1 2020-01-10 Martin Jambor <mjambor@suse.cz>
3 * params.opt (param_ipa_max_agg_items): Mark as Optimization
4 * ipa-cp.c (merge_agg_lats_step): New parameter max_agg_items, use
5 instead of param_ipa_max_agg_items.
6 (merge_aggregate_lattices): Extract param_ipa_max_agg_items from
7 optimization info for the callee.
9 2020-01-09 Kwok Cheung Yeung <kcy@codesourcery.com>
11 * lto-streamer-in.c (input_function): Remove streamed-in inline debug
12 markers if debug_inline_points is false.
14 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
16 * config.gcc (aarch64*-*-*): Add aarch64-sve-builtins-sve2.o to
18 * config/aarch64/t-aarch64 (aarch64-sve-builtins.o): Depend on
19 aarch64-sve-builtins-base.def, aarch64-sve-builtins-sve2.def and
20 aarch64-sve-builtins-sve2.h.
21 (aarch64-sve-builtins-sve2.o): New rule.
22 * config/aarch64/aarch64.h (AARCH64_ISA_SVE2_AES): New macro.
23 (AARCH64_ISA_SVE2_BITPERM, AARCH64_ISA_SVE2_SHA3): Likewise.
24 (AARCH64_ISA_SVE2_SM4, TARGET_SVE2_AES, TARGET_SVE2_BITPERM): Likewise.
25 (TARGET_SVE2_SHA, TARGET_SVE2_SM4): Likewise.
26 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
27 TARGET_SVE2_AES, TARGET_SVE2_BITPERM, TARGET_SVE2_SHA3 and
29 * config/aarch64/aarch64-sve.md: Update comments with SVE2
30 instructions that are handled here.
31 (@cond_asrd<mode>): Generalize to...
32 (@cond_<SVE_INT_SHIFT_IMM:sve_int_op><mode>): ...this.
33 (*cond_asrd<mode>_2): Generalize to...
34 (*cond_<SVE_INT_SHIFT_IMM:sve_int_op><mode>_2): ...this.
35 (*cond_asrd<mode>_z): Generalize to...
36 (*cond_<SVE_INT_SHIFT_IMM:sve_int_op><mode>_z): ...this.
37 * config/aarch64/aarch64.md (UNSPEC_LDNT1_GATHER): New unspec.
38 (UNSPEC_STNT1_SCATTER, UNSPEC_WHILEGE, UNSPEC_WHILEGT): Likewise.
39 (UNSPEC_WHILEHI, UNSPEC_WHILEHS): Likewise.
40 * config/aarch64/aarch64-sve2.md (@aarch64_gather_ldnt<mode>): New
42 (@aarch64_gather_ldnt_<ANY_EXTEND:optab><SVE_FULL_SDI:mode><SVE_PARTIAL_I:mode>)
43 (@aarch64_scatter_stnt<mode>): Likewise.
44 (@aarch64_scatter_stnt_<SVE_FULL_SDI:mode><SVE_PARTIAL_I:mode>)
45 (@aarch64_mul_lane_<mode>): Likewise.
46 (@aarch64_sve_suqadd<mode>_const): Likewise.
47 (*<sur>h<addsub><mode>): Generalize to...
48 (@aarch64_pred_<SVE2_COND_INT_BINARY_REV:sve_int_op><mode>): ...this
50 (@cond_<SVE2_COND_INT_BINARY:sve_int_op><mode>): New expander.
51 (*cond_<SVE2_COND_INT_BINARY:sve_int_op><mode>_2): New pattern.
52 (*cond_<SVE2_COND_INT_BINARY:sve_int_op><mode>_3): Likewise.
53 (*cond_<SVE2_COND_INT_BINARY:sve_int_op><mode>_any): Likewise.
54 (*cond_<SVE2_COND_INT_BINARY_NOREV:sve_int_op><mode>_z): Likewise.
55 (@aarch64_sve_<SVE2_INT_BINARY:sve_int_op><mode>):: Likewise.
56 (@aarch64_sve_<SVE2_INT_BINARY:sve_int_op>_lane_<mode>): Likewise.
57 (@aarch64_pred_<SVE2_COND_INT_SHIFT:sve_int_op><mode>): Likewise.
58 (@cond_<SVE2_COND_INT_SHIFT:sve_int_op><mode>): New expander.
59 (*cond_<SVE2_COND_INT_SHIFT:sve_int_op><mode>_2): New pattern.
60 (*cond_<SVE2_COND_INT_SHIFT:sve_int_op><mode>_3): Likewise.
61 (*cond_<SVE2_COND_INT_SHIFT:sve_int_op><mode>_any): Likewise.
62 (@aarch64_sve_<SVE2_INT_TERNARY:sve_int_op><mode>): Likewise.
63 (@aarch64_sve_<SVE2_INT_TERNARY_LANE:sve_int_op>_lane_<mode>)
64 (@aarch64_sve_add_mul_lane_<mode>): Likewise.
65 (@aarch64_sve_sub_mul_lane_<mode>): Likewise.
66 (@aarch64_sve2_xar<mode>): Likewise.
67 (@aarch64_sve2_bcax<mode>): Likewise.
68 (*aarch64_sve2_eor3<mode>): Rename to...
69 (@aarch64_sve2_eor3<mode>): ...this.
70 (@aarch64_sve2_bsl<mode>): New expander.
71 (@aarch64_sve2_nbsl<mode>): Likewise.
72 (@aarch64_sve2_bsl1n<mode>): Likewise.
73 (@aarch64_sve2_bsl2n<mode>): Likewise.
74 (@aarch64_sve_add_<SHIFTRT:sve_int_op><mode>): Likewise.
75 (*aarch64_sve2_sra<mode>): Add MOVPRFX support.
76 (@aarch64_sve_add_<VRSHR_N:sve_int_op><mode>): New pattern.
77 (@aarch64_sve_<SVE2_INT_SHIFT_INSERT:sve_int_op><mode>): Likewise.
78 (@aarch64_sve2_<USMAX:su>aba<mode>): New expander.
79 (*aarch64_sve2_<USMAX:su>aba<mode>): New pattern.
80 (@aarch64_sve_<SVE2_INT_BINARY_WIDE:sve_int_op><mode>): Likewise.
81 (<su>mull<bt><Vwide>): Generalize to...
82 (@aarch64_sve_<SVE2_INT_BINARY_LONG:sve_int_op><mode>): ...this new
84 (@aarch64_sve_<SVE2_INT_BINARY_LONG_lANE:sve_int_op>_lane_<mode>)
85 (@aarch64_sve_<SVE2_INT_SHIFT_IMM_LONG:sve_int_op><mode>)
86 (@aarch64_sve_add_<SVE2_INT_ADD_BINARY_LONG:sve_int_op><mode>)
87 (@aarch64_sve_add_<SVE2_INT_ADD_BINARY_LONG_LANE:sve_int_op>_lane_<mode>)
88 (@aarch64_sve_qadd_<SVE2_INT_QADD_BINARY_LONG:sve_int_op><mode>)
89 (@aarch64_sve_qadd_<SVE2_INT_QADD_BINARY_LONG_LANE:sve_int_op>_lane_<mode>)
90 (@aarch64_sve_sub_<SVE2_INT_SUB_BINARY_LONG:sve_int_op><mode>)
91 (@aarch64_sve_sub_<SVE2_INT_SUB_BINARY_LONG_LANE:sve_int_op>_lane_<mode>)
92 (@aarch64_sve_qsub_<SVE2_INT_QSUB_BINARY_LONG:sve_int_op><mode>)
93 (@aarch64_sve_qsub_<SVE2_INT_QSUB_BINARY_LONG_LANE:sve_int_op>_lane_<mode>)
94 (@aarch64_sve_<SVE2_FP_TERNARY_LONG:sve_fp_op><mode>): New patterns.
95 (@aarch64_<SVE2_FP_TERNARY_LONG_LANE:sve_fp_op>_lane_<mode>)
96 (@aarch64_sve_<SVE2_INT_UNARY_NARROWB:sve_int_op><mode>): Likewise.
97 (@aarch64_sve_<SVE2_INT_UNARY_NARROWT:sve_int_op><mode>): Likewise.
98 (@aarch64_sve_<SVE2_INT_BINARY_NARROWB:sve_int_op><mode>): Likewise.
99 (@aarch64_sve_<SVE2_INT_BINARY_NARROWT:sve_int_op><mode>): Likewise.
100 (<SHRNB:r>shrnb<mode>): Generalize to...
