1/*
2 *  linux/arch/arm/lib/csumpartialcopygeneric.S
3 *
4 *  Copyright (C) 1995-2001 Russell King
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 as
8 * published by the Free Software Foundation.
9 */
10#include <asm/assembler.h>
11
12/*
13 * unsigned int
14 * csum_partial_copy_xxx(const char *src, char *dst, int len, int sum, )
15 *  r0 = src, r1 = dst, r2 = len, r3 = sum
16 *  Returns : r0 = checksum
17 *
18 * Note that 'tst' and 'teq' preserve the carry flag.
19 */
20
21src	.req	r0
22dst	.req	r1
23len	.req	r2
24sum	.req	r3
25
26.Lzero:		mov	r0, sum
27		load_regs
28
29		/*
30		 * Align an unaligned destination pointer.  We know that
31		 * we have >= 8 bytes here, so we don't need to check
32		 * the length.  Note that the source pointer hasn't been
33		 * aligned yet.
34		 */
35.Ldst_unaligned:
36		tst	dst, #1
37		beq	.Ldst_16bit
38
39		load1b	ip
40		sub	len, len, #1
41		adcs	sum, sum, ip, put_byte_1	@ update checksum
42		strb	ip, [dst], #1
43		tst	dst, #2
44		reteq	lr			@ dst is now 32bit aligned
45
46.Ldst_16bit:	load2b	r8, ip
47		sub	len, len, #2
48		adcs	sum, sum, r8, put_byte_0
49		strb	r8, [dst], #1
50		adcs	sum, sum, ip, put_byte_1
51		strb	ip, [dst], #1
52		ret	lr			@ dst is now 32bit aligned
53
54		/*
55		 * Handle 0 to 7 bytes, with any alignment of source and
56		 * destination pointers.  Note that when we get here, C = 0
57		 */
58.Lless8:	teq	len, #0			@ check for zero count
59		beq	.Lzero
60
61		/* we must have at least one byte. */
62		tst	dst, #1			@ dst 16-bit aligned
63		beq	.Lless8_aligned
64
65		/* Align dst */
66		load1b	ip
67		sub	len, len, #1
68		adcs	sum, sum, ip, put_byte_1	@ update checksum
69		strb	ip, [dst], #1
70		tst	len, #6
71		beq	.Lless8_byteonly
72
731:		load2b	r8, ip
74		sub	len, len, #2
75		adcs	sum, sum, r8, put_byte_0
76		strb	r8, [dst], #1
77		adcs	sum, sum, ip, put_byte_1
78		strb	ip, [dst], #1
79.Lless8_aligned:
80		tst	len, #6
81		bne	1b
82.Lless8_byteonly:
83		tst	len, #1
84		beq	.Ldone
85		load1b	r8
86		adcs	sum, sum, r8, put_byte_0	@ update checksum
87		strb	r8, [dst], #1
88		b	.Ldone
89
90FN_ENTRY
91		save_regs
92
93		cmp	len, #8			@ Ensure that we have at least
94		blo	.Lless8			@ 8 bytes to copy.
95
96		adds	sum, sum, #0		@ C = 0
97		tst	dst, #3			@ Test destination alignment
98		blne	.Ldst_unaligned		@ align destination, return here
99
100		/*
101		 * Ok, the dst pointer is now 32bit aligned, and we know
102		 * that we must have more than 4 bytes to copy.  Note
103		 * that C contains the carry from the dst alignment above.
