xref: /linux/arch/riscv/lib/memcpy.S (revision 22d55f02b8922a097cd4be1e2f131dfa7ef65901) 
1/* SPDX-License-Identifier: GPL-2.0-only */
2/*
3 * Copyright (C) 2013 Regents of the University of California
4 */
5
6#include <linux/linkage.h>
7#include <asm/asm.h>
8
9/* void *memcpy(void *, const void *, size_t) */
10ENTRY(memcpy)
11	move t6, a0  /* Preserve return value */
12
13	/* Defer to byte-oriented copy for small sizes */
14	sltiu a3, a2, 128
15	bnez a3, 4f
16	/* Use word-oriented copy only if low-order bits match */
17	andi a3, t6, SZREG-1
18	andi a4, a1, SZREG-1
19	bne a3, a4, 4f
20
21	beqz a3, 2f  /* Skip if already aligned */
22	/*
23	 * Round to nearest double word-aligned address
24	 * greater than or equal to start address
25	 */
26	andi a3, a1, ~(SZREG-1)
27	addi a3, a3, SZREG
28	/* Handle initial misalignment */
29	sub a4, a3, a1
301:
31	lb a5, 0(a1)
32	addi a1, a1, 1
33	sb a5, 0(t6)
34	addi t6, t6, 1
35	bltu a1, a3, 1b
36	sub a2, a2, a4  /* Update count */
37
382:
39	andi a4, a2, ~((16*SZREG)-1)
40	beqz a4, 4f
41	add a3, a1, a4
423:
43	REG_L a4,       0(a1)
44	REG_L a5,   SZREG(a1)
45	REG_L a6, 2*SZREG(a1)
46	REG_L a7, 3*SZREG(a1)
47	REG_L t0, 4*SZREG(a1)
48	REG_L t1, 5*SZREG(a1)
49	REG_L t2, 6*SZREG(a1)
50	REG_L t3, 7*SZREG(a1)
51	REG_L t4, 8*SZREG(a1)
52	REG_L t5, 9*SZREG(a1)
53	REG_S a4,       0(t6)
54	REG_S a5,   SZREG(t6)
55	REG_S a6, 2*SZREG(t6)
56	REG_S a7, 3*SZREG(t6)
57	REG_S t0, 4*SZREG(t6)
58	REG_S t1, 5*SZREG(t6)
59	REG_S t2, 6*SZREG(t6)
60	REG_S t3, 7*SZREG(t6)
61	REG_S t4, 8*SZREG(t6)
62	REG_S t5, 9*SZREG(t6)
63	REG_L a4, 10*SZREG(a1)
64	REG_L a5, 11*SZREG(a1)
65	REG_L a6, 12*SZREG(a1)
66	REG_L a7, 13*SZREG(a1)
67	REG_L t0, 14*SZREG(a1)
68	REG_L t1, 15*SZREG(a1)
69	addi a1, a1, 16*SZREG
70	REG_S a4, 10*SZREG(t6)
71	REG_S a5, 11*SZREG(t6)
72	REG_S a6, 12*SZREG(t6)
73	REG_S a7, 13*SZREG(t6)
74	REG_S t0, 14*SZREG(t6)
75	REG_S t1, 15*SZREG(t6)
76	addi t6, t6, 16*SZREG
77	bltu a1, a3, 3b
78	andi a2, a2, (16*SZREG)-1  /* Update count */
79
804:
81	/* Handle trailing misalignment */
82	beqz a2, 6f
83	add a3, a1, a2
84
85	/* Use word-oriented copy if co-aligned to word boundary */
86	or a5, a1, t6
87	or a5, a5, a3
88	andi a5, a5, 3
89	bnez a5, 5f
907:
91	lw a4, 0(a1)
92	addi a1, a1, 4
93	sw a4, 0(t6)
94	addi t6, t6, 4
95	bltu a1, a3, 7b
96
97	ret
98
995:
100	lb a4, 0(a1)
101	addi a1, a1, 1
102	sb a4, 0(t6)
103	addi t6, t6, 1
104	bltu a1, a3, 5b
1056:
106	ret
107END(memcpy)
108