101 (@aarch64_sve_<SVE2_INT_SHIFT_IMM_NARROWB:sve_int_op><mode>): ...this
103 (<SHRNT:r>shrnt<mode>): Generalize to...
104 (@aarch64_sve_<SVE2_INT_SHIFT_IMM_NARROWT:sve_int_op><mode>): ...this
106 (@aarch64_pred_<SVE2_INT_BINARY_PAIR:sve_int_op><mode>): New pattern.
107 (@aarch64_pred_<SVE2_FP_BINARY_PAIR:sve_fp_op><mode>): Likewise.
108 (@cond_<SVE2_INT_BINARY_PAIR_LONG:sve_int_op><mode>): New expander.
109 (*cond_<SVE2_INT_BINARY_PAIR_LONG:sve_int_op><mode>_2): New pattern.
110 (*cond_<SVE2_INT_BINARY_PAIR_LONG:sve_int_op><mode>_z): Likewise.
111 (@aarch64_sve_<SVE2_INT_CADD:optab><mode>): Likewise.
112 (@aarch64_sve_<SVE2_INT_CMLA:optab><mode>): Likewise.
113 (@aarch64_<SVE2_INT_CMLA:optab>_lane_<mode>): Likewise.
114 (@aarch64_sve_<SVE2_INT_CDOT:optab><mode>): Likewise.
115 (@aarch64_<SVE2_INT_CDOT:optab>_lane_<mode>): Likewise.
116 (@aarch64_pred_<SVE2_COND_FP_UNARY_LONG:sve_fp_op><mode>): Likewise.
117 (@cond_<SVE2_COND_FP_UNARY_LONG:sve_fp_op><mode>): New expander.
118 (*cond_<SVE2_COND_FP_UNARY_LONG:sve_fp_op><mode>): New pattern.
119 (@aarch64_sve2_cvtnt<mode>): Likewise.
120 (@aarch64_pred_<SVE2_COND_FP_UNARY_NARROWB:sve_fp_op><mode>): Likewise.
121 (@cond_<SVE2_COND_FP_UNARY_NARROWB:sve_fp_op><mode>): New expander.
122 (*cond_<SVE2_COND_FP_UNARY_NARROWB:sve_fp_op><mode>_any): New pattern.
123 (@aarch64_sve2_cvtxnt<mode>): Likewise.
124 (@aarch64_pred_<SVE2_U32_UNARY:sve_int_op><mode>): Likewise.
125 (@cond_<SVE2_U32_UNARY:sve_int_op><mode>): New expander.
126 (*cond_<SVE2_U32_UNARY:sve_int_op><mode>): New pattern.
127 (@aarch64_pred_<SVE2_COND_INT_UNARY_FP:sve_fp_op><mode>): Likewise.
128 (@cond_<SVE2_COND_INT_UNARY_FP:sve_fp_op><mode>): New expander.
129 (*cond_<SVE2_COND_INT_UNARY_FP:sve_fp_op><mode>): New pattern.
130 (@aarch64_sve2_pmul<mode>): Likewise.
131 (@aarch64_sve_<SVE2_PMULL:optab><mode>): Likewise.
132 (@aarch64_sve_<SVE2_PMULL_PAIR:optab><mode>): Likewise.
133 (@aarch64_sve2_tbl2<mode>): Likewise.
134 (@aarch64_sve2_tbx<mode>): Likewise.
135 (@aarch64_sve_<SVE2_INT_BITPERM:sve_int_op><mode>): Likewise.
136 (@aarch64_sve2_histcnt<mode>): Likewise.
137 (@aarch64_sve2_histseg<mode>): Likewise.
138 (@aarch64_pred_<SVE2_MATCH:sve_int_op><mode>): Likewise.
139 (*aarch64_pred_<SVE2_MATCH:sve_int_op><mode>_cc): Likewise.
140 (*aarch64_pred_<SVE2_MATCH:sve_int_op><mode>_ptest): Likewise.
141 (aarch64_sve2_aes<CRYPTO_AES:aes_op>): Likewise.
142 (aarch64_sve2_aes<CRYPTO_AESMC:aesmc_op>): Likewise.
143 (*aarch64_sve2_aese_fused, *aarch64_sve2_aesd_fused): Likewise.
144 (aarch64_sve2_rax1, aarch64_sve2_sm4e, aarch64_sve2_sm4ekey): Likewise.
145 (<su>mulh<r>s<mode>3): Update after above pattern name changes.
146 * config/aarch64/iterators.md (VNx16QI_ONLY, VNx4SF_ONLY)
147 (SVE_STRUCT2, SVE_FULL_BHI, SVE_FULL_HSI, SVE_FULL_HDI)
148 (SVE2_PMULL_PAIR_I): New mode iterators.
149 (UNSPEC_ADCLB, UNSPEC_ADCLT, UNSPEC_ADDHNB, UNSPEC_ADDHNT, UNSPEC_BDEP)
150 (UNSPEC_BEXT, UNSPEC_BGRP, UNSPEC_CADD90, UNSPEC_CADD270, UNSPEC_CDOT)
151 (UNSPEC_CDOT90, UNSPEC_CDOT180, UNSPEC_CDOT270, UNSPEC_CMLA)
152 (UNSPEC_CMLA90, UNSPEC_CMLA180, UNSPEC_CMLA270, UNSPEC_COND_FCVTLT)
153 (UNSPEC_COND_FCVTNT, UNSPEC_COND_FCVTX, UNSPEC_COND_FCVTXNT)
154 (UNSPEC_COND_FLOGB, UNSPEC_EORBT, UNSPEC_EORTB, UNSPEC_FADDP)
155 (UNSPEC_FMAXP, UNSPEC_FMAXNMP, UNSPEC_FMLALB, UNSPEC_FMLALT)
156 (UNSPEC_FMLSLB, UNSPEC_FMLSLT, UNSPEC_FMINP, UNSPEC_FMINNMP)
157 (UNSPEC_HISTCNT, UNSPEC_HISTSEG, UNSPEC_MATCH, UNSPEC_NMATCH)
158 (UNSPEC_PMULLB, UNSPEC_PMULLB_PAIR, UNSPEC_PMULLT, UNSPEC_PMULLT_PAIR)
159 (UNSPEC_RADDHNB, UNSPEC_RADDHNT, UNSPEC_RSUBHNB, UNSPEC_RSUBHNT)
160 (UNSPEC_SLI, UNSPEC_SRI, UNSPEC_SABDLB, UNSPEC_SABDLT, UNSPEC_SADDLB)
161 (UNSPEC_SADDLBT, UNSPEC_SADDLT, UNSPEC_SADDWB, UNSPEC_SADDWT)
162 (UNSPEC_SBCLB, UNSPEC_SBCLT, UNSPEC_SMAXP, UNSPEC_SMINP)
163 (UNSPEC_SQCADD90, UNSPEC_SQCADD270, UNSPEC_SQDMULLB, UNSPEC_SQDMULLBT)
164 (UNSPEC_SQDMULLT, UNSPEC_SQRDCMLAH, UNSPEC_SQRDCMLAH90)
165 (UNSPEC_SQRDCMLAH180, UNSPEC_SQRDCMLAH270, UNSPEC_SQRSHRNB)
166 (UNSPEC_SQRSHRNT, UNSPEC_SQRSHRUNB, UNSPEC_SQRSHRUNT, UNSPEC_SQSHRNB)
167 (UNSPEC_SQSHRNT, UNSPEC_SQSHRUNB, UNSPEC_SQSHRUNT, UNSPEC_SQXTNB)
168 (UNSPEC_SQXTNT, UNSPEC_SQXTUNB, UNSPEC_SQXTUNT, UNSPEC_SSHLLB)
169 (UNSPEC_SSHLLT, UNSPEC_SSUBLB, UNSPEC_SSUBLBT, UNSPEC_SSUBLT)
170 (UNSPEC_SSUBLTB, UNSPEC_SSUBWB, UNSPEC_SSUBWT, UNSPEC_SUBHNB)
171 (UNSPEC_SUBHNT, UNSPEC_TBL2, UNSPEC_UABDLB, UNSPEC_UABDLT)
172 (UNSPEC_UADDLB, UNSPEC_UADDLT, UNSPEC_UADDWB, UNSPEC_UADDWT)
173 (UNSPEC_UMAXP, UNSPEC_UMINP, UNSPEC_UQRSHRNB, UNSPEC_UQRSHRNT)
174 (UNSPEC_UQSHRNB, UNSPEC_UQSHRNT, UNSPEC_UQXTNB, UNSPEC_UQXTNT)
175 (UNSPEC_USHLLB, UNSPEC_USHLLT, UNSPEC_USUBLB, UNSPEC_USUBLT)
176 (UNSPEC_USUBWB, UNSPEC_USUBWT): New unspecs.