104		 */
105
106		tst	src, #3			@ Test source alignment
107		bne	.Lsrc_not_aligned
108
109		/* Routine for src & dst aligned */
110
111		bics	ip, len, #15
112		beq	2f
113
1141:		load4l	r4, r5, r6, r7
115		stmia	dst!, {r4, r5, r6, r7}
116		adcs	sum, sum, r4
117		adcs	sum, sum, r5
118		adcs	sum, sum, r6
119		adcs	sum, sum, r7
120		sub	ip, ip, #16
121		teq	ip, #0
122		bne	1b
123
1242:		ands	ip, len, #12
125		beq	4f
126		tst	ip, #8
127		beq	3f
128		load2l	r4, r5
129		stmia	dst!, {r4, r5}
130		adcs	sum, sum, r4
131		adcs	sum, sum, r5
132		tst	ip, #4
133		beq	4f
134
1353:		load1l	r4
136		str	r4, [dst], #4
137		adcs	sum, sum, r4
138
1394:		ands	len, len, #3
140		beq	.Ldone
141		load1l	r4
142		tst	len, #2
143		mov	r5, r4, get_byte_0
144		beq	.Lexit
145		adcs	sum, sum, r4, lspush #16
146		strb	r5, [dst], #1
147		mov	r5, r4, get_byte_1
148		strb	r5, [dst], #1
149		mov	r5, r4, get_byte_2
150.Lexit:		tst	len, #1
151		strneb	r5, [dst], #1
152		andne	r5, r5, #255
153		adcnes	sum, sum, r5, put_byte_0
154
155		/*
156		 * If the dst pointer was not 16-bit aligned, we
157		 * need to rotate the checksum here to get around
158		 * the inefficient byte manipulations in the
159		 * architecture independent code.
160		 */
161.Ldone:		adc	r0, sum, #0
162		ldr	sum, [sp, #0]		@ dst
163		tst	sum, #1
164		movne	r0, r0, ror #8
165		load_regs
166
167.Lsrc_not_aligned:
168		adc	sum, sum, #0		@ include C from dst alignment
169		and	ip, src, #3
170		bic	src, src, #3
171		load1l	r5
172		cmp	ip, #2
173		beq	.Lsrc2_aligned
174		bhi	.Lsrc3_aligned
175		mov	r4, r5, lspull #8		@ C = 0
176		bics	ip, len, #15
177		beq	2f
1781:		load4l	r5, r6, r7, r8
179		orr	r4, r4, r5, lspush #24
180		mov	r5, r5, lspull #8
181		orr	r5, r5, r6, lspush #24
182		mov	r6, r6, lspull #8
183		orr	r6, r6, r7, lspush #24
184		mov	r7, r7, lspull #8
185		orr	r7, r7, r8, lspush #24
186		stmia	dst!, {r4, r5, r6, r7}
187		adcs	sum, sum, r4
188		adcs	sum, sum, r5
189		adcs	sum, sum, r6
190		adcs	sum, sum, r7
191		mov	r4, r8, lspull #8
192		sub	ip, ip, #16
193		teq	ip, #0
194		bne	1b
1952:		ands	ip, len, #12
196		beq	4f
197		tst	ip, #8
198		beq	3f
199		load2l	r5, r6
200		orr	r4, r4, r5, lspush #24
201		mov	r5, r5, lspull #8
202		orr	r5, r5, r6, lspush #24
203		stmia	dst!, {r4, r5}
204		adcs	sum, sum, r4
205		adcs	sum, sum, r5
206		mov	r4, r6, lspull #8
207		tst	ip, #4
208		beq	4f
2093:		load1l	r5
210		orr	r4, r4, r5, lspush #24
211		str	r4, [dst], #4
212		adcs	sum, sum, r4
213		mov	r4, r5, lspull #8
2144:		ands	len, len, #3
215		beq	.Ldone
216		mov	r5, r4, get_byte_0
217		tst	len, #2
218		beq	.Lexit
219		adcs	sum, sum, r4, lspush #16
