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
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
|
#include <sys/sysmacros.h>
#include "config.h"
/*
* XXX: currently borrrowed from libsolcompat until this
* can be adapted to the linux kernel interfaces.
*/
#if 0
/*
* =========================================================================
* vnode operations
* =========================================================================
*/
/*
* Note: for the xxxat() versions of these functions, we assume that the
* starting vp is always rootdir (which is true for spa_directory.c, the only
* ZFS consumer of these interfaces). We assert this is true, and then emulate
* them by adding '/' in front of the path.
*/
/*ARGSUSED*/
int
vn_open(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3)
{
int fd;
vnode_t *vp;
int old_umask;
char realpath[MAXPATHLEN];
struct stat64 st;
/*
* If we're accessing a real disk from userland, we need to use
* the character interface to avoid caching. This is particularly
* important if we're trying to look at a real in-kernel storage
* pool from userland, e.g. via zdb, because otherwise we won't
* see the changes occurring under the segmap cache.
* On the other hand, the stupid character device returns zero
* for its size. So -- gag -- we open the block device to get
* its size, and remember it for subsequent VOP_GETATTR().
*/
#if defined(__sun__) || defined(__sun)
if (strncmp(path, "/dev/", 5) == 0) {
#else
if (0) {
#endif
char *dsk;
fd = open64(path, O_RDONLY);
if (fd == -1)
return (errno);
if (fstat64(fd, &st) == -1) {
close(fd);
return (errno);
}
close(fd);
(void) sprintf(realpath, "%s", path);
dsk = strstr(path, "/dsk/");
if (dsk != NULL)
(void) sprintf(realpath + (dsk - path) + 1, "r%s",
dsk + 1);
} else {
(void) sprintf(realpath, "%s", path);
if (!(flags & FCREAT) && stat64(realpath, &st) == -1)
return (errno);
}
#ifdef __linux__
if (!(flags & FCREAT) && S_ISBLK(st.st_mode)) {
flags |= O_DIRECT;
if (flags & FWRITE)
flags |= O_EXCL;
}
#endif
if (flags & FCREAT)
old_umask = umask(0);
/*
* The construct 'flags - FREAD' conveniently maps combinations of
* FREAD and FWRITE to the corresponding O_RDONLY, O_WRONLY, and O_RDWR.
*/
fd = open64(realpath, flags - FREAD, mode);
if (flags & FCREAT)
(void) umask(old_umask);
if (fd == -1)
return (errno);
if (fstat64(fd, &st) == -1) {
close(fd);
return (errno);
}
(void) fcntl(fd, F_SETFD, FD_CLOEXEC);
*vpp = vp = umem_zalloc(sizeof (vnode_t), UMEM_NOFAIL);
vp->v_fd = fd;
vp->v_size = st.st_size;
vp->v_mode = st.st_mode;
vp->v_path = spa_strdup(path);
return (0);
}
/*ARGSUSED*/
int
vn_openat(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2,
int x3, vnode_t *startvp, int fd)
{
char *realpath = umem_alloc(strlen(path) + 2, UMEM_NOFAIL);
int ret;
ASSERT(startvp == rootdir);
(void) sprintf(realpath, "/%s", path);
/* fd ignored for now, need if want to simulate nbmand support */
ret = vn_open(realpath, x1, flags, mode, vpp, x2, x3);
umem_free(realpath, strlen(path) + 2);
return (ret);
}
/*ARGSUSED*/
int
vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len, offset_t offset,
int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp)
{
ssize_t iolen, split;
if (uio == UIO_READ) {
iolen = pread64(vp->v_fd, addr, len, offset);
} else {
/*
* To simulate partial disk writes, we split writes into two
* system calls so that the process can be killed in between.
*/
#ifdef ZFS_DEBUG
if (!S_ISBLK(vp->v_mode) && !S_ISCHR(vp->v_mode)) {
split = (len > 0 ? rand() % len : 0);
iolen = pwrite64(vp->v_fd, addr, split, offset);
iolen += pwrite64(vp->v_fd, (char *)addr + split,
len - split, offset + split);
} else
iolen = pwrite64(vp->v_fd, addr, len, offset);
#else
iolen = pwrite64(vp->v_fd, addr, len, offset);
#endif
}
if (iolen < 0)
return (errno);
if (residp)
*residp = len - iolen;
else if (iolen != len)
return (EIO);
return (0);
}
void
vn_close(vnode_t *vp)
{
close(vp->v_fd);
spa_strfree(vp->v_path);
umem_free(vp, sizeof (vnode_t));
}
#endif
|