177 (UNSPEC_SMULLB, UNSPEC_SMULLT, UNSPEC_UMULLB, UNSPEC_UMULLT)
178 (UNSPEC_SMULHS, UNSPEC_SMULHRS, UNSPEC_UMULHS, UNSPEC_UMULHRS)
179 (UNSPEC_RSHRNB, UNSPEC_RSHRNT, UNSPEC_SHRNB, UNSPEC_SHRNT): Move
181 (VNARROW, Ventype): New mode attributes.
182 (Vewtype): Handle VNx2DI. Fix typo in comment.
183 (VDOUBLE): New mode attribute.
184 (sve_lane_con): Handle VNx8HI.
185 (SVE_INT_UNARY): Include ss_abs and ss_neg for TARGET_SVE2.
186 (SVE_INT_BINARY): Likewise ss_plus, us_plus, ss_minus and us_minus.
187 (sve_int_op, sve_int_op_rev): Handle the above codes.
188 (sve_pred_int_rhs2_operand): Likewise.
189 (MULLBT, SHRNB, SHRNT): Delete.
190 (SVE_INT_SHIFT_IMM): New int iterator.
191 (SVE_WHILE): Add UNSPEC_WHILEGE, UNSPEC_WHILEGT, UNSPEC_WHILEHI
192 and UNSPEC_WHILEHS for TARGET_SVE2.
193 (SVE2_U32_UNARY, SVE2_INT_UNARY_NARROWB, SVE2_INT_UNARY_NARROWT)
194 (SVE2_INT_BINARY, SVE2_INT_BINARY_LANE, SVE2_INT_BINARY_LONG)
195 (SVE2_INT_BINARY_LONG_LANE, SVE2_INT_BINARY_NARROWB)
196 (SVE2_INT_BINARY_NARROWT, SVE2_INT_BINARY_PAIR, SVE2_FP_BINARY_PAIR)
197 (SVE2_INT_BINARY_PAIR_LONG, SVE2_INT_BINARY_WIDE): New int iterators.
198 (SVE2_INT_SHIFT_IMM_LONG, SVE2_INT_SHIFT_IMM_NARROWB): Likewise.
199 (SVE2_INT_SHIFT_IMM_NARROWT, SVE2_INT_SHIFT_INSERT, SVE2_INT_CADD)
200 (SVE2_INT_BITPERM, SVE2_INT_TERNARY, SVE2_INT_TERNARY_LANE): Likewise.
201 (SVE2_FP_TERNARY_LONG, SVE2_FP_TERNARY_LONG_LANE, SVE2_INT_CMLA)
202 (SVE2_INT_CDOT, SVE2_INT_ADD_BINARY_LONG, SVE2_INT_QADD_BINARY_LONG)
203 (SVE2_INT_SUB_BINARY_LONG, SVE2_INT_QSUB_BINARY_LONG): Likewise.
204 (SVE2_INT_ADD_BINARY_LONG_LANE, SVE2_INT_QADD_BINARY_LONG_LANE)
205 (SVE2_INT_SUB_BINARY_LONG_LANE, SVE2_INT_QSUB_BINARY_LONG_LANE)
206 (SVE2_COND_INT_UNARY_FP, SVE2_COND_FP_UNARY_LONG): Likewise.
207 (SVE2_COND_FP_UNARY_NARROWB, SVE2_COND_INT_BINARY): Likewise.
208 (SVE2_COND_INT_BINARY_NOREV, SVE2_COND_INT_BINARY_REV): Likewise.
209 (SVE2_COND_INT_SHIFT, SVE2_MATCH, SVE2_PMULL): Likewise.
210 (optab): Handle the new unspecs.
211 (su, r): Remove entries for UNSPEC_SHRNB, UNSPEC_SHRNT, UNSPEC_RSHRNB
213 (lr): Handle the new unspecs.
215 (cmp_op, while_optab_cmp, sve_int_op): Handle the new unspecs.
216 (sve_int_op_rev, sve_int_add_op, sve_int_qadd_op, sve_int_sub_op)
217 (sve_int_qsub_op): New int attributes.
218 (sve_fp_op, rot): Handle the new unspecs.
219 * config/aarch64/aarch64-sve-builtins.h
220 (function_resolver::require_matching_pointer_type): Declare.
221 (function_resolver::resolve_unary): Add an optional boolean argument.
222 (function_resolver::finish_opt_n_resolution): Add an optional
223 type_suffix_index argument.
224 (gimple_folder::redirect_call): Declare.
225 (gimple_expander::prepare_gather_address_operands): Add an optional
227 * config/aarch64/aarch64-sve-builtins.cc: Include
228 aarch64-sve-builtins-sve2.h.
229 (TYPES_b_unsigned, TYPES_b_integer, TYPES_bh_integer): New macros.
230 (TYPES_bs_unsigned, TYPES_hs_signed, TYPES_hs_integer): Likewise.
231 (TYPES_hd_unsigned, TYPES_hsd_signed): Likewise.
232 (TYPES_hsd_integer): Use TYPES_hsd_signed.
233 (TYPES_s_float_hsd_integer, TYPES_s_float_sd_integer): New macros.
234 (TYPES_s_unsigned): Likewise.
235 (TYPES_s_integer): Use TYPES_s_unsigned.
236 (TYPES_sd_signed, TYPES_sd_unsigned): New macros.
237 (TYPES_sd_integer): Use them.
238 (TYPES_d_unsigned): New macro.
239 (TYPES_d_integer): Use it.
240 (TYPES_d_data, TYPES_cvt_long, TYPES_cvt_narrow_s): New macros.
241 (TYPES_cvt_narrow): Likewise.
242 (DEF_SVE_TYPES_ARRAY): Include the new types macros above.
243 (preds_mx): New variable.
244 (function_builder::add_overloaded_function): Allow the new feature
245 set to be more restrictive than the original one.
246 (function_resolver::infer_pointer_type): Remove qualifiers from
247 the pointer type before printing it.
248 (function_resolver::require_matching_pointer_type): New function.
249 (function_resolver::resolve_sv_displacement): Handle functions
250 that don't support 32-bit vector indices or svint32_t vector offsets.
251 (function_resolver::finish_opt_n_resolution): Take the inferred type
252 as a separate argument.
253 (function_resolver::resolve_unary): Optionally treat all forms in
254 the same way as normal merging functions.
255 (gimple_folder::redirect_call): New function.
256 (function_expander::prepare_gather_address_operands): Add an argument
257 that says whether scaled forms are available. If they aren't,
258 handle scaling of vector indices and don't add the extension and
260 (function_expander::map_to_unspecs): If aarch64_sve isn't available,
261 fall back to using cond_* instead.
262 * config/aarch64/aarch64-sve-builtins-functions.h (rtx_code_function):
263 Split out the member variables into...
264 (rtx_code_function_base): ...this new base class.
265 (rtx_code_function_rotated): Inherit rtx_code_function_base.
266 (unspec_based_function): Split out the member variables into...
267 (unspec_based_function_base): ...this new base class.
268 (unspec_based_function_rotated): Inherit unspec_based_function_base.
269 (unspec_based_function_exact_insn): New class.
270 (unspec_based_add_function, unspec_based_add_lane_function)
271 (unspec_based_lane_function, unspec_based_pred_function)
272 (unspec_based_qadd_function, unspec_based_qadd_lane_function)
273 (unspec_based_qsub_function, unspec_based_qsub_lane_function)
274 (unspec_based_sub_function, unspec_based_sub_lane_function): New
276 (unspec_based_fused_function): New class.
277 (unspec_based_mla_function, unspec_based_mls_function): New typedefs.
278 (unspec_based_fused_lane_function): New class.
279 (unspec_based_mla_lane_function, unspec_based_mls_lane_function): New
281 (CODE_FOR_MODE1): New macro.
282 (fixed_insn_function): New class.
283 (while_comparison): Likewise.
284 * config/aarch64/aarch64-sve-builtins-shapes.h (binary_long_lane)
285 (binary_long_opt_n, binary_narrowb_opt_n, binary_narrowt_opt_n)
286 (binary_to_uint, binary_wide, binary_wide_opt_n, compare, compare_ptr)
287 (load_ext_gather_index_restricted, load_ext_gather_offset_restricted)
288 (load_gather_sv_restricted, shift_left_imm_long): Declare.
