int main(void)
{
int p, s, r;
struct rusage rr;
long long tms = 0;
fprintf(stderr, "t6: checking millisecond time limits\n");
if ((linux_version = get_linux_version()) < 0)
return 1;
if (!linux_version) {
fprintf(stderr, "failed: not Linux or unknown linux version\n");
return 1;
}
fprintf(stderr, "t6: linux version %d\n", linux_version);
if (linux_version >= 3000000) {
fprintf(stderr, "t6: patch for linux >= 3.0.0 does not support millisecond time limits\n");
return 0;
}
if ((p = fork()) < 0) {
fprintf(stderr, "failed: unexpected fork() error: %s\n", strerror(errno));
return 1;
}
if (!p) do_son();
sleep(5);
r = wait4(-1, &s, WNOHANG, &rr);
if (!r) {
fprintf(stderr, "failed: child did not terminate in 5 seconds\n");
kill(p, SIGKILL);
wait(0);
return 1;
}
if (WIFEXITED(s) && WEXITSTATUS(s) == 111) {
fprintf(stderr, "failed: child failed to setup limitations\n");
return 1;
}
if (WIFEXITED(s)) {
fprintf(stderr, "info: child exited: %d\n", WEXITSTATUS(s));
} else if (WIFSIGNALED(s)) {
fprintf(stderr, "info: child signaled: %d\n", WTERMSIG(s));
}
tms = 0;
tms += rr.ru_utime.tv_sec * 1000;
tms += rr.ru_stime.tv_sec * 1000;
tms += rr.ru_utime.tv_usec / 1000;
tms += rr.ru_stime.tv_usec / 1000;
fprintf(stderr, "info: child user+sys ms: %lld\n", tms);
if (tms > 900) {
fprintf(stderr, "failed: millisecond timelimit does not work\n");
return 1;
}
fprintf(stderr, "ok\n");
return 0;
}
static int system_supports_ext4_ext2(void)
{
#ifdef __linux__
return get_linux_version() >= EXT4_SUPPORTS_EXT2;
#else
return 0;
#endif
}
/*
* Check if the kernel supports partitioned loop devices.
*
* Notes:
* - kernels < 3.2 support partitioned loop devices and PT scanning
* only if max_part= module paremeter is non-zero
*
* - kernels >= 3.2 always support partitioned loop devices
*
* - kernels >= 3.2 always support BLKPG_{ADD,DEL}_PARTITION ioctls
*
* - kernels >= 3.2 enable PT scanner only if max_part= is non-zero or if the
* LO_FLAGS_PARTSCAN flag is set for the device. The PT scanner is disabled
* by default.
*
* See kernel commit e03c8dd14915fabc101aa495828d58598dc5af98.
*/
int loopmod_supports_partscan(void)
{
int rc, ret = 0;
FILE *f;
if (get_linux_version() >= KERNEL_VERSION(3,2,0))
return 1;
f = fopen("/sys/module/loop/parameters/max_part", "r");
if (!f)
return 0;
rc = fscanf(f, "%d", &ret);
fclose(f);
return rc == 1 ? ret : 0;
}
int main(int argc, char **argv)
{
int p, s, d = 0;
snprintf(progname, sizeof(progname), "%s_helper", argv[0]);
fprintf(stderr, "t5: checking memory limit error for program size\n");
if ((linux_version = get_linux_version()) < 0)
return 1;
if (!linux_version) {
fprintf(stderr, "failed: not Linux or unknown linux version\n");
return 1;
}
fprintf(stderr, "t5: linux version %d\n", linux_version);
if ((p = fork()) < 0) {
fprintf(stderr, "failed: unexpected fork() error: %s\n", strerror(errno));
return 1;
}
if (!p) do_son();
wait(&s);
if ((s & 0x10000)) {
fprintf(stderr, "info: 0x10000 detected\n");
s &= 0xffff;
d = 1;
}
if (WIFEXITED(s) && WEXITSTATUS(s) == 111) {
fprintf(stderr, "failed: child failed to setup limitations\n");
return 1;
}
if (WIFEXITED(s)) {
fprintf(stderr, "info: child exited: %d\n", WEXITSTATUS(s));
} else if (WIFSIGNALED(s)) {
fprintf(stderr, "info: child signaled: %d\n", WTERMSIG(s));
}
if (!d) {
fprintf(stderr, "failed: memory limit error not detected\n");
return 1;
}
fprintf(stderr, "ok\n");
return 0;
}
int main(void)
{
int p, s, d = 0;
fprintf(stderr, "t3: checking memory limit error for stack\n");
if ((linux_version = get_linux_version()) < 0)
return 1;
if (!linux_version) {
fprintf(stderr, "failed: not Linux or unknown linux version\n");
return 1;
}
fprintf(stderr, "t3: linux version %d\n", linux_version);
if ((p = fork()) < 0) {
fprintf(stderr, "failed: unexpected fork() error: %s\n", strerror(errno));
return 1;
}
if (!p) do_son();
wait(&s);
if ((s & 0x10000)) {
fprintf(stderr, "\ninfo: 0x10000 detected\n");
s &= 0xffff;
d = 1;
}
if (WIFEXITED(s) && WEXITSTATUS(s) == 111) {
fprintf(stderr, "failed: child failed to setup limitations\n");
return 1;
}
if (WIFEXITED(s)) {
fprintf(stderr, "info: child exited: %d\n", WEXITSTATUS(s));
} else if (WIFSIGNALED(s)) {
fprintf(stderr, "info: child signaled: %d\n", WTERMSIG(s));
}
if (!d) {
fprintf(stderr, "failed: memory limit error not detected\n");
return 1;
}
fprintf(stderr, "ok\n");
return 0;
}
void
_initialize_arm_linux_nat (void)
{
struct target_ops *t;
os_version = get_linux_version (&os_major, &os_minor, &os_release);
/* Fill in the generic GNU/Linux methods. */
t = linux_target ();
/* Add our register access methods. */
t->to_fetch_registers = arm_linux_fetch_inferior_registers;
t->to_store_registers = arm_linux_store_inferior_registers;
t->to_read_description = arm_linux_read_description;
/* Register the target. */
linux_nat_add_target (t);
}
/*
* @lc: context
* @flags: LOOPDEV_FL_* flags
*
* Initilize loop handler.
*
* We have two sets of the flags:
*
* * LOOPDEV_FL_* flags control loopcxt_* API behavior
*
* * LO_FLAGS_* are kernel flags used for LOOP_{SET,GET}_STAT64 ioctls
*
* Note about LOOPDEV_FL_{RDONLY,RDWR} flags. These flags are used for open(2)
* syscall to open loop device. By default is the device open read-only.
*
* The expection is loopcxt_setup_device(), where the device is open read-write
* if LO_FLAGS_READ_ONLY flags is not set (see loopcxt_set_flags()).
*
* Returns: <0 on error, 0 on success.
*/
int loopcxt_init(struct loopdev_cxt *lc, int flags)
{
struct stat st;
if (!lc)
return -EINVAL;
memset(lc, 0, sizeof(*lc));
lc->flags = flags;
loopcxt_set_device(lc, NULL);
if (!(lc->flags & LOOPDEV_FL_NOSYSFS) &&
get_linux_version() >= KERNEL_VERSION(2,6,37))
/*
* Use only sysfs for basic information about loop devices
*/
lc->flags |= LOOPDEV_FL_NOIOCTL;
if (!(lc->flags & LOOPDEV_FL_CONTROL) && !stat(_PATH_DEV_LOOPCTL, &st))
lc->flags |= LOOPDEV_FL_CONTROL;
return 0;
}
int mnt_context_setup_loopdev(struct libmnt_context *cxt)
{
const char *backing_file, *optstr, *loopdev = NULL;
char *val = NULL, *loopval = NULL;
size_t len;
struct loopdev_cxt lc;
int rc = 0, lo_flags = 0;
uint64_t offset = 0, sizelimit = 0;
assert(cxt);
assert(cxt->fs);
assert((cxt->flags & MNT_FL_MOUNTFLAGS_MERGED));
backing_file = mnt_fs_get_srcpath(cxt->fs);
if (!backing_file)
return -EINVAL;
DBG(LOOP, ul_debugobj(cxt, "trying to setup device for %s", backing_file));
if (cxt->mountflags & MS_RDONLY) {
DBG(LOOP, ul_debugobj(cxt, "enabling READ-ONLY flag"));
lo_flags |= LO_FLAGS_READ_ONLY;
}
optstr = mnt_fs_get_user_options(cxt->fs);
/*
* loop=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_LOOP) &&
mnt_optstr_get_option(optstr, "loop", &val, &len) == 0 && val) {
loopval = strndup(val, len);
rc = loopval ? 0 : -ENOMEM;
}
/*
* offset=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_OFFSET) &&
mnt_optstr_get_option(optstr, "offset", &val, &len) == 0) {
rc = mnt_parse_offset(val, len, &offset);
if (rc) {
DBG(LOOP, ul_debugobj(cxt, "failed to parse offset="));
rc = -MNT_ERR_MOUNTOPT;
}
}
/*
* sizelimit=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_SIZELIMIT) &&
mnt_optstr_get_option(optstr, "sizelimit", &val, &len) == 0) {
rc = mnt_parse_offset(val, len, &sizelimit);
if (rc) {
DBG(LOOP, ul_debugobj(cxt, "failed to parse sizelimit="));
rc = -MNT_ERR_MOUNTOPT;
}
}
/*
* encryption=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_ENCRYPTION) &&
mnt_optstr_get_option(optstr, "encryption", &val, &len) == 0) {
DBG(LOOP, ul_debugobj(cxt, "encryption no longer supported"));
rc = -MNT_ERR_MOUNTOPT;
}
if (rc == 0 && is_mounted_same_loopfile(cxt,
mnt_context_get_target(cxt),
backing_file, offset))
rc = -EBUSY;
if (rc)
goto done_no_deinit;
/* It is possible to mount the same file more times. If we set more
* than one loop device referring to the same file, kernel has no
* mechanism to detect it. To prevent data corruption, the same loop
* device has to be recycled.
