Require equal shift amounts for IFN_DIV_POW2
[gcc.git] / gcc / ChangeLog
1 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
2
3 * tree-vect-slp.c (vect_build_slp_tree_1): Require all shifts
4 in an IFN_DIV_POW2 node to be equal.
5
6 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
7
8 * tree-vect-stmts.c (vect_check_load_store_mask): Rename to...
9 (vect_check_scalar_mask): ...this.
10 (vectorizable_store, vectorizable_load): Update call accordingly.
11 (vectorizable_call): Use vect_check_scalar_mask to check the mask
12 argument in calls to conditional internal functions.
13
14 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
15
16 * config/gcn/gcn-valu.md (subv64di3): Use separate alternatives for
17 '0' matching inputs.
18 (subv64di3_exec): Likewise.
19
20 2020-01-06 Bryan Stenson <bryan@siliconvortex.com>
21
22 * config/mips/mips.c (vr4130_align_insns): Fix typo.
23 * doc/md.texi (movstr): Likewise.
24
25 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
26
27 * config/gcn/gcn-valu.md (vec_extract<mode><scalar_mode>): Add early
28 clobber.
29
30 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
31
32 * config/aarch64/t-aarch64 ($(srcdir)/config/aarch64/aarch64-tune.md):
33 Depend on...
34 (s-aarch64-tune-md): ...this new stamp file. Pipe the new contents
35 to a temporary file and use move-if-change to update the real
36 file where necessary.
37
38 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
39
40 * config/aarch64/aarch64-sve.md (@aarch64_sel_dup<mode>): Use Upl
41 rather than Upa for CPY /M.
42
43 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
44
45 * config/gcn/gcn.c (gcn_inline_constant_p): Allow 64 as an inline
46 immediate.
47
48 2020-01-06 Martin Liska <mliska@suse.cz>
49
50 PR tree-optimization/92860
51 * params.opt: Mark param_max_combine_insns with Optimization
52 keyword.
53
54 2020-01-05 Jakub Jelinek <jakub@redhat.com>
55
56 PR target/93141
57 * config/i386/i386.md (SWIDWI): New mode iterator.
58 (DWI, dwi): Add TImode variants.
59 (addv<mode>4): Use SWIDWI iterator instead of SWI. Use
60 <general_hilo_operand> instead of <general_operand>. Use
61 CONST_SCALAR_INT_P instead of CONST_INT_P.
62 (*addv<mode>4_1): Rename to ...
63 (addv<mode>4_1): ... this.
64 (QWI): New mode attribute.
65 (*addv<dwi>4_doubleword, *addv<dwi>4_doubleword_1): New
66 define_insn_and_split patterns.
67 (*addv<mode>4_overflow_1, *addv<mode>4_overflow_2): New define_insn
68 patterns.
69 (uaddv<mode>4): Use SWIDWI iterator instead of SWI. Use
70 <general_hilo_operand> instead of <general_operand>.
71 (*addcarry<mode>_1): New define_insn.
72 (*add<dwi>3_doubleword_cc_overflow_1): New define_insn_and_split.
73
74 2020-01-03 Konstantin Kharlamov <Hi-Angel@yandex.ru>
75
76 * gdbinit.in (pr, prl, pt, pct, pgg, pgq, pgs, pge, pmz, pdd, pbs, pbm):
77 Use "call" instead of "set".
78
79 2020-01-03 Martin Jambor <mjambor@suse.cz>
80
81 PR ipa/92917
82 * ipa-cp.c (print_all_lattices): Skip functions without info.
83
84 2020-01-03 Jakub Jelinek <jakub@redhat.com>
85
86 PR target/93089
87 * config/i386/i386-options.c (ix86_simd_clone_adjust): If
88 TARGET_PREFER_AVX128, use prefer-vector-width=256 for 'c' and 'd'
89 simd clones. If TARGET_PREFER_AVX256, use prefer-vector-width=512
90 for 'e' simd clones.
91
92 PR target/93089
93 * config/i386/i386.opt (x_prefer_vector_width_type): Remove TargetSave
94 entry.
95 (mprefer-vector-width=): Add Save.
96 * config/i386/i386-options.c (ix86_target_string): Add PVW argument, print
97 -mprefer-vector-width= if non-zero. Fix up -mfpmath= comment.
98 (ix86_debug_options, ix86_function_specific_print): Adjust
99 ix86_target_string callers.
100 (ix86_valid_target_attribute_inner_p): Handle prefer-vector-width=.
101 (ix86_valid_target_attribute_tree): Likewise.