289 (shift_left_imm_to_uint, shift_right_imm_narrowb): Likewise.
290 (shift_right_imm_narrowt, shift_right_imm_narrowb_to_uint): Likewise.
291 (shift_right_imm_narrowt_to_uint, store_scatter_index_restricted)
292 (store_scatter_offset_restricted, tbl_tuple, ternary_long_lane)
293 (ternary_long_opt_n, ternary_qq_lane_rotate, ternary_qq_rotate)
294 (ternary_shift_left_imm, ternary_shift_right_imm, ternary_uint)
295 (unary_convert_narrowt, unary_long, unary_narrowb, unary_narrowt)
296 (unary_narrowb_to_uint, unary_narrowt_to_uint, unary_to_int): Likewise.
297 * config/aarch64/aarch64-sve-builtins-shapes.cc (apply_predication):
298 Also add an initial argument for unary_convert_narrowt, regardless
299 of the predication type.
300 (build_32_64): Allow loads and stores to specify MODE_none.
301 (build_sv_index64, build_sv_uint_offset): New functions.
302 (long_type_suffix): New function.
303 (binary_imm_narrowb_base, binary_imm_narrowt_base): New classes.
304 (binary_imm_long_base, load_gather_sv_base): Likewise.
305 (shift_right_imm_narrow_wrapper, ternary_shift_imm_base): Likewise.
306 (ternary_resize2_opt_n_base, ternary_resize2_lane_base): Likewise.
307 (unary_narrowb_base, unary_narrowt_base): Likewise.
308 (binary_long_lane_def, binary_long_lane): New shape.
309 (binary_long_opt_n_def, binary_long_opt_n): Likewise.
310 (binary_narrowb_opt_n_def, binary_narrowb_opt_n): Likewise.
311 (binary_narrowt_opt_n_def, binary_narrowt_opt_n): Likewise.
312 (binary_to_uint_def, binary_to_uint): Likewise.
313 (binary_wide_def, binary_wide): Likewise.
314 (binary_wide_opt_n_def, binary_wide_opt_n): Likewise.
315 (compare_def, compare): Likewise.
316 (compare_ptr_def, compare_ptr): Likewise.
317 (load_ext_gather_index_restricted_def,
318 load_ext_gather_index_restricted): Likewise.
319 (load_ext_gather_offset_restricted_def,
320 load_ext_gather_offset_restricted): Likewise.
321 (load_gather_sv_def): Inherit from load_gather_sv_base.
322 (load_gather_sv_restricted_def, load_gather_sv_restricted): New shape.
323 (shift_left_imm_def, shift_left_imm): Likewise.
324 (shift_left_imm_long_def, shift_left_imm_long): Likewise.
325 (shift_left_imm_to_uint_def, shift_left_imm_to_uint): Likewise.
326 (store_scatter_index_restricted_def,
327 store_scatter_index_restricted): Likewise.
328 (store_scatter_offset_restricted_def,
329 store_scatter_offset_restricted): Likewise.
330 (tbl_tuple_def, tbl_tuple): Likewise.
331 (ternary_long_lane_def, ternary_long_lane): Likewise.
332 (ternary_long_opt_n_def, ternary_long_opt_n): Likewise.
333 (ternary_qq_lane_def): Inherit from ternary_resize2_lane_base.
334 (ternary_qq_lane_rotate_def, ternary_qq_lane_rotate): New shape
335 (ternary_qq_opt_n_def): Inherit from ternary_resize2_opt_n_base.
336 (ternary_qq_rotate_def, ternary_qq_rotate): New shape.
337 (ternary_shift_left_imm_def, ternary_shift_left_imm): Likewise.
338 (ternary_shift_right_imm_def, ternary_shift_right_imm): Likewise.
339 (ternary_uint_def, ternary_uint): Likewise.
340 (unary_convert): Fix typo in comment.
341 (unary_convert_narrowt_def, unary_convert_narrowt): New shape.
342 (unary_long_def, unary_long): Likewise.
343 (unary_narrowb_def, unary_narrowb): Likewise.
344 (unary_narrowt_def, unary_narrowt): Likewise.
345 (unary_narrowb_to_uint_def, unary_narrowb_to_uint): Likewise.
346 (unary_narrowt_to_uint_def, unary_narrowt_to_uint): Likewise.
347 (unary_to_int_def, unary_to_int): Likewise.
348 * config/aarch64/aarch64-sve-builtins-base.cc (unspec_cmla)
349 (unspec_fcmla, unspec_cond_fcmla, expand_mla_mls_lane): New functions.
350 (svasrd_impl): Delete.
351 (svcadd_impl::expand): Handle integer operations too.
352 (svcmla_impl::expand, svcmla_lane::expand): Likewise, using the
353 new functions to derive the unspec numbers.
354 (svmla_svmls_lane_impl): Replace with...
355 (svmla_lane_impl, svmls_lane_impl): ...these new classes. Handle
356 integer operations too.
357 (svwhile_impl): Rename to...
358 (svwhilelx_impl): ...this and inherit from while_comparison.
359 (svasrd): Use unspec_based_function.
360 (svmla_lane): Use svmla_lane_impl.
361 (svmls_lane): Use svmls_lane_impl.
362 (svrecpe, svrsqrte): Handle unsigned integer operations too.
363 (svwhilele, svwhilelt): Use svwhilelx_impl.
364 * config/aarch64/aarch64-sve-builtins-sve2.h: New file.
365 * config/aarch64/aarch64-sve-builtins-sve2.cc: Likewise.
366 * config/aarch64/aarch64-sve-builtins-sve2.def: Likewise.
367 * config/aarch64/aarch64-sve-builtins.def: Include
368 aarch64-sve-builtins-sve2.def.
370 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
372 * config/aarch64/aarch64-protos.h (aarch64_sve_arith_immediate_p)
373 (aarch64_sve_sqadd_sqsub_immediate_p): Add a machine_mode argument.
374 * config/aarch64/aarch64.c (aarch64_sve_arith_immediate_p)
375 (aarch64_sve_sqadd_sqsub_immediate_p): Likewise. Handle scalar
376 immediates as well as vector ones.
377 * config/aarch64/predicates.md (aarch64_sve_arith_immediate)
378 (aarch64_sve_sub_arith_immediate, aarch64_sve_qadd_immediate)
379 (aarch64_sve_qsub_immediate): Update calls accordingly.
381 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
383 * config/aarch64/aarch64-sve2.md: Add banner comments.
384 (<su>mulh<r>s<mode>3): Move further up file.
385 (<su>mull<bt><Vwide>, <r>shrnb<mode>, <r>shrnt<mode>)
386 (*aarch64_sve2_sra<mode>): Move further down file.
387 * config/aarch64/t-aarch64 (s-check-sve-md): Check aarch64-sve2.md too.
389 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
391 * config/aarch64/iterators.md (SVE_WHILE): Add UNSPEC_WHILERW
393 (while_optab_cmp): Handle them.
394 * config/aarch64/aarch64-sve.md
395 (*while_<while_optab_cmp><GPI:mode><PRED_ALL:mode>_ptest): Make public
396 and add a "@" marker.
397 * config/aarch64/aarch64-sve2.md (check_<raw_war>_ptrs<mode>): Use it
398 instead of gen_aarch64_sve2_while_ptest.
399 (@aarch64_sve2_while<cmp_op><GPI:mode><PRED_ALL:mode>_ptest): Delete.
401 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
403 * config/aarch64/aarch64.md (UNSPEC_WHILE_LE): Rename to...
404 (UNSPEC_WHILELE): ...this.
405 (UNSPEC_WHILE_LO): Rename to...
406 (UNSPEC_WHILELO): ...this.
407 (UNSPEC_WHILE_LS): Rename to...
408 (UNSPEC_WHILELS): ...this.
409 (UNSPEC_WHILE_LT): Rename to...
410 (UNSPEC_WHILELT): ...this.
411 * config/aarch64/iterators.md (SVE_WHILE): Update accordingly.
412 (cmp_op, while_optab_cmp): Likewise.
413 * config/aarch64/aarch64.c (aarch64_sve_move_pred_via_while): Likewise.
414 * config/aarch64/aarch64-sve-builtins-base.cc (svwhilele): Likewise.
415 (svwhilelt): Likewise.
417 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
419 * config/aarch64/aarch64-sve-builtins-shapes.h (unary_count): Delete.