*/
if (backing_file) {
rc = loopcxt_init(&lc, 0);
if (rc)
goto done_no_deinit;
rc = loopcxt_find_overlap(&lc, backing_file, offset, sizelimit);
switch (rc) {
case 0: /* not found */
DBG(LOOP, ul_debugobj(cxt, "not found overlaping loopdev"));
loopcxt_deinit(&lc);
break;
case 1: /* overlap */
DBG(LOOP, ul_debugobj(cxt, "overlaping %s detected",
loopcxt_get_device(&lc)));
rc = -MNT_ERR_LOOPOVERLAP;
goto done;
case 2: /* overlap -- full size and offset match (reuse) */
{
uint32_t lc_encrypt_type;
DBG(LOOP, ul_debugobj(cxt, "re-using existing loop device %s",
loopcxt_get_device(&lc)));
/* Once a loop is initialized RO, there is no
* way to change its parameters. */
if (loopcxt_is_readonly(&lc)
&& !(lo_flags & LO_FLAGS_READ_ONLY)) {
DBG(LOOP, ul_debugobj(cxt, "%s is read-only",
loopcxt_get_device(&lc)));
rc = -EROFS;
goto done;
}
/* This is no more supported, but check to be safe. */
if (loopcxt_get_encrypt_type(&lc, &lc_encrypt_type) == 0
&& lc_encrypt_type != LO_CRYPT_NONE) {
DBG(LOOP, ul_debugobj(cxt, "encryption no longer supported for device %s",
loopcxt_get_device(&lc)));
rc = -MNT_ERR_LOOPOVERLAP;
goto done;
}
rc = 0;
goto success;
}
default: /* error */
goto done;
}
}
DBG(LOOP, ul_debugobj(cxt, "not found; create a new loop device"));
rc = loopcxt_init(&lc, 0);
if (rc)
goto done_no_deinit;
if (loopval) {
rc = loopcxt_set_device(&lc, loopval);
if (rc == 0)
loopdev = loopcxt_get_device(&lc);
}
if (rc)
goto done;
/* since 2.6.37 we don't have to store backing filename to mtab
* because kernel provides the name in /sys.
*/
if (get_linux_version() >= KERNEL_VERSION(2, 6, 37) ||
!mnt_context_mtab_writable(cxt)) {
DBG(LOOP, ul_debugobj(cxt, "enabling AUTOCLEAR flag"));
lo_flags |= LO_FLAGS_AUTOCLEAR;
}
do {
/* found free device */
if (!loopdev) {
rc = loopcxt_find_unused(&lc);
if (rc)
goto done;
DBG(LOOP, ul_debugobj(cxt, "trying to use %s",
loopcxt_get_device(&lc)));
}
/* set device attributes
* -- note that loopcxt_find_unused() resets "lc"
*/
rc = loopcxt_set_backing_file(&lc, backing_file);
if (!rc && offset)
rc = loopcxt_set_offset(&lc, offset);
if (!rc && sizelimit)
rc = loopcxt_set_sizelimit(&lc, sizelimit);
if (!rc)
loopcxt_set_flags(&lc, lo_flags);
if (rc) {
DBG(LOOP, ul_debugobj(cxt, "failed to set loop attributes"));
goto done;
}
/* setup the device */
rc = loopcxt_setup_device(&lc);
if (!rc)
break; /* success */
if (loopdev || rc != -EBUSY) {
DBG(LOOP, ul_debugobj(cxt, "failed to setup device"));
rc = -MNT_ERR_LOOPDEV;
goto done;
}
DBG(LOOP, ul_debugobj(cxt, "device stolen...trying again"));
} while (1);
success:
if (!rc)
rc = mnt_fs_set_source(cxt->fs, loopcxt_get_device(&lc));
if (!rc) {
/* success */
cxt->flags |= MNT_FL_LOOPDEV_READY;
if ((cxt->user_mountflags & MNT_MS_LOOP) &&
loopcxt_is_autoclear(&lc)) {
/*
* autoclear flag accepted by the kernel, don't store
* the "loop=" option to mtab.
*/
DBG(LOOP, ul_debugobj(cxt, "removing unnecessary loop= from mtab"));
cxt->user_mountflags &= ~MNT_MS_LOOP;
mnt_optstr_remove_option(&cxt->fs->user_optstr, "loop");
}
if (!(cxt->mountflags & MS_RDONLY) &&
loopcxt_is_readonly(&lc))
/*
* mount planned read-write, but loopdev is read-only,
* let's fix mount options...
*/
mnt_context_set_mflags(cxt, cxt->mountflags | MS_RDONLY);
/* we have to keep the device open until mount(1),
* otherwise it will be auto-cleared by kernel
*/
cxt->loopdev_fd = loopcxt_get_fd(&lc);
if (cxt->loopdev_fd < 0) {
DBG(LOOP, ul_debugobj(cxt, "failed to get loopdev FD"));
rc = -errno;
} else
loopcxt_set_fd(&lc, -1, 0);
}
done:
loopcxt_deinit(&lc);
done_no_deinit:
free(loopval);
return rc;
}
/* get size in bytes */
int
blkdev_get_size(int fd, unsigned long long *bytes)
{
#ifdef DKIOCGETBLOCKCOUNT
/* Apple Darwin */
if (ioctl(fd, DKIOCGETBLOCKCOUNT, bytes) >= 0) {
*bytes <<= 9;
return 0;
}
#endif
#ifdef BLKGETSIZE64
{
#ifdef __linux__
int ver = get_linux_version();
/* kernels 2.4.15-2.4.17, had a broken BLKGETSIZE64 */
if (ver >= KERNEL_VERSION (2,6,0) ||
(ver >= KERNEL_VERSION (2,4,18) && ver < KERNEL_VERSION (2,5,0)))
#endif
if (ioctl(fd, BLKGETSIZE64, bytes) >= 0)
return 0;
}
#endif /* BLKGETSIZE64 */
#ifdef BLKGETSIZE
{
unsigned long size;
if (ioctl(fd, BLKGETSIZE, &size) >= 0) {
*bytes = ((unsigned long long)size << 9);
return 0;
}
}
#endif /* BLKGETSIZE */
#ifdef DIOCGMEDIASIZE
/* FreeBSD */
if (ioctl(fd, DIOCGMEDIASIZE, bytes) >= 0)
return 0;
#endif
#ifdef FDGETPRM
{
struct floppy_struct this_floppy;
if (ioctl(fd, FDGETPRM, &this_floppy) >= 0) {
*bytes = ((unsigned long long) this_floppy.size) << 9;
return 0;
}
}
#endif /* FDGETPRM */
#ifdef HAVE_SYS_DISKLABEL_H
{
/*
* This code works for FreeBSD 4.11 i386, except for the full device
* (such as /dev/ad0). It doesn't work properly for newer FreeBSD
* though. FreeBSD >= 5.0 should be covered by the DIOCGMEDIASIZE
* above however.
*
* Note that FreeBSD >= 4.0 has disk devices as unbuffered (raw,
* character) devices, so we need to check for S_ISCHR, too.
*/
int part = -1;
struct disklabel lab;
struct partition *pp;
struct stat st;
if ((fstat(fd, &st) >= 0) &&
(S_ISBLK(st.st_mode) || S_ISCHR(st.st_mode)))
part = st.st_rdev & 7;
if (part >= 0 && (ioctl(fd, DIOCGDINFO, (char *)&lab) >= 0)) {
pp = &lab.d_partitions[part];
if (pp->p_size) {
*bytes = pp->p_size << 9;
return 0;
}
}
}
#endif /* HAVE_SYS_DISKLABEL_H */
{
struct stat st;
if (fstat(fd, &st) == 0 && S_ISREG(st.st_mode)) {
*bytes = st.st_size;
return 0;
}
if (!S_ISBLK(st.st_mode))
return -1;
}
*bytes = blkdev_find_size(fd);
return 0;
}
void
_initialize_arm_linux_nat (void)
{
os_version = get_linux_version (&os_major, &os_minor, &os_release);
}
int Manage_subdevs(char *devname, int fd,
struct mddev_dev *devlist, int verbose, int test,
char *update)
{
/* do something to each dev.
