1 2024-02-04 Takayuki 'January June' Suwa <jjsuwa_sys3175@yahoo.co.jp>
3 * config/xtensa/xtensa.md (*eqne_zero_masked_bits):
4 Add missing ":SI" to the match_operator.
6 2024-02-04 Takayuki 'January June' Suwa <jjsuwa_sys3175@yahoo.co.jp>
8 * config/xtensa/xtensa.md (SHI): New mode iterator.
9 (2 split patterns related to constsynth):
10 Change to also accept HImode operands.
12 2024-02-04 Jeff Law <jlaw@ventanamicro.com>
14 * config/riscv/riscv.cc (riscv_rtx_costs): Handle SUBREG and REG
17 2024-02-04 Xi Ruoyao <xry111@xry111.site>
19 * config/loongarch/lsx.md (neg<mode:FLSX>2): Remove the
21 * config/loongarch/simd.md (simdfmt_as_i): New define_mode_attr.
22 (elmsgnbit): Likewise.
23 (neg<mode:FVEC>2): New define_insn.
24 * config/loongarch/lasx.md (negv4df2, negv8sf2): Remove as they
25 are now instantiated in simd.md.
27 2024-02-04 Xi Ruoyao <xry111@xry111.site>
29 * config/loongarch/loongarch.cc (loongarch_symbol_insns): Do not
30 use LSX_SUPPORTED_MODE_P or LASX_SUPPORTED_MODE_P if mode is
33 2024-02-04 Li Wei <liwei@loongson.cn>
35 * config/loongarch/loongarch.cc (loongarch_expand_vselect): Adjust.
36 (loongarch_expand_vselect_vconcat): Ditto.
37 (loongarch_try_expand_lsx_vshuf_const): New, use vshuf to implement
38 all 128-bit constant permutation situations.
39 (loongarch_expand_lsx_shuffle): Adjust and rename function name.
40 (loongarch_is_imm_set_shuffle): Renamed function name.
41 (loongarch_expand_vec_perm_even_odd): Function forward declaration.
42 (loongarch_expand_vec_perm_even_odd_1): Add implement for 128-bit
43 extract-even and extract-odd permutations.
44 (loongarch_is_odd_extraction): Delete.
45 (loongarch_is_even_extraction): Ditto.
46 (loongarch_expand_vec_perm_const): Adjust.
48 2024-02-03 Jakub Jelinek <jakub@redhat.com>
51 * wide-int.cc (wi::bswap_large): Rename third argument from
52 len to xlen and adjust use in safe_uhwi. Add len variable, set
53 it to BLOCKS_NEEDED (precision) and use it for clearing of val
54 and as canonize argument. Clear val using memset instead of
57 2024-02-03 Jakub Jelinek <jakub@redhat.com>
59 * ggc-common.cc (gt_pch_save): Allow addr to be equal to
60 mmi.preferred_base + mmi.size - sizeof (void *).
62 2024-02-03 Xi Ruoyao <xry111@xry111.site>
64 * config/loongarch/loongarch-def.h (abi_minimal_isa): Declare.
65 * config/loongarch/loongarch-opts.cc (abi_minimal_isa): Remove
66 the ODR-violating locale declaration.
68 2024-02-02 Tamar Christina <tamar.christina@arm.com>
70 PR tree-optimization/113588
71 PR tree-optimization/113467
72 * tree-vect-data-refs.cc
73 (vect_analyze_data_ref_dependence): Choose correct dest and fix checks.
74 (vect_analyze_early_break_dependences): Update comments.
76 2024-02-02 John David Anglin <danglin@gcc.gnu.org>
79 * config/pa/pa.cc (enum pa_builtins): Add PA_BUILTIN_GET_FPSR
80 and PA_BUILTIN_SET_FPSR builtins.
81 * (pa_builtins_icode): Declare.
82 * (def_builtin, pa_fpu_init_builtins): New.
83 * (pa_init_builtins): Initialize FPU builtins.
84 * (pa_builtin_decl, pa_expand_builtin_1): New.
85 * (pa_expand_builtin): Handle PA_BUILTIN_GET_FPSR and
86 PA_BUILTIN_SET_FPSR builtins.
87 * (pa_atomic_assign_expand_fenv): New.
88 * config/pa/pa.md (UNSPECV_GET_FPSR, UNSPECV_SET_FPSR): New
90 (get_fpsr, put_fpsr): New expanders.
91 (get_fpsr_32, get_fpsr_64, set_fpsr_32, set_fpsr_64): New
94 2024-02-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
97 * config/riscv/riscv-v.cc (expand_reduction): Pass VLMAX avl to scalar move.
99 2024-02-02 Jonathan Wakely <jwakely@redhat.com>
101 * doc/extend.texi (Common Type Attributes): Fix typo in
102 description of hardbool.
104 2024-02-02 Jakub Jelinek <jakub@redhat.com>
106 PR tree-optimization/113692
107 * gimple-lower-bitint.cc (bitint_large_huge::lower_stmt): Handle casts
108 from large/huge BITINT_TYPEs to POINTER_TYPE/REFERENCE_TYPE as
111 2024-02-02 Jakub Jelinek <jakub@redhat.com>
114 * gimple-lower-bitint.cc (bitint_large_huge::lower_asm): Handle
115 uninitialized large/huge _BitInt SSA_NAME inputs.
117 2024-02-02 Jakub Jelinek <jakub@redhat.com>
120 * tree-ssa-math-opts.cc (is_widening_mult_rhs_p): Use wide_int_from
121 around wi::to_wide in order to compare value in prec precision.
123 2024-02-02 Lehua Ding <lehua.ding@rivai.ai>
126 2024-02-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
128 * config/riscv/riscv.cc (riscv_legitimize_move): Fix poly_int dest generation.
130 2024-02-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
132 * config/riscv/riscv.cc (riscv_legitimize_move): Fix poly_int dest generation.
134 2024-02-02 Pan Li <pan2.li@intel.com>
136 * config/riscv/riscv.cc (riscv_get_arg_info): Cleanup comments.
137 (riscv_pass_by_reference): Ditto.
138 (riscv_fntype_abi): Ditto.
140 2024-02-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
142 * config/riscv/riscv-vsetvl.cc (vsetvl_pre_insn_p): New function.
143 (pre_vsetvl::cleaup): Remove vsetvl_pre.
144 (pre_vsetvl::remove_vsetvl_pre_insns): New function.
146 2024-02-02 Jiahao Xu <xujiahao@loongson.cn>
148 * config/loongarch/larchintrin.h
149 (__frecipe_s): Update function return type.
150 (__frecipe_d): Ditto.
151 (__frsqrte_s): Ditto.
152 (__frsqrte_d): Ditto.
154 2024-02-02 Li Wei <liwei@loongson.cn>
156 * config/loongarch/loongarch.cc (loongarch_multiply_add_p): New.
157 (loongarch_vector_costs::add_stmt_cost): Adjust.
159 2024-02-02 Xi Ruoyao <xry111@xry111.site>
161 * config/loongarch/loongarch.md (unspec): Add
162 UNSPEC_LA_PCREL_64_PART1 and UNSPEC_LA_PCREL_64_PART2.
163 (la_pcrel64_two_parts): New define_insn.
164 * config/loongarch/loongarch.cc (loongarch_tls_symbol): Fix a
166 (loongarch_call_tls_get_addr): If -mcmodel=extreme
167 -mexplicit-relocs={always,auto}, use la_pcrel64_two_parts for
168 addressing the TLS symbol and __tls_get_addr. Emit an REG_EQUAL
169 note to allow CSE addressing __tls_get_addr.
170 (loongarch_legitimize_tls_address): If -mcmodel=extreme
171 -mexplicit-relocs={always,auto}, address TLS IE symbols with
172 la_pcrel64_two_parts.
173 (loongarch_split_symbol): If -mcmodel=extreme
174 -mexplicit-relocs={always,auto}, address symbols with
175 la_pcrel64_two_parts.
176 (loongarch_output_mi_thunk): Clean up unreachable code. If
177 -mcmodel=extreme -mexplicit-relocs={always,auto}, address the MI
178 thunks with la_pcrel64_two_parts.
180 2024-02-02 Lulu Cheng <chenglulu@loongson.cn>
182 * config/loongarch/loongarch.cc (loongarch_call_tls_get_addr):
183 Add support for call36.
185 2024-02-02 Lulu Cheng <chenglulu@loongson.cn>
187 * config/loongarch/loongarch.cc (loongarch_explicit_relocs_p):
188 When the code model of the symbol is extreme and -mexplicit-relocs=auto,
189 the macro instruction loading symbol address is not applicable.
190 (loongarch_call_tls_get_addr): Adjust code.
191 (loongarch_legitimize_tls_address): Likewise.
193 2024-02-02 Lulu Cheng <chenglulu@loongson.cn>
195 * config/loongarch/loongarch-protos.h (loongarch_symbol_extreme_p):
196 Add function declaration.
197 * config/loongarch/loongarch.cc (loongarch_symbolic_constant_p):
198 For SYMBOL_PCREL64, non-zero addend of "la.local $rd,$rt,sym+addend"
200 (loongarch_load_tls): Added macro support in extreme mode.
201 (loongarch_call_tls_get_addr): Likewise.
202 (loongarch_legitimize_tls_address): Likewise.
203 (loongarch_force_address): Likewise.
204 (loongarch_legitimize_move): Likewise.
205 (loongarch_output_mi_thunk): Likewise.
206 (loongarch_option_override_internal): Remove the code that detects
207 explicit relocs status.
208 (loongarch_handle_model_attribute): Likewise.
209 * config/loongarch/loongarch.md (movdi_symbolic_off64): New template.
210 * config/loongarch/predicates.md (symbolic_off64_operand): New predicate.
211 (symbolic_off64_or_reg_operand): Likewise.
213 2024-02-02 Lulu Cheng <chenglulu@loongson.cn>
215 * config/loongarch/loongarch.cc (loongarch_load_tls):
216 Load all types of tls symbols through one function.
217 (loongarch_got_load_tls_gd): Delete.
218 (loongarch_got_load_tls_ld): Delete.
219 (loongarch_got_load_tls_ie): Delete.
220 (loongarch_got_load_tls_le): Delete.
221 (loongarch_call_tls_get_addr): Modify the called function name.
222 (loongarch_legitimize_tls_address): Likewise.
223 * config/loongarch/loongarch.md (@got_load_tls_gd<mode>): Delete.
224 (@load_tls<mode>): New template.
225 (@got_load_tls_ld<mode>): Delete.
226 (@got_load_tls_le<mode>): Delete.
227 (@got_load_tls_ie<mode>): Delete.
229 2024-02-02 Lulu Cheng <chenglulu@loongson.cn>
231 * config/loongarch/loongarch.cc (mem_shadd_or_shadd_rtx_p): New function.
232 (loongarch_legitimize_address): Add logical transformation code.
234 2024-02-01 Marek Polacek <polacek@redhat.com>
236 * doc/invoke.texi: Update -Wdangling-reference documentation.
238 2024-02-01 Uros Bizjak <ubizjak@gmail.com>
241 * config/i386/i386.md (*cmp<dwi>_doubleword):
242 Do not force SUBREG pieces to pseudos.
244 2024-02-01 John David Anglin <danglin@gcc.gnu.org>
246 * config/pa/pa.md (atomic_storedi_1): Fix bug in
249 2024-02-01 Georg-Johann Lay <avr@gjlay.de>
251 * config/avr/avr.cc: Tabify.
253 2024-02-01 Richard Ball <richard.ball@arm.com>
255 PR tree-optimization/111268
256 * tree-vect-slp.cc (vectorizable_slp_permutation_1):
257 Add variable-length check for vector input arguments
260 2024-02-01 Thomas Schwinge <tschwinge@baylibre.com>
262 * config/gcn/gcn.cc (gcn_hsa_declare_function_name): Don't
263 hard-code number of SGPR/VGPR/AVGPR registers.
264 * config/gcn/gcn.h: Add a 'STATIC_ASSERT's for number of
265 SGPR/VGPR/AVGPR registers.
267 2024-02-01 Monk Chiang <monk.chiang@sifive.com>
269 * config/riscv/riscv.md: Add "fcvt_i2f", "fcvt_f2i" type
270 attribute, and include sifive-p600.md.
271 * config/riscv/generic-ooo.md: Update type attribute.
272 * config/riscv/generic.md: Update type attribute.
273 * config/riscv/sifive-7.md: Update type attribute.
274 * config/riscv/sifive-p600.md: New file.
275 * config/riscv/riscv-cores.def (RISCV_TUNE): Add parameter.
276 * config/riscv/riscv-opts.h (enum riscv_microarchitecture_type):
278 * config/riscv/riscv.cc (sifive_p600_tune_info): New.
279 * config/riscv/riscv.h (TARGET_SFB_ALU): Update.
280 * doc/invoke.texi (RISC-V Options): Add sifive-p600-series
282 2024-02-01 Monk Chiang <monk.chiang@sifive.com>
284 * common/config/riscv/riscv-common.cc: Add Za64rs, Za128rs,
285 Ziccif, Ziccrse, Ziccamoa, Zicclsm, Zic64b items.
286 * config/riscv/riscv.opt: New macro for 7 new unprivileged
288 * doc/invoke.texi (RISC-V Options): Add Za64rs, Za128rs,
289 Ziccif, Ziccrse, Ziccamoa, Zicclsm, Zic64b extensions.
291 2024-02-01 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
293 * config/sol2.h (LIBASAN_EARLY_SPEC): Add -z now unless
294 -static-libasan. Add missing whitespace.
296 2024-02-01 Thomas Schwinge <tschwinge@baylibre.com>
298 * config/gcn/gcn.md (FIRST_SGPR_REG, LAST_SGPR_REG)
299 (FIRST_VGPR_REG, LAST_VGPR_REG, FIRST_AVGPR_REG, LAST_AVGPR_REG):
300 Don't 'define_constants'.
302 2024-02-01 Thomas Schwinge <tschwinge@baylibre.com>
304 * config/gcn/gcn.h (SGPR_OR_VGPR_REGNO_P): Remove.
306 2024-02-01 Thomas Schwinge <tschwinge@baylibre.com>
308 * config/gcn/gcn.md (sync_compare_and_swap<mode>_lds_insn)
309 [TARGET_RDNA3]: Adjust.
311 2024-02-01 Richard Biener <rguenther@suse.de>
313 PR tree-optimization/113693
314 * tree-ssa-sccvn.cc (rpo_elim::eliminate_avail): Honor avail
317 2024-02-01 Jakub Jelinek <jakub@redhat.com>
318 Jason Merrill <jason@redhat.com>
321 * gimple-low.cc (lower_stmt): Remove .ASAN_MARK calls
322 on variables which were promoted to TREE_STATIC.
324 2024-02-01 Roger Sayle <roger@nextmovesoftware.com>
325 Richard Biener <rguenther@suse.de>
328 * tree-ssa-math-opts.cc (is_widening_mult_rhs_p): Use range
329 information via tree_non_zero_bits to check if this operand
330 is suitably extended for a widening (or highpart) multiplication.
331 (convert_mult_to_widen): Insert explicit casts if the RHS or LHS
332 isn't already of the claimed type.
334 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
337 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
339 * config/riscv/generic-ooo.md (generic_ooo_sfb_alu): Add reservation
340 (generic_ooo_branch): ditto
341 * config/riscv/generic.md (generic_sfb_alu): ditto
342 (generic_fmul_half): ditto
343 * config/riscv/riscv.md: Remove cbo, pushpop, and rdfrm types
344 * config/riscv/sifive-7.md (sifive_7_hfma):Add reservation
345 (sifive_7_popcount): ditto
346 * config/riscv/vector.md: change rdfrm to fmove
347 * config/riscv/zc.md: change pushpop to load/store
349 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
352 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
353 Robin Dapp <rdapp.gcc@gmail.com>
355 * config/riscv/generic-ooo.md (generic_ooo): Move reservation
356 (generic_ooo_vec_load): ditto
357 (generic_ooo_vec_store): ditto
358 (generic_ooo_vec_loadstore_seg): ditto
359 (generic_ooo_vec_alu): ditto
360 (generic_ooo_vec_fcmp): ditto
361 (generic_ooo_vec_imul): ditto
362 (generic_ooo_vec_fadd): ditto
363 (generic_ooo_vec_fmul): ditto
364 (generic_ooo_crypto): ditto
365 (generic_ooo_perm): ditto
366 (generic_ooo_vec_reduction): ditto
367 (generic_ooo_vec_ordered_reduction): ditto
368 (generic_ooo_vec_idiv): ditto
369 (generic_ooo_vec_float_divsqrt): ditto
370 (generic_ooo_vec_mask): ditto
371 (generic_ooo_vec_vesetvl): ditto
372 (generic_ooo_vec_setrm): ditto
373 (generic_ooo_vec_readlen): ditto
374 * config/riscv/riscv.md: include generic-vector-ooo
375 * config/riscv/generic-vector-ooo.md: New file. to here
377 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
380 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
382 * config/riscv/riscv.cc (riscv_sched_variable_issue): enable assert
384 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
386 * config/riscv/riscv.cc (riscv_sched_variable_issue): enable assert
388 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
389 Robin Dapp <rdapp.gcc@gmail.com>
391 * config/riscv/generic-ooo.md (generic_ooo): Move reservation
392 (generic_ooo_vec_load): ditto
393 (generic_ooo_vec_store): ditto
394 (generic_ooo_vec_loadstore_seg): ditto
395 (generic_ooo_vec_alu): ditto
396 (generic_ooo_vec_fcmp): ditto
397 (generic_ooo_vec_imul): ditto
398 (generic_ooo_vec_fadd): ditto
399 (generic_ooo_vec_fmul): ditto
400 (generic_ooo_crypto): ditto
401 (generic_ooo_perm): ditto
402 (generic_ooo_vec_reduction): ditto
403 (generic_ooo_vec_ordered_reduction): ditto
404 (generic_ooo_vec_idiv): ditto
405 (generic_ooo_vec_float_divsqrt): ditto
406 (generic_ooo_vec_mask): ditto
407 (generic_ooo_vec_vesetvl): ditto
408 (generic_ooo_vec_setrm): ditto
409 (generic_ooo_vec_readlen): ditto
410 * config/riscv/riscv.md: include generic-vector-ooo
411 * config/riscv/generic-vector-ooo.md: New file. to here
413 2024-02-01 Edwin Lu <ewlu@rivosinc.com>
415 * config/riscv/generic-ooo.md (generic_ooo_sfb_alu): Add reservation
416 (generic_ooo_branch): ditto
417 * config/riscv/generic.md (generic_sfb_alu): ditto
418 (generic_fmul_half): ditto
419 * config/riscv/riscv.md: Remove cbo, pushpop, and rdfrm types
420 * config/riscv/sifive-7.md (sifive_7_hfma):Add reservation
421 (sifive_7_popcount): ditto
422 * config/riscv/vector.md: change rdfrm to fmove
423 * config/riscv/zc.md: change pushpop to load/store
425 2024-02-01 Andrew Pinski <quic_apinski@quicinc.com>
428 * config/aarch64/aarch64-simd.md (split for movv8di):
429 For strict aligned mode, use DImode instead of TImode.
431 2024-01-31 Robin Dapp <rdapp@ventanamicro.com>
434 * match.pd: Make sure else values match when folding a
435 vec_cond into a conditional operation.
437 2024-01-31 Marek Polacek <polacek@redhat.com>
439 * doc/invoke.texi: Mention that -fconcepts-ts was deprecated in GCC 14.
441 2024-01-31 Tamar Christina <tamar.christina@arm.com>
442 Matthew Malcomson <matthew.malcomson@arm.com>
445 * asan.h (asan_intercepted_p): Incercept memset, memmove, memcpy and
447 * builtins.cc (expand_builtin): Include HWASAN when checking for
450 2024-01-31 Richard Biener <rguenther@suse.de>
453 * match.pd (zext (bool) <= (int) 4294967295u): Make sure
454 to match INTEGER_CST only without outstanding conversion.
456 2024-01-31 Alex Coplan <alex.coplan@arm.com>
459 * config/aarch64/aarch64.cc (aarch64_reg_save_mode): Use
460 V16QImode for the full 16-byte FPR saves in the vector PCS case.
462 2024-01-31 Richard Biener <rguenther@suse.de>
464 PR tree-optimization/111444
465 * tree-ssa-sccvn.cc (vn_reference_lookup_3): Do not use
466 vn_reference_lookup_2 when optimistically skipping may-defs.
468 2024-01-31 Richard Biener <rguenther@suse.de>
470 PR tree-optimization/113630
471 * tree-ssa-pre.cc (compute_avail): Avoid registering a
472 reference with a representation with not matching base
475 2024-01-31 Jakub Jelinek <jakub@redhat.com>
477 PR rtl-optimization/113656
478 * simplify-rtx.cc (simplify_context::simplify_unary_operation_1)
479 <case FLOAT_TRUNCATE>: Fix up last argument to simplify_gen_unary.
481 2024-01-31 Jakub Jelinek <jakub@redhat.com>
484 * dwarf2out.cc (loc_list_from_tree_1): Assume integral types
485 with BLKmode are larger than DWARF2_ADDR_SIZE.
487 2024-01-31 Jakub Jelinek <jakub@redhat.com>
489 PR tree-optimization/113639
490 * gimple-lower-bitint.cc (bitint_large_huge::handle_operand_addr):
491 For VIEW_CONVERT_EXPR set rhs1 to its operand.
493 2024-01-31 Richard Biener <rguenther@suse.de>
495 PR tree-optimization/113670
496 * tree-vect-data-refs.cc (vect_check_gather_scatter):
497 Make sure we can take the address of the reference base.
499 2024-01-31 Georg-Johann Lay <avr@gjlay.de>
501 * config/avr/avr-mcus.def: Add AVR64DU28, AVR64DU32, ATA5787,
502 ATA5835, ATtiny64AUTO, ATA5700M322.
503 * doc/avr-mmcu.texi: Rebuild.
505 2024-01-31 Alexandre Oliva <oliva@adacore.com>
508 * ipa-strub.cc (build_ref_type_for): Drop nonaliased. Adjust
511 2024-01-31 Alexandre Oliva <oliva@adacore.com>
515 * builtins.cc (expand_builtin_stack_address): Use
516 STACK_ADDRESS_OFFSET.
517 * doc/extend.texi (__builtin_stack_address): Adjust.
518 * config/sparc/sparc.h (STACK_ADDRESS_OFFSET): Define.
519 * doc/tm.texi.in (STACK_ADDRESS_OFFSET): Document.
520 * doc/tm.texi: Rebuilt.
522 2024-01-31 Juzhe-Zhong <juzhe.zhong@rivai.ai>
525 * config/riscv/riscv-vsetvl.cc (extract_single_source): Remove.
526 (pre_vsetvl::compute_vsetvl_def_data): Fix compile time issue.
527 (pre_vsetvl::compute_transparent): New function.
528 (pre_vsetvl::compute_lcm_local_properties): Fix compile time time issue.
530 2024-01-30 Fangrui Song <maskray@google.com>
533 * config/i386/constraints.md: Define constraint "Ws".
534 * doc/md.texi: Document it.
536 2024-01-30 Marek Polacek <polacek@redhat.com>
540 * doc/invoke.texi: Update -Wdangling-reference description.
