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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353
|
/*
* devtab.c
*
* Mapping of disk device numbers to ids used in pseudo inode numbers.
*
* Pseudo inode numbers (psi's) are made up of a combination of an index,
* representing a device number, and the inode. Generation of psi's
* happens like this:
*
* bit prefix index inode # inode max
* 0 3bits 28bits 268435456
* 10 3bits 27bits 134217728
* 110 3bits 26bits 67108864
* 1110 3bits 25bits 33554432
* ...
*
* Even for people exporting 24 different partitions, the smallest
* range of inode numbers we can represent is 25bits.
*
* For mapping device numbers to indices, we use a file
* (/var/state/nfs/devtab), containing the names of the device files.
* The reason to make this a human-readable file is that this allows
* the administrator to reorder entries so that the biggest partitions
* are named first, and thus receive a 28bit inode range.
*
* Beware that when modifying the device table, you must first kill
* mountd and nfsd. Also make sure that no client has mounted a file
* system from your server, otherwise they might be accessing random
* files, with the possibility of hosing your entire system.
*
* Copyright (C) 1998, Olaf Kirch <okir@monad.swb.de>
*/
#include "system.h"
#include "logging.h"
#include "auth.h"
#ifdef ENABLE_DEVTAB
#ifndef PATH_DEVTAB
# define PATH_DEVTAB "/var/state/nfs/devtab"
#endif
#define PATH_DEVTAB_LOCK PATH_DEVTAB ".lock"
static void devtab_read(void);
static void devtab_lock(void);
static void devtab_unlock(void);
static FILE * devtab_open(const char *mode, struct stat *sbp);
static unsigned int devtab_add(dev_t dev);
static const char * devtab_getname(dev_t dev);
static dev_t * devtab;
static unsigned int nrdevs;
static time_t devtab_mtime;
static int devtab_locked = 0;
/*
* Locate the index associated with the given device number
*/
unsigned int
devtab_index(dev_t dev)
{
unsigned int index;
struct stat stb;
FILE *fp;
int oldmask;
/* First, try to find entry in device table */
for (index = 0; index < nrdevs; index++) {
if (devtab[index] == dev)
return index;
}
if (logging_enabled(D_DEVTAB)) {
Dprintf(D_DEVTAB, "Can't find device 0x%x (%s) in devtab.",
dev, devtab_getname(dev));
}
/* Entry not found. We need to create a new entry. */
/* Set proper credentials and umask */
auth_override_uid(ROOT_UID);
oldmask = umask(0);
/* First, lock devtab file */
devtab_lock();
fp = devtab_open("a", &stb);
if (stb.st_mtime != devtab_mtime) {
Dprintf(D_DEVTAB, "%sreading devtab file.\n",
devtab_mtime? "re" : "");
fclose(fp);
devtab_read();
for (index = 0; index < nrdevs; index++) {
if (devtab[index] == dev)
goto done;
}
/* re-open devtab */
fp = devtab_open("a", &stb);
}
/* Add the device to the in-core table */
index = devtab_add(dev);
/* Find the device name and append to devtab file */
fprintf(fp, "%s\n", devtab_getname(dev));
fclose(fp);
done:
devtab_unlock();
auth_override_uid(auth_uid);
umask(oldmask);
return index;
}
static void
devtab_lock(void)
{
char tempname[sizeof(PATH_DEVTAB)+16], buffer[64];
unsigned int retry = 0, maxretry = 10;
int fd, n;
int pid = getpid();
if (devtab_locked)
return;
/* Create temporary lock file, and write our PID to it. */
sprintf(tempname, "%s.%d", PATH_DEVTAB, pid);
if ((fd = open(tempname, O_WRONLY|O_EXCL|O_CREAT, 0600)) < 0) {
Dprintf(L_FATAL, "Unable to create %s: %s",
tempname, strerror(errno));
/* notreached */
}
sprintf(buffer, "%d", pid);
write(fd, buffer, strlen(buffer));
close(fd);
while (1) {
Dprintf(D_DEVTAB, "Trying to lock %s", PATH_DEVTAB);
if (link(tempname, PATH_DEVTAB_LOCK) >= 0) {
devtab_locked = 1;
break;
}
if (errno != EEXIST) {
Dprintf(L_ERROR, "Unable to lock %s: %s",
PATH_DEVTAB, strerror(errno));
break;
}
if ((fd = open(PATH_DEVTAB_LOCK, O_RDONLY)) < 0) {
Dprintf(L_ERROR, "Unable to open %s: %s",
PATH_DEVTAB_LOCK, strerror(errno));
break;
}
if ((n = read(fd, buffer, sizeof(buffer))) < 0) {