* devmode can be
* 'a' - add the device
* try HOT_ADD_DISK
* If that fails EINVAL, try ADD_NEW_DISK
* 'r' - remove the device HOT_REMOVE_DISK
* device can be 'faulty' or 'detached' in which case all
* matching devices are removed.
* 'f' - set the device faulty SET_DISK_FAULTY
* device can be 'detached' in which case any device that
* is inaccessible will be marked faulty.
* For 'f' and 'r', the device can also be a kernel-internal
* name such as 'sdb'.
*/
struct mddev_dev *add_devlist = NULL;
mdu_array_info_t array;
mdu_disk_info_t disc;
unsigned long long array_size;
struct mddev_dev *dv, *next = NULL;
struct stat stb;
int j, jnext = 0;
int tfd = -1;
struct supertype *st, *tst;
char *subarray = NULL;
int duuid[4];
int ouuid[4];
int lfd = -1;
int sysfd = -1;
int count = 0; /* number of actions taken */
if (ioctl(fd, GET_ARRAY_INFO, &array)) {
fprintf(stderr, Name ": cannot get array info for %s\n",
devname);
return 1;
}
/* array.size is only 32 bit and may be truncated.
* So read from sysfs if possible, and record number of sectors
*/
array_size = get_component_size(fd);
if (array_size <= 0)
array_size = array.size * 2;
tst = super_by_fd(fd, &subarray);
if (!tst) {
fprintf(stderr, Name ": unsupport array - version %d.%d\n",
array.major_version, array.minor_version);
return 1;
}
stb.st_rdev = 0;
for (dv = devlist, j=0 ; dv; dv = next, j = jnext) {
unsigned long long ldsize;
char dvname[20];
char *dnprintable = dv->devname;
char *add_dev = dv->devname;
int err;
int re_add_failed = 0;
next = dv->next;
jnext = 0;
if (strcmp(dv->devname, "failed")==0 ||
strcmp(dv->devname, "faulty")==0) {
int remaining_disks = array.nr_disks;
if (dv->disposition != 'r') {
fprintf(stderr, Name ": %s only meaningful "
"with -r, not -%c\n",
dv->devname, dv->disposition);
return 1;
}
for (; j < 1024 && remaining_disks > 0; j++) {
unsigned dev;
disc.number = j;
if (ioctl(fd, GET_DISK_INFO, &disc))
continue;
if (disc.major == 0 && disc.minor == 0)
continue;
remaining_disks --;
if ((disc.state & 1) == 0) /* faulty */
continue;
dev = makedev(disc.major, disc.minor);
if (stb.st_rdev == dev)
/* already did that one */
continue;
stb.st_rdev = dev;
next = dv;
/* same slot again next time - things might
* have reshuffled */
jnext = j;
sprintf(dvname,"%d:%d", disc.major, disc.minor);
dnprintable = dvname;
break;
}
if (next != dv)
continue;
} else if (strcmp(dv->devname, "detached") == 0) {
int remaining_disks = array.nr_disks;
if (dv->disposition != 'r' && dv->disposition != 'f') {
fprintf(stderr, Name ": %s only meaningful "
"with -r of -f, not -%c\n",
dv->devname, dv->disposition);
return 1;
}
for (; j < 1024 && remaining_disks > 0; j++) {
int sfd;
unsigned dev;
disc.number = j;
if (ioctl(fd, GET_DISK_INFO, &disc))
continue;
if (disc.major == 0 && disc.minor == 0)
continue;
remaining_disks --;
sprintf(dvname,"%d:%d", disc.major, disc.minor);
sfd = dev_open(dvname, O_RDONLY);
if (sfd >= 0) {
close(sfd);
continue;
}
if (dv->disposition == 'f' &&
(disc.state & 1) == 1) /* already faulty */
continue;
if (errno != ENXIO)
continue;
dev = makedev(disc.major, disc.minor);
if (stb.st_rdev == dev)
/* already did that one */
continue;
stb.st_rdev = dev;
next = dv;
/* same slot again next time - things might
* have reshuffled */
jnext = j;
dnprintable = dvname;
break;
}
if (next != dv)
continue;
} else if (strcmp(dv->devname, "missing") == 0) {
if (dv->disposition != 'a' || dv->re_add == 0) {
fprintf(stderr, Name ": 'missing' only meaningful "
"with --re-add\n");
return 1;
}
if (add_devlist == NULL)
add_devlist = conf_get_devs();
if (add_devlist == NULL) {
fprintf(stderr, Name ": no devices to scan for missing members.");
continue;
}
add_dev = add_devlist->devname;
add_devlist = add_devlist->next;
if (add_devlist != NULL)
next = dv;
if (stat(add_dev, &stb) < 0)
continue;
} else if (strchr(dv->devname, '/') == NULL &&
strchr(dv->devname, ':') == NULL &&
strlen(dv->devname) < 50) {
/* Assume this is a kernel-internal name like 'sda1' */
int found = 0;
char dname[55];
if (dv->disposition != 'r' && dv->disposition != 'f') {
fprintf(stderr, Name ": %s only meaningful "
"with -r or -f, not -%c\n",
dv->devname, dv->disposition);
return 1;
}
sprintf(dname, "dev-%s", dv->devname);
sysfd = sysfs_open(fd2devnum(fd), dname, "block/dev");
if (sysfd >= 0) {
char dn[20];
int mj,mn;
if (sysfs_fd_get_str(sysfd, dn, 20) > 0 &&
sscanf(dn, "%d:%d", &mj,&mn) == 2) {
stb.st_rdev = makedev(mj,mn);
found = 1;
}
close(sysfd);
sysfd = -1;
}
if (!found) {
sysfd = sysfs_open(fd2devnum(fd), dname, "state");
if (sysfd < 0) {
fprintf(stderr, Name ": %s does not appear "
"to be a component of %s\n",
dv->devname, devname);
return 1;
}
}
} else {
j = 0;
tfd = dev_open(dv->devname, O_RDONLY);
if (tfd < 0 && dv->disposition == 'r' &&
lstat(dv->devname, &stb) == 0)
/* Be happy, the lstat worked, that is
* enough for --remove
*/
;
else {
if (tfd < 0 || fstat(tfd, &stb) != 0) {
fprintf(stderr, Name ": cannot find %s: %s\n",
dv->devname, strerror(errno));
if (tfd >= 0)
close(tfd);
return 1;
}
close(tfd);
tfd = -1;
}
if ((stb.st_mode & S_IFMT) != S_IFBLK) {
fprintf(stderr, Name ": %s is not a "
"block device.\n",
dv->devname);
return 1;
}
}
switch(dv->disposition){
default:
fprintf(stderr, Name ": internal error - devmode[%s]=%d\n",
dv->devname, dv->disposition);
return 1;
case 'a':
/* add the device */
if (subarray) {
fprintf(stderr, Name ": Cannot add disks to a"
" \'member\' array, perform this"
" operation on the parent container\n");
return 1;
}
/* Make sure it isn't in use (in 2.6 or later) */
tfd = dev_open(add_dev, O_RDONLY|O_EXCL|O_DIRECT);
if (tfd < 0 && add_dev != dv->devname)
continue;
if (tfd < 0) {
fprintf(stderr, Name ": Cannot open %s: %s\n",
dv->devname, strerror(errno));
return 1;
}
st = dup_super(tst);
if (array.not_persistent==0)
st->ss->load_super(st, tfd, NULL);
if (add_dev == dv->devname) {
if (!get_dev_size(tfd, dv->devname, &ldsize)) {
close(tfd);
return 1;
}
} else if (!get_dev_size(tfd, NULL, &ldsize)) {
close(tfd);
tfd = -1;
continue;
}
if (!tst->ss->external &&
array.major_version == 0 &&
md_get_version(fd)%100 < 2) {
close(tfd);
tfd = -1;
if (ioctl(fd, HOT_ADD_DISK,
(unsigned long)stb.st_rdev)==0) {
if (verbose >= 0)
fprintf(stderr, Name ": hot added %s\n",
add_dev);
continue;
}
fprintf(stderr, Name ": hot add failed for %s: %s\n",
add_dev, strerror(errno));
return 1;
}
if (array.not_persistent == 0 || tst->ss->external) {
/* need to find a sample superblock to copy, and
* a spare slot to use.
* For 'external' array (well, container based),
* We can just load the metadata for the array.
