1 2020-01-07 Martin Liska <mliska@suse.cz>
3 PR tree-optimization/92860
4 * common.opt: Make in Optimization option
5 as it is affected by -O0, which is an Optimization
7 * tree-inline.c (tree_inlinable_function_p):
8 Use opt_for_fn for warn_inline.
9 (expand_call_inline): Likewise.
11 2020-01-07 Martin Liska <mliska@suse.cz>
13 PR tree-optimization/92860
14 * common.opt: Make flag_ree as optimization
17 2020-01-07 Martin Liska <mliska@suse.cz>
20 * params.opt: Mark param_min_crossjump_insns with Optimization
23 2020-01-07 Luo Xiong Hu <luoxhu@linux.ibm.com>
25 * ipa-inline-analysis.c (estimate_growth): Fix typo.
26 * ipa-inline.c (caller_growth_limits): Use OR instead of AND.
28 2020-01-06 Michael Meissner <meissner@linux.ibm.com>
30 * config/rs6000/rs6000.c (hard_reg_and_mode_to_addr_mask): New
31 helper function to return the valid addressing formats for a given
32 hard register and mode.
33 (rs6000_adjust_vec_address): Call hard_reg_and_mode_to_addr_mask.
35 * config/rs6000/constraints.md (Q constraint): Update
37 * doc/md.texi (RS/6000 constraints): Update 'Q' cosntraint
40 * config/rs6000/vsx.md (vsx_extract_<mode>_var, VSX_D iterator):
41 Use 'Q' for doing vector extract from memory.
42 (vsx_extract_v4sf_var): Use 'Q' for doing vector extract from
44 (vsx_extract_<mode>_var, VSX_EXTRACT_I iterator): Use 'Q' for
45 doing vector extract from memory.
46 (vsx_extract_<mode>_<VS_scalar>mode_var): Use 'Q' for doing vector
49 * config/rs6000/rs6000.c (rs6000_adjust_vec_address): Add support
50 for the offset being 34-bits when -mcpu=future is used.
52 2020-01-06 John David Anglin <danglin@gcc.gnu.org>
54 * config/pa/pa.md: Revert change to use ordered_comparison_operator
55 instead of cmpib_comparison_operator in cmpib patterns.
56 * config/pa/predicates.md (cmpib_comparison_operator): Revert removal
57 of cmpib_comparison_operator. Revise comment.
59 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
61 * tree-vect-slp.c (vect_build_slp_tree_1): Require all shifts
62 in an IFN_DIV_POW2 node to be equal.
64 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
66 * tree-vect-stmts.c (vect_check_load_store_mask): Rename to...
67 (vect_check_scalar_mask): ...this.
68 (vectorizable_store, vectorizable_load): Update call accordingly.
69 (vectorizable_call): Use vect_check_scalar_mask to check the mask
70 argument in calls to conditional internal functions.
72 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
74 * config/gcn/gcn-valu.md (subv64di3): Use separate alternatives for
76 (subv64di3_exec): Likewise.
78 2020-01-06 Bryan Stenson <bryan@siliconvortex.com>
80 * config/mips/mips.c (vr4130_align_insns): Fix typo.
81 * doc/md.texi (movstr): Likewise.
83 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
85 * config/gcn/gcn-valu.md (vec_extract<mode><scalar_mode>): Add early
88 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
90 * config/aarch64/t-aarch64 ($(srcdir)/config/aarch64/aarch64-tune.md):
92 (s-aarch64-tune-md): ...this new stamp file. Pipe the new contents
93 to a temporary file and use move-if-change to update the real
96 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
98 * config/aarch64/aarch64-sve.md (@aarch64_sel_dup<mode>): Use Upl
99 rather than Upa for CPY /M.
101 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
103 * config/gcn/gcn.c (gcn_inline_constant_p): Allow 64 as an inline
106 2020-01-06 Martin Liska <mliska@suse.cz>
108 PR tree-optimization/92860
109 * params.opt: Mark param_max_combine_insns with Optimization
112 2020-01-05 Jakub Jelinek <jakub@redhat.com>
115 * config/i386/i386.md (SWIDWI): New mode iterator.
116 (DWI, dwi): Add TImode variants.
117 (addv<mode>4): Use SWIDWI iterator instead of SWI. Use
118 <general_hilo_operand> instead of <general_operand>. Use
119 CONST_SCALAR_INT_P instead of CONST_INT_P.
120 (*addv<mode>4_1): Rename to ...
121 (addv<mode>4_1): ... this.
122 (QWI): New mode attribute.
123 (*addv<dwi>4_doubleword, *addv<dwi>4_doubleword_1): New
124 define_insn_and_split patterns.
125 (*addv<mode>4_overflow_1, *addv<mode>4_overflow_2): New define_insn
127 (uaddv<mode>4): Use SWIDWI iterator instead of SWI. Use
128 <general_hilo_operand> instead of <general_operand>.
129 (*addcarry<mode>_1): New define_insn.
130 (*add<dwi>3_doubleword_cc_overflow_1): New define_insn_and_split.
132 2020-01-03 Konstantin Kharlamov <Hi-Angel@yandex.ru>
134 * gdbinit.in (pr, prl, pt, pct, pgg, pgq, pgs, pge, pmz, pdd, pbs, pbm):
135 Use "call" instead of "set".
137 2020-01-03 Martin Jambor <mjambor@suse.cz>
140 * ipa-cp.c (print_all_lattices): Skip functions without info.