420 (unary_to_uint): Define.
421 * config/aarch64/aarch64-sve-builtins-shapes.cc (unary_count_def)
422 (unary_count): Rename to...
423 (unary_to_uint_def, unary_to_uint): ...this.
424 * config/aarch64/aarch64-sve-builtins-base.def: Update accordingly.
426 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
428 * config/aarch64/aarch64-sve-builtins-functions.h
429 (code_for_mode_function): New class.
430 (CODE_FOR_MODE0, QUIET_CODE_FOR_MODE0): New macros.
431 * config/aarch64/aarch64-sve-builtins-base.cc (svcompact_impl)
432 (svext_impl, svmul_lane_impl, svsplice_impl, svtmad_impl): Delete.
433 (svcompact, svext, svsplice): Use QUIET_CODE_FOR_MODE0.
434 (svmul_lane, svtmad): Use CODE_FOR_MODE0.
436 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
438 * config/aarch64/iterators.md (addsub): New code attribute.
439 * config/aarch64/aarch64-simd.md (aarch64_<su_optab><optab><mode>):
441 (aarch64_<su_optab>q<addsub><mode>): ...this, making the same change
442 in the asm string and attributes. Fix indentation.
443 * config/aarch64/aarch64-sve.md (@aarch64_<su_optab><optab><mode>):
445 (@aarch64_sve_<optab><mode>): ...this.
446 * config/aarch64/aarch64-sve-builtins.h
447 (function_expander::expand_signed_unpred_op): Delete.
448 * config/aarch64/aarch64-sve-builtins.cc
449 (function_expander::expand_signed_unpred_op): Likewise.
450 (function_expander::map_to_rtx_codes): If the optab isn't defined,
451 try using code_for_aarch64_sve instead.
452 * config/aarch64/aarch64-sve-builtins-base.cc (svqadd_impl): Delete.
453 (svqsub_impl): Likewise.
454 (svqadd, svqsub): Use rtx_code_function instead.
456 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
458 * config/aarch64/iterators.md (SRHSUB, URHSUB): Delete.
459 (HADDSUB, sur, addsub): Remove them.
461 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
463 * tree-nrv.c (pass_return_slot::execute): Handle all internal
464 functions the same way, rather than singling out those that
465 aren't mapped directly to optabs.
467 2020-01-09 Richard Sandiford <richard.sandiford@arm.com>
469 * target.def (compatible_vector_types_p): New target hook.
470 * hooks.h (hook_bool_const_tree_const_tree_true): Declare.
471 * hooks.c (hook_bool_const_tree_const_tree_true): New function.
472 * doc/tm.texi.in (TARGET_COMPATIBLE_VECTOR_TYPES_P): New hook.
473 * doc/tm.texi: Regenerate.
474 * gimple-expr.c: Include target.h.
475 (useless_type_conversion_p): Use targetm.compatible_vector_types_p.
476 * config/aarch64/aarch64.c (aarch64_compatible_vector_types_p): New
478 (TARGET_COMPATIBLE_VECTOR_TYPES_P): Define.
479 * config/aarch64/aarch64-sve-builtins.cc (gimple_folder::convert_pred):
480 Use the original predicate if it already has a suitable type.
482 2020-01-09 Martin Jambor <mjambor@suse.cz>
484 * cgraph.h (cgraph_edge): Make remove, set_call_stmt, make_direct,
485 resolve_speculation and redirect_call_stmt_to_callee static. Change
486 return type of set_call_stmt to cgraph_edge *.
487 * auto-profile.c (afdo_indirect_call): Adjust call to
488 redirect_call_stmt_to_callee.
489 * cgraph.c (cgraph_edge::set_call_stmt): Make return cgraph-edge *,
490 make the this pointer explicit, adjust self-recursive calls and the
491 call top make_direct. Return the resulting edge.
492 (cgraph_edge::remove): Make this pointer explicit.
493 (cgraph_edge::resolve_speculation): Likewise, adjust call to remove.
494 (cgraph_edge::make_direct): Likewise, adjust call to
496 (cgraph_edge::redirect_call_stmt_to_callee): Likewise, also adjust
497 call to set_call_stmt.
498 (cgraph_update_edges_for_call_stmt_node): Update call to
499 set_call_stmt and remove.
500 * cgraphclones.c (cgraph_node::set_call_stmt_including_clones):
501 Renamed edge to master_edge. Adjusted calls to set_call_stmt.
502 (cgraph_node::create_edge_including_clones): Moved "first" definition
503 of edge to the block where it was used. Adjusted calls to
505 (cgraph_node::remove_symbol_and_inline_clones): Adjust call to
507 * cgraphunit.c (walk_polymorphic_call_targets): Adjusted calls to
508 make_direct and redirect_call_stmt_to_callee.
509 * ipa-fnsummary.c (redirect_to_unreachable): Adjust calls to
510 resolve_speculation and make_direct.
511 * ipa-inline-transform.c (inline_transform): Adjust call to
512 redirect_call_stmt_to_callee.
513 (check_speculations_1):: Adjust call to resolve_speculation.
514 * ipa-inline.c (resolve_noninline_speculation): Adjust call to
516 (inline_small_functions): Adjust call to resolve_speculation.
517 (ipa_inline): Likewise.
518 * ipa-prop.c (ipa_make_edge_direct_to_target): Adjust call to
520 * ipa-visibility.c (function_and_variable_visibility): Make iteration
521 safe with regards to edge removal, adjust calls to
522 redirect_call_stmt_to_callee.
523 * ipa.c (walk_polymorphic_call_targets): Adjust calls to make_direct
524 and redirect_call_stmt_to_callee.
525 * multiple_target.c (create_dispatcher_calls): Adjust call to
526 redirect_call_stmt_to_callee
527 (redirect_to_specific_clone): Likewise.
528 * tree-cfgcleanup.c (delete_unreachable_blocks_update_callgraph):
529 Adjust calls to cgraph_edge::remove.
530 * tree-inline.c (copy_bb): Adjust call to set_call_stmt.
531 (redirect_all_calls): Adjust call to redirect_call_stmt_to_callee.
532 (expand_call_inline): Adjust call to cgraph_edge::remove.
534 2020-01-09 Martin Liska <mliska@suse.cz>
536 * params.opt: Set Optimization for
537 param_max_speculative_devirt_maydefs.
539 2020-01-09 Martin Sebor <msebor@redhat.com>
543 * builtins.c (compute_objsize): Avoid handling MEM_REFs of vector type.
545 2020-01-09 Martin Liska <mliska@suse.cz>
547 * auto-profile.c (auto_profile): Use opt_for_fn
549 * ipa-cp.c (ipcp_lattice::add_value): Likewise.
550 (propagate_vals_across_arith_jfunc): Likewise.
551 (hint_time_bonus): Likewise.
552 (incorporate_penalties): Likewise.
553 (good_cloning_opportunity_p): Likewise.
554 (perform_estimation_of_a_value): Likewise.
555 (estimate_local_effects): Likewise.
556 (ipcp_propagate_stage): Likewise.
557 * ipa-fnsummary.c (decompose_param_expr): Likewise.
558 (set_switch_stmt_execution_predicate): Likewise.
559 (analyze_function_body): Likewise.
560 * ipa-inline-analysis.c (offline_size): Likewise.
561 * ipa-inline.c (early_inliner): Likewise.
562 * ipa-prop.c (ipa_analyze_node): Likewise.
563 (ipcp_transform_function): Likewise.
564 * ipa-sra.c (process_scan_results): Likewise.
565 (ipa_sra_summarize_function): Likewise.
566 * params.opt: Rename ipcp-unit-growth to
567 ipa-cp-unit-growth. Add Optimization for various
568 IPA-related parameters.
570 2020-01-09 Richard Biener <rguenther@suse.de>
573 * gimplify.c (gimplify_expr): Deal with NOP definitions.
575 2020-01-09 Richard Biener <rguenther@suse.de>
577 PR tree-optimization/93040
578 * gimple-ssa-store-merging.c (find_bswap_or_nop): Raise search limit.
580 2020-01-09 Georg-Johann Lay <avr@gjlay.de>
582 * common/config/avr/avr-common.c (avr_option_optimization_table)
583 [OPT_LEVELS_1_PLUS]: Set -fsplit-wide-types-early.