*/
if (tst->sb)
/* already loaded */;
else if (tst->ss->external) {
tst->ss->load_container(tst, fd, NULL);
} else for (j = 0; j < tst->max_devs; j++) {
char *dev;
int dfd;
disc.number = j;
if (ioctl(fd, GET_DISK_INFO, &disc))
continue;
if (disc.major==0 && disc.minor==0)
continue;
if ((disc.state & 4)==0) continue; /* sync */
/* Looks like a good device to try */
dev = map_dev(disc.major, disc.minor, 1);
if (!dev) continue;
dfd = dev_open(dev, O_RDONLY);
if (dfd < 0) continue;
if (tst->ss->load_super(tst, dfd,
NULL)) {
close(dfd);
continue;
}
close(dfd);
break;
}
/* FIXME this is a bad test to be using */
if (!tst->sb) {
close(tfd);
fprintf(stderr, Name ": cannot load array metadata from %s\n", devname);
return 1;
}
/* Make sure device is large enough */
if (tst->ss->avail_size(tst, ldsize/512) <
array_size) {
close(tfd);
tfd = -1;
if (add_dev != dv->devname)
continue;
fprintf(stderr, Name ": %s not large enough to join array\n",
dv->devname);
return 1;
}
/* Possibly this device was recently part of the array
* and was temporarily removed, and is now being re-added.
* If so, we can simply re-add it.
*/
tst->ss->uuid_from_super(tst, duuid);
if (st->sb) {
struct mdinfo mdi;
st->ss->getinfo_super(st, &mdi, NULL);
st->ss->uuid_from_super(st, ouuid);
if ((mdi.disk.state & (1<<MD_DISK_ACTIVE)) &&
!(mdi.disk.state & (1<<MD_DISK_FAULTY)) &&
memcmp(duuid, ouuid, sizeof(ouuid))==0) {
/* look like it is worth a try. Need to
* make sure kernel will accept it though.
*/
/* re-add doesn't work for version-1 superblocks
* before 2.6.18 :-(
*/
if (array.major_version == 1 &&
get_linux_version() <= 2006018)
goto skip_re_add;
disc.number = mdi.disk.number;
if (ioctl(fd, GET_DISK_INFO, &disc) != 0
|| disc.major != 0 || disc.minor != 0
|| !enough_fd(fd))
goto skip_re_add;
disc.major = major(stb.st_rdev);
disc.minor = minor(stb.st_rdev);
disc.number = mdi.disk.number;
disc.raid_disk = mdi.disk.raid_disk;
disc.state = mdi.disk.state;
if (dv->writemostly == 1)
disc.state |= 1 << MD_DISK_WRITEMOSTLY;
if (dv->writemostly == 2)
disc.state &= ~(1 << MD_DISK_WRITEMOSTLY);
remove_partitions(tfd);
close(tfd);
tfd = -1;
if (update) {
int rv = -1;
tfd = dev_open(dv->devname, O_RDWR);
if (tfd >= 0)
rv = st->ss->update_super(
st, NULL, update,
devname, verbose, 0, NULL);
if (rv == 0)
rv = st->ss->store_super(st, tfd);
close(tfd);
tfd = -1;
if (rv != 0) {
fprintf(stderr, Name ": failed to update"
" superblock during re-add\n");
return 1;
}
}
/* don't even try if disk is marked as faulty */
errno = 0;
if (ioctl(fd, ADD_NEW_DISK, &disc) == 0) {
if (verbose >= 0)
fprintf(stderr, Name ": re-added %s\n", add_dev);
count++;
continue;
}
if (errno == ENOMEM || errno == EROFS) {
fprintf(stderr, Name ": add new device failed for %s: %s\n",
add_dev, strerror(errno));
if (add_dev != dv->devname)
continue;
return 1;
}
skip_re_add:
re_add_failed = 1;
}
st->ss->free_super(st);
}
if (add_dev != dv->devname) {
if (verbose > 0)
fprintf(stderr, Name
": --re-add for %s to %s is not possible\n",
add_dev, devname);
if (tfd >= 0) {
close(tfd);
tfd = -1;
}
continue;
}
if (dv->re_add) {
if (tfd >= 0)
close(tfd);
fprintf(stderr, Name
": --re-add for %s to %s is not possible\n",
dv->devname, devname);
return 1;
}
if (re_add_failed) {
fprintf(stderr, Name ": %s reports being an active member for %s, but a --re-add fails.\n",
dv->devname, devname);
fprintf(stderr, Name ": not performing --add as that would convert %s in to a spare.\n",
dv->devname);
fprintf(stderr, Name ": To make this a spare, use \"mdadm --zero-superblock %s\" first.\n",
dv->devname);
if (tfd >= 0)
close(tfd);
return 1;
}
} else {
/* non-persistent. Must ensure that new drive
* is at least array.size big.
*/
if (ldsize/512 < array_size) {
fprintf(stderr, Name ": %s not large enough to join array\n",
dv->devname);
if (tfd >= 0)
close(tfd);
return 1;
}
}
/* committed to really trying this device now*/
if (tfd >= 0) {
remove_partitions(tfd);
close(tfd);
tfd = -1;
}
/* in 2.6.17 and earlier, version-1 superblocks won't
* use the number we write, but will choose a free number.
* we must choose the same free number, which requires
* starting at 'raid_disks' and counting up
*/
for (j = array.raid_disks; j< tst->max_devs; j++) {
disc.number = j;
if (ioctl(fd, GET_DISK_INFO, &disc))
break;
if (disc.major==0 && disc.minor==0)
break;
if (disc.state & 8) /* removed */
break;
}
disc.major = major(stb.st_rdev);
disc.minor = minor(stb.st_rdev);
disc.number =j;
disc.state = 0;
if (array.not_persistent==0) {
int dfd;
if (dv->writemostly == 1)
disc.state |= 1 << MD_DISK_WRITEMOSTLY;
dfd = dev_open(dv->devname, O_RDWR | O_EXCL|O_DIRECT);
if (tst->ss->add_to_super(tst, &disc, dfd,
dv->devname)) {
close(dfd);
return 1;
}
if (tst->ss->write_init_super(tst)) {
close(dfd);
return 1;
}
} else if (dv->re_add) {
/* this had better be raid1.
* As we are "--re-add"ing we must find a spare slot
* to fill.
*/
char *used = malloc(array.raid_disks);
memset(used, 0, array.raid_disks);
for (j=0; j< tst->max_devs; j++) {
mdu_disk_info_t disc2;
disc2.number = j;
if (ioctl(fd, GET_DISK_INFO, &disc2))
continue;
if (disc2.major==0 && disc2.minor==0)
continue;
if (disc2.state & 8) /* removed */
continue;
if (disc2.raid_disk < 0)
continue;
if (disc2.raid_disk > array.raid_disks)
continue;
used[disc2.raid_disk] = 1;
}
for (j=0 ; j<array.raid_disks; j++)
if (!used[j]) {
disc.raid_disk = j;
disc.state |= (1<<MD_DISK_SYNC);
break;
}
free(used);
}
if (dv->writemostly == 1)
disc.state |= (1 << MD_DISK_WRITEMOSTLY);
if (tst->ss->external) {
/* add a disk
* to an external metadata container */
struct mdinfo new_mdi;
struct mdinfo *sra;
int container_fd;
int devnum = fd2devnum(fd);
int dfd;
container_fd = open_dev_excl(devnum);
if (container_fd < 0) {
fprintf(stderr, Name ": add failed for %s:"
" could not get exclusive access to container\n",
dv->devname);
tst->ss->free_super(tst);
return 1;
}
dfd = dev_open(dv->devname, O_RDWR | O_EXCL|O_DIRECT);
if (mdmon_running(tst->container_dev))
tst->update_tail = &tst->updates;
if (tst->ss->add_to_super(tst, &disc, dfd,
dv->devname)) {
close(dfd);
close(container_fd);
return 1;
}
if (tst->update_tail)
flush_metadata_updates(tst);
else
tst->ss->sync_metadata(tst);
sra = sysfs_read(container_fd, -1, 0);
if (!sra) {
fprintf(stderr, Name ": add failed for %s: sysfs_read failed\n",
dv->devname);
close(container_fd);
tst->ss->free_super(tst);
return 1;
}
sra->array.level = LEVEL_CONTAINER;
/* Need to set data_offset and component_size */
tst->ss->getinfo_super(tst, &new_mdi, NULL);
new_mdi.disk.major = disc.major;
new_mdi.disk.minor = disc.minor;
new_mdi.recovery_start = 0;
/* Make sure fds are closed as they are O_EXCL which
* would block add_disk */
tst->ss->free_super(tst);
if (sysfs_add_disk(sra, &new_mdi, 0) != 0) {
fprintf(stderr, Name ": add new device to external metadata"
" failed for %s\n", dv->devname);
close(container_fd);
sysfs_free(sra);
return 1;
}
ping_monitor_by_id(devnum);
sysfs_free(sra);
close(container_fd);
} else {
tst->ss->free_super(tst);
if (ioctl(fd, ADD_NEW_DISK, &disc)) {
fprintf(stderr, Name ": add new device failed for %s as %d: %s\n",
dv->devname, j, strerror(errno));
return 1;
}
}
if (verbose >= 0)
fprintf(stderr, Name ": added %s\n", dv->devname);
break;
case 'r':
/* hot remove */
if (subarray) {
fprintf(stderr, Name ": Cannot remove disks from a"
" \'member\' array, perform this"
" operation on the parent container\n");
if (sysfd >= 0)
close(sysfd);
return 1;
}
if (tst->ss->external) {
/* To remove a device from a container, we must
* check that it isn't in use in an array.
