.
9401         PR tree-optimization/59622
9402         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
9403         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
9404         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
9405         Don't devirtualize for inplace at all.  For targets.length () == 1,
9406         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
9408 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
9410         * config/i386/i386.md (cpu): Remove the unused btver1.
9412 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
9414         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
9416 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
9418         PR target/58115
9419         * tree-core.h (struct target_globals): New forward declaration.
9420         (struct tree_target_option): Add globals field.
9421         * tree.h (TREE_TARGET_GLOBALS): Define.
9422         (prepare_target_option_nodes_for_pch): New prototype.
9423         * target-globals.h (struct target_globals): Define even if
9424         !SWITCHABLE_TARGET.
9425         * tree.c (prepare_target_option_node_for_pch,
9426         prepare_target_option_nodes_for_pch): New functions.
9427         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
9428         * config/i386/i386.c: Include target-globals.h.
9429         (ix86_set_current_function): Instead of doing target_reinit
9430         unconditionally, use save_target_globals_default_opts and
9431         restore_target_globals.
9433 2014-01-09  Richard Biener  <rguenther@suse.de>
9435         PR tree-optimization/59715
9436         * tree-cfg.h (split_critical_edges): Declare.
9437         * tree-cfg.c (split_critical_edges): Export.
9438         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
9440 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
9442         * cfgexpand.c (expand_stack_vars): Optionally disable
9443         asan stack protection.
9444         (expand_used_vars): Likewise.
9445         (partition_stack_vars): Likewise.
9446         * asan.c (asan_emit_stack_protection): Optionally disable
9447         after return stack usage.
9448         (instrument_derefs): Optionally disable memory access instrumentation.
9449         (instrument_builtin_call): Likewise.
9450         (instrument_strlen_call): Likewise.
9451         (asan_protect_global): Optionally disable global variables protection.
9452         * doc/invoke.texi: Added doc for new options.
9453         * params.def: Added new options.
9454         * params.h: Likewise.
9456 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
9458         PR rtl-optimization/59724
9459         * ifcvt.c (cond_exec_process_if_block): Don't call
9460         flow_find_head_matching_sequence with 0 longest_match.
9461         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
9462         non-active insns if !stop_after.
9463         (try_head_merge_bb): Revert 2014-01-07 changes.
9465 2014-01-08  Jeff Law  <law@redhat.com>
9467         * ree.c (get_sub_rtx): New function, extracted from...
9468         (merge_def_and_ext): Here.
9469         (combine_reaching_defs): Use get_sub_rtx.
9471 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
9473         * cgraph.h (varpool_variable_node): Do not choke on null node.
9475 2014-01-08  Catherine Moore  <clm@codesourcery.com>
9477         * config/mips/mips.md (simple_return): Attempt to use JRC
9478         for microMIPS.
9479         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
9481 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
9483         PR rtl-optimization/59137
9484         * reorg.c (steal_delay_list_from_target): Call update_block for
9485         elided insns.
9486         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
9488 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9490         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
9491         two duplicate entries.
9493 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
9495         Revert:
9496         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
9498         * config/mips/mips.c (mips_truncated_op_cost): New function.
9499         (mips_rtx_costs): Adjust test for BADDU.
9500         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
9502         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
9504         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
9505         (*baddu_si): ...this new pattern.
9507 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
9509         PR ipa/59722
9510         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
9512 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9514         PR middle-end/57748
9515         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
9516         inner_reference_p.
9517         (expand_expr, expand_normal): Adjust.
9518         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
9519         inner_reference_p. Use inner_reference_p to expand inner references.
9520         (store_expr): Adjust.
9521         * cfgexpand.c (expand_call_stmt): Adjust.
9523 2014-01-08  Rong Xu  <xur@google.com>
9525         * gcov-io.c (gcov_var): Move from gcov-io.h.
9526         (gcov_position): Ditto.
9527         (gcov_is_error): Ditto.
9528         (gcov_rewrite): Ditto.
9529         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
9530         only part to libgcc/libgcov.h.
9532 2014-01-08  Marek Polacek  <polacek@redhat.com>
9534         PR middle-end/59669
9535         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
9537 2014-01-08  Marek Polacek  <polacek@redhat.com>
9539         PR sanitizer/59667
9540         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
9542 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
9544         PR rtl-optimization/59649
9545         * stor-layout.c (get_mode_bounds): For BImode return
9546         0 and STORE_FLAG_VALUE.
9548 2014-01-08  Richard Biener  <rguenther@suse.de>
9550         PR middle-end/59630
9551         * gimple.h (is_gimple_builtin_call): Remove.
9552         (gimple_builtin_call_types_compatible_p): New.
9553         (gimple_call_builtin_p): New overload.
9554         * gimple.c (is_gimple_builtin_call): Remove.
9555         (validate_call): Rename to ...