585 2020-01-09 Martin Liska <mliska@suse.cz>
587 * cgraphclones.c (symbol_table::materialize_all_clones):
588 Use cgraph_node::dump_name.
590 2020-01-09 Jakub Jelinek <jakub@redhat.com>
593 * config/riscv/riscv.c (riscv_print_operand_reloc): Use
594 output_operand_lossage instead of gcc_unreachable.
595 * doc/md.texi (riscv f constraint): Fix typo.
598 * config/i386/i386.md (subv<mode>4): Use SWIDWI iterator instead of
599 SWI. Use <general_hilo_operand> instead of <general_operand>. Use
600 CONST_SCALAR_INT_P instead of CONST_INT_P.
601 (*subv<mode>4_1): Rename to ...
602 (subv<mode>4_1): ... this.
603 (*subv<dwi>4_doubleword, *addv<dwi>4_doubleword_1): New
604 define_insn_and_split patterns.
605 (*subv<mode>4_overflow_1, *addv<mode>4_overflow_2): New define_insn
608 2020-01-08 David Malcolm <dmalcolm@redhat.com>
610 * vec.c (class selftest::count_dtor): New class.
611 (selftest::test_auto_delete_vec): New test.
612 (selftest::vec_c_tests): Call it.
613 * vec.h (class auto_delete_vec): New class template.
614 (auto_delete_vec<T>::~auto_delete_vec): New dtor.
616 2020-01-08 David Malcolm <dmalcolm@redhat.com>
618 * sbitmap.h (auto_sbitmap): Add operator const_sbitmap.
620 2020-01-08 Jim Wilson <jimw@sifive.com>
622 * config/riscv/riscv.c (riscv_legitimize_tls_address): Ifdef out
623 use of TLS_MODEL_LOCAL_EXEC when not pic.
625 2020-01-08 David Malcolm <dmalcolm@redhat.com>
627 * hash-map-tests.c (selftest::test_map_of_strings_to_int): Fix
630 2020-01-08 Jakub Jelinek <jakub@redhat.com>
633 * config/i386/i386.md (*stack_protect_set_2_<mode> peephole2,
634 *stack_protect_set_3 peephole2): Also check that the second
635 insns source is general_operand.
638 * config/i386/i386.md (addcarry<mode>_0): Use nonimmediate_operand
639 predicate for output operand instead of register_operand.
640 (addcarry<mode>, addcarry<mode>_1): Likewise. Add alternative with
641 memory destination and non-memory operands[2].
643 2020-01-08 Martin Liska <mliska@suse.cz>
645 * cgraph.c (cgraph_node::dump): Use ::dump_name or
646 ::dump_asm_name instead of (::name or ::asm_name).
647 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
648 * cgraphunit.c (walk_polymorphic_call_targets): Likewise.
649 (analyze_functions): Likewise.
650 (expand_all_functions): Likewise.
651 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
652 (propagate_bits_across_jump_function): Likewise.
653 (dump_profile_updates): Likewise.
654 (ipcp_store_bits_results): Likewise.
655 (ipcp_store_vr_results): Likewise.
656 * ipa-devirt.c (dump_targets): Likewise.
657 * ipa-fnsummary.c (analyze_function_body): Likewise.
658 * ipa-hsa.c (check_warn_node_versionable): Likewise.
659 (process_hsa_functions): Likewise.
660 * ipa-icf.c (sem_item_optimizer::merge_classes): Likewise.
661 (set_alias_uids): Likewise.
662 * ipa-inline-transform.c (save_inline_function_body): Likewise.
663 * ipa-inline.c (recursive_inlining): Likewise.
664 (inline_to_all_callers_1): Likewise.
665 (ipa_inline): Likewise.
666 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
667 (ipa_propagate_frequency): Likewise.
668 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
669 (remove_described_reference): Likewise.
670 * ipa-pure-const.c (worse_state): Likewise.
671 (check_retval_uses): Likewise.
672 (analyze_function): Likewise.
673 (propagate_pure_const): Likewise.
674 (propagate_nothrow): Likewise.
675 (dump_malloc_lattice): Likewise.
676 (propagate_malloc): Likewise.
677 (pass_local_pure_const::execute): Likewise.
678 * ipa-visibility.c (optimize_weakref): Likewise.
679 (function_and_variable_visibility): Likewise.
680 * ipa.c (symbol_table::remove_unreachable_nodes): Likewise.
681 (ipa_discover_variable_flags): Likewise.
682 * lto-streamer-out.c (output_function): Likewise.
683 (output_constructor): Likewise.
684 * tree-inline.c (copy_bb): Likewise.
685 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
686 * varpool.c (symbol_table::remove_unreferenced_decls): Likewise.
688 2020-01-08 Richard Biener <rguenther@suse.de>
691 * tree-eh.c (sink_clobbers): Update virtual operands for
692 the first and last stmt only. Add a dry-run capability.
693 (pass_lower_eh_dispatch::execute): Perform clobber sinking
694 after CFG manipulations and in RPO order to catch all
695 secondary opportunities reliably.
697 2020-01-08 Georg-Johann Lay <avr@gjlay.de>
700 * doc/invoke.texi (AVR Options) <-nodevicespecs>: Document.
702 2019-01-08 Richard Biener <rguenther@suse.de>
705 * gimple-fold.c (rewrite_to_defined_overflow): Mark stmt modified.
706 * tree-ssa-loop-im.c (move_computations_worker): Properly adjust
707 virtual operand, also updating SSA use.
708 * gimple-loop-interchange.cc (loop_cand::undo_simple_reduction):
709 Update stmt after resetting virtual operand.
710 (tree_loop_interchange::move_code_to_inner_loop): Likewise.
711 * gimple-iterator.c (gsi_remove): When not removing the stmt
712 permanently do not delink immediate uses or mark the stmt modified.
714 2020-01-08 Martin Liska <mliska@suse.cz>
716 * ipa-fnsummary.c (dump_ipa_call_summary): Use symtab_node::dump_name.
717 (ipa_call_context::estimate_size_and_time): Likewise.
718 (inline_analyze_function): Likewise.
720 2020-01-08 Martin Liska <mliska@suse.cz>
722 * cgraph.c (cgraph_node::dump): Use systematically
725 2020-01-08 Georg-Johann Lay <avr@gjlay.de>
727 Add -nodevicespecs option for avr.
730 * config/avr/avr.opt (-nodevicespecs): New driver option.
731 * config/avr/driver-avr.c (avr_devicespecs_file): Only issue
732 "-specs=device-specs/..." if that option is not set.
733 * doc/invoke.texi (AVR Options) <-nodevicespecs>: Document.
735 2020-01-08 Georg-Johann Lay <avr@gjlay.de>
737 Implement 64-bit double functions for avr.
740 * config.gcc (tm_defines) [target=avr]: Support --with-libf7,
741 --with-double-comparison.
742 * doc/install.texi: Document them.
743 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
744 <WITH_LIBF7_LIBGCC, WITH_LIBF7_MATH, WITH_LIBF7_MATH_SYMBOLS>
745 <WITH_DOUBLE_COMPARISON>: New built-in defines.
746 * doc/invoke.texi (AVR Built-in Macros): Document them.
747 * config/avr/avr-protos.h (avr_float_lib_compare_returns_bool): New.
748 * config/avr/avr.c (avr_float_lib_compare_returns_bool): New function.
749 * config/avr/avr.h (FLOAT_LIB_COMPARE_RETURNS_BOOL): New macro.
751 2020-01-08 Richard Earnshaw <rearnsha@arm.com>
754 * config/arm/t-multilib (MULTILIB_MATCHES): Add rules to match
755 armv7-a{+mp,+sec,+mp+sec} to appropriate armv7 multilib variants
756 when only building rm-profile multilibs.
758 2020-01-08 Feng Xue <fxue@os.amperecomputing.com>
761 * ipa-cp.c (self_recursively_generated_p): Find matched aggregate
762 lattice for a value to check.
763 (propagate_vals_across_arith_jfunc): Add an assertion to ensure
764 finite propagation in self-recursive scc.
766 2020-01-08 Luo Xiong Hu <luoxhu@linux.ibm.com>
768 * ipa-inline.c (caller_growth_limits): Restore the AND.
770 2020-01-07 Andrew Stubbs <ams@codesourcery.com>
772 * config/gcn/gcn-valu.md (VEC_1REG_INT_ALT): Delete iterator.
773 (VEC_ALLREG_ALT): New iterator.
774 (VEC_ALLREG_INT_MODE): New iterator.
775 (VCMP_MODE): New iterator.
776 (VCMP_MODE_INT): New iterator.