142 2020-01-03 Jakub Jelinek <jakub@redhat.com>
145 * config/i386/i386-options.c (ix86_simd_clone_adjust): If
146 TARGET_PREFER_AVX128, use prefer-vector-width=256 for 'c' and 'd'
147 simd clones. If TARGET_PREFER_AVX256, use prefer-vector-width=512
151 * config/i386/i386.opt (x_prefer_vector_width_type): Remove TargetSave
153 (mprefer-vector-width=): Add Save.
154 * config/i386/i386-options.c (ix86_target_string): Add PVW argument, print
155 -mprefer-vector-width= if non-zero. Fix up -mfpmath= comment.
156 (ix86_debug_options, ix86_function_specific_print): Adjust
157 ix86_target_string callers.
158 (ix86_valid_target_attribute_inner_p): Handle prefer-vector-width=.
159 (ix86_valid_target_attribute_tree): Likewise.
160 * config/i386/i386-options.h (ix86_target_string): Add PVW argument.
161 * config/i386/i386-expand.c (ix86_expand_builtin): Adjust
162 ix86_target_string caller.
165 * config/i386/i386.md (abs<mode>2): Use expand_simple_binop instead of
166 emitting ASHIFTRT, XOR and MINUS by hand. Use gen_int_mode with QImode
167 instead of gen_int_shift_amount + convert_modes.
169 PR rtl-optimization/93088
170 * loop-iv.c (find_single_def_src): Punt after looking through
171 128 reg copies for regs with single definitions. Move definitions
174 2020-01-02 Dennis Zhang <dennis.zhang@arm.com>
176 * config/arm/arm-c.c (arm_cpu_builtins): Define
177 __ARM_FEATURE_MATMUL_INT8, __ARM_FEATURE_BF16_VECTOR_ARITHMETIC,
178 __ARM_FEATURE_BF16_SCALAR_ARITHMETIC, and
179 __ARM_BF16_FORMAT_ALTERNATIVE when enabled.
180 * config/arm/arm-cpus.in (armv8_6, i8mm, bf16): New features.
181 * config/arm/arm-tables.opt: Regenerated.
182 * config/arm/arm.c (arm_option_reconfigure_globals): Initialize
183 arm_arch_i8mm and arm_arch_bf16 when enabled.
184 * config/arm/arm.h (TARGET_I8MM): New macro.
185 (TARGET_BF16_FP, TARGET_BF16_SIMD): Likewise.
186 * config/arm/t-aprofile: Add matching rules for -march=armv8.6-a.
187 * config/arm/t-arm-elf (all_v8_archs): Add armv8.6-a.
188 * config/arm/t-multilib: Add matching rules for -march=armv8.6-a.
189 (v8_6_a_simd_variants): New.
190 (v8_*_a_simd_variants): Add i8mm and bf16.
191 * doc/invoke.texi (armv8.6-a, i8mm, bf16): Document new options.
193 2020-01-02 Jakub Jelinek <jakub@redhat.com>
196 * predict.c (compute_function_frequency): Don't call
197 warn_function_cold on functions that already have cold attribute.
199 2020-01-01 John David Anglin <danglin@gcc.gnu.org>
202 * config/pa/pa.c (pa_elf_select_rtx_section): New. Put references to
203 COMDAT group function labels in .data.rel.ro.local section.
204 * config/pa/pa32-linux.h (TARGET_ASM_SELECT_RTX_SECTION): Define.
207 * config/pa/pa.md (scc): Use ordered_comparison_operator instead of
208 comparison_operator in B and S integer comparisons. Likewise, use
209 ordered_comparison_operator instead of cmpib_comparison_operator in
211 * config/pa/predicates.md (cmpib_comparison_operator): Remove.
213 2020-01-01 Jakub Jelinek <jakub@redhat.com>
215 Update copyright years.
217 * gcc.c (process_command): Update copyright notice dates.
218 * gcov-dump.c (print_version): Ditto.
219 * gcov.c (print_version): Ditto.
220 * gcov-tool.c (print_version): Ditto.
221 * gengtype.c (create_file): Ditto.
222 * doc/cpp.texi: Bump @copying's copyright year.
223 * doc/cppinternals.texi: Ditto.
224 * doc/gcc.texi: Ditto.
225 * doc/gccint.texi: Ditto.
226 * doc/gcov.texi: Ditto.
227 * doc/install.texi: Ditto.
228 * doc/invoke.texi: Ditto.
230 2020-01-01 Jan Hubicka <hubicka@ucw.cz>
232 * ipa.c (walk_polymorphic_call_targets): Fix updating of overall
235 2020-01-01 Jakub Jelinek <jakub@redhat.com>
237 PR tree-optimization/93098
238 * match.pd (popcount): For shift amounts, use integer_onep
239 or wi::to_widest () == cst instead of tree_to_uhwi () == cst
240 tests. Make sure that precision is power of two larger than or equal
241 to 16. Ensure shift is never negative. Use HOST_WIDE_INT_UC macro
242 instead of ULL suffixed constants. Formatting fixes.
244 Copyright (C) 2020 Free Software Foundation, Inc.
246 Copying and distribution of this file, with or without modification,
247 are permitted in any medium without royalty provided the copyright
248 notice and this notice are preserved.