9556         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
9557         check return types.
9558         (validate_type): New static function.
9559         (gimple_call_builtin_p): New overload and adjust.
9560         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
9561         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
9562         (gimple_fold_stmt_to_constant_1): Likewise.
9563         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
9565 2014-01-08  Richard Biener  <rguenther@suse.de>
9567         PR middle-end/59471
9568         * gimplify.c (gimplify_expr): Gimplify register-register type
9569         VIEW_CONVERT_EXPRs to separate stmts.
9571 2014-01-07  Jeff Law  <law@redhat.com>
9573         PR middle-end/53623
9574         * ree.c (combine_set_extension): Handle case where source
9575         and destination registers in an extension insn are different.
9576         (combine_reaching_defs): Allow source and destination registers
9577         in extension to be different under limited circumstances.
9578         (add_removable_extension): Remove restriction that the
9579         source and destination registers in the extension are the same.
9580         (find_and_remove_re): Emit a copy from the extension's
9581         destination to its source after the defining insn if
9582         the source and destination registers are different.
9584         PR middle-end/59285
9585         * ifcvt.c (merge_if_block): If we are merging a block with more than
9586         one successor with a block with no successors, remove any BARRIER
9587         after the second block.
9589 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
9591         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
9593 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
9595         PR target/59652
9596         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
9597         for 14-bit register offsets when INT14_OK_STRICT is false.
9599 2014-01-07  Roland Stigge  <stigge@antcom.de>
9600             Michael Meissner  <meissner@linux.vnet.ibm.com>
9602         PR 57386/target
9603         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
9604         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
9606 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
9608         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
9609         -mcpu.
9611 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
9613         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
9614         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
9615         rtx is const0_rtx or not.
9617 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
9619         PR target/58115
9620         * target-globals.c (save_target_globals): Remove this_fn_optab
9621         handling.
9622         * toplev.c: Include optabs.h.
9623         (target_reinit): Temporarily restore the global options if another
9624         set of options are in force.
9626 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
9628         PR rtl-optimization/58668
9629         * cfgcleanup.c (flow_find_cross_jump): Don't count
9630         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
9631         to determine what is counted.
9632         (flow_find_head_matching_sequence): Use active_insn_p to determine
9633         what is counted.
9634         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
9635         counting change.
9636         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
9637         determine what is counted.
9639         PR tree-optimization/59643
9640         * tree-predcom.c (split_data_refs_to_components): If one dr is
9641         read and one write, determine_offset fails and the write isn't
9642         in the bad component, just put the read into the bad component.
9644 2014-01-07  Mike Stump  <mikestump@comcast.net>
9645             Jakub Jelinek  <jakub@redhat.com>
9647         PR pch/59436
9648         * tree-core.h (struct tree_optimization_option): Change optabs
9649         type from unsigned char * to void *.
9650         * optabs.c (init_tree_optimization_optabs): Adjust
9651         TREE_OPTIMIZATION_OPTABS initialization.
9653 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
9655         PR target/59644
9656         * config/i386/i386.h (struct machine_function): Add
9657         no_drap_save_restore field.
9658         * config/i386/i386.c (ix86_save_reg): Use
9659         !cfun->machine->no_drap_save_restore instead of
9660         crtl->stack_realign_needed.
9661         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
9662         this function clears frame_pointer_needed.  Set
9663         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
9664         and DRAP reg is needed.
9666 2014-01-06  Marek Polacek  <polacek@redhat.com>
9668         PR c/57773
9669         * doc/implement-c.texi: Mention that other integer types are
9670         permitted as bit-field types in strictly conforming mode.
9672 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
9674         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
9675         is newly allocated.
9677 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
9679         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
9681 2014-01-06  Martin Jambor  <mjambor@suse.cz>
9683         PR ipa/59008
9684         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
9685         to int.
9686         * ipa-prop.c (ipa_print_node_params): Fix indentation.
9688 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
9690         PR debug/59350
9691         PR debug/59510
9692         * var-tracking.c (add_stores): Preserve the value of the source even if
9693         we don't record the store.
9695 2014-01-06  Terry Guo  <terry.guo@arm.com>
9697         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
9699 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
9701         PR bootstrap/59541
9702         * config/darwin.c (darwin_function_section): Adjust return values to
9703         correspond to optimisation changes made in r206070.
9705 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
9707         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
9708         from prefetch_block tune setting.
9709         (nocona_cost): Correct size of prefetch block to 64.
9711 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
9713         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
9714         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
9715         used to save the static chain register in the computation of the offset
9716         from which the FP registers need to be restored.
9718 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
9720         PR tree-optimization/59519
9721         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
9722         ICE if get_current_def (current_new_name) is already non-NULL, as long
9723         as it is a phi result of some other phi in *new_exit_bb that has
9724         the same argument.
