arm64/lib/strncmp.S

1 /* SPDX-License-Identifier: GPL-2.0-only */
3  * Copyright (c) 2013-2022, Arm Limited.
6  * https://github.com/ARM-software/optimized-routines/blob/189dfefe37d54c5b/string/aarch64/strncmp.S
14  * ARMv8-a, AArch64.
49    On big-endian early bytes are at MSB and on little-endian LSB.
70 	/* NUL detection works on the principle that (X - 1) & (~X) & 0x80
71 	   (=> (X - 1) & ~(X | 0x7f)) is non-zero iff a byte is zero, and
75 	ldr	data1, [src1], #8
76 	ldr	data2, [src2], #8
78 	subs	limit, limit, #8
81 	eor	diff, data1, data2	/* Non-zero if differences found.  */
83 	bics	has_nul, tmp1, tmp2	/* Non-zero if NUL terminator.  */
91 	add	limit, limit, 8	/* Rewind limit to before last subs. */
102 	/* But we need to zero-extend (char is unsigned) the value and then
103 	   perform a signed 32-bit subtraction.  */
109 	/* Not reached the limit, must have found the end or a diff.  */
111 	add	tmp1, limit, 8
114 	lsl	limit, tmp1, #3	/* Bits -> bytes.  */
124 	/* For big-endian we cannot use the trick with the syndrome value
125 	   as carry-propagation can corrupt the upper bits if the trailing
136 	/* Re-compute the NUL-byte detection, using a byte-reversed value.  */
144 	/* The most-significant-non-zero bit of the syndrome marks either the
151 	/* But we need to zero-extend (char is unsigned) the value and then
152 	   perform a signed 32-bit subtraction.  */
166 	ldr	data1, [src1], #8
167 	neg	tmp3, count, lsl #3	/* 64 - bits(bytes beyond align). */
168 	ldr	data2, [src2], #8
195 	/* Align the SRC1 to a dword by doing a bytewise compare and then do
214 	   The bytes are shown in natural order. For little-endian, it is
217 	   src1     | a a a a a a a a | b b b c c c c c | . . .
218 	   src2     | x x x x x a a a   a a a a a b b b | c c c c c . . .
222 	   data1    a a a a a a a a     b b b c c c c c     b b b c c c c c
223 	   data2    a a a a a a a a     b b b 0 0 0 0 0     0 0 0 c c c c c
227 	   Align SRC2 down to 16 bytes. This way we can read 16 bytes at a
231 	/* Calculate offset from 8 byte alignment to string start in bits. No
235 	mov	mask, -1
237 	ldr	data1, [src1], #8
245 	/* STEP_A: Compare full 8 bytes when there is enough data from SRC2.*/
248 	subs	limit, limit, #8
249 	orr	data2, data2, tmp1	/* 8 bytes from SRC2 combined from two regs.*/
251 	eor	diff, data1, data2	/* Non-zero if differences found.  */
254 	bic	has_nul, has_nul, tmp3	/* Non-zero if NUL byte found in SRC1. */
258 	ldr	data1, [src1], #8
263 	/* For big-endian we do a byte reverse to avoid carry-propagation
273 	eor	diff, data2, data1	/* Non-zero if differences found.  */
274 	bic	has_nul, has_nul, tmp3	/* Non-zero if NUL terminator.  */
286 	cmp	limit, #8
288 	eor	diff, data2, data1	/* Non-zero if differences found.  */
294 	ldr	data1, [src1], #8
295 	sub	limit, limit, #8