[glibc] AArch64: Improve strlen_asimd

[Wilco Dijkstra <wilco@sourceware.org>]

https://sourceware.org/git/gitweb.cgi?p=glibc.git;h=1bbb1a2022e126f21810d3d0ebe0a975d5243e43

commit 1bbb1a2022e126f21810d3d0ebe0a975d5243e43
Author: Wilco Dijkstra <wilco.dijkstra@arm.com>
Date:   Wed Jan 11 13:51:48 2023 +0000

    AArch64: Improve strlen_asimd
    
    Use shrn for the mask, merge tst+bne into cbnz, and tweak code alignment.
    Performance improves slightly as a result.
    
    Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>

Diff:
---
 sysdeps/aarch64/multiarch/strlen_asimd.S | 16 ++++------------
 1 file changed, 4 insertions(+), 12 deletions(-)

diff --git a/sysdeps/aarch64/multiarch/strlen_asimd.S b/sysdeps/aarch64/multiarch/strlen_asimd.S
index ca6ab96ecf..490439491d 100644
--- a/sysdeps/aarch64/multiarch/strlen_asimd.S
+++ b/sysdeps/aarch64/multiarch/strlen_asimd.S
@@ -48,6 +48,7 @@
 #define tmp	x2
 #define tmpw	w2
 #define synd	x3
+#define syndw	w3
 #define shift	x4
 
 /* For the first 32 bytes, NUL detection works on the principle that
@@ -87,7 +88,6 @@
 
 ENTRY (__strlen_asimd)
 	PTR_ARG (0)
-
 	and	tmp1, srcin, MIN_PAGE_SIZE - 1
 	cmp	tmp1, MIN_PAGE_SIZE - 32
 	b.hi	L(page_cross)
@@ -123,7 +123,6 @@ ENTRY (__strlen_asimd)
 	add	len, len, tmp1, lsr 3
 	ret
 
-	.p2align 3
 	/* Look for a NUL byte at offset 16..31 in the string.  */
 L(bytes16_31):
 	ldp	data1, data2, [srcin, 16]
@@ -151,6 +150,7 @@ L(bytes16_31):
 	add	len, len, tmp1, lsr 3
 	ret
 
+	nop
 L(loop_entry):
 	bic	src, srcin, 31
 
@@ -166,18 +166,12 @@ L(loop):
 	/* Low 32 bits of synd are non-zero if a NUL was found in datav1.  */
 	cmeq	maskv.16b, datav1.16b, 0
 	sub	len, src, srcin
-	tst	synd, 0xffffffff
-	b.ne	1f
+	cbnz	syndw, 1f
 	cmeq	maskv.16b, datav2.16b, 0
 	add	len, len, 16
 1:
 	/* Generate a bitmask and compute correct byte offset.  */
-#ifdef __AARCH64EB__
-	bic	maskv.8h, 0xf0
-#else
-	bic	maskv.8h, 0x0f, lsl 8
-#endif
-	umaxp	maskv.16b, maskv.16b, maskv.16b
+	shrn	maskv.8b, maskv.8h, 4
 	fmov	synd, maskd
 #ifndef __AARCH64EB__
 	rbit	synd, synd
@@ -186,8 +180,6 @@ L(loop):
 	add	len, len, tmp, lsr 2
 	ret
 
-        .p2align 4
-
 L(page_cross):
 	bic	src, srcin, 31
 	mov	tmpw, 0x0c03