* This involves looking in the 'holders'
* directory - there must be just one entry,
* the container.
* To ensure that it doesn't get used as a
* hold spare while we are checking, we
* get an O_EXCL open on the container
*/
int dnum = fd2devnum(fd);
lfd = open_dev_excl(dnum);
if (lfd < 0) {
fprintf(stderr, Name
": Cannot get exclusive access "
" to container - odd\n");
if (sysfd >= 0)
close(sysfd);
return 1;
}
/* in the detached case it is not possible to
* check if we are the unique holder, so just
* rely on the 'detached' checks
*/
if (strcmp(dv->devname, "detached") == 0 ||
sysfd >= 0 ||
sysfs_unique_holder(dnum, stb.st_rdev))
/* pass */;
else {
fprintf(stderr, Name
": %s is %s, cannot remove.\n",
dnprintable,
errno == EEXIST ? "still in use":
"not a member");
close(lfd);
return 1;
}
}
/* FIXME check that it is a current member */
if (sysfd >= 0) {
/* device has been removed and we don't know
* the major:minor number
*/
int n = write(sysfd, "remove", 6);
if (n != 6)
err = -1;
else
err = 0;
close(sysfd);
sysfd = -1;
} else {
err = ioctl(fd, HOT_REMOVE_DISK, (unsigned long)stb.st_rdev);
if (err && errno == ENODEV) {
/* Old kernels rejected this if no personality
* registered */
struct mdinfo *sra = sysfs_read(fd, 0, GET_DEVS);
struct mdinfo *dv = NULL;
if (sra)
dv = sra->devs;
for ( ; dv ; dv=dv->next)
if (dv->disk.major == (int)major(stb.st_rdev) &&
dv->disk.minor == (int)minor(stb.st_rdev))
break;
if (dv)
err = sysfs_set_str(sra, dv,
"state", "remove");
else
err = -1;
if (sra)
sysfs_free(sra);
}
}
if (err) {
fprintf(stderr, Name ": hot remove failed "
"for %s: %s\n", dnprintable,
strerror(errno));
if (lfd >= 0)
close(lfd);
return 1;
}
if (tst->ss->external) {
/*
* Before dropping our exclusive open we make an
* attempt at preventing mdmon from seeing an
* 'add' event before reconciling this 'remove'
* event.
*/
char *name = devnum2devname(fd2devnum(fd));
if (!name) {
fprintf(stderr, Name ": unable to get container name\n");
return 1;
}
ping_manager(name);
free(name);
}
if (lfd >= 0)
close(lfd);
count++;
if (verbose >= 0)
fprintf(stderr, Name ": hot removed %s from %s\n",
dnprintable, devname);
break;
case 'f': /* set faulty */
/* FIXME check current member */
if ((sysfd >= 0 && write(sysfd, "faulty", 6) != 6) ||
(sysfd < 0 && ioctl(fd, SET_DISK_FAULTY,
(unsigned long) stb.st_rdev))) {
fprintf(stderr, Name ": set device faulty failed for %s: %s\n",
dnprintable, strerror(errno));
if (sysfd >= 0)
close(sysfd);
return 1;
}
if (sysfd >= 0)
close(sysfd);
sysfd = -1;
count++;
if (verbose >= 0)
fprintf(stderr, Name ": set %s faulty in %s\n",
dnprintable, devname);
break;
}
}
if (test && count == 0)
return 2;
return 0;
}
int mnt_context_setup_loopdev(struct libmnt_context *cxt)
{
const char *backing_file, *optstr, *loopdev = NULL;
char *val = NULL;
size_t len;
struct loopdev_cxt lc;
int rc = 0, lo_flags = 0;
uint64_t offset = 0, sizelimit = 0;
assert(cxt->fs);
assert((cxt->flags & MNT_FL_MOUNTFLAGS_MERGED));
if (!cxt)
return -EINVAL;
backing_file = mnt_fs_get_srcpath(cxt->fs);
if (!backing_file)
return -EINVAL;
DBG(CXT, ul_debugobj(cxt, "trying to setup loopdev for %s", backing_file));
if (cxt->mountflags & MS_RDONLY) {
DBG(CXT, ul_debugobj(cxt, "enabling READ-ONLY flag"));
lo_flags |= LO_FLAGS_READ_ONLY;
}
rc = loopcxt_init(&lc, 0);
if (rc)
return rc;
ON_DBG(CXT, loopcxt_enable_debug(&lc, 1));
optstr = mnt_fs_get_user_options(cxt->fs);
/*
* loop=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_LOOP) &&
mnt_optstr_get_option(optstr, "loop", &val, &len) == 0 && val) {
val = strndup(val, len);
rc = val ? loopcxt_set_device(&lc, val) : -ENOMEM;
free(val);
if (rc == 0)
loopdev = loopcxt_get_device(&lc);
}
/*
* offset=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_OFFSET) &&
mnt_optstr_get_option(optstr, "offset", &val, &len) == 0) {
rc = mnt_parse_offset(val, len, &offset);
if (rc) {
DBG(CXT, ul_debugobj(cxt, "failed to parse offset="));
rc = -MNT_ERR_MOUNTOPT;
}
}
/*
* sizelimit=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_SIZELIMIT) &&
mnt_optstr_get_option(optstr, "sizelimit", &val, &len) == 0) {
rc = mnt_parse_offset(val, len, &sizelimit);
if (rc) {
DBG(CXT, ul_debugobj(cxt, "failed to parse sizelimit="));
rc = -MNT_ERR_MOUNTOPT;
}
}
/*
* encryption=
*/
if (rc == 0 && (cxt->user_mountflags & MNT_MS_ENCRYPTION) &&
mnt_optstr_get_option(optstr, "encryption", &val, &len) == 0) {
DBG(CXT, ul_debugobj(cxt, "encryption no longer supported"));
rc = -MNT_ERR_MOUNTOPT;
}
if (rc == 0 && is_mounted_same_loopfile(cxt,
mnt_context_get_target(cxt),
backing_file, offset))
rc = -EBUSY;
if (rc)
goto done;
/* since 2.6.37 we don't have to store backing filename to mtab
* because kernel provides the name in /sys.
*/
if (get_linux_version() >= KERNEL_VERSION(2, 6, 37) ||
!mnt_context_mtab_writable(cxt)) {
DBG(CXT, ul_debugobj(cxt, "enabling AUTOCLEAR flag"));
lo_flags |= LO_FLAGS_AUTOCLEAR;
}
do {
/* found free device */
if (!loopdev) {
rc = loopcxt_find_unused(&lc);
if (rc)
goto done;
DBG(CXT, ul_debugobj(cxt, "trying to use %s",
loopcxt_get_device(&lc)));
}
/* set device attributes
* -- note that loopcxt_find_unused() resets "lc"
*/
rc = loopcxt_set_backing_file(&lc, backing_file);
if (!rc && offset)
rc = loopcxt_set_offset(&lc, offset);
if (!rc && sizelimit)
rc = loopcxt_set_sizelimit(&lc, sizelimit);
if (!rc)
loopcxt_set_flags(&lc, lo_flags);
if (rc) {
DBG(CXT, ul_debugobj(cxt, "failed to set loopdev attributes"));
goto done;
}
/* setup the device */
rc = loopcxt_setup_device(&lc);
if (!rc)
break; /* success */
if (loopdev || rc != -EBUSY) {
DBG(CXT, ul_debugobj(cxt, "failed to setup device"));
rc = -MNT_ERR_LOOPDEV;
goto done;
}
DBG(CXT, ul_debugobj(cxt, "loopdev stolen...trying again"));
} while (1);
if (!rc)
rc = mnt_fs_set_source(cxt->fs, loopcxt_get_device(&lc));
if (!rc) {
/* success */
cxt->flags |= MNT_FL_LOOPDEV_READY;
if ((cxt->user_mountflags & MNT_MS_LOOP) &&
loopcxt_is_autoclear(&lc)) {
/*
* autoclear flag accepted by the kernel, don't store
* the "loop=" option to mtab.
*/
cxt->user_mountflags &= ~MNT_MS_LOOP;
mnt_optstr_remove_option(&cxt->fs->user_optstr, "loop");
}
if (!(cxt->mountflags & MS_RDONLY) &&
loopcxt_is_readonly(&lc))
/*
* mount planned read-write, but loopdev is read-only,
* let's fix mount options...