220		strb	r5, [dst], #1
221		mov	r5, r4, get_byte_1
222		strb	r5, [dst], #1
223		mov	r5, r4, get_byte_2
224		b	.Lexit
225
226.Lsrc2_aligned:	mov	r4, r5, lspull #16
227		adds	sum, sum, #0
228		bics	ip, len, #15
229		beq	2f
2301:		load4l	r5, r6, r7, r8
231		orr	r4, r4, r5, lspush #16
232		mov	r5, r5, lspull #16
233		orr	r5, r5, r6, lspush #16
234		mov	r6, r6, lspull #16
235		orr	r6, r6, r7, lspush #16
236		mov	r7, r7, lspull #16
237		orr	r7, r7, r8, lspush #16
238		stmia	dst!, {r4, r5, r6, r7}
239		adcs	sum, sum, r4
240		adcs	sum, sum, r5
241		adcs	sum, sum, r6
242		adcs	sum, sum, r7
243		mov	r4, r8, lspull #16
244		sub	ip, ip, #16
245		teq	ip, #0
246		bne	1b
2472:		ands	ip, len, #12
248		beq	4f
249		tst	ip, #8
250		beq	3f
251		load2l	r5, r6
252		orr	r4, r4, r5, lspush #16
253		mov	r5, r5, lspull #16
254		orr	r5, r5, r6, lspush #16
255		stmia	dst!, {r4, r5}
256		adcs	sum, sum, r4
257		adcs	sum, sum, r5
258		mov	r4, r6, lspull #16
259		tst	ip, #4
260		beq	4f
2613:		load1l	r5
262		orr	r4, r4, r5, lspush #16
263		str	r4, [dst], #4
264		adcs	sum, sum, r4
265		mov	r4, r5, lspull #16
2664:		ands	len, len, #3
267		beq	.Ldone
268		mov	r5, r4, get_byte_0
269		tst	len, #2
270		beq	.Lexit
271		adcs	sum, sum, r4
272		strb	r5, [dst], #1
273		mov	r5, r4, get_byte_1
274		strb	r5, [dst], #1
275		tst	len, #1
276		beq	.Ldone
277		load1b	r5
278		b	.Lexit
279
280.Lsrc3_aligned:	mov	r4, r5, lspull #24
281		adds	sum, sum, #0
282		bics	ip, len, #15
283		beq	2f
2841:		load4l	r5, r6, r7, r8
285		orr	r4, r4, r5, lspush #8
286		mov	r5, r5, lspull #24
287		orr	r5, r5, r6, lspush #8
288		mov	r6, r6, lspull #24
289		orr	r6, r6, r7, lspush #8
290		mov	r7, r7, lspull #24
291		orr	r7, r7, r8, lspush #8
292		stmia	dst!, {r4, r5, r6, r7}
293		adcs	sum, sum, r4
294		adcs	sum, sum, r5
295		adcs	sum, sum, r6
296		adcs	sum, sum, r7
297		mov	r4, r8, lspull #24
298		sub	ip, ip, #16
299		teq	ip, #0
300		bne	1b
3012:		ands	ip, len, #12
302		beq	4f
303		tst	ip, #8
304		beq	3f
305		load2l	r5, r6
306		orr	r4, r4, r5, lspush #8
307		mov	r5, r5, lspull #24
308		orr	r5, r5, r6, lspush #8
309		stmia	dst!, {r4, r5}
310		adcs	sum, sum, r4
311		adcs	sum, sum, r5
312		mov	r4, r6, lspull #24
313		tst	ip, #4
314		beq	4f
3153:		load1l	r5
316		orr	r4, r4, r5, lspush #8
317		str	r4, [dst], #4
318		adcs	sum, sum, r4
319		mov	r4, r5, lspull #24
3204:		ands	len, len, #3
321		beq	.Ldone
322		mov	r5, r4, get_byte_0
323		tst	len, #2
324		beq	.Lexit
325		strb	r5, [dst], #1
326		adcs	sum, sum, r4
327		load1l	r4
328		mov	r5, r4, get_byte_0
329		strb	r5, [dst], #1
330		adcs	sum, sum, r4, lspush #24
331		mov	r5, r4, get_byte_1
332		b	.Lexit
333FN_EXIT
334