diff options
Diffstat (limited to 'sysdeps/loongarch/lp64/multiarch/strrchr-aligned.S')
-rw-r--r-- | sysdeps/loongarch/lp64/multiarch/strrchr-aligned.S | 170 |
1 files changed, 170 insertions, 0 deletions
diff --git a/sysdeps/loongarch/lp64/multiarch/strrchr-aligned.S b/sysdeps/loongarch/lp64/multiarch/strrchr-aligned.S new file mode 100644 index 0000000000..a73deb7840 --- /dev/null +++ b/sysdeps/loongarch/lp64/multiarch/strrchr-aligned.S @@ -0,0 +1,170 @@ +/* Optimized strrchr implementation using basic LoongArch instructions. + Copyright (C) 2023 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library. If not, see + <https://www.gnu.org/licenses/>. */ + +#include <sysdep.h> +#include <sys/regdef.h> +#include <sys/asm.h> + +#if IS_IN (libc) +# define STRRCHR __strrchr_aligned +#else +# define STRRCHR strrchr +#endif + +LEAF(STRRCHR, 6) + slli.d t0, a0, 3 + bstrins.d a0, zero, 2, 0 + lu12i.w a2, 0x01010 + ld.d t2, a0, 0 + + andi a1, a1, 0xff + ori a2, a2, 0x101 + li.d t3, -1 + bstrins.d a2, a2, 63, 32 + + sll.d t5, t3, t0 + slli.d a3, a2, 7 + orn t4, t2, t5 + mul.d a1, a1, a2 + + sub.d t0, t4, a2 + andn t1, a3, t4 + and t1, t0, t1 + beqz t1, L(find_tail) + + + ctz.d t0, t1 + orn t0, zero, t0 + xor t2, t4, a1 + srl.d t0, t3, t0 + + orn t2, t2, t0 + orn t2, t2, t5 + revb.d t2, t2 + sub.d t1, t2, a2 + + andn t0, a3, t2 + and t1, t0, t1 + ctz.d t0, t1 + srli.d t0, t0, 3 + + addi.d a0, a0, 7 + sub.d a0, a0, t0 + maskeqz a0, a0, t1 + jr ra + + +L(find_tail): + addi.d a4, a0, 8 + addi.d a0, a0, 8 +L(loop_ascii): + ld.d t2, a0, 0 + sub.d t1, t2, a2 + + and t0, t1, a3 + bnez t0, L(more_check) + ld.d t2, a0, 8 + sub.d t1, t2, a2 + + and t0, t1, a3 + addi.d a0, a0, 16 + beqz t0, L(loop_ascii) + addi.d a0, a0, -8 + +L(more_check): + andn t0, a3, t2 + and t1, t1, t0 + bnez t1, L(tail) + addi.d a0, a0, 8 + + +L(loop_nonascii): + ld.d t2, a0, 0 + sub.d t1, t2, a2 + andn t0, a3, t2 + and t1, t0, t1 + + bnez t1, L(tail) + ld.d t2, a0, 8 + addi.d a0, a0, 16 + sub.d t1, t2, a2 + + andn t0, a3, t2 + and t1, t0, t1 + beqz t1, L(loop_nonascii) + addi.d a0, a0, -8 + +L(tail): + ctz.d t0, t1 + orn t0, zero, t0 + xor t2, t2, a1 + srl.d t0, t3, t0 + + + orn t2, t2, t0 + revb.d t2, t2 + sub.d t1, t2, a2 + andn t0, a3, t2 + + and t1, t0, t1 + bnez t1, L(count_pos) +L(find_loop): + beq a0, a4, L(find_end) + ld.d t2, a0, -8 + + addi.d a0, a0, -8 + xor t2, t2, a1 + sub.d t1, t2, a2 + andn t0, a3, t2 + + and t1, t0, t1 + beqz t1, L(find_loop) + revb.d t2, t2 + sub.d t1, t2, a2 + + + andn t0, a3, t2 + and t1, t0, t1 +L(count_pos): + ctz.d t0, t1 + addi.d a0, a0, 7 + + srli.d t0, t0, 3 + sub.d a0, a0, t0 + jr ra + nop + +L(find_end): + xor t2, t4, a1 + orn t2, t2, t5 + revb.d t2, t2 + sub.d t1, t2, a2 + + + andn t0, a3, t2 + and t1, t0, t1 + ctz.d t0, t1 + srli.d t0, t0, 3 + + addi.d a0, a4, -1 + sub.d a0, a0, t0 + maskeqz a0, a0, t1 + jr ra +END(STRRCHR) + +libc_hidden_builtin_def(STRRCHR) |