542 2024-01-30 Takayuki 'January June' Suwa <jjsuwa_sys3175@yahoo.co.jp>
544 * config/xtensa/constraints.md (R, T, U):
545 Change define_constraint to define_memory_constraint.
546 * config/xtensa/predicates.md (move_operand): Don't check that a
547 constant pool operand size is a multiple of UNITS_PER_WORD.
548 * config/xtensa/xtensa.cc
549 (xtensa_lra_p, TARGET_LRA_P): Remove.
550 (xtensa_emit_move_sequence): Remove "if (reload_in_progress)"
551 clause as it can no longer be true.
552 (fixup_subreg_mem): Drop function.
553 (xtensa_output_integer_literal_parts): Consider 16-bit wide
555 (xtensa_legitimate_constant_p): Add short-circuit path for
556 integer load instructions. Don't check that mode size is
557 at least UNITS_PER_WORD.
558 * config/xtensa/xtensa.md (movsf): Use can_create_pseudo_p()
559 rather reload_in_progress and reload_completed.
560 (doloop_end): Drop operand 2.
561 (movhi_internal): Add alternative loading constant from a
563 (define_split for DI register_operand): Don't limit to
564 !TARGET_AUTO_LITPOOLS.
565 * config/xtensa/xtensa.opt (mlra): Change to no effect.
567 2024-01-30 Pan Li <pan2.li@intel.com>
569 * config/riscv/riscv.cc (riscv_v_vls_mode_aggregate_gpr_count): New function to
570 calculate the gpr count required by vls mode.
571 (riscv_v_vls_to_gpr_mode): New function convert vls mode to gpr mode.
572 (riscv_pass_vls_aggregate_in_gpr): New function to return the rtx of gpr
574 (riscv_get_arg_info): Add vls mode handling.
575 (riscv_pass_by_reference): Return false if arg info has no zero gpr count.
577 2024-01-30 Richard Biener <rguenther@suse.de>
579 PR tree-optimization/113659
580 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
581 Handle main exit without virtual use.
583 2024-01-30 Christoph Müllner <christoph.muellner@vrull.eu>
585 * config/riscv/riscv.md: Move UNSPEC_XTHEADFMV* to unspec enum.
587 2024-01-30 Iain Sandoe <iain@sandoe.co.uk>
590 * config/darwin.h (DARWIN_SHARED_WEAK_ADDS, DARWIN_WEAK_CRTS): New.
591 (REAL_LIBGCC_SPEC): Move weak CRT handling to separate spec.
592 * config/i386/darwin.h (DARWIN_HEAP_T_LIB): New.
593 * config/i386/darwin32-biarch.h (DARWIN_HEAP_T_LIB): New.
594 * config/i386/darwin64-biarch.h (DARWIN_HEAP_T_LIB): New.
595 * config/rs6000/darwin.h (DARWIN_HEAP_T_LIB): New.
597 2024-01-30 Richard Sandiford <richard.sandiford@arm.com>
600 * config/aarch64/aarch64-early-ra.cc (early_ra::preprocess_insns):
601 Mark all registers that occur in addresses as needing a GPR.
603 2024-01-30 Richard Sandiford <richard.sandiford@arm.com>
606 * config/aarch64/aarch64-early-ra.cc (early_ra::replace_regs): Take
607 the containing insn as an extra parameter. Reset debug instructions
608 if they reference a register that is no longer used by real insns.
609 (early_ra::apply_allocation): Update calls accordingly.
611 2024-01-30 Jakub Jelinek <jakub@redhat.com>
613 PR tree-optimization/113603
614 * tree-ssa-strlen.cc (strlen_pass::handle_store): After
615 count_nonzero_bytes call refetch si using get_strinfo in case it
616 has been unshared in the meantime.
618 2024-01-30 Jakub Jelinek <jakub@redhat.com>
621 * except.cc (expand_builtin_eh_return_data_regno): If which doesn't
622 fit into unsigned HOST_WIDE_INT, return constm1_rtx.
624 2024-01-30 Jin Ma <jinma@linux.alibaba.com>
626 * config/riscv/thead.cc (th_print_operand_address): Change %ld
629 2024-01-29 Manos Anagnostakis <manos.anagnostakis@vrull.eu>
630 Manolis Tsamis <manolis.tsamis@vrull.eu>
631 Philipp Tomsich <philipp.tomsich@vrull.eu>
633 * config/aarch64/aarch64-ldpstp.md: Remove unused mode.
634 * config/aarch64/aarch64-protos.h (aarch64_operands_ok_for_ldpstp):
636 * config/aarch64/aarch64.cc (aarch64_operands_ok_for_ldpstp):
637 Call on framework moved later.
639 2024-01-29 Jose E. Marchesi <jose.marchesi@oracle.com>
641 * config/bpf/bpf.cc (bpf_expand_epilogue): Do not emit a return
642 instruction in naked function epilogues.
644 2024-01-29 YunQiang Su <syq@gcc.gnu.org>
647 * configure.ac: Fix typo gcc_cv_as_mips_explicit should be
648 gcc_cv_as_mips_explicit_relocs.
649 * configure: Regnerated.
651 2024-01-29 Matthieu Longo <matthieu.longo@arm.com>
654 * config/arm/arm.md (arm_rev16si2): Convert to define_insn.
655 Correct generated RTL.
656 (arm_rev16si2_alt1): Correctly handle conditional execution.
657 (arm_rev16si2_alt2): Likewise.
659 2024-01-29 Richard Biener <rguenther@suse.de>
662 * expr.cc (expand_assignment): Spill hard registers if
663 we index them with a variable offset.
665 2024-01-29 Richard Biener <rguenther@suse.de>
668 * gimple-isel.cc (gimple_expand_vec_set_extract_expr):
669 Also allow DECL_HARD_REGISTER variables.
671 2024-01-29 Alex Coplan <alex.coplan@arm.com>
674 * config/aarch64/aarch64-ldp-fusion.cc (fixup_debug_uses_trailing_add):
675 Use iterate_safely when iterating over debug uses.
676 (fixup_debug_uses): Likewise.
677 (ldp_bb_info::cleanup_tombstones): Use iterate_safely to iterate
678 over nondebug insns instead of manually maintaining the next insn.
679 * iterator-utils.h (class safe_iterator): New.
680 (iterate_safely): New.
682 2024-01-29 H.J. Lu <hjl.tools@gmail.com>
685 * config/i386/i386-options.cc (ix86_set_func_type): Save
686 callee-saved registers in noreturn functions for -O0/-Og.
688 2024-01-29 Tobias Burnus <tburnus@baylibre.com>
691 * config/gcn/gcn-valu.md (fold_left_plus_<mode>): Only
692 define for !TARGET_RDNA2_PLUS.
694 2024-01-29 Richard Sandiford <richard.sandiford@arm.com>
697 * tree-vect-patterns.cc (vect_recog_over_widening_pattern): Remove
698 workaround for right shifts.
699 (vect_truncatable_operation_p): Handle NEGATE_EXPR and BIT_NOT_EXPR.
700 (vect_determine_precisions_from_range): Be more selective about
701 which codes can be narrowed based on their input and output ranges.
702 For shifts, require at least one more bit of precision than the
703 maximum shift amount.
705 2024-01-29 Tobias Burnus <tburnus@baylibre.com>
707 * config/nvptx/nvptx.opt (march-map=): Add sm_89 and sm_90a.
709 2024-01-29 Tobias Burnus <tburnus@baylibre.com>
711 * doc/install.texi (amdgcn): Recommend LLVM 15+ and newlib 4.4+,
712 but keep requiring only newlib 4.3+ and, if gfx1100 is disabled,
715 2024-01-29 Tobias Burnus <tburnus@baylibre.com>
718 * config/gcn/mkoffload.cc (SET_XNACK_UNSET, TEST_SRAM_ECC_UNSET): New.
719 (SET_SRAM_ECC_UNSUPPORTED): Renamed to ...
720 (SET_SRAM_ECC_UNSET): ... this.
721 (copy_early_debug_info): Remove gfx900 special case, now handled as
722 part of the generic handling.
723 (main): Update SRAM_ECC and XNACK for the -march as done in gcn-hsa.h.
725 2024-01-29 Jakub Jelinek <jakub@redhat.com>
727 PR tree-optimization/110603
728 * tree-ssa-strlen.cc (get_range_strlen_dynamic): Remove incorrect
729 setting of pdata->maxlen to vr.upper_bound (which is unconditionally
730 overwritten anyway). Avoid creating invalid range with minlen
731 larger than maxlen. Formatting fix.
733 2024-01-29 Richard Biener <rguenther@suse.de>
736 * tree-inline.cc (initialize_inlined_parameters): Reverse
737 the decl chain of inlined parameters.
739 2024-01-28 Iain Sandoe <iain@sandoe.co.uk>
741 * config/darwin.cc (darwin_build_constant_cfstring): Prevent over-
742 alignment of CFString constants by setting DECL_USER_ALIGN.
744 2024-01-28 Iain Sandoe <iain@sandoe.co.uk>
745 Jakub Jelinek <jakub@redhat.com>
748 * builtins.cc (expand_builtin): Handle BUILT_IN_GCC_NESTED_PTR_CREATED
749 and BUILT_IN_GCC_NESTED_PTR_DELETED.
750 * builtins.def (BUILT_IN_GCC_NESTED_PTR_CREATED,
751 BUILT_IN_GCC_NESTED_PTR_DELETED): Make these builtins LIB-EXT and
752 rename the library fallbacks to __gcc_nested_func_ptr_created and
753 __gcc_nested_func_ptr_deleted.
754 * doc/invoke.texi: Rename these to __gcc_nested_func_ptr_created
755 and __gcc_nested_func_ptr_deleted.
756 * tree-nested.cc (finalize_nesting_tree_1): Use builtin_explicit for
757 BUILT_IN_GCC_NESTED_PTR_CREATED and BUILT_IN_GCC_NESTED_PTR_DELETED.
758 * tree.cc (build_common_builtin_nodes): Build the
759 BUILT_IN_GCC_NESTED_PTR_CREATED and BUILT_IN_GCC_NESTED_PTR_DELETED local
760 builtins only for non-explicit.
762 2024-01-28 YunQiang Su <syq@gcc.gnu.org>
764 * doc/invoke.texi: Remove duplicate MIPS explicit-relocs option.
766 2024-01-27 H.J. Lu <hjl.tools@gmail.com>
769 * config/i386/i386-options.cc (ix86_set_func_type): Don't
770 save and restore callee saved registers for a noreturn function
771 with nothrow or compiled with -fno-exceptions.
773 2024-01-27 H.J. Lu <hjl.tools@gmail.com>
777 * config/i386/i386-expand.cc (ix86_expand_call): Replace
778 no_caller_saved_registers check with call_saved_registers check.
779 Clobber all registers that are not used by the callee with
780 no_callee_saved_registers attribute.
781 * config/i386/i386-options.cc (ix86_set_func_type): Set
782 call_saved_registers to TYPE_NO_CALLEE_SAVED_REGISTERS for
783 noreturn function. Disallow no_callee_saved_registers with
784 interrupt or no_caller_saved_registers attributes together.
785 (ix86_set_current_function): Replace no_caller_saved_registers
786 check with call_saved_registers check.
787 (ix86_handle_no_caller_saved_registers_attribute): Renamed to ...
788 (ix86_handle_call_saved_registers_attribute): This.
789 (ix86_gnu_attributes): Add
790 ix86_handle_call_saved_registers_attribute.
791 * config/i386/i386.cc (ix86_conditional_register_usage): Replace
792 no_caller_saved_registers check with call_saved_registers check.
793 (ix86_function_ok_for_sibcall): Don't allow callee with
794 no_callee_saved_registers attribute when the calling function
795 has callee-saved registers.
796 (ix86_comp_type_attributes): Also check
797 no_callee_saved_registers.
798 (ix86_epilogue_uses): Replace no_caller_saved_registers check
799 with call_saved_registers check.
800 (ix86_hard_regno_scratch_ok): Likewise.
801 (ix86_save_reg): Replace no_caller_saved_registers check with
802 call_saved_registers check. Don't save any registers for
803 TYPE_NO_CALLEE_SAVED_REGISTERS. Save all registers with
804 TYPE_DEFAULT_CALL_SAVED_REGISTERS if function with
805 no_callee_saved_registers attribute is called.
806 (find_drap_reg): Replace no_caller_saved_registers check with
807 call_saved_registers check.
808 * config/i386/i386.h (call_saved_registers_type): New enum.
809 (machine_function): Replace no_caller_saved_registers with
810 call_saved_registers.
811 * doc/extend.texi: Document no_callee_saved_registers attribute.
813 2024-01-27 Jakub Jelinek <jakub@redhat.com>
815 PR tree-optimization/113614
816 * gimple-lower-bitint.cc (gimple_lower_bitint): Don't merge
817 widening casts from signed to unsigned types with TRUNC_DIV_EXPR,
818 TRUNC_MOD_EXPR or FLOAT_EXPR uses.
820 2024-01-27 Jakub Jelinek <jakub@redhat.com>
822 PR tree-optimization/113568
823 * gimple-lower-bitint.cc (bitint_large_huge::lower_mergeable_stmt):
824 For VIEW_CONVERT_EXPR use first operand of rhs1 instead of rhs1
825 in the widening extension checks.
827 2024-01-27 Jakub Jelinek <jakub@redhat.com>
829 * gimple-lower-bitint.cc (gimple_lower_bitint): For
830 TDF_DETAILS dump mapping of SSA_NAMEs to decls.
832 2024-01-26 Hans-Peter Nilsson <hp@axis.com>
834 * cgraphunit.cc (process_function_and_variable_attributes): Tweak
835 the warning for an attribute-always_inline without inline declaration.
837 2024-01-26 Robin Dapp <rdapp@ventanamicro.com>
840 * genopinit.cc (main): Split init_all_optabs into functions
841 of 1000 patterns each.
843 2024-01-26 Tobias Burnus <tburnus@baylibre.com>
845 * config.gcc (amdgcn-*-*): Add gfx1030 and gfx1100 to
847 * doc/install.texi (Configuration amdgcn-*-*): Mention gfx1030/gfx1100.
848 * doc/invoke.texi (AMD GCN Options): Add gfx1030 and gfx1100 to
851 2024-01-26 Andrew Stubbs <ams@baylibre.com>
853 * config/gcn/gcn-opts.h (TARGET_PACKED_WORK_ITEMS): Add TARGET_RDNA3.
854 * config/gcn/gcn-valu.md (all_convert): New iterator.
855 (<convop><V_INT_1REG_ALT:mode><V_INT_1REG:mode>2<exec>): New
856 define_expand, and rename the old one to ...
857 (*<convop><V_INT_1REG_ALT:mode><V_INT_1REG:mode>_sdwa<exec>): ... this.
858 (extend<V_INT_1REG_ALT:mode><V_INT_1REG:mode>2<exec>): Likewise, to ...
859 (extend<V_INT_1REG_ALT:mode><V_INT_1REG:mode>_sdwa<exec>): .. this.
860 (*<convop><V_INT_1REG_ALT:mode><V_INT_1REG:mode>_shift<exec>): New.
861 * config/gcn/gcn.cc (gcn_global_address_p): Use "offsetbits" correctly.
862 (gcn_hsa_declare_function_name): Update the vgpr counting for gfx1100.
863 * config/gcn/gcn.md (<u>mulhisi3): Disable on RDNA3.
864 (<u>mulqihi3_scalar): Likewise.
866 2024-01-26 Richard Biener <rguenther@suse.de>
868 PR tree-optimization/113602
869 * tree-data-ref.cc (dr_analyze_innermost): Fail when
870 the base object isn't addressable.
872 2024-01-26 Tobias Burnus <tburnus@baylibre.com>
874 * config/gcn/gcn-hsa.h (ABI_VERSION_SPEC): New; creates the
875 "--amdhsa-code-object-version=" argument.
876 (ASM_SPEC): Use it; replace previous version of it.
878 2024-01-26 Juzhe-Zhong <juzhe.zhong@rivai.ai>
880 * config/riscv/riscv-vsetvl.cc (pre_vsetvl::earliest_fuse_vsetvl_info): Refine some codes.
881 (pre_vsetvl::emit_vsetvl): Ditto.
883 2024-01-26 Jiahao Xu <xujiahao@loongson.cn>
885 * config/loongarch/lasx.md (vec_extract<mode>_0):
886 New define_insn_and_split patten.
888 2024-01-26 Jiahao Xu <xujiahao@loongson.cn>
890 * config/loongarch/loongarch.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
892 2024-01-26 Li Wei <liwei@loongson.cn>
894 * config/loongarch/loongarch.cc (loongarch_emit_swdivsf): Adjust.
896 2024-01-26 Juzhe-Zhong <juzhe.zhong@rivai.ai>
899 * config/riscv/riscv-vsetvl.cc (pre_vsetvl::compute_lcm_local_properties): Fix bug.
901 2024-01-26 Andrew Pinski <quic_apinski@quicinc.com>
904 * config/aarch64/aarch64.cc (aarch64_classify_index): Avoid
905 undefined shift after the call to exact_log2.
907 2024-01-25 Andrew Pinski <quic_apinski@quicinc.com>
910 * config/aarch64/constraints.md (J): Cast to `unsigned HOST_WIDE_INT`
911 before taking the negative of it.
913 2024-01-25 Vladimir N. Makarov <vmakarov@redhat.com>
916 * lra-constraints.cc (curr_insn_transform): Change class even for
917 spilled pseudo successfully matched with with NO_REGS.
919 2024-01-25 Georg-Johann Lay <avr@gjlay.de>
922 * config/avr/avr-mcus.def (atmega3208, atmega3209): Fix data_section_start.
924 2024-01-25 Szabolcs Nagy <szabolcs.nagy@arm.com>
927 * config/aarch64/aarch64.cc (aarch64_gen_compare_zero_and_branch): New.
928 (aarch64_expand_epilogue): Use the new function.
929 (aarch64_split_compare_and_swap): Likewise.
930 (aarch64_split_atomic_op): Likewise.
932 2024-01-25 Robin Dapp <rdapp.gcc@gmail.com>
935 * fold-const.cc (simplify_const_binop): New function for binop
936 simplification of two constant vectors when element-wise
937 handling is not necessary.
938 (const_binop): Call new function.
940 2024-01-25 Mary Bennett <mary.bennett@embecosm.com>
942 * common/config/riscv/riscv-common.cc: Add XCVbitmanip.
943 * config/riscv/constraints.md: Likewise.
944 * config/riscv/corev.def: Likewise.
945 * config/riscv/corev.md: Likewise.
946 * config/riscv/predicates.md: Likewise.
947 * config/riscv/riscv-builtins.cc (AVAIL): Likewise.
948 * config/riscv/riscv-ftypes.def: Likewise.
949 * config/riscv/riscv.opt: Likewise.
950 * config/riscv/riscv.cc (riscv_print_operand): Add new operand 'Y'.
951 * doc/extend.texi: Add XCVbitmanip builtin documentation.
952 * doc/sourcebuild.texi: Likewise.
954 2024-01-25 Tobias Burnus <tburnus@baylibre.com>
956 * config/gcn/gcn-hsa.h (ASM_SPEC): Add space after -mxnack= argument.
958 2024-01-25 Yanzhang Wang <yanzhang.wang@intel.com>
961 * config/riscv/riscv.cc (riscv_get_arg_info): Remove the flag.
962 (riscv_fntype_abi): Ditto.
963 * config/riscv/riscv.opt: Ditto.
965 2024-01-25 Jakub Jelinek <jakub@redhat.com>
968 * convert.cc (convert_to_integer_1) <case LSHIFT_EXPR>: Compare shift
969 count against TYPE_PRECISION rather than TYPE_SIZE.
971 2024-01-25 Richard Sandiford <richard.sandiford@arm.com>
974 * config/aarch64/aarch64-sve-builtins.cc (vector_cst_all_same):
975 Check VECTOR_CST_ELT instead of VECTOR_CST_ENCODED_ELT
977 2024-01-25 Richard Sandiford <richard.sandiford@arm.com>
980 * config/aarch64/aarch64-simd.md: In the movv8di splitter, check
981 whether each split instruction is a load that clobbers the source
982 address. Emit that instruction last if so.
984 2024-01-25 Richard Sandiford <richard.sandiford@arm.com>
987 * config/aarch64/aarch64-simd.md (aarch64_zip1<mode>_low): New
989 (<optab><Vnarrowq><mode>2): Use it instead of generating a
990 paradoxical subreg for the input.
992 2024-01-25 Juzhe-Zhong <juzhe.zhong@rivai.ai>
994 * config/riscv/riscv-vsetvl.cc (get_all_predecessors): New function.
995 (pre_vsetvl::pre_global_vsetvl_info): Add LCM delete block all
996 predecessors dump information.
998 2024-01-25 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1000 * config/riscv/riscv-vsetvl.cc (pre_vsetvl::compute_vsetvl_def_data): Remove
1001 redundant full available computation.
1002 (pre_vsetvl::pre_global_vsetvl_info): Ditto.
1004 2024-01-25 Jakub Jelinek <jakub@redhat.com>
1006 * doc/generic.texi (VECTOR_CST): Fix typo - petterns -> patterns.
1007 * doc/rtl.texi (CONST_VECTOR): Likewise.
1009 2024-01-25 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1011 * config/riscv/riscv-opts.h (enum vsetvl_strategy_enum): Add optim-no-fusion option.
1012 * config/riscv/riscv-vsetvl.cc (pass_vsetvl::lazy_vsetvl): Ditto.
1013 (pass_vsetvl::execute): Ditto.
1014 * config/riscv/riscv.opt: Ditto.
1016 2024-01-25 Jiahao Xu <xujiahao@loongson.cn>
1018 * config/loongarch/lasx.md (@vec_concatz<mode>): Remove this define_insn pattern.
1019 * config/loongarch/loongarch.cc (loongarch_expand_vector_group_init): Use vec_concat<mode>.
1021 2024-01-25 Richard Biener <rguenther@suse.de>
1023 PR tree-optimization/113576
1024 * tree-vect-loop.cc (vec_init_loop_exit_info): Only allow
1025 exits with may_be_zero niters when its the last one.
1027 2024-01-25 Lulu Cheng <chenglulu@loongson.cn>
1029 * config/loongarch/loongarch.cc (loongarch_symbolic_constant_p):
1030 For symbols of type tls, non-zero Offset is not generated.
1032 2024-01-25 Haochen Gui <guihaoc@gcc.gnu.org>
1034 * config/rs6000/rs6000-string.cc (expand_block_compare): Enable
1035 P9 with m32 and mpowerpc64.
1037 2024-01-25 liuhongt <hongtao.liu@intel.com>
1039 * config/i386/i386-options.cc (ix86_option_override_internal):
1040 Enable -mlam=u57 by default when compiled with
1041 -fsanitize=hwaddress.
1043 2024-01-25 Palmer Dabbelt <palmer@rivosinc.com>
1045 * common/config/riscv/riscv-common.cc (riscv_implied_info):
1046 Remove {"ztso", "a"}.
1048 2024-01-24 Martin Jambor <mjambor@suse.cz>
1052 * cgraph.h (cgraph_edge): Add a parameter to
1053 redirect_call_stmt_to_callee.
