blob: 1cbfd04fc496bb0c60b5984a2d932ca97b5b0697 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
|
/*
* Copyright (C) 2004 Joakim Tjernlund
* Copyright (C) 2000-2005 Erik Andersen <andersen@uclibc.org>
*
* Licensed under the LGPL v2.1, see the file COPYING.LIB in this tarball.
*/
/* These are carefully optimized mem*() functions for PPC written in C.
* Don't muck around with these function without checking the generated
* assembler code.
* It is possible to optimize these significantly more by using specific
* data cache instructions(mainly dcbz). However that requires knownledge
* about the CPU's cache line size.
*
* BUG ALERT!
* The cache instructions on MPC8xx CPU's are buggy(they don't update
* the DAR register when causing a DTLB Miss/Error) and cannot be
* used on 8xx CPU's without a kernel patch to work around this
* problem.
*/
#include <string.h>
/* Experimentally off - libc_hidden_proto(memset) */
static __inline__ int expand_byte_word(int c){
/* this does:
c = c << 8 | c;
c = c << 16 | c ;
*/
__asm__("rlwimi %0,%0,8,16,23\n"
"\trlwimi %0,%0,16,0,15\n"
: "=r" (c) : "0" (c));
return c;
}
void *memset(void *to, int c, size_t n)
{
unsigned long rem, chunks;
unsigned char *tmp_to;
chunks = n / 8;
tmp_to = to - 4;
c = expand_byte_word(c);
if (!chunks)
goto lessthan8;
rem = (unsigned long )tmp_to % 4;
if (rem)
goto align;
copy_chunks:
do {
*(unsigned long *)(tmp_to+4) = c;
tmp_to += 4;
*(unsigned long *)(tmp_to+4) = c;
tmp_to += 4;
} while (--chunks);
lessthan8:
n = n % 8;
if (n >= 4) {
*(unsigned long *)(tmp_to+4) = c;
tmp_to += 4;
n = n-4;
}
if (!n ) return to;
tmp_to += 3;
do {
*++tmp_to = c;
} while (--n);
return to;
align:
rem = 4 - rem;
n = n-rem;
do {
*(tmp_to+4) = c;
++tmp_to;
} while (--rem);
chunks = n / 8;
if (chunks)
goto copy_chunks;
goto lessthan8;
}
libc_hidden_def(memset)
|