*/
mnt_context_set_mflags(cxt, cxt->mountflags | MS_RDONLY);
/* we have to keep the device open until mount(1),
* otherwise it will be auto-cleared by kernel
*/
cxt->loopdev_fd = loopcxt_get_fd(&lc);
loopcxt_set_fd(&lc, -1, 0);
}
done:
loopcxt_deinit(&lc);
return rc;
}
int Assemble(struct supertype *st, char *mddev, int mdfd,
mddev_ident_t ident, char *conffile,
mddev_dev_t devlist,
int readonly, int runstop,
char *update,
int verbose, int force)
{
/*
* The task of Assemble is to find a collection of
* devices that should (according to their superblocks)
* form an array, and to give this collection to the MD driver.
* In Linux-2.4 and later, this involves submitting a
* SET_ARRAY_INFO ioctl with no arg - to prepare
* the array - and then submit a number of
* ADD_NEW_DISK ioctls to add disks into
* the array. Finally RUN_ARRAY might
* be submitted to start the array.
*
* Much of the work of Assemble is in finding and/or
* checking the disks to make sure they look right.
*
* If mddev is not set, then scan must be set and we
* read through the config file for dev+uuid mapping
* We recurse, setting mddev, for each device that
* - isn't running
* - has a valid uuid (or any uuid if !uuidset)
*
* If mddev is set, we try to determine state of md.
* check version - must be at least 0.90.0
* check kernel version. must be at least 2.4.
* If not, we can possibly fall back on START_ARRAY
* Try to GET_ARRAY_INFO.
* If possible, give up
* If not, try to STOP_ARRAY just to make sure
*
* If !uuidset and scan, look in conf-file for uuid
* If not found, give up
* If !devlist and scan and uuidset, get list of devs from conf-file
*
* For each device:
* Check superblock - discard if bad
* Check uuid (set if we don't have one) - discard if no match
* Check superblock similarity if we have a superblock - discard if different
* Record events, devicenum
* This should give us a list of devices for the array
* We should collect the most recent event number
*
* Count disks with recent enough event count
* While force && !enough disks
* Choose newest rejected disks, update event count
* mark clean and rewrite superblock
* If recent kernel:
* SET_ARRAY_INFO
* foreach device with recent events : ADD_NEW_DISK
* if runstop == 1 || "enough" disks and runstop==0 -> RUN_ARRAY
* If old kernel:
* Check the device numbers in superblock are right
* update superblock if any changes
* START_ARRAY
*
*/
int old_linux = 0;
int vers;
void *first_super = NULL, *super = NULL;
struct {
char *devname;
unsigned int major, minor;
unsigned int oldmajor, oldminor;
long long events;
int uptodate;
int state;
int raid_disk;
int disk_nr;
} *devices;
int *best = NULL; /* indexed by raid_disk */
unsigned int bestcnt = 0;
int devcnt = 0;
unsigned int okcnt, sparecnt;
unsigned int req_cnt;
unsigned int i;
int most_recent = 0;
int chosen_drive;
int change = 0;
int inargv = 0;
int start_partial_ok = force || devlist==NULL;
unsigned int num_devs;
mddev_dev_t tmpdev;
struct mdinfo info;
struct mddev_ident_s ident2;
char *avail;
int nextspare = 0;
vers = md_get_version(mdfd);
if (vers <= 0) {
fprintf(stderr, Name ": %s appears not to be an md device.\n", mddev);
return 1;
}
if (vers < 9000) {
fprintf(stderr, Name ": Assemble requires driver version 0.90.0 or later.\n"
" Upgrade your kernel or try --build\n");
return 1;
}
if (get_linux_version() < 2004000)
old_linux = 1;
if (ioctl(mdfd, GET_ARRAY_INFO, &info.array)>=0) {
fprintf(stderr, Name ": device %s already active - cannot assemble it\n",
mddev);
return 1;
}
ioctl(mdfd, STOP_ARRAY, NULL); /* just incase it was started but has no content */
/*
* If any subdevs are listed, then any that don't
* match ident are discarded. Remainder must all match and
* become the array.
* If no subdevs, then we scan all devices in the config file, but
* there must be something in the identity
*/
if (!devlist &&
ident->uuid_set == 0 &&
ident->super_minor < 0 &&
ident->devices == NULL) {
fprintf(stderr, Name ": No identity information available for %s - cannot assemble.\n",
mddev);
return 1;
}
if (devlist == NULL)
devlist = conf_get_devs(conffile);
else inargv = 1;
tmpdev = devlist; num_devs = 0;
while (tmpdev) {
num_devs++;
tmpdev = tmpdev->next;
}
devices = malloc(num_devs * sizeof(*devices));
if (!st && ident->st) st = ident->st;
if (verbose>0)
fprintf(stderr, Name ": looking for devices for %s\n",
mddev);
while ( devlist) {
char *devname;
int dfd;
struct stat stb;
struct supertype *tst = st;
devname = devlist->devname;
devlist = devlist->next;
if (ident->devices &&
!match_oneof(ident->devices, devname)) {
if ((inargv && verbose>=0) || verbose > 0)
fprintf(stderr, Name ": %s is not one of %s\n", devname, ident->devices);
continue;
}
if (super) {
free(super);
super = NULL;
}
dfd = dev_open(devname, O_RDONLY|O_EXCL);
if (dfd < 0) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": cannot open device %s: %s\n",
devname, strerror(errno));
} else if (fstat(dfd, &stb)< 0) {
/* Impossible! */
fprintf(stderr, Name ": fstat failed for %s: %s\n",
devname, strerror(errno));
} else if ((stb.st_mode & S_IFMT) != S_IFBLK) {
fprintf(stderr, Name ": %s is not a block device.\n",
devname);
} else if (!tst && (tst = guess_super(dfd)) == NULL) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": no recogniseable superblock\n");
} else if (tst->ss->load_super(tst,dfd, &super, NULL)) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf( stderr, Name ": no RAID superblock on %s\n",
devname);
} else {
tst->ss->getinfo_super(&info, &ident2, super);
}
if (dfd >= 0) close(dfd);
if (ident->uuid_set && (!update || strcmp(update, "uuid")!= 0) &&
(!super || same_uuid(info.uuid, ident->uuid, tst->ss->swapuuid)==0)) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": %s has wrong uuid.\n",
devname);
continue;
}
if (ident->name[0] &&
(!super || strncmp(ident2.name, ident->name, 32)!=0)) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": %s has wrong name.\n",
devname);
continue;
}
if (ident->super_minor != UnSet &&
(!super || ident->super_minor != info.array.md_minor)) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": %s has wrong super-minor.\n",
devname);
continue;
}
if (ident->level != UnSet &&
(!super|| ident->level != info.array.level)) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": %s has wrong raid level.\n",
devname);
continue;
}
if (ident->raid_disks != UnSet &&
(!super || ident->raid_disks!= info.array.raid_disks)) {
if ((inargv && verbose >= 0) || verbose > 0)
fprintf(stderr, Name ": %s requires wrong number of drives.\n",
devname);
continue;
}
/* If we are this far, then we are commited to this device.
* If the super_block doesn't exist, or doesn't match others,
* then we cannot continue
*/
if (!super) {
fprintf(stderr, Name ": %s has no superblock - assembly aborted\n",
devname);
free(first_super);
return 1;
}
st = tst; /* commit to this format, if haven't already */
if (st->ss->compare_super(&first_super, super)) {
fprintf(stderr, Name ": superblock on %s doesn't match others - assembly aborted\n",
devname);
free(super);
free(first_super);
return 1;
}
/* looks like a good enough match to update the super block if needed */
if (update) {
/* prepare useful information in info structures */
struct stat stb2;
fstat(mdfd, &stb2);
info.array.md_minor = minor(stb2.st_rdev);
if (strcmp(update, "uuid")==0 &&
!ident->uuid_set) {
int rfd;
if ((rfd = open("/dev/urandom", O_RDONLY)) < 0 ||
read(rfd, ident->uuid, 16) != 16) {
*(__u32*)(ident->uuid) = random();
*(__u32*)(ident->uuid+1) = random();
*(__u32*)(ident->uuid+2) = random();
*(__u32*)(ident->uuid+3) = random();
}
if (rfd >= 0) close(rfd);
ident->uuid_set = 1;
}
memcpy(info.uuid, ident->uuid, 16);
st->ss->update_super(&info, super, update, devname, verbose);
dfd = dev_open(devname, O_RDWR|O_EXCL);
if (dfd < 0)
fprintf(stderr, Name ": Cannot open %s for superblock update\n",
devname);
else if (st->ss->store_super(st, dfd, super))
fprintf(stderr, Name ": Could not re-write superblock on %s.\n",
devname);
if (dfd >= 0)
close(dfd);
}
if (verbose > 0)
fprintf(stderr, Name ": %s is identified as a member of %s, slot %d.\n",
devname, mddev, info.disk.raid_disk);
devices[devcnt].devname = devname;
devices[devcnt].major = major(stb.st_rdev);
devices[devcnt].minor = minor(stb.st_rdev);
devices[devcnt].oldmajor = info.disk.major;
devices[devcnt].oldminor = info.disk.minor;
devices[devcnt].events = info.events;
devices[devcnt].raid_disk = info.disk.raid_disk;
devices[devcnt].disk_nr = info.disk.number;
devices[devcnt].uptodate = 0;
devices[devcnt].state = info.disk.state;
if (most_recent < devcnt) {
if (devices[devcnt].events
> devices[most_recent].events)
most_recent = devcnt;
}
if (info.array.level == -4)
/* with multipath, the raid_disk from the superblock is meaningless */
i = devcnt;
else
i = devices[devcnt].raid_disk;
if (i+1 == 0) {
if (nextspare < info.array.raid_disks)
nextspare = info.array.raid_disks;
i = nextspare++;
}
if (i < 10000) {
if (i >= bestcnt) {
unsigned int newbestcnt = i+10;
int *newbest = malloc(sizeof(int)*newbestcnt);
unsigned int c;
for (c=0; c < newbestcnt; c++)
if (c < bestcnt)
newbest[c] = best[c];
else
newbest[c] = -1;
if (best)free(best);
best = newbest;
bestcnt = newbestcnt;
}
if (best[i] == -1
|| devices[best[i]].events < devices[devcnt].events)
best[i] = devcnt;
}
devcnt++;
}
if (super)
free(super);
super = NULL;
if (update && strcmp(update, "byteorder")==0)
st->minor_version = 90;
if (devcnt == 0) {
fprintf(stderr, Name ": no devices found for %s\n",
mddev);
free(first_super);
return 1;
}
st->ss->getinfo_super(&info, &ident2, first_super);
/* now we have some devices that might be suitable.