777 (vec_cmpu<mode>di): Use VCMP_MODE_INT.
778 (vec_cmp<u>v64qidi): New define_expand.
779 (vec_cmp<mode>di_exec): Use VCMP_MODE.
780 (vec_cmpu<mode>di_exec): New define_expand.
781 (vec_cmp<u>v64qidi_exec): New define_expand.
782 (vec_cmp<mode>di_dup): Use VCMP_MODE.
783 (vec_cmp<mode>di_dup_exec): Use VCMP_MODE.
784 (vcond<VEC_ALL1REG_MODE:mode><VEC_1REG_ALT:mode>): Rename ...
785 (vcond<VEC_ALLREG_MODE:mode><VEC_ALLREG_ALT:mode>): ... to this.
786 (vcond<VEC_ALL1REG_MODE:mode><VEC_1REG_ALT:mode>_exec): Rename ...
787 (vcond<VEC_ALLREG_MODE:mode><VEC_ALLREG_ALT:mode>_exec): ... to this.
788 (vcondu<VEC_ALL1REG_MODE:mode><VEC_1REG_INT_ALT:mode>): Rename ...
789 (vcondu<VEC_ALLREG_MODE:mode><VEC_ALLREG_INT_MODE:mode>): ... to this.
790 (vcondu<VEC_ALL1REG_MODE:mode><VEC_1REG_INT_ALT:mode>_exec): Rename ...
791 (vcondu<VEC_ALLREG_MODE:mode><VEC_ALLREG_INT_MODE:mode>_exec): ... to
793 * config/gcn/gcn.c (print_operand): Fix 8 and 16 bit suffixes.
794 * config/gcn/gcn.md (expander): Add sign_extend and zero_extend.
796 2020-01-07 Andrew Stubbs <ams@codesourcery.com>
798 * config/gcn/constraints.md (DA): Update description and match.
800 (Db): New constraint.
801 * config/gcn/gcn-protos.h (gcn_inline_constant64_p): Add second
803 * config/gcn/gcn.c (gcn_inline_constant64_p): Add 'mixed' parameter.
804 Implement 'Db' mixed immediate type.
805 * config/gcn/gcn-valu.md (addcv64si3<exec_vcc>): Rework constraints.
806 (addcv64si3_dup<exec_vcc>): Delete.
807 (subcv64si3<exec_vcc>): Rework constraints.
808 (addv64di3): Rework constraints.
809 (addv64di3_exec): Rework constraints.
810 (subv64di3): Rework constraints.
811 (addv64di3_dup): Delete.
812 (addv64di3_dup_exec): Delete.
813 (addv64di3_zext): Rework constraints.
814 (addv64di3_zext_exec): Rework constraints.
815 (addv64di3_zext_dup): Rework constraints.
816 (addv64di3_zext_dup_exec): Rework constraints.
817 (addv64di3_zext_dup2): Rework constraints.
818 (addv64di3_zext_dup2_exec): Rework constraints.
819 (addv64di3_sext_dup2): Rework constraints.
820 (addv64di3_sext_dup2_exec): Rework constraints.
822 2020-01-07 Andre Vieira <andre.simoesdiasvieira@arm.com>
824 * doc/sourcebuild.texi (arm_little_endian, arm_nothumb): Documented
825 existing target checks.
827 2020-01-07 Richard Biener <rguenther@suse.de>
829 * doc/install.texi: Bump minimal supported MPC version.
831 2020-01-07 Richard Sandiford <richard.sandiford@arm.com>
833 * langhooks-def.h (lhd_simulate_enum_decl): Declare.
834 (LANG_HOOKS_SIMULATE_ENUM_DECL): Use it.
835 * langhooks.c: Include stor-layout.h.
836 (lhd_simulate_enum_decl): New function.
837 * config/aarch64/aarch64-sve-builtins.cc (init_builtins): Call
838 handle_arm_sve_h for the LTO frontend.
839 (register_vector_type): Cope with null returns from pushdecl.
841 2020-01-07 Richard Sandiford <richard.sandiford@arm.com>
843 * config/aarch64/aarch64-protos.h (aarch64_sve::svbool_type_p)
844 (aarch64_sve::nvectors_if_data_type): Replace with...
845 (aarch64_sve::builtin_type_p): ...this.
846 * config/aarch64/aarch64-sve-builtins.cc: Include attribs.h.
847 (find_vector_type): Delete.
848 (add_sve_type_attribute): New function.
849 (lookup_sve_type_attribute): Likewise.
850 (register_builtin_types): Add an "SVE type" attribute to each type.
851 (register_tuple_type): Likewise.
852 (svbool_type_p, nvectors_if_data_type): Delete.
853 (mangle_builtin_type): Use lookup_sve_type_attribute.
854 (builtin_type_p): Likewise. Add an overload that returns the
855 number of constituent vector and predicate registers.
856 * config/aarch64/aarch64.c (aarch64_sve_argument_p): Delete.
857 (aarch64_returns_value_in_sve_regs_p): Use aarch64_sve::builtin_type_p
858 instead of aarch64_sve_argument_p.
859 (aarch64_takes_arguments_in_sve_regs_p): Likewise.
860 (aarch64_pass_by_reference): Likewise.
861 (aarch64_function_value_1): Likewise.
862 (aarch64_return_in_memory): Likewise.
863 (aarch64_layout_arg): Likewise.
865 2020-01-07 Jakub Jelinek <jakub@redhat.com>
867 PR tree-optimization/93156
868 * tree-ssa-ccp.c (bit_value_binop): For x * x note that the second
869 least significant bit is always clear.
871 PR tree-optimization/93118
872 * match.pd ((x >> c) << c -> x & (-1<<c)): Add nop_convert?. Add new
873 simplifier with two intermediate conversions.
875 2020-01-07 Martin Liska <mliska@suse.cz>
877 * params.opt: Add Optimization for various parameters.
879 2020-01-07 Martin Liska <mliska@suse.cz>
882 * doc/extend.texi: Explain cloning for target_clone
885 2020-01-07 Martin Liska <mliska@suse.cz>
887 PR tree-optimization/92860
888 * common.opt: Make in Optimization option
889 as it is affected by -O0, which is an Optimization
891 * tree-inline.c (tree_inlinable_function_p):
892 Use opt_for_fn for warn_inline.
893 (expand_call_inline): Likewise.
895 2020-01-07 Martin Liska <mliska@suse.cz>
897 PR tree-optimization/92860
898 * common.opt: Make flag_ree as optimization
901 2020-01-07 Martin Liska <mliska@suse.cz>
903 PR optimization/92860
904 * params.opt: Mark param_min_crossjump_insns with Optimization
907 2020-01-07 Luo Xiong Hu <luoxhu@linux.ibm.com>
909 * ipa-inline-analysis.c (estimate_growth): Fix typo.
910 * ipa-inline.c (caller_growth_limits): Use OR instead of AND.
912 2020-01-06 Michael Meissner <meissner@linux.ibm.com>
914 * config/rs6000/rs6000.c (hard_reg_and_mode_to_addr_mask): New
915 helper function to return the valid addressing formats for a given
916 hard register and mode.
917 (rs6000_adjust_vec_address): Call hard_reg_and_mode_to_addr_mask.
919 * config/rs6000/constraints.md (Q constraint): Update
921 * doc/md.texi (RS/6000 constraints): Update 'Q' cosntraint
924 * config/rs6000/vsx.md (vsx_extract_<mode>_var, VSX_D iterator):
925 Use 'Q' for doing vector extract from memory.
926 (vsx_extract_v4sf_var): Use 'Q' for doing vector extract from
928 (vsx_extract_<mode>_var, VSX_EXTRACT_I iterator): Use 'Q' for
929 doing vector extract from memory.
930 (vsx_extract_<mode>_<VS_scalar>mode_var): Use 'Q' for doing vector
933 * config/rs6000/rs6000.c (rs6000_adjust_vec_address): Add support
934 for the offset being 34-bits when -mcpu=future is used.
936 2020-01-06 John David Anglin <danglin@gcc.gnu.org>
938 * config/pa/pa.md: Revert change to use ordered_comparison_operator
939 instead of cmpib_comparison_operator in cmpib patterns.
940 * config/pa/predicates.md (cmpib_comparison_operator): Revert removal
941 of cmpib_comparison_operator. Revise comment.
943 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
945 * tree-vect-slp.c (vect_build_slp_tree_1): Require all shifts
946 in an IFN_DIV_POW2 node to be equal.
948 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
950 * tree-vect-stmts.c (vect_check_load_store_mask): Rename to...