9726         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
9727         or vmovdqu* for misaligned_operand.
9728         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
9729         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
9730         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
9731         aligned_mem for AVX512F masked aligned load and store builtins and for
9732         non-temporal moves.
9734 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
9736         PR tree-optimization/59651
9737         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
9738         Address range for negative step should be added by TYPE_SIZE_UNIT.
9740 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
9742         * config/m68k/m68k.c (handle_move_double): Handle pushes with
9743         overlapping registers also for registers other than the stack pointer.
9745 2014-01-03  Marek Polacek  <polacek@redhat.com>
9747         PR other/59661
9748         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
9749         __builtin_FILE.
9751 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
9753         PR target/59625
9754         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
9755         asm goto as jump.
9757         * config/i386/i386.md (MODE_SIZE): New mode attribute.
9758         (push splitter): Use <P:MODE_SIZE> instead of
9759         GET_MODE_SIZE (<P:MODE>mode).
9760         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
9761         (mov -1, reg peephole2): Likewise.
9762         * config/i386/sse.md (*mov<mode>_internal,
9763         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
9764         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
9765         *<code><mode>3, *andnot<mode>3<mask_name>,
9766         <mask_codefor><code><mode>3<mask_name>): Likewise.
9767         * config/i386/subst.md (mask_mode512bit_condition,
9768         sd_mask_mode512bit_condition): Likewise.
9770 2014-01-02  Xinliang David Li  <davidxl@google.com>
9772         PR tree-optimization/59303
9773         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
9774         (dump_predicates): Better output format.
9775         (pred_equal_p): New function.
9776         (is_neq_relop_p): Ditto.
9777         (is_neq_zero_form_p): Ditto.
9778         (pred_expr_equal_p): Ditto.
9779         (pred_neg_p): Ditto.
9780         (simplify_pred): Ditto.
9781         (simplify_preds_2): Ditto.
9782         (simplify_preds_3): Ditto.
9783         (simplify_preds_4): Ditto.
9784         (simplify_preds): Ditto.
9785         (push_pred): Ditto.
9786         (push_to_worklist): Ditto.
9787         (get_pred_info_from_cmp): Ditto.
9788         (is_degenerated_phi): Ditto.
9789         (normalize_one_pred_1): Ditto.
9790         (normalize_one_pred): Ditto.
9791         (normalize_one_pred_chain): Ditto.
9792         (normalize_preds): Ditto.
9793         (normalize_cond_1): Remove function.
9794         (normalize_cond): Ditto.
9795         (is_gcond_subset_of): Ditto.
9796         (is_subset_of_any): Ditto.
9797         (is_or_set_subset_of): Ditto.
9798         (is_and_set_subset_of): Ditto.
9799         (is_norm_cond_subset_of): Ditto.
9800         (pred_chain_length_cmp): Ditto.
9801         (convert_control_dep_chain_into_preds): Type change.
9802         (find_predicates): Ditto.
9803         (find_def_preds): Ditto.
9804         (destroy_predicates_vecs): Ditto.
9805         (find_matching_predicates_in_rest_chains): Ditto.
9806         (use_pred_not_overlap_with_undef_path_pred): Ditto.
9807         (is_pred_expr_subset): Ditto.
9808         (is_pred_chain_subset_of): Ditto.
9809         (is_included_in): Ditto.
9810         (is_superset_of): Ditto.
9812 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
9814         Update copyright years.
9816 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
9818         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
9819         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
9820         config/arc/arc.md, config/arc/arc.opt,
9821         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
9822         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
9823         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
9824         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
9825         config/linux-protos.h, config/linux.c, config/winnt-c.c,
9826         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
9827         vtable-verify.c, vtable-verify.h: Use the standard form for the
9828         copyright notice.
9830 2014-01-02  Tobias Burnus  <burnus@net-b.de>
9832         * gcc.c (process_command): Update copyright notice dates.
9833         * gcov-dump.c: Ditto.
9834         * gcov.c: Ditto.
9835         * doc/cpp.texi: Bump @copying's copyright year.
9836         * doc/cppinternals.texi: Ditto.
9837         * doc/gcc.texi: Ditto.
9838         * doc/gccint.texi: Ditto.
9839         * doc/gcov.texi: Ditto.
9840         * doc/install.texi: Ditto.
9841         * doc/invoke.texi: Ditto.
9843 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9845         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
9847 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
9849         * config/i386/sse.md (*mov<mode>_internal): Guard
9850         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
9852         PR rtl-optimization/59647
9853         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
9854         new_rtx into UNSIGNED_FLOAT rtxes.
9856 Copyright (C) 2014 Free Software Foundation, Inc.
9858 Copying and distribution of this file, with or without modification,
9859 are permitted in any medium without royalty provided the copyright
9860 notice and this notice are preserved.