1054 * ipa-param-manipulation.h (ipa_param_adjustments): Add a
1055 parameter to modify_call.
1056 (ipa_release_ssas_in_hash): Declare.
1057 * cgraph.cc (cgraph_edge::redirect_call_stmt_to_callee): New
1058 parameter killed_ssas, pass it to padjs->modify_call.
1059 * ipa-param-manipulation.cc (purge_all_uses): New function.
1060 (ipa_param_adjustments::modify_call): New parameter killed_ssas.
1061 Instead of substituting uses, invoke purge_all_uses. If
1062 hash of killed SSAs has not been provided, create a temporary one
1063 and release SSAs that have been added to it.
1064 (compare_ssa_versions): New function.
1065 (ipa_release_ssas_in_hash): Likewise.
1066 * tree-inline.cc (redirect_all_calls): Create
1067 id->killed_new_ssa_names earlier, pass it to edge redirection,
1069 (copy_body): Release SSAs in id->killed_new_ssa_names.
1071 2024-01-24 Andrew Pinski <quic_apinski@quicinc.com>
1074 * config/aarch64/aarch64.cc (aarch64_get_reg_raw_mode): For
1075 TARGET_GENERAL_REGS_ONLY, return VOIDmode for non-GP_REGNUM_P regno.
1077 2024-01-24 Monk Chiang <monk.chiang@sifive.com>
1080 * config/riscv/sfb.md: New splitters to rewrite single bit
1081 sign extension as the condition to SFB instructions.
1083 2024-01-24 Jan Hubicka <jh@suse.cz>
1086 * common.opt: (flimit-function-alignment): Reorder alphabeticaly
1087 (fmin-function-alignment): New parameter.
1088 * doc/invoke.texi: (-fmin-function-alignment): Document.
1089 (-falign-functions,-falign-loops,-falign-labels): Mention that
1090 aglinments are ignored in cold code.
1091 * varasm.cc (assemble_start_function): Handle min-function-alignment.
1093 2024-01-24 Tamar Christina <tamar.christina@arm.com>
1096 * config/aarch64/aarch64-simd.md (<su_optab>div<mode>3,
1098 * config/aarch64/iterators.md (VQDIV): Remove.
1099 (SVE_FULL_SDI_SIMD, SVE_FULL_HSDI_SIMD_DI,
1100 SVE_I_SIMD_DI): New.
1101 (VPRED, sve_lane_con): Add V4SI and V2DI.
1102 * config/aarch64/aarch64-sve.md (<optab><mode>3,
1103 @aarch64_pred_<optab><mode>): Support Advanced SIMD types.
1104 (mul<mode>3): New, split from <optab><mode>3.
1105 (@aarch64_pred_<optab><mode>, *post_ra_<optab><mode>3): New.
1106 * config/aarch64/aarch64-sve2.md (@aarch64_mul_lane_<mode>,
1107 *aarch64_mul_unpredicated_<mode>): Change SVE_FULL_HSDI to
1108 SVE_FULL_HSDI_SIMD_DI.
1110 2024-01-24 Tamar Christina <tamar.christina@arm.com>
1112 PR tree-optimization/113552
1113 * config/aarch64/aarch64.cc
1114 (aarch64_simd_clone_compute_vecsize_and_simdlen): Block simdlen 1.
1116 2024-01-24 Martin Jambor <mjambor@suse.cz>
1119 * ipa-cp.cc (ipcp_lattice<valtype>::add_value): Bail out if value
1120 count is equal or greater than the limit. Use the limit from the
1123 2024-01-24 YunQiang Su <syq@gcc.gnu.org>
1125 * configure.ac: Detect the explicit relocs support for
1126 mips, and define C macro MIPS_EXPLICIT_RELOCS.
1127 * config.in: Regenerated.
1128 * configure: Regenerated.
1129 * doc/invoke.texi(MIPS Options): Add -mexplicit-relocs.
1130 * config/mips/mips-opts.h: Define enum mips_explicit_relocs.
1131 * config/mips/mips.cc(mips_set_compression_mode): Sorry if
1132 !TARGET_EXPLICIT_RELOCS instead of just set it.
1133 * config/mips/mips.h: Define TARGET_EXPLICIT_RELOCS and
1134 TARGET_EXPLICIT_RELOCS_PCREL with mips_opt_explicit_relocs.
1135 * config/mips/mips.opt: Introduce -mexplicit-relocs= option
1136 and define -m(no-)explicit-relocs as aliases.
1138 2024-01-24 Alex Coplan <alex.coplan@arm.com>
1140 * config/aarch64/aarch64.opt (-mearly-ldp-fusion): Set default
1142 (-mlate-ldp-fusion): Likewise.
1144 2024-01-24 Tamar Christina <tamar.christina@arm.com>
1146 * tree-vect-loop.cc (vect_get_vect_def,
1147 vect_create_epilog_for_reduction): Rename main_exit_p to
1150 2024-01-24 Tamar Christina <tamar.christina@arm.com>
1152 PR tree-optimization/113364
1153 * tree-vect-loop.cc (vect_create_epilog_for_reduction): If all exits all
1154 early exits then we must reduce from the first offset for all of them.
1156 2024-01-24 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1159 * config/riscv/riscv-vsetvl.cc (get_expr_id): Remove.
1161 (get_bb_index): Ditto.
1162 (pre_vsetvl::compute_avl_def_data): Ditto.
1163 (pre_vsetvl::earliest_fuse_vsetvl_info): Fix large memory usage.
1164 (pre_vsetvl::pre_global_vsetvl_info): Ditto.
1166 2024-01-23 Andrew Pinski <quic_apinski@quicinc.com>
1167 Richard Sandiford <richard.sandiford@arm.com>
1170 * ccmp.cc (ccmp_candidate_p): Add outer argument.
1171 Allow if the outer is true and the lhs is used more
1173 (expand_ccmp_expr): Update call to ccmp_candidate_p.
1174 * expr.h (expand_expr_real_gassign): Declare.
1175 * expr.cc (expand_expr_real_gassign): New function, split out from...
1176 (expand_expr_real_1): ...here.
1177 * cfgexpand.cc (expand_gimple_stmt_1): Use expand_expr_real_gassign.
1179 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1182 * config/aarch64/aarch64-ldp-fusion.cc (reset_debug_use): New.
1183 (fixup_debug_use): New.
1184 (fixup_debug_uses_trailing_add): New.
1185 (fixup_debug_uses): New. Use it ...
1186 (ldp_bb_info::fuse_pair): ... here.
1187 (try_promote_writeback): Call fixup_debug_uses_trailing_add to
1188 fix up debug uses of the base register that are affected by
1189 folding in the trailing add insn.
1191 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1194 * config/aarch64/aarch64-ldp-fusion.cc (ldp_bb_info::fuse_pair):
1195 Update trailing nondebug uses of the base register in the case
1196 of cancelling writeback.
1198 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1201 * rtl-ssa/accesses.h (use_info::next_debug_insn_use): New.
1202 (debug_insn_use_iterator): New.
1203 (set_info::first_debug_insn_use): New.
1204 (set_info::debug_insn_uses): New.
1205 * rtl-ssa/member-fns.inl (use_info::next_debug_insn_use): New.
1206 (set_info::first_debug_insn_use): New.
1207 (set_info::debug_insn_uses): New.
1209 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1212 * config/aarch64/aarch64-ldp-fusion.cc (ldp_bb_info::try_fuse_pair):
1213 Don't record hazards against the opposite insn in the pair.
1215 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1218 * config/aarch64/aarch64-ldp-fusion.cc
1219 (struct stp_change_builder): New.
1220 (decide_stp_strategy): Reanme to ...
1221 (try_repurpose_store): ... this.
1222 (ldp_bb_info::fuse_pair): Refactor to use stp_change_builder to
1223 construct stp changes. Fix up uses when inserting new stp insns.
1225 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1228 * rtl-ssa.h: Include hash-set.h.
1229 * rtl-ssa/changes.cc (function_info::finalize_new_accesses): Add
1230 new_sets parameter and use it to keep track of new user-created sets.
1231 (function_info::apply_changes_to_insn): Also call add_def on new sets.
1232 (function_info::change_insns): Add hash_set to keep track of new
1233 user-created defs. Plumb it through.
1234 * rtl-ssa/functions.h: Add hash_set parameter to finalize_new_accesses and
1235 apply_changes_to_insn.
1237 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1240 * rtl-ssa/accesses.cc (function_info::create_use): New.
1241 * rtl-ssa/changes.cc (function_info::finalize_new_accesses):
1242 Ensure new uses end up referring to permanent defs.
1243 * rtl-ssa/functions.h (function_info::create_use): Declare.
1245 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1248 * rtl-ssa/changes.cc (function_info::change_insns): Split out the call
1249 to finalize_new_accesses from the backwards placement loop, run it
1250 forwards in a separate loop.
1252 2024-01-23 Richard Biener <rguenther@suse.de>
1254 PR tree-optimization/113552
1255 * tree-vect-stmts.cc (vectorizable_simd_clone_call): Use
1256 floor_log2 instead of exact_log2 on the number of calls.
1258 2024-01-23 Jeff Law <jlaw@ventanamicro.com>
1259 Jakub Jelinek <jakub@redhat.com>
1261 * config/ia64/ia64.cc (ia64_start_function): Add ATTRIBUTE_UNUSED to
1264 2024-01-23 Richard Biener <rguenther@suse.de>
1266 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
1267 Separate single and multi-exit case when creating PHIs between
1268 the main and epilogue.
1270 2024-01-23 Richard Sandiford <richard.sandiford@arm.com>
1273 * config/aarch64/aarch64-sve-builtins-shapes.cc (build_one): Skip
1274 MODE_single variants of functions that don't take tuple arguments.
1276 2024-01-23 Alex Coplan <alex.coplan@arm.com>
1279 * config/aarch64/aarch64-ldp-fusion.cc (try_promote_writeback):
1280 Don't assert recog success, just punt if the writeback pair
1283 2024-01-23 Jakub Jelinek <jakub@redhat.com>
1285 * config/gcn/gcn.cc (gcn_hsa_declare_function_name): Add
1286 ATTRIBUTE_UNUSED to decl.
1288 2024-01-23 Richard Biener <rguenther@suse.de>
1291 * dwarf2out.cc (dwarf2out_die_ref_for_decl): Gracefully
1292 handle unexpected but bogus DIE contexts when not checking
1295 2024-01-23 Jakub Jelinek <jakub@redhat.com>
1297 PR tree-optimization/113462
1298 * fold-const.cc (native_interpret_int): Don't punt if total_bytes
1299 is larger than HOST_BITS_PER_DOUBLE_INT / BITS_PER_UNIT.
1300 (fold_view_convert_expr): Use XALLOCAVEC buffers for types with
1301 sizes between 129 and 8192 bytes.
1303 2024-01-23 Xi Ruoyao <xry111@xry111.site>
1305 * config/loongarch/loongarch.cc (loongarch_explicit_relocs_p):
1306 If la_opt_explicit_relocs is EXPLICIT_RELOCS_AUTO, return false
1307 for SYMBOL_TLS_LDM and SYMBOL_TLS_GD.
1308 (loongarch_call_tls_get_addr): Do not split symbols of
1309 SYMBOL_TLS_LDM or SYMBOL_TLS_GD if la_opt_explicit_relocs is
1310 EXPLICIT_RELOCS_AUTO.
1312 2024-01-23 Richard Biener <rguenther@suse.de>
1314 * alias.cc (known_base_value_p): Remove.
1315 (find_base_value): Remove PLUS/MINUS handling
1316 when both operands are not CONST_INT_P.
1318 2024-01-23 Richard Biener <rguenther@suse.de>
1320 PR rtl-optimization/113255
1321 * alias.cc (find_base_term): Remove PLUS/MINUS handling
1322 when both operands are not CONST_INT_P.
1324 2024-01-23 Richard Biener <rguenther@suse.de>
1327 * dwarf2out.cc (dwarf2out_finish): Reset all type units
1328 for the fat part of an LTO compile.
1330 2024-01-23 chenxiaolong <chenxiaolong@loongson.cn>
1332 * doc/sourcebuild.texi: Add attributes for keywords.
1334 2024-01-23 Sandra Loosemore <sandra@codesourcery.com>
1337 * doc/invoke.texi (Warning Options): Correct lists of options
1338 enabled by -Wall and -Wextra by checking against common.opt
1341 2024-01-22 Andrew Pinski <quic_apinski@quicinc.com>
1344 * config/arm/parsecpu.awk (check_cpu): Use cpu_opt_alias
1345 instead of cpu_optaliases.
1346 (check_arch): Use arch_opt_alias instead of arch_optaliases.
1348 2024-01-22 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1350 * config/riscv/riscv-protos.h (splat_to_scalar_move_p): New function.
1351 * config/riscv/riscv-v.cc (splat_to_scalar_move_p): Ditto.
1352 * config/riscv/vector.md: Simplify vmv.v.x. into vmv.s.x.
1354 2024-01-22 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1357 * config/riscv/riscv.md: Use reg instead of subreg.
1359 2024-01-22 Tobias Burnus <tburnus@baylibre.com>
1362 * config/gcn/mkoffload.cc (elf_arch): Change default to gfx900
1363 to match the compiler default.
1364 (simple_object_copy_lto_debug_sections): Never unlink the outfile
1365 on error as the caller does so.
1366 (maybe_unlink, compile_native): Use %<...%> and %qs in fatal_error.
1367 (main): Likewise. Fix 'mkoffload.dbg.o' cleanup.
1369 2024-01-22 Richard Biener <rguenther@suse.de>
1371 PR tree-optimization/113373
1372 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
1373 Create LC PHIs in the exit blocks where necessary.
1374 * tree-vect-loop.cc (vectorizable_live_operation): Do not try
1375 to handle missing LC PHIs.
1376 (find_connected_edge): Remove.
1377 (vect_create_epilog_for_reduction): Cleanup use of auto_vec.
1379 2024-01-22 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1381 * config/riscv/vector.md: Fix vfirst/vmsbf/vmsof ratio attributes.
1383 2024-01-22 xuli <xuli1@eswincomputing.com>
1386 * config/riscv/riscv-vector-builtins.cc (has_vxrm_or_frm_p):remove.
1387 (registered_function::overloaded_hash):refactor.
1388 (resolve_overloaded_builtin):avoid internal ICE.
1390 2024-01-21 Mikael Pettersson <mikpelinux@gmail.com>
1394 * calls.cc (emit_library_call_value_1): Pass valid TYPE
1396 * expr.cc (emit_push_insn): Likewise.
1398 2024-01-21 Jeff Law <jlaw@ventanamicro.com>
1400 * config/riscv/riscv.cc (riscv_init_cumulative_args): Install
1401 correcction version of last change.
1403 2024-01-21 Jeff Law <jlaw@ventanamicro.com>
1405 * config/riscv/riscv.cc (riscv_init_cumulative_args): Update and
1406 fix bugs in signature.
1408 2024-01-21 Roger Sayle <roger@nextmovesoftware.com>
1409 Richard Biener <rguenther@suse.de>
1411 PR rtl-optimization/111267
1412 * fwprop.cc (fwprop_propagation::profitabe_p): Rename
1413 profitable_p method to likely_profitable_p.
1414 (try_fwprop_subst_node): Update call to likely_profitable_p.
1415 Only bail-out early when !prop.likely_profitable_p for instructions
1416 that are not single sets. When comparing costs, bail-out if the
1417 cost is unchanged and !prop.likely_profitable_p.
1419 2024-01-21 Sandra Loosemore <sandra@codesourcery.com>
1422 * doc/invoke.texi (Warning Options): Document that -Wunused-parameter
1423 isn't enabled by -Wunused unless -Wextra is provided, and that
1424 -Wunused does enable -Wunused-const-variable=1 for C. Clarify that
1425 -Wunused doesn't enable -Wunused-* options documented as behaving
1426 otherwise, and list them explicitly.
1428 2024-01-21 Sandra Loosemore <sandra@codesourcery.com>
1431 * doc/invoke.texi (Warning Options): Fix broken example and
1432 clean up/reorganize the others. Also describe what the short-form
1435 2024-01-20 Sandra Loosemore <sandra@codesourcery.com>
1438 * doc/invoke.texi (Option Summary): Add -Warray-parameter.
1439 (Warning Options): Correct/edit discussion of -Warray-parameter
1440 to make the first example less confusing, and fill in missing info.
1442 2024-01-20 Jakub Jelinek <jakub@redhat.com>
1444 PR tree-optimization/113462
1445 * gimple-lower-bitint.cc (bitint_large_huge::handle_cast):
1446 Handle rhs1 INTEGER_CST like SSA_NAME.
1448 2024-01-20 Jakub Jelinek <jakub@redhat.com>
1450 PR tree-optimization/113491
1451 * tree-switch-conversion.cc (switch_conversion::build_constructors):
1452 If elt.index has precision higher than sizetype, fold_convert it to
1454 (switch_conversion::array_value_type): Return type if type is
1455 BITINT_TYPE with precision above MAX_FIXED_MODE_SIZE or with BLKmode.
1456 (switch_conversion::build_arrays): Use unsigned_type_for rather than
1457 lang_hooks.types.type_for_mode if utype is BITINT_TYPE with precision
1458 above MAX_FIXED_MODE_SIZE or with BLKmode. If utype has precision
1459 higher than sizetype, use sizetype as tidx type and fold_convert the
1460 subtraction to sizetype.
1462 2024-01-20 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1464 * config/riscv/riscv.cc (riscv_init_cumulative_args): Suppress warning.
1465 (riscv_vector_mode_supported_any_target_p): Ditto.
1467 2024-01-19 Mikael Pettersson <mikpelinux@gmail.com>
1470 * config/m68k/m68k.cc (m68k_zero_call_used_regs): New function.
1471 (TARGET_ZERO_CALL_USED_REGS): Define.
1473 2024-01-19 Mikael Pettersson <mikpelinux@gmail.com>
1476 * config/m68k/m68k.cc (output_andsi3): Use QImode for
1477 address adjusted for 1-byte RMW access.
1478 (output_iorsi3): Likewise.
1479 (output_xorsi3): Likewise.
1481 2024-01-19 Kito Cheng <kito.cheng@sifive.com>
1483 * doc/invoke.texi (RISC-V Options): Add list of supported
1486 2024-01-19 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1489 * config/riscv/riscv-protos.h (RVV_VLMAX): Change to regno_reg_rtx[X0_REGNUM].
1490 (RVV_VUNDEF): Ditto.
1491 * config/riscv/riscv-vsetvl.cc: Add timevar.
1493 2024-01-19 Richard Biener <rguenther@suse.de>
1496 * lto-streamer-in.cc (lto_read_tree_1): When there isn't
1497 an early DIE but there should be, do not pretend there is.
1499 2024-01-19 Richard Biener <rguenther@suse.de>
1501 PR tree-optimization/113494
1502 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
1503 Handle endless loop on exit. Handle re-allocated PHI.
1505 2024-01-19 Jakub Jelinek <jakub@redhat.com>
1507 PR tree-optimization/113464
1508 * gimple-lower-bitint.cc (gimple_lower_bitint): Don't try to
1509 optimize loads into GIMPLE_ASM stmts.
1511 2024-01-19 Jakub Jelinek <jakub@redhat.com>
1513 PR tree-optimization/113463
1514 * gimple-ssa-warn-restrict.cc (builtin_memref::extend_offset_range):
1515 Only look through NOP_EXPRs if rhs1 doesn't have wider type than
1518 2024-01-19 Jakub Jelinek <jakub@redhat.com>
1520 PR tree-optimization/113459
1521 * tree-ssa-sccvn.cc (vn_walk_cb_data::push_partial_def): Use
1522 TREE_INT_CST_LOW of TYPE_SIZE_UNIT rather than GET_MODE_SIZE
1523 of SCALAR_INT_TYPE_MODE if type has BLKmode.
1524 (vn_reference_lookup_3): Likewise. Formatting fix.
1526 2024-01-19 Jakub Jelinek <jakub@redhat.com>
1527 Richard Biener <rguenther@suse.de>
1529 * cfgexpand.cc (discover_nonconstant_array_refs_r): Force non-BLKmode
1530 VAR_DECLs referenced in BLKmode VIEW_CONVERT_EXPRs into memory.
1531 * expr.cc (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: Do nothing
1532 but adjust_address also for BLKmode mode and MEM op0.
1534 2024-01-19 Palmer Dabbelt <palmer@rivosinc.com>
1536 * common/config/riscv/riscv-common.cc: Add Zihpm and Zicnttr
1539 2024-01-19 Kito Cheng <kito.cheng@sifive.com>
1541 * doc/invoke.texi (RISC-V Options): Document the syntax of -march.
1543 2024-01-19 Kito Cheng <kito.cheng@sifive.com>
1545 * common/config/riscv/riscv-common.cc
1546 (riscv_subset_list::parse_std_ext): Remove.
1547 (riscv_subset_list::parse_multiletter_ext): Remove.
1548 * config/riscv/riscv-subset.h
1549 (riscv_subset_list::parse_std_ext): Remove.
1550 (riscv_subset_list::parse_multiletter_ext): Remove.
1552 2024-01-19 Kito Cheng <kito.cheng@sifive.com>
1554 * common/config/riscv/riscv-common.cc
1555 (riscv_subset_list::parse_single_std_ext): New parameter.
1556 (riscv_subset_list::parse_single_multiletter_ext): Ditto.
1557 (riscv_subset_list::parse_single_ext): Ditto.
1558 (riscv_subset_list::parse): Relax the order for the input of ISA
1560 * config/riscv/riscv-subset.h
1561 (riscv_subset_list::parse_single_std_ext): New parameter.
1562 (riscv_subset_list::parse_single_multiletter_ext): Ditto.
1563 (riscv_subset_list::parse_single_ext): Ditto.
1565 2024-01-19 Kito Cheng <kito.cheng@sifive.com>
1567 * common/config/riscv/riscv-common.cc
1568 (riscv_subset_list::parse_base_ext): New.
1569 (riscv_subset_list::parse): Extract part of logic into
1570 riscv_subset_list::parse_base_ext.
1571 * config/riscv/riscv-subset.h (riscv_subset_list::parse_base_ext):
1574 2024-01-19 Kito Cheng <kito.cheng@sifive.com>
1576 * config/riscv/riscv.cc (riscv_override_options_internal): Tweak
1579 2024-01-19 Kuan-Lin Chen <rufus@andestech.com>
1581 * config/riscv/vector-crypto.md (UNSPEC_CLMUL): Rename to
1584 2024-01-19 Sandra Loosemore <sandra@codesourcery.com>
1587 * doc/extend.texi (Common Variable Attributes): Explain what
1588 happens when multiple variables with cleanups are in the same scope.
1590 2024-01-18 Sandra Loosemore <sandra@codesourcery.com>
1593 * doc/extend.texi (Common Function Attributes): Document that
1594 noinline also disables some interprocedural optimizations and
1595 improve flow to the part about using inline asm instead to
1596 disable calls from being optimized away completely. Remove the
1597 sentence that says noipa is mainly for internal compiler testing.
1599 2024-01-18 John David Anglin <danglin@gcc.gnu.org>
1601 PR tree-optimization/69807
1602 * config/pa/pa.cc (pa_option_override): Set flag_pie on TARGET_64BIT.