951 (vect_check_scalar_mask): ...this.
952 (vectorizable_store, vectorizable_load): Update call accordingly.
953 (vectorizable_call): Use vect_check_scalar_mask to check the mask
954 argument in calls to conditional internal functions.
956 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
958 * config/gcn/gcn-valu.md (subv64di3): Use separate alternatives for
960 (subv64di3_exec): Likewise.
962 2020-01-06 Bryan Stenson <bryan@siliconvortex.com>
964 * config/mips/mips.c (vr4130_align_insns): Fix typo.
965 * doc/md.texi (movstr): Likewise.
967 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
969 * config/gcn/gcn-valu.md (vec_extract<mode><scalar_mode>): Add early
972 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
974 * config/aarch64/t-aarch64 ($(srcdir)/config/aarch64/aarch64-tune.md):
976 (s-aarch64-tune-md): ...this new stamp file. Pipe the new contents
977 to a temporary file and use move-if-change to update the real
978 file where necessary.
980 2020-01-06 Richard Sandiford <richard.sandiford@arm.com>
982 * config/aarch64/aarch64-sve.md (@aarch64_sel_dup<mode>): Use Upl
983 rather than Upa for CPY /M.
985 2020-01-06 Andrew Stubbs <ams@codesourcery.com>
987 * config/gcn/gcn.c (gcn_inline_constant_p): Allow 64 as an inline
990 2020-01-06 Martin Liska <mliska@suse.cz>
992 PR tree-optimization/92860
993 * params.opt: Mark param_max_combine_insns with Optimization
996 2020-01-05 Jakub Jelinek <jakub@redhat.com>
999 * config/i386/i386.md (SWIDWI): New mode iterator.
1000 (DWI, dwi): Add TImode variants.
1001 (addv<mode>4): Use SWIDWI iterator instead of SWI. Use
1002 <general_hilo_operand> instead of <general_operand>. Use
1003 CONST_SCALAR_INT_P instead of CONST_INT_P.
1004 (*addv<mode>4_1): Rename to ...
1005 (addv<mode>4_1): ... this.
1006 (QWI): New mode attribute.
1007 (*addv<dwi>4_doubleword, *addv<dwi>4_doubleword_1): New
1008 define_insn_and_split patterns.
1009 (*addv<mode>4_overflow_1, *addv<mode>4_overflow_2): New define_insn
1011 (uaddv<mode>4): Use SWIDWI iterator instead of SWI. Use
1012 <general_hilo_operand> instead of <general_operand>.
1013 (*addcarry<mode>_1): New define_insn.
1014 (*add<dwi>3_doubleword_cc_overflow_1): New define_insn_and_split.
1016 2020-01-03 Konstantin Kharlamov <Hi-Angel@yandex.ru>
1018 * gdbinit.in (pr, prl, pt, pct, pgg, pgq, pgs, pge, pmz, pdd, pbs, pbm):
1019 Use "call" instead of "set".
1021 2020-01-03 Martin Jambor <mjambor@suse.cz>
1024 * ipa-cp.c (print_all_lattices): Skip functions without info.
1026 2020-01-03 Jakub Jelinek <jakub@redhat.com>
1029 * config/i386/i386-options.c (ix86_simd_clone_adjust): If
1030 TARGET_PREFER_AVX128, use prefer-vector-width=256 for 'c' and 'd'
1031 simd clones. If TARGET_PREFER_AVX256, use prefer-vector-width=512
1032 for 'e' simd clones.
1035 * config/i386/i386.opt (x_prefer_vector_width_type): Remove TargetSave
1037 (mprefer-vector-width=): Add Save.
1038 * config/i386/i386-options.c (ix86_target_string): Add PVW argument, print
1039 -mprefer-vector-width= if non-zero. Fix up -mfpmath= comment.
1040 (ix86_debug_options, ix86_function_specific_print): Adjust
1041 ix86_target_string callers.
1042 (ix86_valid_target_attribute_inner_p): Handle prefer-vector-width=.
1043 (ix86_valid_target_attribute_tree): Likewise.
1044 * config/i386/i386-options.h (ix86_target_string): Add PVW argument.
1045 * config/i386/i386-expand.c (ix86_expand_builtin): Adjust
1046 ix86_target_string caller.
1049 * config/i386/i386.md (abs<mode>2): Use expand_simple_binop instead of
1050 emitting ASHIFTRT, XOR and MINUS by hand. Use gen_int_mode with QImode
1051 instead of gen_int_shift_amount + convert_modes.
1053 PR rtl-optimization/93088
1054 * loop-iv.c (find_single_def_src): Punt after looking through
1055 128 reg copies for regs with single definitions. Move definitions
1058 2020-01-02 Dennis Zhang <dennis.zhang@arm.com>
1060 * config/arm/arm-c.c (arm_cpu_builtins): Define
1061 __ARM_FEATURE_MATMUL_INT8, __ARM_FEATURE_BF16_VECTOR_ARITHMETIC,
1062 __ARM_FEATURE_BF16_SCALAR_ARITHMETIC, and
1063 __ARM_BF16_FORMAT_ALTERNATIVE when enabled.
1064 * config/arm/arm-cpus.in (armv8_6, i8mm, bf16): New features.
1065 * config/arm/arm-tables.opt: Regenerated.
1066 * config/arm/arm.c (arm_option_reconfigure_globals): Initialize
1067 arm_arch_i8mm and arm_arch_bf16 when enabled.
1068 * config/arm/arm.h (TARGET_I8MM): New macro.
1069 (TARGET_BF16_FP, TARGET_BF16_SIMD): Likewise.
1070 * config/arm/t-aprofile: Add matching rules for -march=armv8.6-a.
1071 * config/arm/t-arm-elf (all_v8_archs): Add armv8.6-a.
1072 * config/arm/t-multilib: Add matching rules for -march=armv8.6-a.
1073 (v8_6_a_simd_variants): New.
1074 (v8_*_a_simd_variants): Add i8mm and bf16.
1075 * doc/invoke.texi (armv8.6-a, i8mm, bf16): Document new options.
1077 2020-01-02 Jakub Jelinek <jakub@redhat.com>
1080 * predict.c (compute_function_frequency): Don't call
1081 warn_function_cold on functions that already have cold attribute.
1083 2020-01-01 John David Anglin <danglin@gcc.gnu.org>
1086 * config/pa/pa.c (pa_elf_select_rtx_section): New. Put references to
1087 COMDAT group function labels in .data.rel.ro.local section.
1088 * config/pa/pa32-linux.h (TARGET_ASM_SELECT_RTX_SECTION): Define.
1091 * config/pa/pa.md (scc): Use ordered_comparison_operator instead of
1092 comparison_operator in B and S integer comparisons. Likewise, use
1093 ordered_comparison_operator instead of cmpib_comparison_operator in
1095 * config/pa/predicates.md (cmpib_comparison_operator): Remove.
1097 2020-01-01 Jakub Jelinek <jakub@redhat.com>
1099 Update copyright years.
1101 * gcc.c (process_command): Update copyright notice dates.
1102 * gcov-dump.c (print_version): Ditto.
1103 * gcov.c (print_version): Ditto.
1104 * gcov-tool.c (print_version): Ditto.
1105 * gengtype.c (create_file): Ditto.
1106 * doc/cpp.texi: Bump @copying's copyright year.
1107 * doc/cppinternals.texi: Ditto.
1108 * doc/gcc.texi: Ditto.
1109 * doc/gccint.texi: Ditto.
1110 * doc/gcov.texi: Ditto.
1111 * doc/install.texi: Ditto.
1112 * doc/invoke.texi: Ditto.
1114 2020-01-01 Jan Hubicka <hubicka@ucw.cz>
1116 * ipa.c (walk_polymorphic_call_targets): Fix updating of overall
1119 2020-01-01 Jakub Jelinek <jakub@redhat.com>
1121 PR tree-optimization/93098
1122 * match.pd (popcount): For shift amounts, use integer_onep
1123 or wi::to_widest () == cst instead of tree_to_uhwi () == cst
1124 tests. Make sure that precision is power of two larger than or equal
1125 to 16. Ensure shift is never negative. Use HOST_WIDE_INT_UC macro
1126 instead of ULL suffixed constants. Formatting fixes.
1128 Copyright (C) 2020 Free Software Foundation, Inc.
1130 Copying and distribution of this file, with or without modification,
1131 are permitted in any medium without royalty provided the copyright
1132 notice and this notice are preserved.