0928368f62
This adds the Arm Optimized Routines (see https://github.com/ARM-software/optimized-routines) source code under the the LLVM license. The version of the code provided in this patch is v20.02 of the Arm Optimized Routines project. This entire contribution is being committed as is even though it does not currently fit the LLVM libc model and does not follow the LLVM coding style. In the near future, implementations from this patch will be moved over to their right place in the LLVM-libc tree. This will be done over many small patches, all of which will go through the normal LLVM code review process. See this libc-dev post for the plan: http://lists.llvm.org/pipermail/libc-dev/2020-March/000044.html Differential revision of the original upload: https://reviews.llvm.org/D75355
61 lines
1.7 KiB
ArmAsm
61 lines
1.7 KiB
ArmAsm
/*
|
|
* __strcmp_aarch64_sve - compare two strings
|
|
*
|
|
* Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
* See https://llvm.org/LICENSE.txt for license information.
|
|
* SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
*/
|
|
|
|
#if __ARM_FEATURE_SVE
|
|
/* Assumptions:
|
|
*
|
|
* ARMv8-a, AArch64
|
|
* SVE Available.
|
|
*/
|
|
|
|
.arch armv8-a+sve
|
|
.text
|
|
|
|
.globl __strcmp_aarch64_sve
|
|
.type __strcmp_aarch64_sve, %function
|
|
.p2align 4
|
|
__strcmp_aarch64_sve:
|
|
setffr /* initialize FFR */
|
|
ptrue p1.b, all /* all ones; loop invariant */
|
|
mov x2, 0 /* initialize offset */
|
|
nop
|
|
|
|
/* Read a vector's worth of bytes, stopping on first fault. */
|
|
0: ldff1b z0.b, p1/z, [x0, x2]
|
|
ldff1b z1.b, p1/z, [x1, x2]
|
|
rdffrs p0.b, p1/z
|
|
b.nlast 2f
|
|
|
|
/* First fault did not fail: the whole vector is valid.
|
|
Avoid depending on the contents of FFR beyond the branch. */
|
|
incb x2, all /* skip bytes for next round */
|
|
cmpeq p2.b, p1/z, z0.b, z1.b /* compare strings */
|
|
cmpne p3.b, p1/z, z0.b, 0 /* search for ~zero */
|
|
nands p2.b, p1/z, p2.b, p3.b /* ~(eq & ~zero) -> ne | zero */
|
|
b.none 0b
|
|
|
|
/* Found end-of-string or inequality. */
|
|
1: brkb p2.b, p1/z, p2.b /* find first such */
|
|
lasta w0, p2, z0.b /* extract each char */
|
|
lasta w1, p2, z1.b
|
|
sub x0, x0, x1 /* return comparison */
|
|
ret
|
|
|
|
/* First fault failed: only some of the vector is valid.
|
|
Perform the comparison only on the valid bytes. */
|
|
2: incp x2, p0.b /* skip bytes for next round */
|
|
setffr /* re-init FFR for next round */
|
|
cmpeq p2.b, p0/z, z0.b, z1.b /* compare strings, as above */
|
|
cmpne p3.b, p0/z, z0.b, 0
|
|
nands p2.b, p0/z, p2.b, p3.b
|
|
b.none 0b
|
|
b 1b
|
|
|
|
.size __strcmp_aarch64_sve, . - __strcmp_aarch64_sve
|
|
#endif
|