1604 2024-01-18 Brian Inglis <Brian.Inglis@Shaw.ca>
1607 * doc/invoke.texi (Option Summary): Remove -mcygwin and -mno-cygwin
1608 from x86 Windows Options.
1610 2024-01-18 Sandra Loosemore <sandra@codesourcery.com>
1613 * doc/extend.texi (C Extensions): Add new section to menu.
1614 (Function Attributes): Move dangling index entries to....
1615 (Const and Volatile Functions): New section.
1617 2024-01-18 David Malcolm <dmalcolm@redhat.com>
1619 PR middle-end/112684
1620 * toplev.cc (toplev::main): Don't ICE in
1621 -fdiagnostics-generate-patch when exiting after options,
1622 since no edit context will have been created.
1624 2024-01-18 Richard Biener <rguenther@suse.de>
1626 * tree-vect-stmts.cc (vectorizable_store): Do not pre-allocate
1629 2024-01-18 Iain Sandoe <iain@sandoe.co.uk>
1631 * Makefile.in: Emit ENABLE_DARWIN_AT_RPATH into site.exp
1632 when ENABLE_DARWIN_AT_RPATH_TRUE is not '#'.
1634 2024-01-18 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
1635 Jin Ma <jinma@linux.alibaba.com>
1636 Xianmiao Qu <cooper.qu@linux.alibaba.com>
1637 Christoph Müllner <christoph.muellner@vrull.eu>
1639 * config/riscv/thead.cc
1640 (th_asm_output_opcode): Rewrite some instructions.
1642 2024-01-18 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
1643 Jin Ma <jinma@linux.alibaba.com>
1644 Xianmiao Qu <cooper.qu@linux.alibaba.com>
1645 Christoph Müllner <christoph.muellner@vrull.eu>
1647 * config/riscv/riscv.md (none,thv,rvv): New attribute.
1648 (no,yes): Add an attribute to disable alternative
1649 for xtheadvector or RVV1.0.
1650 * config/riscv/vector.md:
1651 Disable alternatives that destination register overlaps
1652 source register group for xtheadvector.
1654 2024-01-18 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
1655 Jin Ma <jinma@linux.alibaba.com>
1656 Xianmiao Qu <cooper.qu@linux.alibaba.com>
1657 Christoph Müllner <christoph.muellner@vrull.eu>
1659 * config/riscv/riscv-vector-builtins-bases.cc
1660 (class th_loadstore_width): Define new builtin bases.
1661 (class th_extract): Define new builtin bases.
1662 (BASE): Define new builtin bases.
1663 * config/riscv/riscv-vector-builtins-bases.h:
1664 Define new builtin class.
1665 * config/riscv/riscv-vector-builtins-shapes.cc
1666 (struct th_loadstore_width_def): Define new builtin shapes.
1667 (struct th_indexed_loadstore_width_def):
1668 Define new builtin shapes.
1669 (struct th_extract_def): Define new builtin shapes.
1670 (SHAPE): Define new builtin shapes.
1671 * config/riscv/riscv-vector-builtins-shapes.h:
1672 Define new builtin shapes.
1673 * config/riscv/riscv-vector-builtins.cc (DEF_RVV_FUNCTION):
1674 Redefine DEF_RVV_FUNCTION for XTheadVector special intrinsics.
1675 * config/riscv/riscv-vector-builtins.h
1676 (enum required_ext): Add new XTheadVector member.
1677 (struct function_group_info): Likewise.
1678 * config/riscv/t-riscv:
1679 Add thead-vector-builtins-functions.def
1680 * config/riscv/thead-vector.md
1681 (@pred_mov_width<vlmem_op_attr><mode>): Add new patterns.
1682 (*pred_mov_width<vlmem_op_attr><mode>): Likewise.
1683 (@pred_store_width<vlmem_op_attr><mode>): Likewise.
1684 (@pred_strided_load_width<vlmem_op_attr><mode>): Likewise.
1685 (@pred_strided_store_width<vlmem_op_attr><mode>): Likewise.
1686 (@pred_indexed_load_width<vlmem_op_attr><mode>): Likewise.
1687 (@pred_th_extract<mode>): Likewise.
1688 (*pred_th_extract<mode>): Likewise.
1689 * config/riscv/thead-vector-builtins-functions.def: New file.
1691 2024-01-18 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
1692 Jin Ma <jinma@linux.alibaba.com>
1693 Xianmiao Qu <cooper.qu@linux.alibaba.com>
1694 Christoph Müllner <christoph.muellner@vrull.eu>
1696 * config.gcc: Add files for XTheadVector intrinsics.
1697 * config/riscv/autovec.md: Guard XTheadVector.
1698 * config/riscv/predicates.md: Disable immediate vl
1700 * config/riscv/riscv-c.cc (riscv_pragma_intrinsic):
1701 Add pragma for XTheadVector.
1702 * config/riscv/riscv-string.cc (riscv_expand_block_move):
1704 * config/riscv/riscv-v.cc (vls_mode_valid_p):
1706 * config/riscv/riscv-vector-builtins-bases.cc:
1707 Do not normalize vsetvl instructions for XTheadVector.
1708 * config/riscv/riscv-vector-builtins-shapes.cc (check_type):
1709 New check type function.
1710 (build_one): Adjust for XTheadVector.
1711 * config/riscv/riscv-vector-switch.def (ENTRY):
1712 Disable fractional mode for the XTheadVector extension.
1713 (TUPLE_ENTRY): Likewise.
1714 * config/riscv/riscv.cc (riscv_v_adjust_bytesize):
1716 (riscv_preferred_simd_mode): Likewsie.
1717 (riscv_autovectorize_vector_modes): Likewise.
1718 (riscv_vector_mode_supported_any_target_p): Likewise.
1719 (TARGET_VECTOR_MODE_SUPPORTED_ANY_TARGET_P): Likewise.
1720 * config/riscv/thead.cc (th_asm_output_opcode):
1721 Rewrite vsetvl instructions.
1722 * config/riscv/vector.md:
1723 Include thead-vector.md and change fractional LMUL
1725 * config/riscv/riscv_th_vector.h: New file.
1726 * config/riscv/thead-vector.md: New file.
1728 2024-01-18 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
1729 Jin Ma <jinma@linux.alibaba.com>
1730 Xianmiao Qu <cooper.qu@linux.alibaba.com>
1731 Christoph Müllner <christoph.muellner@vrull.eu>
1733 * config/riscv/riscv-protos.h (riscv_asm_output_opcode):
1734 Add new function to add assembler insn code prefix/suffix.
1735 (th_asm_output_opcode):
1736 Add Thead function to add assembler insn code prefix/suffix.
1737 * config/riscv/riscv.cc (riscv_asm_output_opcode):
1738 Implement function to add assembler insn code prefix/suffix.
1739 * config/riscv/riscv.h (ASM_OUTPUT_OPCODE):
1740 Add new function to add assembler insn code prefix/suffix.
1741 * config/riscv/thead.cc (th_asm_output_opcode):
1742 Implement Thead function to add assembler insn code
1745 2024-01-18 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
1746 Jin Ma <jinma@linux.alibaba.com>
1747 Xianmiao Qu <cooper.qu@linux.alibaba.com>
1748 Christoph Müllner <christoph.muellner@vrull.eu>
1750 * common/config/riscv/riscv-common.cc
1751 (riscv_subset_list::parse): Add new vendor extension.
1752 * config/riscv/riscv-c.cc (riscv_cpu_cpp_builtins):
1754 * config/riscv/riscv.opt: Add new mask.
1756 2024-01-18 Iain Sandoe <iain@sandoe.co.uk>
1758 * config/darwin.h (DARWIN_RPATH_SPEC): Arrange for the %P spec
1759 to be conditional on macosx-version-min.
1761 2024-01-18 Iain Sandoe <iain@sandoe.co.uk>
1763 * config/darwin.cc (darwin_objc1_section): Use the correct
1764 meta-data version for constant strings.
1765 (machopic_select_section): Assert if we fail to handle CFString
1766 sections as Obejctive-C meta-data or drectly.
1768 2024-01-18 Iain Sandoe <iain@sandoe.co.uk>
1770 * lto-section-names.h (OFFLOAD_SECTION_NAME_PREFIX,
1771 OFFLOAD_VAR_TABLE_SECTION_NAME, OFFLOAD_FUNC_TABLE_SECTION_NAME,
1772 OFFLOAD_IND_FUNC_TABLE_SECTION_NAME): Provide Mach-O syntax
1773 versions when the object format is Mach-O.
1775 2024-01-18 Iain Sandoe <iain@sandoe.co.uk>
1778 * config/darwin.cc (machopic_select_section): Handle C and C++
1780 (darwin_rename_builtins): Move this out of the CFString code.
1781 (darwin_libc_has_function): Likewise.
1782 (darwin_build_constant_cfstring): Create an anonymous var to
1784 * config/darwin.h (ASM_OUTPUT_LABELREF): Handle constant
1787 2024-01-18 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
1790 * haifa-sched.cc (dep_list_size): Make global.
1791 * sched-deps.cc (find_inc): Use instead of sd_lists_size().
1792 * sched-int.h (dep_list_size): Declare.
1794 2024-01-18 Martin Jambor <mjambor@suse.cz>
1796 PR tree-optimization/110422
1797 * tree-sra.cc (scan_function): Disqualify bases of operands of asm
1800 2024-01-18 Richard Biener <rguenther@suse.de>
1802 PR tree-optimization/113475
1803 * gimple-range-phi.h (phi_analyzer::m_phi_groups): New.
1804 * gimple-range-phi.cc (phi_analyzer::phi_analyzer): Initialize.
1805 (phi_analyzer::~phi_analyzer): Deallocate and free collected
1807 (phi_analyzer::process_phi): Record allocated phi_groups.
1809 2024-01-18 Richard Biener <rguenther@suse.de>
1811 * tree-vect-stmts.cc (vectorizable_store): Do not allocate
1812 storage for gvec_oprnds elements.
1814 2024-01-18 Richard Biener <rguenther@suse.de>
1816 * tree-vect-loop.cc (vec_init_loop_exit_info): Adjust comment,
1817 prefer all later exits we can handle.
1818 (vect_analyze_loop_form): Free the allocated loop body.
1821 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1823 * config/avr/avr-log.cc: Tabify.
1825 2024-01-18 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1827 * config/riscv/autovec.md: Support vi variant.
1829 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1831 * config/avr/avr-devices.cc: Tabify.
1833 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1835 * config/avr/avr-c.cc: Tabify.
1837 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1839 * config/avr/driver-avr.cc: Tabify.
1841 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1843 * config/avr/gen-avr-mmcu-texi.cc: Tabify.
1845 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1847 * config/avr/gen-avr-mmcu-specs.cc: Tabify.
1849 2024-01-18 Jakub Jelinek <jakub@redhat.com>
1851 * config/riscv/riscv.opt (mshorten-memrefs, mrelax, mcsr-check,
1852 minline-strcmp, minline-strncmp, minline-strlen,
1853 -param=riscv-vector-abi): Remove Bool keywords.
1855 2024-01-18 Jakub Jelinek <jakub@redhat.com>
1858 * config/i386/i386.cc (x86_function_profiler): Add -masm=intel
1859 support. Add missing space after , in emitted assembly in some
1860 cases. Formatting fixes.
1862 2024-01-18 Xi Ruoyao <xry111@xry111.site>
1864 * config/loongarch/loongarch.md (movsi_internal): Remove
1867 2024-01-18 Georg-Johann Lay <avr@gjlay.de>
1869 * config/avr/gen-avr-mmcu-specs.cc (diagnose_rodata_in_ram): Fix typo
1870 in the diagnostic, and capitalize the device name.
1871 (print_mcu): Generate specs such that:
1872 <*check_rodata_in_ram>: New.
1873 <*cc1_misc>: Use check_rodata_in_ram instead of cc1_rodata_in_ram.
1874 <*link_misc>: Use check_rodata_in_ram instead of link_rodata_in_ram.
1875 <*cc1_rodata_in_ram, *link_rodata_in_ram>: Remove.
1877 2024-01-18 Jakub Jelinek <jakub@redhat.com>
1880 * common.opt (ffold-mem-offsets): Remove Target and Bool keywords, add
1881 Common and Optimization.
1883 2024-01-18 Richard Biener <rguenther@suse.de>
1885 PR tree-optimization/113431
1886 * tree-vect-data-refs.cc (vect_preserves_scalar_order_p):
1887 When there is an invariant load we might not preserve
1890 2024-01-18 Richard Biener <rguenther@suse.de>
1892 PR tree-optimization/113374
1893 * tree-ssa-operands.h (SET_PHI_ARG_DEF_ON_EDGE): New.
1894 * tree-vect-loop.cc (move_early_exit_stmts): Update
1896 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
1897 Refactor. Preserve virtual LC PHIs on all exits.
1899 2024-01-18 Lulu Cheng <chenglulu@loongson.cn>
1901 * config/loongarch/loongarch.cc (loongarch_split_symbol):
1902 Assign the '/u' attribute to the mem.
1904 2024-01-18 Sandra Loosemore <sandra@codesourcery.com>
1906 PR middle-end/110847
1907 * doc/invoke.texi (Option Summary): Document negative forms of
1908 -Wtsan and -Wxor-used-as-pow.
1909 (Warning Options): Likewise.
1911 2024-01-18 Juzhe-Zhong <juzhe.zhong@rivai.ai>
1914 * config/riscv/riscv-vsetvl.cc (pre_vsetvl::earliest_fuse_vsetvl_info): Fix bug.
1916 2024-01-18 Sandra Loosemore <sandra@codesourcery.com>
1918 * doc/extend.texi (Common Function Attributes): Re-alphabetize
1920 (Common Variable Attributes): Likewise.
1921 (Common Type Attributes): Likewise.
1923 2024-01-17 Sandra Loosemore <sandra@codesourcery.com>
1925 PR middle-end/111659
1926 * doc/extend.texi (Common Variable Attributes): Fix long lines
1927 in documentation of strict_flex_array + other minor copy-editing.
1928 Add a cross-reference to -Wstrict-flex-arrays.
1929 * doc/invoke.texi (Option Summary): Fix whitespace in tables
1930 before -fstrict-flex-arrays and -Wstrict-flex-arrays.
1931 (C Dialect Options): Combine the docs for the two
1932 -fstrict-flex-arrays forms into a single entry. Note this option
1933 is for C/C++ only. Add a cross-reference to -Wstrict-flex-arrays.
1934 (Warning Options): Note -Wstrict-flex-arrays is for C/C++ only.
1935 Minor copy-editing. Add cross references to the strict_flex_array
1936 attribute and -fstrict-flex-arrays option. Add note that this
1937 option depends on -ftree-vrp.
1939 2024-01-17 Andrew Pinski <quic_apinski@quicinc.com>
1942 * config/aarch64/predicates.md (aarch64_ldp_reg_operand): For subreg,
1943 only allow REG operands instead of allowing all.
1945 2024-01-17 Vineet Gupta <vineetg@rivosinc.com>
1947 * config/riscv/riscv-vsetvl.cc (earliest_fuse_vsetvl_info):
1948 Remove redundant checks in else condition for readablity.
1949 (earliest_fuse_vsetvl_info) Print iteration count in debug
1951 (earliest_fuse_vsetvl_info) Fix misleading vsetvl info
1952 dump details in certain cases.
1954 2024-01-17 Vineet Gupta <vineetg@rivosinc.com>
1956 * config/riscv/riscv.opt: New -param=vsetvl-strategy.
1957 * config/riscv/riscv-opts.h: New enum vsetvl_strategy_enum.
1958 * config/riscv/riscv-vsetvl.cc
1959 (pre_vsetvl::pre_global_vsetvl_info): Use vsetvl_strategy.
1960 (pass_vsetvl::execute): Use vsetvl_strategy.
1962 2024-01-17 Jan Hubicka <jh@suse.cz>
1964 * ipa-polymorphic-call.cc (ipa_polymorphic_call_context::set_by_invariant): Remove
1965 accidental hack reseting offset.
1967 2024-01-17 Jan Hubicka <jh@suse.cz>
1969 * config/i386/i386-options.cc (ix86_option_override_internal): Fix
1970 handling of X86_TUNE_AVOID_512FMA_CHAINS.
1972 2024-01-17 Jan Hubicka <jh@suse.cz>
1973 Jakub Jelinek <jakub@redhat.com>
1975 PR tree-optimization/110852
1976 * predict.cc (expr_expected_value_1): Fix profile merging of PHI and
1978 (get_predictor_value): Handle PRED_COMBINED_VALUE_PREDICTIONS and
1979 PRED_COMBINED_VALUE_PREDICTIONS_PHI
1980 * predict.def (PRED_COMBINED_VALUE_PREDICTIONS): New predictor.
1981 (PRED_COMBINED_VALUE_PREDICTIONS_PHI): New predictor.
1983 2024-01-17 Jakub Jelinek <jakub@redhat.com>
1985 PR tree-optimization/113421
1986 * gimple-lower-bitint.cc (stmt_needs_operand_addr): Adjust function
1988 (bitint_dom_walker::before_dom_children): Add g temporary to simplify
1989 formatting. Start at vop rather than cvop even if stmt is a store
1990 and needs_operand_addr.
1992 2024-01-17 Jakub Jelinek <jakub@redhat.com>
1994 PR middle-end/113410
1995 * gimple-ssa-warn-access.cc (pass_waccess::maybe_check_access_sizes):
1996 If access_nelts is integral with larger precision than sizetype,
1997 fold_convert it to sizetype.
1999 2024-01-17 Jakub Jelinek <jakub@redhat.com>
2001 PR tree-optimization/113408
2002 * gimple-lower-bitint.cc (bitint_large_huge::handle_stmt): For
2003 VIEW_CONVERT_EXPR, pass TREE_OPERAND (rhs1, 0) rather than rhs1
2006 2024-01-17 Jakub Jelinek <jakub@redhat.com>
2008 PR middle-end/113406
2009 * ipa-strub.cc (pass_ipa_strub::execute): Check aggregate_value_p
2010 regardless of whether is_gimple_reg_type (restype) or not.
2012 2024-01-17 Jakub Jelinek <jakub@redhat.com>
2014 * tree-into-ssa.cc (pass_build_ssa::gate): Fix comment typo,
2015 funcions -> functions, and use were instead of was.
2016 * gengtype.cc (dump_typekind): Fix comment typos, funcion -> function
2017 and guaranteee -> guarantee.
2018 * attribs.h (struct attr_access): Fix comment typo funcion -> function.
2020 2024-01-17 Jakub Jelinek <jakub@redhat.com>
2022 PR middle-end/113409
2023 * omp-general.cc (omp_adjust_for_condition): Handle BITINT_TYPE like
2025 (omp_extract_for_data): Use build_bitint_type rather than
2026 build_nonstandard_integer_type if either iter_type or loop->v type
2028 * omp-expand.cc (expand_omp_for_generic,
2029 expand_omp_taskloop_for_outer, expand_omp_taskloop_for_inner): Handle
2030 BITINT_TYPE like INTEGER_TYPE.
2032 2024-01-17 Richard Biener <rguenther@suse.de>
2034 PR tree-optimization/113371
2035 * tree-vect-data-refs.cc (vect_enhance_data_refs_alignment):
2036 Do not peel when LOOP_VINFO_EARLY_BREAKS_VECT_PEELED.
2037 * tree-vect-loop-manip.cc (vect_do_peeling): Assert we do
2038 not perform prologue peeling when LOOP_VINFO_EARLY_BREAKS_VECT_PEELED.
2040 2024-01-17 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
2042 PR rtl-optimization/96388
2043 PR rtl-optimization/111554
2044 * sched-deps.cc (find_inc): Avoid exponential behavior.
2046 2024-01-17 Sandra Loosemore <sandra@codesourcery.com>
2049 * doc/invoke.texi (Option Summary): Move -Wuseless-cast
2050 from C++ Language Options to Warning Options. Add entry for
2052 (C++ Dialect Options): Move -Wuse-after-free and -Wuseless-cast
2054 (Warning Options): ...to here. Minor copy-editing to fix typo
2057 2024-01-17 YunQiang Su <syq@gcc.gnu.org>
2059 * config/mips/mips.cc (mips_compute_frame_info): If another
2060 register is used as global_pointer, mark $GP live false.
2062 2024-01-17 Sandra Loosemore <sandra@codesourcery.com>
2065 * doc/extend.texi (BPF Built-in Functions): Wrap long lines and
2066 give the section a light copy-editing pass.
2068 2024-01-16 Wilco Dijkstra <wilco.dijkstra@arm.com>
2070 * config/aarch64/aarch64-cores.def (AARCH64_CORE): Add 'cobalt-100' CPU.
2071 * config/aarch64/aarch64-tune.md: Regenerated.
2072 * doc/invoke.texi (-mcpu): Add cobalt-100 core.
2074 2024-01-16 Wilco Dijkstra <wilco.dijkstra@arm.com>
2077 * config/aarch64/aarch64.cc (aarch64_legitimize_address): Reassociate
2078 badly formed CONST expressions.
2080 2024-01-16 Daniel Cederman <cederman@gaisler.com>
2082 * config/sparc/sparc.cc (next_active_non_empty_insn): Length 0 treated as empty
2084 2024-01-16 Daniel Cederman <cederman@gaisler.com>
2086 * config/sparc/sparc.cc (atomic_insn_for_leon3_p): Treat membar_storeload as atomic
2087 * config/sparc/sync.md (membar_storeload): Turn into named insn
2088 and add GR712RC errata workaround.
2089 (membar_v8): Add GR712RC errata workaround.
2091 2024-01-16 Andreas Larsson <andreas@gaisler.com>
2093 * config/sparc/sync.md (*membar_storeload_leon3): Remove
2094 (*membar_storeload): Enable for LEON
2096 2024-01-16 Jakub Jelinek <jakub@redhat.com>
2098 PR tree-optimization/113372
2100 PR middle-end/110115
2101 PR middle-end/111422
2102 * cfgexpand.cc (add_scope_conflicts_2): New function.
2103 (add_scope_conflicts_1): Use it.
2105 2024-01-16 Georg-Johann Lay <avr@gjlay.de>
2107 * config/avr/avr-mcus.def (avr16eb14, avr16eb20, avr16eb28, avr16eb32)
2108 (avr16ea28, avr16ea32, avr16ea48, avr32ea28, avr32ea32, avr32ea48): Add.
2109 * doc/avr-mmcu.texi: Regenerate.
2111 2024-01-16 Feng Xue <fxue@os.amperecomputing.com>
2113 PR tree-optimization/113091
2114 * tree-vect-slp.cc (vect_slp_has_scalar_use): New function.
2115 (vect_bb_slp_mark_live_stmts): New parameter scalar_use_map, check
2116 scalar use with new function.
2117 (vect_bb_slp_mark_live_stmts): New function as entry to existing
2118 overriden functions with same name.
2119 (vect_slp_analyze_operations): Call new entry function to mark
2122 2024-01-16 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2125 * config/riscv/riscv.cc (riscv_override_options_internal): Report sorry
2126 for RVV in big-endian mode.
2128 2024-01-16 Yanzhang Wang <yanzhang.wang@intel.com>
2130 * config/riscv/riscv.cc (riscv_arg_has_vector): Delete.