* I wonder how many
*/
avail = malloc(info.array.raid_disks);
memset(avail, 0, info.array.raid_disks);
okcnt = 0;
sparecnt=0;
for (i=0; i< bestcnt ;i++) {
int j = best[i];
int event_margin = 1; /* always allow a difference of '1'
* like the kernel does
*/
if (j < 0) continue;
/* note: we ignore error flags in multipath arrays
* as they don't make sense
*/
if (info.array.level != -4)
if (!(devices[j].state & (1<<MD_DISK_SYNC))) {
if (!(devices[j].state & (1<<MD_DISK_FAULTY)))
sparecnt++;
continue;
}
if (devices[j].events+event_margin >=
devices[most_recent].events) {
devices[j].uptodate = 1;
if (i < info.array.raid_disks) {
okcnt++;
avail[i]=1;
} else
sparecnt++;
}
}
while (force && !enough(info.array.level, info.array.raid_disks,
info.array.layout,
avail, okcnt)) {
/* Choose the newest best drive which is
* not up-to-date, update the superblock
* and add it.
*/
int fd;
chosen_drive = -1;
for (i=0; i<info.array.raid_disks && i < bestcnt; i++) {
int j = best[i];
if (j>=0 &&
!devices[j].uptodate &&
devices[j].events > 0 &&
(chosen_drive < 0 ||
devices[j].events > devices[chosen_drive].events))
chosen_drive = j;
}
if (chosen_drive < 0)
break;
if (verbose >= 0)
fprintf(stderr, Name ": forcing event count in %s(%d) from %d upto %d\n",
devices[chosen_drive].devname, devices[chosen_drive].raid_disk,
(int)(devices[chosen_drive].events),
(int)(devices[most_recent].events));
fd = dev_open(devices[chosen_drive].devname, O_RDWR|O_EXCL);
if (fd < 0) {
fprintf(stderr, Name ": Couldn't open %s for write - not updating\n",
devices[chosen_drive].devname);
devices[chosen_drive].events = 0;
continue;
}
if (st->ss->load_super(st,fd, &super, NULL)) {
close(fd);
fprintf(stderr, Name ": RAID superblock disappeared from %s - not updating.\n",
devices[chosen_drive].devname);
devices[chosen_drive].events = 0;
continue;
}
info.events = devices[most_recent].events;
st->ss->update_super(&info, super, "force", devices[chosen_drive].devname, verbose);
if (st->ss->store_super(st, fd, super)) {
close(fd);
fprintf(stderr, Name ": Could not re-write superblock on %s\n",
devices[chosen_drive].devname);
devices[chosen_drive].events = 0;
free(super);
continue;
}
close(fd);
devices[chosen_drive].events = devices[most_recent].events;
devices[chosen_drive].uptodate = 1;
avail[chosen_drive] = 1;
okcnt++;
free(super);
}
/* Now we want to look at the superblock which the kernel will base things on
* and compare the devices that we think are working with the devices that the
* superblock thinks are working.
* If there are differences and --force is given, then update this chosen
* superblock.
*/
chosen_drive = -1;
super = NULL;
for (i=0; chosen_drive < 0 && i<bestcnt; i++) {
int j = best[i];
int fd;
if (j<0)
continue;
if (!devices[j].uptodate)
continue;
chosen_drive = j;
if ((fd=dev_open(devices[j].devname, O_RDONLY|O_EXCL))< 0) {
fprintf(stderr, Name ": Cannot open %s: %s\n",
devices[j].devname, strerror(errno));
return 1;
}
if (st->ss->load_super(st,fd, &super, NULL)) {
close(fd);
fprintf(stderr, Name ": RAID superblock has disappeared from %s\n",
devices[j].devname);
return 1;
}
close(fd);
}
if (super == NULL) {
fprintf(stderr, Name ": No suitable drives found for %s\n", mddev);
return 1;
}
st->ss->getinfo_super(&info, &ident2, super);
for (i=0; i<bestcnt; i++) {
int j = best[i];
unsigned int desired_state;
if (i < info.array.raid_disks)
desired_state = (1<<MD_DISK_ACTIVE) | (1<<MD_DISK_SYNC);
else
desired_state = 0;
if (j<0)
continue;
if (!devices[j].uptodate)
continue;
info.disk.number = devices[j].disk_nr;
info.disk.raid_disk = i;
info.disk.state = desired_state;
if (devices[j].uptodate &&
st->ss->update_super(&info, super, "assemble", NULL, verbose)) {
if (force) {
if (verbose >= 0)
fprintf(stderr, Name ": "
"clearing FAULTY flag for device %d in %s for %s\n",
j, mddev, devices[j].devname);
change = 1;
} else {
if (verbose >= -1)
fprintf(stderr, Name ": "
"device %d in %s has wrong state in superblock, but %s seems ok\n",
i, mddev, devices[j].devname);
}
}
#if 0
if (!devices[j].uptodate &&
!(super.disks[i].state & (1 << MD_DISK_FAULTY))) {
fprintf(stderr, Name ": devices %d of %s is not marked FAULTY in superblock, but cannot be found\n",
i, mddev);
}
#endif
}
if (force && okcnt == info.array.raid_disks-1) {
/* FIXME check event count */
change += st->ss->update_super(&info, super, "force",
devices[chosen_drive].devname, verbose);
}
if (change) {
int fd;
fd = dev_open(devices[chosen_drive].devname, O_RDWR|O_EXCL);
if (fd < 0) {
fprintf(stderr, Name ": Could open %s for write - cannot Assemble array.\n",
devices[chosen_drive].devname);
return 1;
}
if (st->ss->store_super(st, fd, super)) {
close(fd);
fprintf(stderr, Name ": Could not re-write superblock on %s\n",
devices[chosen_drive].devname);
return 1;
}
close(fd);
}
/* count number of in-sync devices according to the superblock.