102 * config/i386/i386-options.h (ix86_target_string): Add PVW argument.
103 * config/i386/i386-expand.c (ix86_expand_builtin): Adjust
104 ix86_target_string caller.
105
106 PR target/93110
107 * config/i386/i386.md (abs<mode>2): Use expand_simple_binop instead of
108 emitting ASHIFTRT, XOR and MINUS by hand. Use gen_int_mode with QImode
109 instead of gen_int_shift_amount + convert_modes.
110
111 PR rtl-optimization/93088
112 * loop-iv.c (find_single_def_src): Punt after looking through
113 128 reg copies for regs with single definitions. Move definitions
114 to first uses.
115
116 2020-01-02 Dennis Zhang <dennis.zhang@arm.com>
117
118 * config/arm/arm-c.c (arm_cpu_builtins): Define
119 __ARM_FEATURE_MATMUL_INT8, __ARM_FEATURE_BF16_VECTOR_ARITHMETIC,
120 __ARM_FEATURE_BF16_SCALAR_ARITHMETIC, and
121 __ARM_BF16_FORMAT_ALTERNATIVE when enabled.
122 * config/arm/arm-cpus.in (armv8_6, i8mm, bf16): New features.
123 * config/arm/arm-tables.opt: Regenerated.
124 * config/arm/arm.c (arm_option_reconfigure_globals): Initialize
125 arm_arch_i8mm and arm_arch_bf16 when enabled.
126 * config/arm/arm.h (TARGET_I8MM): New macro.
127 (TARGET_BF16_FP, TARGET_BF16_SIMD): Likewise.
128 * config/arm/t-aprofile: Add matching rules for -march=armv8.6-a.
129 * config/arm/t-arm-elf (all_v8_archs): Add armv8.6-a.
130 * config/arm/t-multilib: Add matching rules for -march=armv8.6-a.
131 (v8_6_a_simd_variants): New.
132 (v8_*_a_simd_variants): Add i8mm and bf16.
133 * doc/invoke.texi (armv8.6-a, i8mm, bf16): Document new options.
134
135 2020-01-02 Jakub Jelinek <jakub@redhat.com>
136
137 PR ipa/93087
138 * predict.c (compute_function_frequency): Don't call
139 warn_function_cold on functions that already have cold attribute.
140
141 2020-01-01 John David Anglin <danglin@gcc.gnu.org>
142
143 PR target/67834
144 * config/pa/pa.c (pa_elf_select_rtx_section): New. Put references to
145 COMDAT group function labels in .data.rel.ro.local section.
146 * config/pa/pa32-linux.h (TARGET_ASM_SELECT_RTX_SECTION): Define.
147
148 PR target/93111
149 * config/pa/pa.md (scc): Use ordered_comparison_operator instead of
150 comparison_operator in B and S integer comparisons. Likewise, use
151 ordered_comparison_operator instead of cmpib_comparison_operator in
152 cmpib patterns.
153 * config/pa/predicates.md (cmpib_comparison_operator): Remove.
154
155 2020-01-01 Jakub Jelinek <jakub@redhat.com>
156
157 Update copyright years.
158
159 * gcc.c (process_command): Update copyright notice dates.
160 * gcov-dump.c (print_version): Ditto.
161 * gcov.c (print_version): Ditto.
162 * gcov-tool.c (print_version): Ditto.
163 * gengtype.c (create_file): Ditto.
164 * doc/cpp.texi: Bump @copying's copyright year.
165 * doc/cppinternals.texi: Ditto.
166 * doc/gcc.texi: Ditto.
167 * doc/gccint.texi: Ditto.
168 * doc/gcov.texi: Ditto.
169 * doc/install.texi: Ditto.
170 * doc/invoke.texi: Ditto.
171
172 2020-01-01 Jan Hubicka <hubicka@ucw.cz>
173
174 * ipa.c (walk_polymorphic_call_targets): Fix updating of overall
175 summary.
176
177 2020-01-01 Jakub Jelinek <jakub@redhat.com>
178
179 PR tree-optimization/93098
180 * match.pd (popcount): For shift amounts, use integer_onep
181 or wi::to_widest () == cst instead of tree_to_uhwi () == cst
182 tests. Make sure that precision is power of two larger than or equal
183 to 16. Ensure shift is never negative. Use HOST_WIDE_INT_UC macro
184 instead of ULL suffixed constants. Formatting fixes.
185 \f
186 Copyright (C) 2020 Free Software Foundation, Inc.
187
188 Copying and distribution of this file, with or without modification,
189 are permitted in any medium without royalty provided the copyright
190 notice and this notice are preserved.