2131 (riscv_pass_in_vector_p): Delete.
2132 (riscv_init_cumulative_args): Delete the checking.
2133 (riscv_get_arg_info): Delete the checking.
2134 (riscv_function_value): Delete the checking.
2135 * config/riscv/riscv.h: Delete the member for checking.
2137 2024-01-15 Georg-Johann Lay <avr@gjlay.de>
2139 * doc/invoke.texi (AVR Options) [-mskip-bug]: Add documentation.
2141 2024-01-15 Liao Shihua <shihua@iscas.ac.cn>
2143 * config.gcc: Include riscv_bitmanip.h.
2144 * config/riscv/bitmanip.md: Changed mode form X to GPR in orcb and clmul pattern.
2145 * config/riscv/crypto.md: Changed mode form X to GPR in brev8 pattern.
2146 * config/riscv/riscv-builtins.cc (AVAIL): Adding new bitmanip builtins.
2147 (RISCV_BUILTIN_NO_PREFIX): New helper macro.
2148 * config/riscv/riscv-cmo.def (RISCV_BUILTIN): Add '_32'/'_64' postfix to builtins.
2149 * config/riscv/riscv-ftypes.def (2): New ftypes.
2150 * config/riscv/riscv-scalar-crypto.def (RISCV_BUILTIN): New builtins.
2151 (RISCV_BUILTIN_NO_PREFIX): Likewise.
2152 * config/riscv/riscv_bitmanip.h: New file.
2154 2024-01-15 Liao Shihua <shihua@iscas.ac.cn>
2156 * config.gcc: Include riscv_crypto.h.
2157 * config/riscv/riscv_crypto.h: New file.
2159 2024-01-15 Vladimir N. Makarov <vmakarov@redhat.com>
2161 PR middle-end/113354
2162 * lra-constraints.cc (curr_insn_transform): Spill pseudo only used
2163 in the insn if the corresponding operand does not require hard
2166 2024-01-15 Georg-Johann Lay <avr@gjlay.de>
2169 * config/avr/avr.h (EXTRA_SPEC_FUNCTIONS): Add no-devlib, avr_no_devlib.
2170 * config/avr/driver-avr.cc (avr_no_devlib): New function.
2171 (avr_devicespecs_file): Use it to remove -nodevicelib from the
2172 options for cores only.
2173 * config/avr/avr-arch.h (avr_get_parch): New prototype.
2174 * config/avr/avr-devices.cc (avr_get_parch): New function.
2176 2024-01-15 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2179 * config/riscv/riscv-protos.h (struct regmove_vector_cost): Add vector to scalar regmove.
2180 * config/riscv/riscv-vector-costs.cc (adjust_stmt_cost): Ditto.
2181 * config/riscv/riscv.cc (riscv_builtin_vectorization_cost): Adjust vec_construct cost.
2183 2024-01-15 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2186 * config/riscv/riscv-vector-costs.cc (costs::adjust_vect_cost_per_loop): New function.
2187 (costs::finish_cost): Adjust cost for LOOP LEN with NITERS < VF.
2188 * config/riscv/riscv-vector-costs.h: New function.
2190 2024-01-15 Richard Biener <rguenther@suse.de>
2192 PR tree-optimization/113385
2193 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
2194 First redirect, then split the exit edge.
2196 2024-01-15 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2198 * config/riscv/riscv-vector-costs.cc (costs::analyze_loop_vinfo):
2199 Remove m_num_vector_iterations.
2200 * config/riscv/riscv-vector-costs.h: Ditto.
2202 2024-01-15 Andrew Pinski <quic_apinski@quicinc.com>
2205 * config/avr/avr.opt (-mdouble, -mlong-double): Add "Save" flag.
2206 (-mbranch-cost): Set "Optimization" flag.
2208 2024-01-15 Jakub Jelinek <jakub@redhat.com>
2210 PR tree-optimization/113370
2211 * gimple-lower-bitint.cc (bitint_large_huge::handle_operand): Only
2212 set rem to prec % (2 * limb_prec) if m_upwards_2limb, otherwise
2213 set it to just prec % limb_prec.
2215 2024-01-15 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2218 * config/riscv/vector.md: Fix ternary attributes.
2220 2024-01-14 Georg-Johann Lay <avr@gjlay.de>
2223 * configure.ac [target=avr]: Check availability of emulations
2224 avrxmega2_flmap and avrxmega4_flmap, resulting in new config vars
2225 HAVE_LD_AVR_AVRXMEGA2_FLMAP and HAVE_LD_AVR_AVRXMEGA4_FLMAP.
2226 * configure: Regenerate.
2227 * config.in: Regenerate.
2228 * doc/invoke.texi (AVR Options): Document -mflmap, -mrodata-in-ram,
2229 __AVR_HAVE_FLMAP__, __AVR_RODATA_IN_RAM__.
2230 * config/avr/avr.opt (-mflmap, -mrodata-in-ram): New options.
2231 * config/avr/avr-arch.h (enum avr_device_specific_features):
2233 * config/avr/avr-mcus.def (AVR_MCU) [avr64*, avr128*]: Set isa flag
2235 * config/avr/avr.cc (avr_arch_index, avr_has_rodata_p): New vars.
2236 (avr_set_core_architecture): Set avr_arch_index.
2237 (have_avrxmega2_flmap, have_avrxmega4_flmap)
2238 (have_avrxmega3_rodata_in_flash): Set new static const bool according
2239 to configure results.
2240 (avr_rodata_in_flash_p): New function using them.
2241 (avr_asm_init_sections): Let readonly_data_section->unnamed.callback
2242 track avr_need_copy_data_p only if not avr_rodata_in_flash_p().
2243 (avr_asm_named_section): Track avr_has_rodata_p.
2244 (avr_file_end): Emit __do_copy_data also when avr_has_rodata_p
2245 and not avr_rodata_in_flash_p ().
2246 * config/avr/specs.h (CC1_SPEC): Add %(cc1_rodata_in_ram).
2247 (LINK_SPEC): Add %(link_rodata_in_ram).
2248 (LINK_ARCH_SPEC): Remove.
2249 * config/avr/gen-avr-mmcu-specs.cc (have_avrxmega3_rodata_in_flash)
2250 (have_avrxmega2_flmap, have_avrxmega4_flmap): Set new static
2251 const bool according to configure results.
2252 (diagnose_mrodata_in_ram): New function.
2253 (print_mcu): Generate specs with the following changes:
2254 <*cc1_misc, *asm_misc, *link_misc>: New specs so that we don't
2255 need to extend avr/specs.h each time we add a new bell or whistle.
2256 <*cc1_rodata_in_ram, *link_rodata_in_ram>: New specs to diagnose
2257 -m[no-]rodata-in-ram.
2258 <*cpp_rodata_in_ram>: New. Does -D__AVR_RODATA_IN_RAM__=0/1.
2259 <*cpp_mcu>: Add -D__AVR_AVR_FLMAP__ if it applies.
2260 <*cpp>: Add %(cpp_rodata_in_ram).
2261 <*link_arch>: Use emulation avrxmega2_flmap, avrxmega4_flmap as
2263 <*self_spec>: Add -mflmap or %<mflmap as needed.
2265 2024-01-14 Jeff Law <jlaw@ventanamicro.com>
2267 * config/mips/mips.md (ior<mode>3_mips16_asmacro): Use SImode,
2268 not the GPR iterator. Adjust pattern name and mode attribute
2271 2024-01-13 Jakub Jelinek <jakub@redhat.com>
2273 PR tree-optimization/113361
2274 * gimple-lower-bitint.cc (bitint_large_huge::handle_operand_addr):
2275 Fix up determination of the type for > limb_prec constants.
2277 2024-01-12 Georg-Johann Lay <avr@gjlay.de>
2279 * doc/extend.texi (AVR Named Address Spaces, Limitations and Caveats):
2280 Add web-link to the avr-gcc wiki.
2282 2024-01-12 Georg-Johann Lay <avr@gjlay.de>
2284 * doc/extend.texi (AVR Variable Attributes) [address]: Remove
2285 documentation for a version without argument, which is not supported.
2287 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2289 * config/arm/arm_neon.h
2290 (vld1_u8_x4, vld1_u16_x4, vld1_u32_x4, vld1_u64_x4): New.
2291 (vld1_s8_x4, vld1_s16_x4, vld1_s32_x4, vld1_s64_x4): New.
2292 (vld1_f16_x4, vld1_f32_x4): New.
2293 (vld1_p8_x4, vld1_p16_x4, vld1_p64_x4): New.
2294 (vld1_bf16_x4): New.
2295 (vld1q_types_x4): Updated to use vld1q_x4
2296 from arm_neon_builtins.def
2297 * config/arm/arm_neon_builtins.def
2298 (vld1_x4): Updated entries.
2299 (vld1q_x4): New entries, but comes from the old vld1_x4
2300 * config/arm/neon.md
2301 (neon_vld1q_x4<mode>): Updated from neon_vld1_x4<mode>.
2303 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2305 * config/arm/arm_neon.h
2306 (vld1_u8_x3, vld1_u16_x3, vld1_u32_x3, vld1_u64_x3): New.
2307 (vld1_s8_x3, vld1_s16_x3, vld1_s32_x3, vld1_s64_x3): New.
2308 (vld1_f16_x3, vld1_f32_x3): New.
2309 (vld1_p8_x3, vld1_p16_x3, vld1_p64_x3): New.
2310 (vld1_bf16_x3): New.
2311 (vld1q_types_x3): Updated to use vld1q_x3 from
2312 arm_neon_builtins.def
2313 * config/arm/arm_neon_builtins.def
2314 (vld1_x3): Updated entries.
2315 (vld1q_x3): New entries, but comes from the old vld1_x2
2316 * config/arm/neon.md
2317 (neon_vld1q_x3<mode>): Updated from neon_vld1_x3<mode>.
2319 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2321 * config/arm/arm_neon.h
2322 (vld1_u8_x2, vld1_u16_x2, vld1_u32_x2, vld1_u64_x2): New.
2323 (vld1_s8_x2, vld1_s16_x2, vld1_s32_x2, vld1_s64_x2): New.
2324 (vld1_f16_x2, vld1_f32_x2): New.
2325 (vld1_p8_x2, vld1_p16_x2, vld1_p64_x2): New.
2326 (vld1_bf16_x2): New.
2327 (vld1q_types_x2): Updated to use vld1q_x2 from
2328 arm_neon_builtins.def
2329 * config/arm/arm_neon_builtins.def
2330 (vld1_x2): Updated entries.
2331 (vld1q_x2): New entries, but comes from the old vld1_x2
2332 * config/arm/neon.md
2333 (neon_vld1<VMEMX2_q>_x2<VDQX:mode>): Updated from
2336 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2338 * config/arm/arm_neon.h
2339 (vst1q_u8_x4, vst1q_u16_x4, vst1q_u32_x4, vst1q_u64_x4): New.
2340 (vst1q_s8_x4, vst1q_s16_x4, vst1q_s32_x4, vst1q_s64_x4): New.
2341 (vst1q_f16_x4, vst1q_f32_x4): New.
2342 (vst1q_p8_x4, vst1q_p16_x4, vst1q_p64_x4): New.
2343 (vst1q_bf16_x4): New.
2344 * config/arm/arm_neon_builtins.def (vst1q_x4): New entries.
2345 * config/arm/neon.md
2346 (neon_vst1q_x4<mode>): New.
2347 (neon_vst1x4qa<mode>, neon_vst1x4qb<mode>): New.
2348 * config/arm/unspecs.md
2349 (UNSPEC_VST1X4A, UNSPEC_VST1X4B): New.
2351 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2353 * config/arm/arm_neon.h
2354 (vst1q_u8_x3, vst1q_u16_x3, vst1q_u32_x3, vst1q_u64_x3): New.
2355 (vst1q_s8_x3, vst1q_s16_x3, vst1q_s32_x3, vst1q_s64_x3): New.
2356 (vst1q_f16_x3, vst1q_f32_x3): New.
2357 (vst1q_p8_x3, vst1q_p16_x3, vst1q_p64_x3): New.
2358 (vst1q_bf16_x3): New.
2359 * config/arm/arm_neon_builtins.def (vst1q_x3): New entries.
2360 * config/arm/neon.md
2361 (neon_vst1q_x3<mode>): New.
2362 (neon_vld1x3qa<mode>, neon_vst1x3qb<mode>): New.
2363 * config/arm/unspecs.md
2364 (UNSPEC_VST1X3A, UNSPEC_VST1X3B): New.
2366 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2368 * config/arm/arm_neon.h
2369 (vst1q_u8_x2, vst1q_u16_x2, vst1q_u32_x2, vst1q_u64_x2): New.
2370 (vst1q_s8_x2, vst1q_s16_x2, vst1q_s32_x2, vst1q_s64_x2): New.
2371 (vst1q_f16_x2, vst1q_f32_x2): New.
2372 (vst1q_p8_x2, vst1q_p16_x2, vst1q_p64_x2): New.
2373 (vst1q_bf16_x2): New.
2374 * config/arm/arm_neon_builtins.def (vst1<_x2): New entries.
2375 * config/arm/neon.md
2376 (neon_vst1<VMEMX2_q>_x2<VDQX:mode>): Updated from
2378 * config/arm/iterators.md
2379 (VMEMX2): New mode iterator.
2380 (VMEMX2_q): New mode attribute.
2382 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2384 * config/arm/arm_neon.h
2385 (vst1_u8_x4, vst1_u16_x4, vst1_u32_x4, vst1_u64_x4): New.
2386 (vst1_s8_x4, vst1_s16_x4, vst1_s32_x4, vst1_s64_x4): New.
2387 (vst1_f16_x4, vst1_f32_x4): New.
2388 (vst1_p8_x4, vst1_p16_x4, vst1_p64_x4): New.
2389 (vst1_bf16_x4): New.
2390 * config/arm/arm_neon_builtins.def (vst1_x4): New entries.
2391 * config/arm/neon.md (vst1_x4<mode>): New.
2393 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2395 * config/arm/arm_neon.h
2396 (vst1_u8_x3, vst1_u16_x3, vst1_u32_x3, vst1_u64_x3): New.
2397 (vst1_s8_x3, vst1_s16_x3, vst1_s32_x3, vst1_s64_x3): New.
2398 (vst1_f16_x3, vst1_f32_x3): New.
2399 (vst1_p8_x3, vst1_p16_x3, vst1_p64_x3): New.
2400 (vst1_bf16_x3): New.
2401 * config/arm/arm_neon_builtins.def (vst1_x3): New entries.
2402 * config/arm/neon.md (vst1_x3<mode>): New.
2404 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2406 * config/arm/arm_neon.h
2407 (vst1_u8_x2, vst1_u16_x2, vst1_u32_x2, vst1_u64_x2): New.
2408 (vst1_s8_x2, vst1_s16_x2, vst1_s32_x2, vst1_s64_x2): New.
2409 (vst1_f16_x2, vst1_f32_x2): New.
2410 (vst1_p8_x2, vst1_p16_x2, vst1_p64_x2): New.
2411 (vst1_bf16_x2): New.
2412 * config/arm/arm_neon_builtins.def (vst1_x2): New entries.
2413 * config/arm/neon.md (vst1_x2<mode>): New.
2415 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2417 * config/arm/arm_neon.h
2418 (vld1q_u8_x4, vld1q_u16_x4, vld1q_u32_x4, vld1q_u64_x4): New.
2419 (vld1q_s8_x4, vld1q_s16_x4, vld1q_s32_x4, vld1q_s64_x4): New.
2420 (vld1q_f16_x4, vld1q_f32_x4): New.
2421 (vld1q_p8_x4, vld1q_p16_x4, vld1q_p64_x4): New.
2422 (vld1q_bf16_x4): New.
2423 * config/arm/arm_neon_builtins.def (vld1_x4): New entries.
2424 * config/arm/neon.md
2425 (neon_vld1_x4<mode>): New.
2426 (neon_vld1x4qa<mode>, neon_vld1x4qb<mode>): New
2427 * config/arm/unspecs.md
2428 (UNSPEC_VLD1X4A, UNSPEC_VLD1X4B): New.
2430 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2432 * config/arm/arm_neon.h
2433 (vld1q_u8_x3, vld1q_u16_x3, vld1q_u32_x3, vld1q_u64_x3): New.
2434 (vld1q_s8_x3, vld1q_s16_x3, vld1q_s32_x3, vld1q_s64_x3): New.
2435 (vld1q_f16_x3, vld1q_f32_x3): New.
2436 (vld1q_p8_x3, vld1q_p16_x3, vld1q_p64_x3): New.
2437 (vld1q_bf16_x3): New.
2438 * config/arm/arm_neon_builtins.def (vld1_x3): New entries.
2439 * config/arm/neon.md
2440 (neon_vld1_x3<mode>): New.
2441 (neon_vld1x3qa<mode>, neon_vld1x3qb<mode>): New.
2442 * config/arm/unspecs.md
2443 (UNSPEC_VLD1X3A, UNSPEC_VLD1X3B): New.
2445 2024-01-12 Ezra Sitorus <ezra.sitorus@arm.com>
2447 * config/arm/arm_neon.h
2448 (vld1q_u8_x2, vld1q_u16_x2, vld1q_u32_x2, vld1q_u64_x2): New.
2449 (vld1q_s8_x2, vld1q_s16_x2, vld1q_s32_x2, vld1q_s64_x2): New.
2450 (vld1q_f16_x2, vld1q_f32_x2): New.
2451 (vld1q_p8_x2, vld1q_p16_x2, vld1q_p64_x2): New.
2452 (vld1q_bf16_x2): New.
2453 * config/arm/arm_neon_builtins.def (vld1_x2): New entries.
2454 * config/arm/neon.md (vld1_x2<mode>): New.
2456 2024-01-12 Tamar Christina <tamar.christina@arm.com>
2458 PR tree-optimization/113287
2459 * doc/sourcebuild.texi (check_effective_target_bitint65535): New.
2461 2024-01-12 Tamar Christina <tamar.christina@arm.com>
2463 * tree-vect-loop-manip.cc (vect_loop_versioning): Replace single_exit.
2464 * tree-vect-loop.cc (vect_transform_loop): Likewise.
2466 2024-01-12 Tamar Christina <tamar.christina@arm.com>
2468 PR tree-optimization/113178
2469 * tree-vect-loop.cc (vect_create_epilog_for_reduction): Fill in all
2472 2024-01-12 Tamar Christina <tamar.christina@arm.com>
2474 PR tree-optimization/113237
2475 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg): Use
2476 existing LCSSA variable for exit when all exits are early break.
2478 2024-01-12 Tamar Christina <tamar.christina@arm.com>
2480 PR tree-optimization/113137
2481 PR tree-optimization/113136
2482 PR tree-optimization/113172
2483 PR tree-optimization/113178
2484 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
2485 Maintain PHIs on inverted loops.
2486 (vect_do_peeling): Maintain virtual PHIs on inverted loops.
2487 * tree-vect-loop.cc (vec_init_loop_exit_info): Pick exit closes to
2489 (vect_create_loop_vinfo): Record all conds instead of only alt ones.
2491 2024-01-12 Tamar Christina <tamar.christina@arm.com>
2493 PR tree-optimization/113135
2494 * tree-vect-data-refs.cc (vect_analyze_early_break_dependences): Rework
2495 dependency analysis.
2497 2024-01-12 Iain Sandoe <iain@sandoe.co.uk>
2499 * config/rs6000/host-darwin.cc (segv_handler): Use the revised
2500 diagnostics class member name for abort of error.
2502 2024-01-12 Georg-Johann Lay <avr@gjlay.de>
2504 * config/avr/avr.cc (avr_handle_addr_attribute): Move "..." from
2505 format string to %s argument.
2507 2024-01-12 John David Anglin <danglin@gcc.gnu.org>
2508 Jakub Jelinek <jakub@redhat.com>
2510 PR middle-end/113182
2511 * varasm.cc (process_pending_assemble_externals,
2512 assemble_external_libcall): Use targetm.strip_name_encoding
2513 before calling get_identifier.
2515 2024-01-12 Richard Sandiford <richard.sandiford@arm.com>
2518 * config/aarch64/aarch64.h (machine_function::advsimd_zero_insn):
2519 New member variable.
2520 * config/aarch64/aarch64-protos.h (aarch64_split_simd_shift_p):
2522 * config/aarch64/iterators.md (Vnarrowq2): New mode attribute.
2523 * config/aarch64/aarch64-simd.md
2524 (vec_unpacku_hi_<mode>, vec_unpacks_hi_<mode>): Recombine into...
2525 (vec_unpack<su>_hi_<mode>): ...this. Move the generation of
2526 zip2 for zero-extends to...
2527 (aarch64_simd_vec_unpack<su>_hi_<mode>): ...a split of this
2528 instruction. Fix big-endian handling.
2529 (vec_unpacku_lo_<mode>, vec_unpacks_lo_<mode>): Recombine into...
2530 (vec_unpack<su>_lo_<mode>): ...this. Move the generation of
2531 zip1 for zero-extends to...
2532 (<optab><Vnarrowq><mode>2): ...a split of this instruction.
2533 Fix big-endian handling.
2534 (*aarch64_zip1_uxtl): New pattern.
2535 (aarch64_usubw<mode>_lo_zip, aarch64_uaddw<mode>_lo_zip): Delete
2536 (aarch64_usubw<mode>_hi_zip, aarch64_uaddw<mode>_hi_zip): Likewise.
2537 * config/aarch64/aarch64.cc (aarch64_get_shareable_reg): New function.
2538 (aarch64_gen_shareable_zero): Use it.
2539 (aarch64_split_simd_shift_p): New function.
2541 2024-01-12 Richard Sandiford <richard.sandiford@arm.com>
2543 * emit-rtl.h (rtl_data::x_function_beg_note): New member variable.
2544 (function_beg_insn): New macro.
2545 * function.cc (expand_function_start): Initialize function_beg_insn.
2547 2024-01-12 Richard Sandiford <richard.sandiford@arm.com>
2550 * config/aarch64/aarch64-sve-builtins.h
2551 (function_builder::m_overload_names): Replace with...
2552 * config/aarch64/aarch64-sve-builtins.cc (overload_names): ...this
2554 (add_overloaded_function): Update accordingly, using get_identifier
2555 to get a GGC-friendly record of the name.
2557 2024-01-12 Richard Sandiford <richard.sandiford@arm.com>
2560 * config/aarch64/aarch64-sve-builtins.def: Don't include
2561 aarch64-sve-builtins-sme.def.
2562 (DEF_SME_ZA_FUNCTION_GS, DEF_SME_ZA_FUNCTION): Move to...
2563 * config/aarch64/aarch64-sve-builtins-sme.def: ...here.
2564 (DEF_SME_FUNCTION): New macro. Use it and DEF_SME_FUNCTION_GS
2565 instead of DEF_SVE_*. Add AARCH64_FL_SME to anything that
2566 requires AARCH64_FL_SME2.
2567 * config/aarch64/aarch64-sve-builtins-sve2.def: Make same
2568 AARCH64_FL_SME adjustment here.
2569 * config/aarch64/aarch64-sve-builtins.cc (function_groups): Don't
2570 include SME intrinsics.
2571 (sme_function_groups): New array.