Dprintf(L_ERROR,
"unable to read pid from %s: %s",
PATH_DEVTAB_LOCK, strerror(errno));
break;
}
close(fd);
buffer[n] = '\0';
pid = atoi(buffer);
if (kill(pid, 0) < 0 && errno == ESRCH) {
Dprintf(L_WARNING,
"removing stale lock for %s by pid=%d",
PATH_DEVTAB, pid);
unlink(PATH_DEVTAB_LOCK);
continue;
}
if (retry == 1)
Dprintf(L_ERROR, "%s already locked by pid=%d.",
PATH_DEVTAB, pid);
if (retry >= maxretry) {
if ((maxretry <<= 1) >= 10 * 60)
maxretry = 10 * 60;
retry = 0;
}
retry++;
sleep(1);
}
unlink(tempname);
if (!devtab_locked)
Dprintf(L_FATAL, "Aborting.");
Dprintf(D_DEVTAB, "Successfully locked %s", PATH_DEVTAB);
}
static void
devtab_unlock(void)
{
if (devtab_locked) {
Dprintf(D_DEVTAB, "Unlocking %s", PATH_DEVTAB);
unlink(PATH_DEVTAB_LOCK);
}
devtab_locked = 0;
}
static FILE *
devtab_open(const char *mode, struct stat *sbp)
{
FILE *fp;
if ((fp = fopen(PATH_DEVTAB, mode)) == NULL) {
devtab_unlock();
Dprintf(L_FATAL,
"unable to open %s for %s: %s", PATH_DEVTAB,
(mode[0] == 'r')? "reading" : "writing",
strerror(errno));
}
if (sbp && fstat(fileno(fp), sbp) < 0) {
devtab_unlock();
Dprintf(L_FATAL,
"unable to stat %s: %s", PATH_DEVTAB,
strerror(errno));
}
return fp;
}
static unsigned int
devtab_add(dev_t dev)
{
Dprintf(D_DEVTAB, "Mapping dev 0x%x to index %u\n", dev, nrdevs);
/* Grow device table if needed */
if ((nrdevs % 8) == 0)
devtab = (dev_t *) xrealloc(devtab,
(nrdevs + 8) * sizeof(dev_t));
devtab[nrdevs++] = dev;
return nrdevs-1;
}
static void
devtab_read(void)
{
struct stat stb;
char buffer[1024], *sp;
FILE *fp;
if (devtab)
free(devtab);
devtab = NULL;
nrdevs = 0;
fp = devtab_open("r", &stb);
devtab_mtime = stb.st_mtime;
while (fgets(buffer, sizeof(buffer), fp) != NULL) {
/* shouldn't be there, but try to be nice to
* admins */
if (buffer[0] == '#')
continue;
if ((sp = strchr(buffer, '\n')) != NULL)
*sp = '\0';
if (!strncmp(buffer, "devnum-", 7)) {
stb.st_rdev = strtoul(buffer+7, &sp, 16);
if (*sp) {
Dprintf(L_FATAL,
"invalid device name %s in %s",
buffer, PATH_DEVTAB);
}
} else if (stat(buffer, &stb) < 0) {
Dprintf(L_FATAL,
"can't stat device name %s in %s: %s",
buffer, PATH_DEVTAB,
strerror(errno));
}
devtab_add(stb.st_rdev);
}
fclose(fp);
return;
}
/*
* Search all of /dev for a device file matching the given device number
*
* This routine works recursively in order to accomodate systems
* that keep their device names in something like /dev/dsk
*/
static const char *
devtab_find(const char *dirname, dev_t dev, unsigned int depth)
{
static char fullname[1024];
const char *result = NULL;
unsigned int dirlen;
struct dirent *dp;
struct stat stb;
DIR *dir;
/* Safeguard against infinite recursion (/dev may contain
* strange things like /dev/fd) */
if (depth > 4)
return NULL;
Dprintf(D_DEVTAB, "Looking for dev 0x%x in %s\n", dev, dirname);
if (dirname != fullname)
strcpy(fullname, dirname);
dirlen = strlen(fullname);
fullname[dirlen++] = '/';
fullname[dirlen] = '\0';
if ((dir = opendir(fullname)) == NULL) {
Dprintf(L_WARNING,
"can't open %s for reading: %s",
fullname, strerror(errno));
return NULL;
}
while (!result && (dp = readdir(dir)) != NULL) {
/* skip . and .. */
if (!strcmp(dp->d_name, "..") || !strcmp(dp->d_name, "."))
continue;
/* skip long file names */
if (strlen(dp->d_name) + dirlen + 1 >= sizeof(fullname))
continue;
strcpy(fullname + dirlen, dp->d_name);
if (lstat(fullname, &stb) < 0) {
Dprintf(L_WARNING,
"unable to stat %s: %s (huh?!)",
fullname, strerror(errno));
continue;
}
if (S_ISDIR(stb.st_mode)) {
result = devtab_find(fullname, dev, depth + 1);
} else if (S_ISBLK(stb.st_mode) && stb.st_rdev == dev) {
result = fullname;
}
}
closedir(dir);
return result;
}
static const char *
devtab_getname(dev_t dev)
{
static char fakename[64];
const char *result;
if ((result = devtab_find("/dev", dev, 0)) == NULL) {
/* Uh-oh... fake name */
sprintf(fakename, "devnum-0x%lx", (unsigned long) dev);
result = fakename;
}
return result;
}
#endif /* ENABLE_DEVTAB */
|