Orange Pi5 kernel

^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  1) /* SPDX-License-Identifier: GPL-2.0 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  3)  * strlen.S (c) 1995 David Mosberger (davidm@cs.arizona.edu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  4)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  5)  * Finds length of a 0-terminated string.  Optimized for the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  6)  * Alpha architecture:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  7)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  8)  *	- memory accessed as aligned quadwords only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  9)  *	- uses bcmpge to compare 8 bytes in parallel
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10)  *	- does binary search to find 0 byte in last
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11)  *	  quadword (HAKMEM needed 12 instructions to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12)  *	  do this instead of the 9 instructions that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13)  *	  binary search needs).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) #include <asm/export.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) 	.set noreorder
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) 	.set noat
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) 	.align 3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) 	.globl	strlen
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) 	.ent	strlen
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) strlen:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) 	ldq_u	$1, 0($16)	# load first quadword ($16  may be misaligned)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) 	lda	$2, -1($31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) 	insqh	$2, $16, $2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) 	andnot	$16, 7, $0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) 	or	$2, $1, $1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) 	cmpbge	$31, $1, $2	# $2  <- bitmask: bit i == 1 <==> i-th byte == 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) 	bne	$2, found
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) loop:	ldq	$1, 8($0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) 	addq	$0, 8, $0	# addr += 8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) 	nop			# helps dual issue last two insns
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) 	cmpbge	$31, $1, $2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) 	beq	$2, loop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) found:	blbs	$2, done	# make aligned case fast
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) 	negq	$2, $3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) 	and	$2, $3, $2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) 	and	$2, 0x0f, $1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) 	addq	$0, 4, $3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) 	cmoveq	$1, $3, $0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) 	and	$2, 0x33, $1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) 	addq	$0, 2, $3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) 	cmoveq	$1, $3, $0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) 	and	$2, 0x55, $1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) 	addq	$0, 1, $3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) 	cmoveq	$1, $3, $0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) done:	subq	$0, $16, $0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) 	ret	$31, ($26)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) 	.end	strlen
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) 	EXPORT_SYMBOL(strlen)