2572 (handle_arm_sve_h): Remove check for AARCH64_FL_SME.
2573 (handle_arm_sme_h): Use sme_function_groups instead of function_groups.
2575 2024-01-12 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2578 * config/riscv/riscv-protos.h (struct regmove_vector_cost): New struct.
2579 (struct cpu_vector_cost): Add regmove struct.
2580 (get_vector_costs): Export as global.
2581 * config/riscv/riscv-vector-costs.cc (adjust_stmt_cost): Adjust scalar_to_vec cost.
2582 (costs::add_stmt_cost): Ditto.
2583 * config/riscv/riscv.cc (get_common_costs): Export global function.
2585 2024-01-12 Jakub Jelinek <jakub@redhat.com>
2587 PR tree-optimization/113334
2588 * gimple-lower-bitint.cc (bitint_large_huge::handle_operand): Use
2589 wi::neg_p (wi::to_wide (op)) instead of tree_int_cst_sgn (op) < 0
2590 to determine if number should be extended by all ones rather than zero
2593 2024-01-12 Jakub Jelinek <jakub@redhat.com>
2595 PR tree-optimization/113330
2596 * tree-sra.cc (create_access): Punt for BITINT_TYPE accesses with
2599 2024-01-12 Jakub Jelinek <jakub@redhat.com>
2601 PR tree-optimization/113323
2602 * gimple-lower-bitint.cc (bitint_dom_walker::before_dom_children): Fix
2603 check for lhs being large/huge _BitInt not in m_names.
2605 2024-01-12 Jakub Jelinek <jakub@redhat.com>
2607 PR tree-optimization/113316
2608 * gimple-lower-bitint.cc (bitint_large_huge::lower_call): Handle
2609 uninitialized large/huge _BitInt arguments to calls.
2611 2024-01-12 Jakub Jelinek <jakub@redhat.com>
2613 * gimple-lower-bitint.cc (mergeable_op): Instead of comparing
2614 TYPE_SIZE (t) of large/huge BITINT_TYPEs, compare
2615 CEIL (TYPE_PRECISION (t), limb_prec).
2616 (bitint_large_huge::handle_cast): Likewise.
2618 2024-01-12 Ilya Leoshkevich <iii@linux.ibm.com>
2621 * config/rs6000/rs6000.cc (rs6000_elf_declare_function_name):
2622 Use assemble_function_label_final () for Power ELF V1 ABI.
2623 * output.h (assemble_function_label_final): New function.
2624 * varasm.cc (assemble_function_label_raw): Use
2625 assemble_function_label_final ().
2626 (assemble_function_label_final): New function.
2628 2024-01-12 Richard Biener <rguenther@suse.de>
2630 PR middle-end/113344
2631 * match.pd ((double)float CMP (double)float -> float CMP float):
2632 Perform result type check only for vectors.
2633 * fold-const.cc (fold_binary_loc): Likewise.
2635 2024-01-12 Haochen Jiang <haochen.jiang@intel.com>
2637 * config/i386/sse.md (sdot_prod<mode>): Remove redundant SET.
2638 (usdot_prod<mode>): Ditto.
2639 (sdot_prod<mode>): Ditto.
2640 (udot_prod<mode>): Ditto.
2642 2024-01-12 Haochen Jiang <haochen.jiang@intel.com>
2645 * config/i386/i386-c.cc (ix86_target_macros_internal):
2646 Add __AVX10_1__, __AVX10_1_256__ and __AVX10_1_512__.
2648 2024-01-12 Richard Biener <rguenther@suse.de>
2651 * config/s390/s390.cc (expand_perm_as_a_vlbr_vstbr_candidate):
2652 Do not generate code when d.testing_p.
2654 2024-01-12 liuhongt <hongtao.liu@intel.com>
2657 * doc/invoke.texi (fcf-protection=): Update documents.
2659 2024-01-12 Pan Li <pan2.li@intel.com>
2661 * config/riscv/riscv.cc (riscv_v_ext_mode_p): Update the
2662 comments of predicate func riscv_v_ext_mode_p.
2664 2024-01-12 Feng Wang <wangfeng@eswincomputing.com>
2666 * config/riscv/riscv-vector-builtins.def (vfloat16m8_t):
2667 Modify ABI-name length of vfloat16m8_t
2669 2024-01-12 Li Wei <liwei@loongson.cn>
2671 * config/loongarch/loongarch.cc (loongarch_expand_conditional_move):
2674 2024-01-12 Li Wei <liwei@loongson.cn>
2676 * config/loongarch/loongarch.md (add<mode>3): Removed.
2680 (*addsi3_extended): Removed.
2681 (addsi3_extended): New.
2683 2024-01-11 Jin Ma <jinma@linux.alibaba.com>
2685 * config/riscv/thead.md: Add limits for splits.
2687 2024-01-11 Andrew Pinski <quic_apinski@quicinc.com>
2689 PR middle-end/113322
2690 * expr.cc (do_store_flag): Don't try single bit tests with
2691 comparison on vector types.
2693 2024-01-11 Andrew Pinski <quic_apinski@quicinc.com>
2695 PR tree-optimization/113301
2696 * match.pd (`1/x`): Delay signed case until late.
2698 2024-01-11 Georg-Johann Lay <avr@gjlay.de>
2700 * doc/invoke.texi (AVR Options): Move -mrmw, -mn-flash, -mshort-calls
2702 (AVR Internal Options): ...this new @subsubsection.
2704 2024-01-11 Vladimir N. Makarov <vmakarov@redhat.com>
2706 PR rtl-optimization/112918
2707 * lra-constraints.cc (SMALL_REGISTER_CLASS_P): Move before in_class_p.
2708 (in_class_p): Restrict condition for narrowing class in case of
2709 allow_all_reload_class_changes_p.
2710 (process_alt_operands): Try to match operand without and with
2711 narrowing reg class. Discourage narrowing the class. Finish insn
2712 matching only if there is no class narrowing.
2713 (curr_insn_transform): Pass true to in_class_p for reg operand win.
2715 2024-01-11 Richard Biener <rguenther@suse.de>
2717 PR tree-optimization/112505
2718 * tree-vect-loop.cc (vectorizable_induction): Reject
2719 bit-precision induction.
2721 2024-01-11 Richard Biener <rguenther@suse.de>
2723 PR tree-optimization/113126
2724 * match.pd ((double)float CMP (double)float -> float CMP float):
2725 Make sure the boolean type is the same.
2726 * fold-const.cc (fold_binary_loc): Likewise.
2728 2024-01-11 Richard Biener <rguenther@suse.de>
2730 PR tree-optimization/112636
2731 * tree-ssa-loop-ch.cc (ch_base::copy_headers): Call
2732 estimate_numbers_of_iterations before querying
2733 get_max_loop_iterations_int.
2734 (pass_ch::execute): Initialize SCEV and loops appropriately.
2736 2024-01-11 Georg-Johann Lay <avr@gjlay.de>
2738 * config/avr/avr-devices.cc (avr_texinfo): Adjust documentation for
2740 * config/avr/gen-avr-mmcu-texi.cc (main): Add @anchor for each core.
2741 * doc/extend.texi (AVR Variable Attributes): Improve documentation
2742 of io, io_low and address attributes.
2743 * doc/invoke.texi (AVR Options): Add some anchors for external refs.
2744 * doc/avr-mmcu.texi: Rebuild.
2746 2024-01-11 Yang Yujie <yangyujie@loongson.cn>
2749 * config/loongarch/genopts/loongarch.opt.in: Mark options with
2750 the "Save" property.
2751 * config/loongarch/loongarch.opt: Same.
2752 * config/loongarch/loongarch-opts.cc: Refresh -mcmodel= state
2753 according to la_target.
2754 * config/loongarch/loongarch.cc: Implement TARGET_OPTION_{SAVE,
2755 RESTORE} for the la_target structure; Rename option conditions
2756 to have the same "la_" prefix.
2757 * config/loongarch/loongarch.h: Same.
2759 2024-01-11 Pan Li <pan2.li@intel.com>
2761 * loop-unroll.cc (insert_var_expansion_initialization): Leverage
2762 MODE_HAS_SIGNED_ZEROS for expansion variable initialization.
2764 2024-01-11 Alex Coplan <alex.coplan@arm.com>
2767 * config/aarch64/aarch64-ldp-fusion.cc (filter_notes): Add
2768 fr_expr param to extract REG_FRAME_RELATED_EXPR notes.
2769 (combine_reg_notes): Handle REG_FRAME_RELATED_EXPR notes, and
2770 synthesize these if needed. Update caller ...
2771 (ldp_bb_info::fuse_pair): ... here.
2772 (ldp_bb_info::try_fuse_pair): Punt if either insn has writeback
2773 and either insn is frame-related.
2774 (find_trailing_add): Punt on frame-related insns.
2775 * config/aarch64/aarch64.cc (aarch64_save_callee_saves): Use
2776 REG_FRAME_RELATED_EXPR instead of REG_CFA_OFFSET.
2778 2024-01-11 YunQiang Su <syq@gcc.gnu.org>
2780 * config/mips/mips.cc (mips_start_function_definition):
2781 Add ATTRIBUTE_UNUSED.
2783 2024-01-11 Richard Biener <rguenther@suse.de>
2785 PR middle-end/112740
2786 * expr.cc (store_constructor): Check the integer vector
2787 mask has a single bit per element before using sign-extension
2788 to expand an uniform vector.
2790 2024-01-11 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2792 * config/riscv/riscv-vector-costs.cc (costs::better_main_loop_than_p): VLA
2793 preempt VLS on unknown NITERS loop.
2795 2024-01-11 Haochen Jiang <haochen.jiang@intel.com>
2797 * doc/invoke.texi: Add -mevex512.
2799 2024-01-11 Lulu Cheng <chenglulu@loongson.cn>
2801 * config/loongarch/loongarch.md (one_cmpl<mode>2): Replace GPR with X.
2802 (*nor<mode>3): Likewise.
2803 (nor<mode>3): Likewise.
2804 (*negsi2_extended): New template.
2805 (*<optab>si3_internal): Likewise.
2806 (*one_cmplsi2_internal): Likewise.
2807 (*norsi3_internal): Likewise.
2808 (*<optab>nsi_internal): Likewise.
2809 (bytepick_w_<bytepick_imm>_extend): Modify this template according to the
2810 modified bit operation to make the optimization work.
2812 2024-01-11 liuhongt <hongtao.liu@intel.com>
2815 * match.pd (VEC_COND_EXPR: A < B ? A : B -> MIN_EXPR): New patten match.
2817 2024-01-10 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2819 * config/riscv/riscv.cc (get_common_costs): Switch RVV cost model.
2820 (get_vector_costs): Ditto.
2821 (riscv_builtin_vectorization_cost): Ditto.
2823 2024-01-10 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2825 * config/riscv/riscv-vector-costs.cc (costs::better_main_loop_than_p): Minior tweak.
2827 2024-01-10 Antoni Boucher <bouanto@zoho.com>
2830 * ipa-fnsummary.cc (ipa_fnsummary_cc_finalize): Call
2831 ipa_free_size_summary.
2832 * ipa-icf.cc (ipa_icf_cc_finalize): New function.
2833 * ipa-profile.cc (ipa_profile_cc_finalize): New function.
2834 * ipa-prop.cc (ipa_prop_cc_finalize): New function.
2835 * ipa-prop.h (ipa_prop_cc_finalize): New function.
2836 * ipa-sra.cc (ipa_sra_cc_finalize): New function.
2837 * ipa-utils.h (ipa_profile_cc_finalize, ipa_icf_cc_finalize,
2838 ipa_sra_cc_finalize): New functions.
2839 * toplev.cc (toplev::finalize): Call ipa_icf_cc_finalize,
2840 ipa_prop_cc_finalize, ipa_profile_cc_finalize and
2842 Include ipa-utils.h.
2844 2024-01-10 Jin Ma <jinma@linux.alibaba.com>
2846 * config/riscv/riscv-protos.h (th_int_get_mask): New prototype.
2847 (th_int_get_save_adjustment): Likewise.
2848 (th_int_adjust_cfi_prologue): Likewise.
2849 * config/riscv/riscv.cc (BITSET_P): Moved away from here.
2850 (TH_INT_INTERRUPT): New macro.
2851 (riscv_expand_prologue): Add the processing of XTheadInt.
2852 (riscv_expand_epilogue): Likewise.
2853 * config/riscv/riscv.h (BITSET_P): Moved to here.
2854 * config/riscv/riscv.md: New unspec.
2855 * config/riscv/thead.cc (th_int_get_mask): New function.
2856 (th_int_get_save_adjustment): Likewise.
2857 (th_int_adjust_cfi_prologue): Likewise.
2858 * config/riscv/thead.md (th_int_push): New pattern.
2859 (th_int_pop): new pattern.
2861 2024-01-10 Tamar Christina <tamar.christina@arm.com>
2863 PR tree-optimization/112468
2864 * doc/sourcebuild.texi: Document ifn_copysign.
2865 * match.pd: Only apply transformation if target supports the IFN.
2867 2024-01-10 Andrew Pinski <quic_apinski@quicinc.com>
2869 PR tree-optimization/112581
2870 * gimple-if-to-switch.cc (pass_if_to_switch::execute): Call
2871 mark_ssa_maybe_undefs.
2872 * tree-ssa-reassoc.cc (can_reassociate_op_p): Uninitialized
2873 variables can not be reassociated.
2874 (init_range_entry): Check for uninitialized variables too.
2875 (init_reassoc): Call mark_ssa_maybe_undefs.
2877 2024-01-10 Maciej W. Rozycki <macro@embecosm.com>
2879 * config/riscv/riscv.cc (riscv_noce_conversion_profitable_p):
2880 Also handle sign extension.
2882 2024-01-10 Alex Coplan <alex.coplan@arm.com>
2884 * config/aarch64/aarch64.opt (-mearly-ldp-fusion): Set default
2886 (-mlate-ldp-fusion): Likewise.
2888 2024-01-10 Tamar Christina <tamar.christina@arm.com>
2890 PR tree-optimization/113287
2891 * tree-vect-stmts.cc (vectorizable_early_exit): Check the flags on edge
2892 instead of using BRANCH_EDGE to determine true edge.
2894 2024-01-10 Richard Biener <rguenther@suse.de>
2896 PR tree-optimization/113078
2897 * tree-vect-loop.cc (check_reduction_path): Canonicalize
2898 .COND_SUB to .COND_ADD.
2900 2024-01-10 David Malcolm <dmalcolm@redhat.com>
2902 * gcc-urlifier.cc (gcc_urlifier::get_url_suffix_for_option):
2903 Handle prefix mappings before calling find_opt.
2904 (selftest::gcc_urlifier_cc_tests): Add example of urlifying a
2905 "-fno-"-prefixed command-line option.
2906 * opts-common.cc (get_option_prefix_remapping): New.
2907 * opts.h (get_option_prefix_remapping): New decl.
2909 2024-01-10 David Malcolm <dmalcolm@redhat.com>
2911 * diagnostic.cc (diagnostic_context::report_diagnostic): Pass
2912 m_urlifier to pp_output_formatted_text.
2913 * pretty-print.cc: Add #define of INCLUDE_VECTOR.
2914 (obstack_append_string): New overload, taking a length.
2915 (urlify_quoted_string): Pass in an obstack ptr, rather than using
2916 that of the pp's buffer. Generalize to handle trailing text in
2917 the buffer beyond the run of quoted text.
2918 (class quoting_info): New.
2919 (on_begin_quote): New.
2920 (on_end_quote): New.
2921 (pp_format): Refactor phase 1 and phase 2 quoting support, moving
2922 it to calls to on_begin_quote and on_end_quote.
2923 (struct auto_obstack): New.
2924 (quoting_info::handle_phase_3): New.
2925 (pp_output_formatted_text): Add urlifier param. Use it if there
2926 is deferred urlification. Delete m_quotes.
2927 (selftest::pp_printf_with_urlifier): Pass urlifier to
2928 pp_output_formatted_text.
2929 (selftest::test_urlification): Update results for the existing
2930 case of quoted text stradding chunks; add more such test cases.
2931 * pretty-print.h (class quoting_info): New forward decl.
2932 (chunk_info::m_quotes): New field.
2933 (pp_output_formatted_text): Add optional urlifier param.
2935 2024-01-10 David Malcolm <dmalcolm@redhat.com>
2937 * pretty-print.cc (selftest::test_pp_format): Add selftest
2938 coverage for numbered args.
2940 2024-01-10 Tamar Christina <tamar.christina@arm.com>
2942 PR tree-optimization/113144
2943 PR tree-optimization/113145
2944 * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
2945 Update all BB that the original exits dominated.
2947 2024-01-10 Eric Botcazou <ebotcazou@adacore.com>
2949 * dwarf2out.cc (modified_type_die): Extend the support of reverse
2950 storage order to enumeration types if -gstrict-dwarf is not passed.
2951 (gen_enumeration_type_die): Add REVERSE parameter and generate the
2952 DIE immediately after the existing one if it is true.
2953 (gen_tagged_type_die): Add REVERSE parameter and pass it in the
2954 call to gen_enumeration_type_die.
2955 (gen_type_die_with_usage): Add REVERSE parameter and pass it in the
2956 first recursive call as well as the call to gen_tagged_type_die.
2957 (gen_type_die): Add REVERSE parameter and pass it in the call to
2958 gen_type_die_with_usage.
2960 2024-01-10 Jakub Jelinek <jakub@redhat.com>
2962 PR tree-optimization/113120
2963 * tree-sra.cc (analyze_access_subtree): For BITINT_TYPE
2964 with root->size TYPE_PRECISION don't build anything new.
2965 Otherwise, if root->type is a BITINT_TYPE, use build_bitint_type
2966 rather than build_nonstandard_integer_type.
2968 2024-01-10 Hongyu Wang <hongyu.wang@intel.com>
2970 * config/i386/i386.opt: Adjust document.
2971 * doc/invoke.texi: Add description for
2972 -mapx-inline-asm-use-gpr32.
2974 2024-01-10 Juzhe-Zhong <juzhe.zhong@rivai.ai>
2976 * config/riscv/autovec.md (<u>avg<v_double_trunc>3_floor): Remove.
2977 (avg<v_double_trunc>3_floor): New pattern.
2978 (<u>avg<v_double_trunc>3_ceil): Remove.
2979 (avg<v_double_trunc>3_ceil): New pattern.
2980 (uavg<mode>3_floor): Ditto.
2981 (uavg<mode>3_ceil): Ditto.
2982 * config/riscv/riscv-protos.h (enum insn_flags): Add for average addition.
2983 (enum insn_type): Ditto.
2984 * config/riscv/riscv-v.cc: Ditto.
2985 * config/riscv/vector-iterators.md (ashiftrt): Remove.
2987 * config/riscv/vector.md: Add VLS modes.
2989 2024-01-10 Kewen Lin <linkw@linux.ibm.com>
2992 * config/rs6000/vsx.md (VCZLSBB): New int iterator.
2993 (vczlsbb_char): New int attribute.
2994 (vclzlsbb_<mode>, vctzlsbb_<mode>): Merge to ...
2995 (vc<vczlsbb_char>zlsbb_<mode>): ... this.
2996 (*vctzlsbb_zext_<mode>): Rename to ...
2997 (*vc<vczlsbb_char>zlsbb_zext_<mode>): ... this, and extend it to
3000 2024-01-10 Kewen Lin <linkw@linux.ibm.com>
3003 * config/rs6000/rs6000.md (copysign<mode>3 IEEE128): Change predicate
3004 of the last argument from altivec_register_operand to any_operand. If
3005 operands[2] is CONST_DOUBLE, emit abs or neg abs depending on its sign
3006 otherwise if it doesn't satisfy altivec_register_operand, force it to
3007 REG using copy_to_mode_reg.
3009 2024-01-10 Kewen Lin <linkw@linux.ibm.com>
3011 PR middle-end/113100
3012 * builtins.cc (expand_builtin_stack_address): Guard stack point
3013 adjustment with SPARC_STACK_BOUNDARY_HACK.
3015 2024-01-10 Yang Yujie <yangyujie@loongson.cn>
3017 * config/loongarch/genopts/loongarch-strings: Remove explicit-reloc
3018 argument string definitions.
3019 * config/loongarch/loongarch-str.h: Same.
3020 * config/loongarch/genopts/loongarch.opt.in: Mark -m[no-]explicit-relocs
3021 as aliases to -mexplicit-relocs={always,none}
3022 * config/loongarch/loongarch.opt: Regenerate.
3023 * config/loongarch/loongarch.cc: Same.
3025 2024-01-10 Yang Yujie <yangyujie@loongson.cn>
3027 * config/loongarch/loongarch-def.h: Define constants with
3028 enums instead of Macros.
3030 2024-01-10 Yang Yujie <yangyujie@loongson.cn>
3032 * config/loongarch/genopts/loongarch-strings: Rename.
3033 * config/loongarch/genopts/loongarch.opt.in: Same.
3034 * config/loongarch/loongarch-cpu.cc: Same.
3035 * config/loongarch/loongarch-def.cc: Same.
3036 * config/loongarch/loongarch-def.h: Same.
3037 * config/loongarch/loongarch-opts.cc: Same.
3038 * config/loongarch/loongarch-opts.h: Same.
3039 * config/loongarch/loongarch-str.h: Same.
3040 * config/loongarch/loongarch.opt: Same.
3042 2024-01-10 Yang Yujie <yangyujie@loongson.cn>
3044 * config/loongarch/genopts/genstr.sh: Prepend the isa_evolution
3045 variable with the common la_ prefix.
3046 * config/loongarch/genopts/loongarch.opt.in: Mark ISA evolution
3047 flags as saved using TargetVariable.
3048 * config/loongarch/loongarch.opt: Same.
3049 * config/loongarch/loongarch-def.h: Define evolution_set to
3050 mark changes to the -march default.
3051 * config/loongarch/loongarch-driver.cc: Same.
3052 * config/loongarch/loongarch-opts.cc: Same.
3053 * config/loongarch/loongarch-opts.h: Define and use ISA evolution
3054 conditions around the la_target structure.
3055 * config/loongarch/loongarch.cc: Same.
3056 * config/loongarch/loongarch.md: Same.
3057 * config/loongarch/loongarch-builtins.cc: Same.
3058 * config/loongarch/loongarch-c.cc: Same.
3059 * config/loongarch/lasx.md: Same.
3060 * config/loongarch/lsx.md: Same.
3061 * config/loongarch/sync.md: Same.
3063 2024-01-09 Jeff Law <jlaw@ventanamicro.com>
3065 * config/epiphany/constraints.md (Car): Allow -1024..1023, no more,
3068 2024-01-09 Richard Sandiford <richard.sandiford@arm.com>
3070 * config/mn10300/mn10300.md (subdi3_degenerate): Add isa attribute.
3072 2024-01-09 Tamar Christina <tamar.christina@arm.com>
3074 * tree-vect-loop.cc (vectorizable_live_operation_1): Drop unused
3076 (vectorizable_live_operation): Likewise.
3078 2024-01-09 Tamar Christina <tamar.christina@arm.com>
3080 PR tree-optimization/113199
3081 * tree-vect-loop.cc (vectorizable_live_operation_1): Use
3084 2024-01-09 Jakub Jelinek <jakub@redhat.com>
3087 * config.gcc (aarch64*-*-*): Add aarch64-builtins.h to target_gtfiles.
3088 * config/aarch64/aarch64-builtins.cc (aarch64_simd_types): Add extern
3089 GTY(()) declaration before the definition, drop GTY(()) drom the
3092 2024-01-09 Richard Biener <rguenther@suse.de>
3094 PR tree-optimization/113026
3095 * tree-vect-loop-manip.cc (vect_do_peeling): Remove
3096 redundant and wrong niter bound setting. Move niter
3097 bound adjustment down.
3099 2024-01-09 Tamar Christina <tamar.christina@arm.com>
3101 PR middle-end/113163
3102 * tree-vect-loop-manip.cc (vect_can_peel_nonlinear_iv_p):
3103 Reject non-linear inductions that aren't supported.
3105 2024-01-09 Roger Sayle <roger@nextmovesoftware.com>
3107 * config/arc/arc.cc (arc_shift_alg): New enumerated type for
3108 left shift implementation strategies.
3109 (arc_shift_info): Type for each entry of the shift strategy table.
3110 (arc_shift_context_idx): Return a integer value for each code
3111 generation context, used as an index
3112 (arc_ashl_alg): Table indexed by context and shifted bit count.
3113 (arc_split_ashl): Use the arc_ashl_alg table to select SImode
3114 left shift implementation.
3115 (arc_rtx_costs) <case ASHIFT>: Use the arc_ashl_alg table to
3116 provide accurate costs, when optimizing for speed or size.
3118 2024-01-09 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3120 * config/riscv/riscv-vector-costs.cc (loop_invariant_op_p): Fix loop invariant check.
3122 2024-01-09 Julian Brown <julian@codesourcery.com>
3124 * gimplify.cc (gimplify_expr): Ensure OMP_ARRAY_SECTION has been
3125 processed out before gimplification.
3126 * tree-pretty-print.cc (dump_generic_node): Support OMP_ARRAY_SECTION.
3127 * tree.def (OMP_ARRAY_SECTION): New tree code.
3129 2024-01-09 Jakub Jelinek <jakub@redhat.com>
3131 PR tree-optimization/113210
3132 * tree-vect-loop.cc (vect_get_loop_niters): If non-INTEGER_CST
3133 value in *number_of_iterationsm1 PLUS_EXPR 1 is folded into
3134 INTEGER_CST, recompute *number_of_iterationsm1 as the INTEGER_CST
3137 2024-01-09 Eric Botcazou <ebotcazou@adacore.com>
3139 PR rtl-optimization/113140
3140 * reorg.cc (fill_slots_from_thread): If we are to branch after the
3141 last instruction of the function, create an end label.
3143 2024-01-09 Roger Sayle <roger@nextmovesoftware.com>
3144 Hongtao Liu <hongtao.liu@intel.com>
3147 * config/i386/i386-expand.cc
3148 (ix86_convert_const_wide_int_to_broadcast): Allow call to
3149 ix86_expand_vector_init_duplicate to fail, and return NULL_RTX.
3150 (ix86_broadcast_from_constant): Revert recent change; Return a
3151 suitable MEMREF independently of mode/target combinations.
3152 (ix86_expand_vector_move): Allow ix86_expand_vector_init_duplicate
3153 to decide whether expansion is possible/preferrable. Only try
3154 forcing DImode constants to memory (and trying again) if calling
3155 ix86_expand_vector_init_duplicate fails with an DImode immediate
3157 (ix86_expand_vector_init_duplicate) <case E_V2DImode>: Try using
3158 V4SImode for suitable immediate constants.
3159 <case E_V4DImode>: Try using V8SImode for suitable constants.
3160 <case E_V4HImode>: Fail for CONST_INT_P, i.e. use constant pool.
3161 <case E_V2HImode>: Likewise.
3162 <case E_V8HImode>: For CONST_INT_P try using V4SImode via widen.
3163 <case E_V16QImode>: For CONT_INT_P try using V8HImode via widen.
3164 <label widen>: Handle CONT_INTs via simplify_binary_operation.
3165 Allow recursive calls to ix86_expand_vector_init_duplicate to fail.
3166 <case E_V16HImode>: For CONST_INT_P try V8SImode via widen.
3167 <case E_V32QImode>: For CONST_INT_P try V16HImode via widen.
3168 (ix86_expand_vector_init): Move try using a broadcast for all_same
3169 with ix86_expand_vector_init_duplicate before using constant pool.
3171 2024-01-09 Chung-Ju Wu <jasonwucj@gmail.com>
3173 * doc/invoke.texi (Arm Options): Document Cortex-M52 options.
3175 2024-01-09 Chung-Ju Wu <jasonwucj@gmail.com>
3177 * config/arm/arm-cpus.in (cortex-m52): New cpu.
3178 * config/arm/arm-tables.opt: Regenerate.
3179 * config/arm/arm-tune.md: Regenerate.
3181 2024-01-09 Jiahao Xu <xujiahao@loongson.cn>
3183 * config/loongarch/lasx.md (vec_initv32qiv16qi): Rename to ..
3184 (vec_init<mode><lasxhalf>): .. this, and extend to mode.
3185 (@vec_concatz<mode>): New insn pattern.
3186 * config/loongarch/loongarch.cc (loongarch_expand_vector_group_init):
3187 Handle VALS containing two vectors.
3189 2024-01-09 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3191 * config/riscv/riscv-vector-builtins-functions.def (vleff): Move comments.
3192 (vundefined): Ditto.
3194 2024-01-09 Feng Wang <wangfeng@eswincomputing.com>
3196 * config/riscv/riscv-vector-builtins-bases.cc (class vandn):
3197 Add new function_base for crypto vector.
3198 (class bitmanip): Ditto.
3199 (class b_reverse):Ditto.
3200 (class vwsll): Ditto.
3201 (class clmul): Ditto.
3202 (class vg_nhab): Ditto.
3203 (class crypto_vv):Ditto.
3204 (class crypto_vi):Ditto.
3205 (class vaeskf2_vsm3c):Ditto.
3206 (class vsm3me): Ditto.
3207 (BASE): Add BASE declaration for crypto vector.
3208 * config/riscv/riscv-vector-builtins-bases.h: Ditto.
3209 * config/riscv/riscv-vector-builtins-functions.def (REQUIRED_EXTENSIONS):
3210 Add crypto vector intrinsic definition.
3238 * config/riscv/riscv-vector-builtins-shapes.cc (struct crypto_vv_def):
3239 Add new function_shape for crypto vector.
3240 (struct crypto_vi_def): Ditto.
3241 (struct crypto_vv_no_op_type_def): Ditto.
3242 (SHAPE): Add SHAPE declaration of crypto vector.
3243 * config/riscv/riscv-vector-builtins-shapes.h: Ditto.
3244 * config/riscv/riscv-vector-builtins-types.def (DEF_RVV_CRYPTO_SEW32_OPS):
3245 Add new data type for crypto vector.
3246 (DEF_RVV_CRYPTO_SEW64_OPS): Ditto.
3247 (vuint32mf2_t): Ditto.
3248 (vuint32m1_t): Ditto.
3249 (vuint32m2_t): Ditto.
3250 (vuint32m4_t): Ditto.
3251 (vuint32m8_t): Ditto.
3252 (vuint64m1_t): Ditto.
3253 (vuint64m2_t): Ditto.
3254 (vuint64m4_t): Ditto.
3255 (vuint64m8_t): Ditto.
3256 * config/riscv/riscv-vector-builtins.cc (DEF_RVV_CRYPTO_SEW32_OPS):
3257 Add new data struct for crypto vector.
3258 (DEF_RVV_CRYPTO_SEW64_OPS): Ditto.
3259 (registered_function::overloaded_hash): Processing size_t uimm for C overloaded func.
3260 * config/riscv/riscv-vector-builtins.def (vi): Add vi OP_TYPE.
3262 2024-01-08 Ilya Leoshkevich <iii@linux.ibm.com>
3265 * varasm.cc (assemble_function_label_raw): Do not call
3266 asan_function_start () without the current function.
3268 2024-01-08 Cupertino Miranda <cupertino.miranda@oracle.com>
3271 * btfout.cc (btf_collect_datasec): Skip creating BTF info for
3272 extern and kernel_helper attributed function decls.
3274 2024-01-08 Cupertino Miranda <cupertino.miranda@oracle.com>
3276 * btfout.cc (output_btf_strs): Changed.
3278 2024-01-08 Tobias Burnus <tobias@codesourcery.com>
3280 * config/gcn/mkoffload.cc (main): Handle gfx1100
3281 when setting the default XNACK.
3283 2024-01-08 Tobias Burnus <tobias@codesourcery.com>
3285 * config.gcc (amdgcn-*-amdhsa): Accept --with-arch=gfx1100.
3286 * config/gcn/gcn-hsa.h (NO_XNACK): Add gfx1100:
3287 (ASM_SPEC): Handle gfx1100.
3288 * config/gcn/gcn-opts.h (enum processor_type): Add PROCESSOR_GFX1100.
3289 (enum gcn_isa): Add ISA_RDNA3.
3290 (TARGET_GFX1100, TARGET_RDNA2_PLUS, TARGET_RDNA3): Define.
3291 * config/gcn/gcn-valu.md: Change TARGET_RDNA2 to TARGET_RDNA2_PLUS.
3292 * config/gcn/gcn.cc (gcn_option_override,
3293 gcn_omp_device_kind_arch_isa, output_file_start): Handle gfx1100.
3294 (gcn_global_address_p, gcn_addr_space_legitimate_address_p): Change
3295 TARGET_RDNA2 to TARGET_RDNA2_PLUS.
3296 (gcn_hsa_declare_function_name): Don't use '.amdhsa_reserve_flat_scratch'
3298 * config/gcn/gcn.h (ASSEMBLER_DIALECT): Likewise.
3299 (TARGET_CPU_CPP_BUILTINS): Define __RDNA3__, __gfx1030__ and
3301 * config/gcn/gcn.md: Change TARGET_RDNA2 to TARGET_RDNA2_PLUS.
3302 * config/gcn/gcn.opt (Enum gpu_type): Add gfx1100.
3303 * config/gcn/mkoffload.cc (EF_AMDGPU_MACH_AMDGCN_GFX1100): Define.
3304 (isa_has_combined_avgprs, main): Handle gfx1100.
3305 * config/gcn/t-omp-device (isa): Add gfx1100.
3307 2024-01-08 Richard Biener <rguenther@suse.de>
3309 * doc/invoke.texi (-mmovbe): Clarify.
3311 2024-01-08 Richard Biener <rguenther@suse.de>
3313 PR tree-optimization/113026
3314 * tree-vect-loop.cc (vect_need_peeling_or_partial_vectors_p):
3315 Avoid an epilog in more cases.
3316 * tree-vect-loop-manip.cc (vect_do_peeling): Adjust the
3317 epilogues niter upper bounds and estimates.
3319 2024-01-08 Jakub Jelinek <jakub@redhat.com>
3321 PR tree-optimization/113228
3322 * gimplify.cc (recalculate_side_effects): Do nothing for SSA_NAMEs.
3324 2024-01-08 Jakub Jelinek <jakub@redhat.com>
3326 PR tree-optimization/113120
3327 * gimple-lower-bitint.cc (gimple_lower_bitint): Fix handling of very
3328 large _BitInt zero INTEGER_CST PHI argument.
3330 2024-01-08 Jakub Jelinek <jakub@redhat.com>
3332 PR tree-optimization/113119
3333 * gimple-lower-bitint.cc (optimizable_arith_overflow): Punt if
3334 both REALPART_EXPR and cast from IMAGPART_EXPR appear, but cast
3335 is before REALPART_EXPR.
3337 2024-01-08 Georg-Johann Lay <avr@gjlay.de>
3340 * config/avr/avr.cc (avr_handle_addr_attribute): Also print valid
3341 range when diagnosing attribute "io" and "io_low" are out of range.
3342 (avr_eval_addr_attrib): Don't ICE on empty address at that place.
3343 (avr_insert_attributes): Reject if attribute "address", "io" or "io_low"
3344 in contexts other than static storage.
3345 (avr_asm_output_aligned_decl_common): Move output of decls with
3346 attribute "address", "io", and "io_low" to...
3347 (avr_output_addr_attrib): ...this new function.
3348 (avr_asm_asm_output_aligned_bss): Remove output for decls with
3349 attribute "address", "io", and "io_low".
3350 (avr_encode_section_info): Rectify handling of decls with attribute
3351 "address", "io", and "io_low".
3353 2024-01-08 Andrew Stubbs <ams@codesourcery.com>
3355 * config/gcn/mkoffload.cc (TEST_XNACK_UNSET): New.
3356 (elf_flags): Remove XNACK from the default value.
3357 (main): Set a default XNACK according to the arch.
3359 2024-01-08 Andrew Stubbs <ams@codesourcery.com>
3361 * config/gcn/mkoffload.cc (isa_has_combined_avgprs): Delete.
3362 (process_asm): Don't count avgprs.
3364 2024-01-08 Hongyu Wang <hongyu.wang@intel.com>
3366 * config/i386/i386.opt: Add supported sub-features.
3367 * doc/extend.texi: Add description for target attribute.
3369 2024-01-08 Feng Wang <wangfeng@eswincomputing.com>
3371 * config/riscv/vector.md: Modify avl_type operand index of zvbc ins.
3373 2024-01-07 Roger Sayle <roger@nextmovesoftware.com>
3374 Uros Bizjak <ubizjak@gmail.com>
3377 * config/i386/i386-features.cc (compute_convert_gain): Include
3378 the overhead of explicit load and store (movd) instructions when
3379 converting non-store scalar operations with memory destinations.
3380 Various indentation whitespace fixes.
3382 2024-01-07 Tamar Christina <tamar.christina@arm.com>
3384 * config/arm/neon.md (cbranch<mode>4): New.
3386 2024-01-07 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3388 * config/riscv/riscv-vsetvl.cc: replace std::max by MAX.
3390 2024-01-06 Jiahao Xu <xujiahao@loongson.cn>
3392 * config/loongarch/lasx.md: Set the unused bits in operand[3] to 0.
3394 2024-01-06 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3397 * config/riscv/riscv-vsetvl.cc (pre_vsetvl::fuse_local_vsetvl_info):
3400 2024-01-06 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3402 * config/riscv/riscv-vector-costs.cc (loop_invariant_op_p): New function.
3403 (variable_vectorized_p): Teach loop invariant.
3404 (has_unexpected_spills_p): Ditto.
3406 2024-01-06 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3408 * config/riscv/riscv-protos.h (whole_reg_to_reg_move_p): New function.
3409 * config/riscv/riscv-v.cc (whole_reg_to_reg_move_p): Ditto.
3410 * config/riscv/vector.md: Allow non-vlmax with len = NUNITS simplification.
3412 2024-01-05 Richard Sandiford <richard.sandiford@arm.com>
3415 * doc/invoke.texi (aarch64-sve-compare-costs): Replace with...
3416 (aarch64-vect-compare-costs): ...this.
3417 * config/aarch64/aarch64.opt (-param=aarch64-sve-compare-costs=):
3419 (-param=aarch64-vect-compare-costs=): ...this new param.
3420 * config/aarch64/aarch64.cc (aarch64_override_options_internal):
3421 Don't disable it when vectorizing for Advanced SIMD only.
3422 (aarch64_autovectorize_vector_modes): Apply VECT_COMPARE_COSTS
3423 whenever aarch64_vect_compare_costs is true.
3425 2024-01-05 Lulu Cheng <chenglulu@loongson.cn>
3427 * config/loongarch/lasx.md (lasx_mxld_<lasxfmt_f>):
3428 Modify the method of determining the memory offset of [x]vld/[x]vst.
3429 (lasx_mxst_<lasxfmt_f>): Likewise.
3430 * config/loongarch/loongarch.cc (loongarch_valid_offset_p): Delete.
3431 (loongarch_address_insns): Likewise.
3432 * config/loongarch/lsx.md (lsx_ld_<lsxfmt_f>): Likewise.
3433 (lsx_st_<lsxfmt_f>): Likewise.
3434 * config/loongarch/predicates.md (aq10b_operand): Likewise.
3435 (aq10h_operand): Likewise.
3436 (aq10w_operand): Likewise.
3437 (aq10d_operand): Likewise.
3439 2024-01-05 Alex Coplan <alex.coplan@arm.com>
3442 * config/aarch64/aarch64-ldp-fusion.cc
3443 (ldp_bb_info::try_fuse_pair): If the second access can throw,
3444 narrow the move range to exactly that insn.
3446 2024-01-05 Ilya Leoshkevich <iii@linux.ibm.com>
3448 * asan.cc (asan_function_start): Drop switch_to_section ().
3449 (asan_emit_stack_protection): Set .LASANPC alignment.
3450 * config/i386/i386.cc: Use assemble_function_label_raw ()
3451 instead of ASM_OUTPUT_LABEL ().
3452 * config/s390/s390.cc (s390_asm_output_function_label):
3454 * defaults.h (ASM_OUTPUT_FUNCTION_LABEL): Likewise.
3455 * final.cc (final_start_function_1): Drop
3456 asan_function_start ().
3457 * output.h (assemble_function_label_raw): New function.
3458 * varasm.cc (assemble_function_label_raw): Likewise.
3460 2024-01-05 Ilya Leoshkevich <iii@linux.ibm.com>
3462 * config/aarch64/aarch64.cc (aarch64_declare_function_name):
3463 Use ASM_OUTPUT_FUNCTION_LABEL ().
3464 * config/alpha/alpha.cc (alpha_start_function): Likewise.
3465 * config/arm/aout.h (ASM_DECLARE_FUNCTION_NAME): Likewise.
3466 * config/arm/arm.cc (arm_asm_declare_function_name): Likewise.
3467 * config/bfin/bfin.h (ASM_DECLARE_FUNCTION_NAME): Likewise.
3468 * config/c6x/c6x.h (ASM_DECLARE_FUNCTION_NAME): Likewise.
3469 * config/gcn/gcn.cc (gcn_hsa_declare_function_name): Likewise.
3470 * config/h8300/h8300.h (ASM_DECLARE_FUNCTION_NAME): Likewise.
3471 * config/ia64/ia64.cc (ia64_start_function): Likewise.
3472 * config/mcore/mcore-elf.h (ASM_DECLARE_FUNCTION_NAME):
3474 * config/microblaze/microblaze.cc (microblaze_function_prologue):
3476 * config/mips/mips.cc (mips_start_unique_function): Return the
3478 (mips_start_function_definition): Use
3479 ASM_OUTPUT_FUNCTION_LABEL ().
3480 (mips_finish_stub): Pass the tree to
3481 mips_start_function_definition ().
3482 (mips16_build_function_stub): Likewise.
3483 (mips16_build_call_stub): Likewise.
3484 (mips_output_function_prologue): Likewise.
3485 * config/pa/pa.cc (pa_output_function_label): Use
3486 ASM_OUTPUT_FUNCTION_LABEL ().
3487 * config/riscv/riscv.cc (riscv_declare_function_name): Likewise.
3488 * config/rs6000/rs6000.cc (rs6000_elf_declare_function_name):
3490 (rs6000_xcoff_declare_function_name): Likewise.
3492 2024-01-05 Jakub Jelinek <jakub@redhat.com>
3494 PR tree-optimization/113201
3495 * tree-scalar-evolution.cc (final_value_replacement_loop): Don't call
3496 replace_uses_by on SSA_NAME_OCCURS_IN_ABNORMAL_PHI rslt.
3498 2024-01-05 Jakub Jelinek <jakub@redhat.com>
3500 PR tree-optimization/90693
3501 * tree-ssa-math-opts.cc (match_single_bit_test): If
3502 tree_expr_nonzero_p (arg), remember it in the second argument to
3503 IFN_POPCOUNT or lower it as arg & (arg - 1) == 0 rather than
3504 arg ^ (arg - 1) > arg - 1.
3505 * internal-fn.cc (expand_POPCOUNT): If second argument to
3506 IFN_POPCOUNT suggests arg is non-zero, try to expand it as
3507 arg & (arg - 1) == 0 rather than arg ^ (arg - 1) > arg - 1.
3509 2024-01-05 Kito Cheng <kito.cheng@sifive.com>
3511 * config/riscv/riscv-v.cc (expand_load_store):
3513 (expand_cond_len_op): Ditto.
3514 (expand_gather_scatter): Ditto.
3515 (expand_lanes_load_store): Ditto.
3516 (expand_fold_extract_last): Ditto.
3518 2024-01-05 Pan Li <pan2.li@intel.com>
3521 2024-01-05 Feng Wang <wangfeng@eswincomputing.com>
3523 * config/riscv/riscv-vector-builtins-bases.cc (class vandn):
3524 Add new function_base for crypto vector.
3525 (class bitmanip): Ditto.
3526 (class b_reverse):Ditto.
3527 (class vwsll): Ditto.
3528 (class clmul): Ditto.
3529 (class vg_nhab): Ditto.
3530 (class crypto_vv):Ditto.
3531 (class crypto_vi):Ditto.
3532 (class vaeskf2_vsm3c):Ditto.
3533 (class vsm3me): Ditto.
3534 (BASE): Add BASE declaration for crypto vector.
3535 * config/riscv/riscv-vector-builtins-bases.h: Ditto.
3536 * config/riscv/riscv-vector-builtins-functions.def (REQUIRED_EXTENSIONS):
3537 Add crypto vector intrinsic definition.
3565 * config/riscv/riscv-vector-builtins-shapes.cc (struct crypto_vv_def):
3566 Add new function_shape for crypto vector.
3567 (struct crypto_vi_def): Ditto.
3568 (struct crypto_vv_no_op_type_def): Ditto.
3569 (SHAPE): Add SHAPE declaration of crypto vector.
3570 * config/riscv/riscv-vector-builtins-shapes.h: Ditto.
3571 * config/riscv/riscv-vector-builtins-types.def (DEF_RVV_CRYPTO_SEW32_OPS):
3572 Add new data type for crypto vector.
3573 (DEF_RVV_CRYPTO_SEW64_OPS): Ditto.
3574 (vuint32mf2_t): Ditto.
3575 (vuint32m1_t): Ditto.
3576 (vuint32m2_t): Ditto.
3577 (vuint32m4_t): Ditto.
3578 (vuint32m8_t): Ditto.
3579 (vuint64m1_t): Ditto.
3580 (vuint64m2_t): Ditto.
3581 (vuint64m4_t): Ditto.
3582 (vuint64m8_t): Ditto.
3583 * config/riscv/riscv-vector-builtins.cc (DEF_RVV_CRYPTO_SEW32_OPS):
3584 Add new data struct for crypto vector.
3585 (DEF_RVV_CRYPTO_SEW64_OPS): Ditto.
3586 (registered_function::overloaded_hash): Processing size_t uimm for C overloaded func.
3587 * config/riscv/riscv-vector-builtins.def (vi): Add vi OP_TYPE.
3589 2024-01-05 Feng Wang <wangfeng@eswincomputing.com>
3591 * config/riscv/riscv-vector-builtins-bases.cc (class vandn):
3592 Add new function_base for crypto vector.
3593 (class bitmanip): Ditto.
3594 (class b_reverse):Ditto.
3595 (class vwsll): Ditto.
3596 (class clmul): Ditto.
3597 (class vg_nhab): Ditto.
3598 (class crypto_vv):Ditto.
3599 (class crypto_vi):Ditto.
3600 (class vaeskf2_vsm3c):Ditto.
3601 (class vsm3me): Ditto.
3602 (BASE): Add BASE declaration for crypto vector.
3603 * config/riscv/riscv-vector-builtins-bases.h: Ditto.
3604 * config/riscv/riscv-vector-builtins-functions.def (REQUIRED_EXTENSIONS):
3605 Add crypto vector intrinsic definition.
3633 * config/riscv/riscv-vector-builtins-shapes.cc (struct crypto_vv_def):
3634 Add new function_shape for crypto vector.
3635 (struct crypto_vi_def): Ditto.
3636 (struct crypto_vv_no_op_type_def): Ditto.
3637 (SHAPE): Add SHAPE declaration of crypto vector.
3638 * config/riscv/riscv-vector-builtins-shapes.h: Ditto.
3639 * config/riscv/riscv-vector-builtins-types.def (DEF_RVV_CRYPTO_SEW32_OPS):
3640 Add new data type for crypto vector.
3641 (DEF_RVV_CRYPTO_SEW64_OPS): Ditto.
3642 (vuint32mf2_t): Ditto.
3643 (vuint32m1_t): Ditto.
3644 (vuint32m2_t): Ditto.
3645 (vuint32m4_t): Ditto.
3646 (vuint32m8_t): Ditto.
3647 (vuint64m1_t): Ditto.
3648 (vuint64m2_t): Ditto.
3649 (vuint64m4_t): Ditto.
3650 (vuint64m8_t): Ditto.
3651 * config/riscv/riscv-vector-builtins.cc (DEF_RVV_CRYPTO_SEW32_OPS):
3652 Add new data struct for crypto vector.
3653 (DEF_RVV_CRYPTO_SEW64_OPS): Ditto.
3654 (registered_function::overloaded_hash): Processing size_t uimm for C overloaded func.
3655 * config/riscv/riscv-vector-builtins.def (vi): Add vi OP_TYPE.
3657 2024-01-04 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3659 * config/riscv/riscv-vector-costs.cc (variable_vectorized_p): Teach vi variant.
3661 2024-01-04 Andrew Pinski <quic_apinski@quicinc.com>
3663 PR tree-optimization/113186
3664 * gimple-match-head.cc (gimple_bitwise_inverted_equal_p):
3665 Match `^` with the `==` for 1bit integral types.
3666 * match.pd (maybe_cmp): Allow for bit_xor for 1bit
3669 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3671 * toplev.cc (general_init): Pass lang_mask to urlifier.
3673 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3675 * diagnostic.h (diagnostic_make_option_url_cb): Add lang_mask
3677 (diagnostic_context::make_option_url): Update for lang_mask param.
3678 * gcc-urlifier.cc: Include "opts.h" and "options.h".
3679 (gcc_urlifier::gcc_urlifier): Add lang_mask param.
3680 (gcc_urlifier::m_lang_mask): New field.
3681 (doc_urls): Make static.
3682 (gcc_urlifier::get_url_for_quoted_text): Use label_text.
3683 (gcc_urlifier::get_url_suffix_for_quoted_text): Use label_text.
3684 Look for an option by name before trying a binary search in
3686 (gcc_urlifier::get_url_suffix_for_quoted_text): Use label_text.
3687 (gcc_urlifier::get_url_suffix_for_option): New.
3688 (make_gcc_urlifier): Add lang_mask param.
3689 (selftest::gcc_urlifier_cc_tests): Update for above changes.
3690 Verify that a URL is found for "-fpack-struct".
3691 * gcc-urlifier.def: Drop options "--version" and "-fpack-struct".
3692 * gcc-urlifier.h (make_gcc_urlifier): Add lang_mask param.
3693 * gcc.cc (driver::global_initializations): Pass 0 for lang_mask
3694 to make_gcc_urlifier.
3695 * opts-diagnostic.h (get_option_url): Add lang_mask param.
3696 * opts.cc (get_option_html_page): Remove special-casing for
3698 (get_option_url_suffix): New.
3699 (get_option_url): Reimplement.
3700 (selftest::test_get_option_html_page): Rename to...
3701 (selftest::test_get_option_url_suffix): ...this and update for
3703 (selftest::opts_cc_tests): Update for renaming.
3704 * opts.h: Include "rich-location.h".
3705 (get_option_url_suffix): New decl.
3707 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3709 * Makefile.in (ALL_OPT_URL_FILES): New.
3710 (GCC_OBJS): Add options-urls.o.
3712 (OBJS-libcommon): Likewise.
3713 (s-options): Depend on $(ALL_OPT_URL_FILES), and add this to
3714 inputs to opt-gather.awk.
3715 (options-urls.cc): New Makefile target.
3716 * opt-functions.awk (url_suffix): New function.
3717 (lang_url_suffix): New function.
3718 * options-urls-cc-gen.awk: New file.
3719 * opts.h (get_opt_url_suffix): New decl.
3721 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3723 * params.opt.urls: New file, autogenerated by
3724 regenerate-opt-urls.py.
3726 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3728 * common.opt.urls: New file, autogenerated by
3729 regenerate-opt-urls.py.
3730 * config/aarch64/aarch64.opt.urls: Likewise.
3731 * config/alpha/alpha.opt.urls: Likewise.
3732 * config/alpha/elf.opt.urls: Likewise.
3733 * config/arc/arc-tables.opt.urls: Likewise.
3734 * config/arc/arc.opt.urls: Likewise.
3735 * config/arm/arm-tables.opt.urls: Likewise.
3736 * config/arm/arm.opt.urls: Likewise.
3737 * config/arm/vxworks.opt.urls: Likewise.
3738 * config/avr/avr.opt.urls: Likewise.
3739 * config/bpf/bpf.opt.urls: Likewise.
3740 * config/c6x/c6x-tables.opt.urls: Likewise.
3741 * config/c6x/c6x.opt.urls: Likewise.
3742 * config/cris/cris.opt.urls: Likewise.
3743 * config/cris/elf.opt.urls: Likewise.
3744 * config/csky/csky.opt.urls: Likewise.
3745 * config/csky/csky_tables.opt.urls: Likewise.
3746 * config/darwin.opt.urls: Likewise.
3747 * config/dragonfly.opt.urls: Likewise.
3748 * config/epiphany/epiphany.opt.urls: Likewise.
3749 * config/fr30/fr30.opt.urls: Likewise.
3750 * config/freebsd.opt.urls: Likewise.
3751 * config/frv/frv.opt.urls: Likewise.
3752 * config/ft32/ft32.opt.urls: Likewise.
3753 * config/fused-madd.opt.urls: Likewise.
3754 * config/g.opt.urls: Likewise.
3755 * config/gcn/gcn.opt.urls: Likewise.
3756 * config/gnu-user.opt.urls: Likewise.
3757 * config/h8300/h8300.opt.urls: Likewise.
3758 * config/hpux11.opt.urls: Likewise.
3759 * config/i386/cygming.opt.urls: Likewise.
3760 * config/i386/cygwin.opt.urls: Likewise.
3761 * config/i386/djgpp.opt.urls: Likewise.
3762 * config/i386/i386.opt.urls: Likewise.
3763 * config/i386/mingw-w64.opt.urls: Likewise.
3764 * config/i386/mingw.opt.urls: Likewise.
3765 * config/i386/nto.opt.urls: Likewise.
3766 * config/ia64/ia64.opt.urls: Likewise.
3767 * config/ia64/ilp32.opt.urls: Likewise.
3768 * config/ia64/vms.opt.urls: Likewise.
3769 * config/iq2000/iq2000.opt.urls: Likewise.
3770 * config/linux-android.opt.urls: Likewise.
3771 * config/linux.opt.urls: Likewise.
3772 * config/lm32/lm32.opt.urls: Likewise.
3773 * config/loongarch/loongarch.opt.urls: Likewise.
3774 * config/lynx.opt.urls: Likewise.
3775 * config/m32c/m32c.opt.urls: Likewise.
3776 * config/m32r/m32r.opt.urls: Likewise.
3777 * config/m68k/ieee.opt.urls: Likewise.
3778 * config/m68k/m68k-tables.opt.urls: Likewise.
3779 * config/m68k/m68k.opt.urls: Likewise.
3780 * config/m68k/uclinux.opt.urls: Likewise.
3781 * config/mcore/mcore.opt.urls: Likewise.
3782 * config/microblaze/microblaze.opt.urls: Likewise.
3783 * config/mips/mips-tables.opt.urls: Likewise.
3784 * config/mips/mips.opt.urls: Likewise.
3785 * config/mips/sde.opt.urls: Likewise.
3786 * config/mmix/mmix.opt.urls: Likewise.
3787 * config/mn10300/mn10300.opt.urls: Likewise.
3788 * config/moxie/moxie.opt.urls: Likewise.
3789 * config/msp430/msp430.opt.urls: Likewise.
3790 * config/nds32/nds32-elf.opt.urls: Likewise.
3791 * config/nds32/nds32-linux.opt.urls: Likewise.
3792 * config/nds32/nds32.opt.urls: Likewise.
3793 * config/netbsd-elf.opt.urls: Likewise.
3794 * config/netbsd.opt.urls: Likewise.
3795 * config/nios2/elf.opt.urls: Likewise.
3796 * config/nios2/nios2.opt.urls: Likewise.
3797 * config/nvptx/nvptx-gen.opt.urls: Likewise.
3798 * config/nvptx/nvptx.opt.urls: Likewise.
3799 * config/openbsd.opt.urls: Likewise.
3800 * config/or1k/elf.opt.urls: Likewise.
3801 * config/or1k/or1k.opt.urls: Likewise.
3802 * config/pa/pa-hpux.opt.urls: Likewise.
3803 * config/pa/pa-hpux1010.opt.urls: Likewise.
3804 * config/pa/pa-hpux1111.opt.urls: Likewise.
3805 * config/pa/pa-hpux1131.opt.urls: Likewise.
3806 * config/pa/pa.opt.urls: Likewise.
3807 * config/pa/pa64-hpux.opt.urls: Likewise.
3808 * config/pdp11/pdp11.opt.urls: Likewise.
3809 * config/pru/pru.opt.urls: Likewise.
3810 * config/riscv/riscv.opt.urls: Likewise.
3811 * config/rl78/rl78.opt.urls: Likewise.
3812 * config/rpath.opt.urls: Likewise.
3813 * config/rs6000/476.opt.urls: Likewise.
3814 * config/rs6000/aix64.opt.urls: Likewise.
3815 * config/rs6000/darwin.opt.urls: Likewise.
3816 * config/rs6000/linux64.opt.urls: Likewise.
3817 * config/rs6000/rs6000-tables.opt.urls: Likewise.
3818 * config/rs6000/rs6000.opt.urls: Likewise.
3819 * config/rs6000/sysv4.opt.urls: Likewise.
3820 * config/rtems.opt.urls: Likewise.
3821 * config/rx/elf.opt.urls: Likewise.
3822 * config/rx/rx.opt.urls: Likewise.
3823 * config/s390/s390.opt.urls: Likewise.
3824 * config/s390/tpf.opt.urls: Likewise.
3825 * config/sh/sh.opt.urls: Likewise.
3826 * config/sh/superh.opt.urls: Likewise.
3827 * config/sol2.opt.urls: Likewise.
3828 * config/sparc/long-double-switch.opt.urls: Likewise.
3829 * config/sparc/sparc.opt.urls: Likewise.
3830 * config/stormy16/stormy16.opt.urls: Likewise.
3831 * config/v850/v850.opt.urls: Likewise.
3832 * config/vax/elf.opt.urls: Likewise.
3833 * config/vax/vax.opt.urls: Likewise.
3834 * config/visium/visium.opt.urls: Likewise.
3835 * config/vms/vms.opt.urls: Likewise.
3836 * config/vxworks-smp.opt.urls: Likewise.
3837 * config/vxworks.opt.urls: Likewise.
3838 * config/xtensa/elf.opt.urls: Likewise.
3839 * config/xtensa/uclinux.opt.urls: Likewise.
3840 * config/xtensa/xtensa.opt.urls: Likewise.
3841 * config/bfin/bfin.opt.urls: New file.
3843 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3845 * Makefile.in (OPT_URLS_HTML_DEPS): New.
3846 (regenerate-opt-urls): New target.
3847 (regenerate-opt-urls-unit-test): New target.
3848 * doc/options.texi (Option properties): Add UrlSuffix and
3849 description of regenerate-opt-urls.py. Add LangUrlSuffix_*.
3850 * doc/sourcebuild.texi (Anatomy of a Language Front End): Add
3851 reference to regenerate-opt-urls.py's PER_LANGUAGE_OPTION_INDEXES
3852 and Makefile.in's OPT_URLS_HTML_DEPS.
3853 (Anatomy of a Target Back End): Add
3854 reference to regenerate-opt-urls.py's TARGET_SPECIFIC_PAGES.
3855 * regenerate-opt-urls.py: New file.
3857 2024-01-04 David Malcolm <dmalcolm@redhat.com>
3859 * diagnostic-format-sarif.cc
3860 (sarif_builder::make_logical_location_object): Convert to...
3861 (make_sarif_logical_location_object): ...this.
3862 (sarif_builder::set_any_logical_locs_arr): Update for above
3864 (sarif_builder::make_thread_flow_location_object): Call
3865 maybe_add_sarif_properties on each diagnostic_event.
3866 * diagnostic-format-sarif.h (class logical_location): New forward
3868 (make_sarif_logical_location_object): New decl.
3869 * diagnostic-path.h (class sarif_object): New forward decl.
3870 (diagnostic_event::maybe_add_sarif_properties): New vfunc.
3872 2024-01-04 Kuan-Lin Chen <rufus@andestech.com>
3873 Patrick Lin <patrick@andestech.com>
3874 Rufus Chen <rufus@andestech.com>
3875 Monk Chiang <monk.chiang@sifive.com>
3877 * config/riscv/riscv.cc (riscv_legitimize_move): Expand movfh
3878 with Nan-boxing value.
3879 * config/riscv/riscv.md (*movhf_softfloat_unspec): New pattern.
3881 2024-01-04 Roger Sayle <roger@nextmovesoftware.com>
3882 Jeff Law <jlaw@ventanamicro.com>
3884 PR rtl-optimization/104914
3885 * expr.cc (expand_assignment): When target is SUBREG_PROMOTED_VAR_P
3886 a sign or zero extension is only required if the modified field
3887 overlaps the SUBREG's most significant bit. On MODE_REP_EXTENDED
3888 targets, don't refer to the temporarily incorrectly extended value
3889 using a SUBREG, but instead generate an explicit TRUNCATE rtx.
3891 2024-01-04 Pan Li <pan2.li@intel.com>
3894 2024-01-04 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3896 * config/riscv/riscv-vector-costs.cc (variable_vectorized_p): Teach vi variant.
3898 2024-01-04 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3900 * config/riscv/riscv-vector-costs.cc (variable_vectorized_p): Teach vi variant.
3902 2024-01-04 Kito Cheng <kito.cheng@sifive.com>
3904 * config/riscv/riscv.cc (riscv_for_each_saved_reg): Adjust the
3907 2024-01-04 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3909 * config/riscv/riscv-vector-costs.cc (variable_vectorized_p): New function.
3910 (compute_nregs_for_mode): Refine LMUL.
3911 (max_number_of_live_regs): Ditto.
3912 (compute_estimated_lmul): Ditto.
3913 (has_unexpected_spills_p): Ditto.
3915 2024-01-04 Li Wei <liwei@loongson.cn>
3917 * config/loongarch/loongarch.cc (loongarch_is_odd_extraction):
3918 Remove useless forward declaration.
3919 (loongarch_is_even_extraction): Remove useless forward declaration.
3920 (loongarch_try_expand_lsx_vshuf_const): Removed.
3921 (loongarch_expand_vec_perm_const_1): Merged.
3922 (loongarch_is_double_duplicate): Removed.
3923 (loongarch_is_center_extraction): Ditto.
3924 (loongarch_is_reversing_permutation): Ditto.
3925 (loongarch_is_di_misalign_extract): Ditto.
3926 (loongarch_is_si_misalign_extract): Ditto.
3927 (loongarch_is_lasx_lowpart_extract): Ditto.
3928 (loongarch_is_op_reverse_perm): Ditto.
3929 (loongarch_is_single_op_perm): Ditto.
3930 (loongarch_is_divisible_perm): Ditto.
3931 (loongarch_is_triple_stride_extract): Ditto.
3932 (loongarch_expand_vec_perm_const_2): Merged.
3933 (loongarch_expand_vec_perm_const): New.
3934 (loongarch_vectorize_vec_perm_const): Adjust.
3936 2024-01-04 Sandra Loosemore <sandra@codesourcery.com>
3938 * omp-general.cc: Fix comment typos and misplaced/confusing
3939 comments. Delete redundant include of omp-general.h.
3941 2024-01-04 YunQiang Su <syq@gcc.gnu.org>
3943 PR rtl-optimization/104914
3944 * config/mips/mips.md (insqisi_extended): New patterns.
3945 (inshisi_extended): Ditto.
3947 2024-01-04 YunQiang Su <syq@gcc.gnu.org>
3949 * config/mips/mips.cc (mips_insn_cost): New function.
3951 2024-01-04 YunQiang Su <syq@gcc.gnu.org>
3953 * config/mips/mips.md (perf_ratio): New attribute.
3955 2024-01-04 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3959 * config/riscv/riscv-vsetvl.cc (invalid_opt_bb_p): New function.
3960 (pre_vsetvl::compute_lcm_local_properties): Disable earliest fusion on
3961 blocks belong to infinite loop.
3962 (pre_vsetvl::emit_vsetvl): Remove fake edges.
3963 * config/riscv/t-riscv: Add a new include file.
3965 2024-01-04 Juzhe-Zhong <juzhe.zhong@rivai.ai>
3967 * config/riscv/vector.md: Fix indent.
3969 2024-01-03 Kwok Cheung Yeung <kcy@codesourcery.com>
3971 * tree-core.h (enum omp_clause_code): Move OMP_CLAUSE_INDIRECT to before
3972 OMP_CLAUSE__SIMDUID_.
3973 * tree.cc (omp_clause_num_ops): Update position of entry for
3974 OMP_CLAUSE_INDIRECT to correspond with omp_clause_code.
3975 (omp_clause_code_name): Likewise.
3977 2024-01-03 Kwok Cheung Yeung <kcy@codesourcery.com>
3979 * config/nvptx/nvptx.cc (nvptx_record_offload_symbol): Restucture
3980 printing of FUNC_MAP/IND_FUNC_MAP labels.
3982 2024-01-03 Jakub Jelinek <jakub@redhat.com>
3984 * gcc.cc (process_command): Update copyright notice dates.
3985 * gcov-dump.cc (print_version): Ditto.
3986 * gcov.cc (print_version): Ditto.
3987 * gcov-tool.cc (print_version): Ditto.
3988 * gengtype.cc (create_file): Ditto.
3989 * doc/cpp.texi: Bump @copying's copyright year.
3990 * doc/cppinternals.texi: Ditto.
3991 * doc/gcc.texi: Ditto.
3992 * doc/gccint.texi: Ditto.
3993 * doc/gcov.texi: Ditto.
3994 * doc/install.texi: Ditto.
3995 * doc/invoke.texi: Ditto.
3997 2024-01-03 Xi Ruoyao <xry111@xry111.site>
3999 * config/loongarch/simd.md (fmax<mode>3): New define_insn.
4000 (fmin<mode>3): Likewise.
4001 (reduc_fmax_scal_<mode>3): New define_expand.
4002 (reduc_fmin_scal_<mode>3): Likewise.
4004 2024-01-03 Juzhe-Zhong <juzhe.zhong@rivai.ai>
4007 * config/riscv/riscv-vector-costs.cc (compute_nregs_for_mode): Add rgroup info.
4008 (max_number_of_live_regs): Ditto.
4009 (has_unexpected_spills_p): Ditto.
4011 2024-01-02 Jun Sha (Joshua) <cooper.joshua@linux.alibaba.com>
4012 Jin Ma <jinma@linux.alibaba.com>
4013 Xianmiao Qu <cooper.qu@linux.alibaba.com>
4014 Christoph Müllner <christoph.muellner@vrull.eu>
4016 * config/riscv/vector.md:
4017 Use vector_length_operand for vsetvl patterns.
4019 2024-01-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
4021 * config/riscv/riscv-v.cc (is_vlmax_len_p): Remove satisfies_constraint_K.
4022 (expand_cond_len_op): Add simplification of dummy len and dummy mask.
4024 2024-01-02 Di Zhao <dizhao@os.amperecomputing.com>
4026 * config/aarch64/aarch64-tuning-flags.def
4027 (AARCH64_EXTRA_TUNING_OPTION): New tuning option
4028 AARCH64_EXTRA_TUNE_FULLY_PIPELINED_FMA.
4029 * config/aarch64/aarch64.cc
4030 (aarch64_override_options_internal): Set
4031 param_fully_pipelined_fma according to tuning option.
4032 * config/aarch64/tuning_models/ampere1.h: Add
4033 AARCH64_EXTRA_TUNE_FULLY_PIPELINED_FMA to tune_flags.
4034 * config/aarch64/tuning_models/ampere1a.h: Likewise.
4035 * config/aarch64/tuning_models/ampere1b.h: Likewise.
4037 2024-01-02 Feng Wang <wangfeng@eswincomputing.com>
4039 * config/riscv/vector-crypto.md: Modify copyright year.
4041 2024-01-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
4043 * config/riscv/riscv-vector-costs.cc: Move STMT_VINFO_TYPE (...) to local.
4045 2024-01-02 Lulu Cheng <chenglulu@loongson.cn>
4047 * config.in: Regenerate.
4048 * config/loongarch/loongarch-opts.h (HAVE_AS_TLS_LE_RELAXATION): Define.
4049 * config/loongarch/loongarch.cc (loongarch_legitimize_tls_address):
4050 Added TLS Le Relax support.
4051 (loongarch_print_operand_reloc): Add the output string of TLS Le Relax.
4052 * config/loongarch/loongarch.md (@add_tls_le_relax<mode>): New template.
4053 * configure: Regenerate.
4054 * configure.ac: Check if binutils supports TLS le relax.
4056 2024-01-02 Feng Wang <wangfeng@eswincomputing.com>
4058 * config/riscv/iterators.md: Add rotate insn name.
4059 * config/riscv/riscv.md: Add new insns name for crypto vector.
4060 * config/riscv/vector-iterators.md: Add new iterators for crypto vector.
4061 * config/riscv/vector.md: Add the corresponding attr for crypto vector.
4062 * config/riscv/vector-crypto.md: New file.The machine descriptions for crypto vector.
4064 2024-01-02 Juzhe-Zhong <juzhe.zhong@rivai.ai>
4067 * config/riscv/riscv-vector-costs.cc (compute_nregs_for_mode): Fix
4068 pointer type liveness count.
4070 Copyright (C) 2024 Free Software Foundation, Inc.
4072 Copying and distribution of this file, with or without modification,
4073 are permitted in any medium without royalty provided the copyright
4074 notice and this notice are preserved.