* We must have this number to start the array without -s or -R
*/
req_cnt = info.array.working_disks;
/* Almost ready to actually *do* something */
if (!old_linux) {
int rv;
if ((vers % 100) >= 1) { /* can use different versions */
mdu_array_info_t inf;
memset(&inf, 0, sizeof(inf));
inf.major_version = st->ss->major;
inf.minor_version = st->minor_version;
rv = ioctl(mdfd, SET_ARRAY_INFO, &inf);
} else
rv = ioctl(mdfd, SET_ARRAY_INFO, NULL);
if (rv) {
fprintf(stderr, Name ": SET_ARRAY_INFO failed for %s: %s\n",
mddev, strerror(errno));
return 1;
}
if (ident->bitmap_fd >= 0) {
if (ioctl(mdfd, SET_BITMAP_FILE, ident->bitmap_fd) != 0) {
fprintf(stderr, Name ": SET_BITMAP_FILE failed.\n");
return 1;
}
}
/* First, add the raid disks, but add the chosen one last */
for (i=0; i<= bestcnt; i++) {
int j;
if (i < bestcnt) {
j = best[i];
if (j == chosen_drive)
continue;
} else
j = chosen_drive;
if (j >= 0 /* && devices[j].uptodate */) {
mdu_disk_info_t disk;
memset(&disk, 0, sizeof(disk));
disk.major = devices[j].major;
disk.minor = devices[j].minor;
if (ioctl(mdfd, ADD_NEW_DISK, &disk)!=0) {
fprintf(stderr, Name ": failed to add %s to %s: %s\n",
devices[j].devname,
mddev,
strerror(errno));
if (i < info.array.raid_disks || i == bestcnt)
okcnt--;
else
sparecnt--;
} else if (verbose > 0)
fprintf(stderr, Name ": added %s to %s as %d\n",
devices[j].devname, mddev, devices[j].raid_disk);
} else if (verbose > 0 && i < info.array.raid_disks)
fprintf(stderr, Name ": no uptodate device for slot %d of %s\n",
i, mddev);
}
if (runstop == 1 ||
(runstop == 0 &&
( enough(info.array.level, info.array.raid_disks, info.array.layout, avail, okcnt) &&
(okcnt >= req_cnt || start_partial_ok)
))) {
if (ioctl(mdfd, RUN_ARRAY, NULL)==0) {
if (verbose >= 0) {
fprintf(stderr, Name ": %s has been started with %d drive%s",
mddev, okcnt, okcnt==1?"":"s");
if (okcnt < info.array.raid_disks)
fprintf(stderr, " (out of %d)", info.array.raid_disks);
if (sparecnt)
fprintf(stderr, " and %d spare%s", sparecnt, sparecnt==1?"":"s");
fprintf(stderr, ".\n");
}
return 0;
}
fprintf(stderr, Name ": failed to RUN_ARRAY %s: %s\n",
mddev, strerror(errno));
return 1;
}
if (runstop == -1) {
fprintf(stderr, Name ": %s assembled from %d drive%s, but not started.\n",
mddev, okcnt, okcnt==1?"":"s");
return 0;
}
if (verbose >= 0) {
fprintf(stderr, Name ": %s assembled from %d drive%s", mddev, okcnt, okcnt==1?"":"s");
if (sparecnt)
fprintf(stderr, " and %d spare%s", sparecnt, sparecnt==1?"":"s");
if (!enough(info.array.level, info.array.raid_disks, info.array.layout, avail, okcnt))
fprintf(stderr, " - not enough to start the array.\n");
else {
if (req_cnt == info.array.raid_disks)
fprintf(stderr, " - need all %d to start it", req_cnt);
else
fprintf(stderr, " - need %d of %d to start", req_cnt, info.array.raid_disks);
fprintf(stderr, " (use --run to insist).\n");
}
}
return 1;
} else {
/* The "chosen_drive" is a good choice, and if necessary, the superblock has
* been updated to point to the current locations of devices.
* so we can just start the array
*/
unsigned long dev;
dev = makedev(devices[chosen_drive].major,
devices[chosen_drive].minor);
if (ioctl(mdfd, START_ARRAY, dev)) {
fprintf(stderr, Name ": Cannot start array: %s\n",
strerror(errno));
}
}
return 0;
}
int
main(int argc, char ** argv) {
struct stat statbuf;
struct swap_header_v1_2 *hdr;
int i;
unsigned long long maxpages;
unsigned long long goodpages;
unsigned long long sz;
off_t offset;
int force = 0;
int version = 1;
char *block_count = 0;
char *pp;
char *opt_label = NULL;
unsigned char *uuid = NULL;
#ifdef HAVE_LIBUUID
const char *opt_uuid = NULL;
uuid_t uuid_dat;
#endif
program_name = (argc && *argv) ? argv[0] : "mkswap";
if ((pp = strrchr(program_name, '/')) != NULL)
program_name = pp+1;
setlocale(LC_ALL, "");
bindtextdomain(PACKAGE, LOCALEDIR);
textdomain(PACKAGE);
if (argc == 2 &&
(!strcmp(argv[1], "-V") || !strcmp(argv[1], "--version"))) {
printf(_("%s (%s)\n"), program_name, PACKAGE_STRING);
exit(0);
}
for (i=1; i<argc; i++) {
if (argv[i][0] == '-') {
switch (argv[i][1]) {
case 'c':
check=1;
break;
case 'f':
force=1;
break;
case 'p':
pp = argv[i]+2;
if (!*pp && i+1 < argc)
pp = argv[++i];
if (isnzdigit(*pp))
user_pagesize = atoi(pp);
else
usage();
break;
case 'L':
pp = argv[i]+2;
if (!*pp && i+1 < argc)
pp = argv[++i];
opt_label = pp;
break;
case 'v':
version = atoi(argv[i]+2);
break;
case 'U':
#ifdef HAVE_LIBUUID
opt_uuid = argv[i]+2;
if (!*opt_uuid && i+1 < argc)
opt_uuid = argv[++i];
#else
fprintf(stderr, _("%1$s: warning: ignore -U (UUIDs are unsupported by %1$s)\n"),
program_name);
#endif
break;
default:
usage();
}
} else if (!device_name) {
device_name = argv[i];
} else if (!block_count) {
block_count = argv[i];
} else
usage();
}
if (version != 1) {
fprintf(stderr, _("%s: does not support swapspace version %d.\n"),
program_name, version);
exit(EXIT_FAILURE);
}
#ifdef HAVE_LIBUUID
if(opt_uuid) {
if (uuid_parse(opt_uuid, uuid_dat) != 0)
die(_("error: UUID parsing failed"));
} else
uuid_generate(uuid_dat);
uuid = uuid_dat;
#endif
init_signature_page(); /* get pagesize */
atexit(deinit_signature_page);
if (!device_name) {
fprintf(stderr,
_("%s: error: Nowhere to set up swap on?\n"),
program_name);
usage();
}
if (block_count) {
/* this silly user specified the number of blocks explicitly */
char *tmp = NULL;
long long blks;
errno = 0;
blks = strtoll(block_count, &tmp, 0);
if ((tmp == block_count) ||
(tmp && *tmp) ||
(errno != 0 && (blks == LLONG_MAX || blks == LLONG_MIN)) ||
blks < 0)
usage();
PAGES = blks / (pagesize / 1024);
}
sz = get_size(device_name);
if (!PAGES) {
PAGES = sz;
} else if (PAGES > sz && !force) {
fprintf(stderr,
_("%s: error: "
"size %llu KiB is larger than device size %llu KiB\n"),
program_name,
PAGES*(pagesize/1024), sz*(pagesize/1024));
exit(1);
}
if (PAGES < MIN_GOODPAGES) {
fprintf(stderr,
_("%s: error: swap area needs to be at least %ld KiB\n"),
program_name, (long)(MIN_GOODPAGES * pagesize/1024));
usage();
}
#ifdef __linux__
if (get_linux_version() >= KERNEL_VERSION(2,3,4))
maxpages = UINT_MAX + 1ULL;
else if (get_linux_version() >= KERNEL_VERSION(2,2,1))
maxpages = V1_MAX_PAGES;
else
#endif
maxpages = V1_OLD_MAX_PAGES;
if (PAGES > maxpages) {
PAGES = maxpages;
fprintf(stderr,
_("%s: warning: truncating swap area to %llu KiB\n"),
program_name, PAGES * pagesize / 1024);
}
if (stat(device_name, &statbuf) < 0) {
perror(device_name);
exit(EXIT_FAILURE);
}
if (S_ISBLK(statbuf.st_mode))
DEV = open(device_name, O_RDWR | O_EXCL);
else
DEV = open(device_name, O_RDWR);
if (DEV < 0) {
perror(device_name);
exit(1);
}
/* Want a block device. Probably not /dev/hda or /dev/hdb. */
if (!S_ISBLK(statbuf.st_mode))
check=0;
else if (statbuf.st_rdev == 0x0300 || statbuf.st_rdev == 0x0340) {
fprintf(stderr,
_("%s: error: "
"will not try to make swapdevice on '%s'\n"),
program_name, device_name);
exit(1);
} else if (check_mount()) {
fprintf(stderr,
_("%s: error: "
"%s is mounted; will not make swapspace.\n"),
program_name, device_name);
exit(1);
}
if (check)
check_blocks();
zap_bootbits(DEV, device_name, force, S_ISBLK(statbuf.st_mode));
hdr = (struct swap_header_v1_2 *) signature_page;
hdr->version = 1;
hdr->last_page = PAGES - 1;
hdr->nr_badpages = badpages;
if (badpages > PAGES - MIN_GOODPAGES)
die(_("Unable to set up swap-space: unreadable"));
goodpages = PAGES - badpages - 1;
printf(_("Setting up swapspace version 1, size = %llu KiB\n"),
goodpages * pagesize / 1024);
write_signature("SWAPSPACE2");
write_uuid_and_label(uuid, opt_label);
offset = 1024;
if (lseek(DEV, offset, SEEK_SET) != offset)
die(_("unable to rewind swap-device"));
if (write_all(DEV, (char *) signature_page + offset,
pagesize - offset) == -1) {
fprintf(stderr, _("%s: %s: unable to write signature page: %s"),
program_name, device_name, strerror(errno));
exit(1);
}
/*
* A subsequent swapon() will fail if the signature
* is not actually on disk. (This is a kernel bug.)
*/
#ifdef HAVE_FSYNC
if (fsync(DEV))
die(_("fsync failed"));
#endif
#ifdef HAVE_LIBSELINUX
if (S_ISREG(statbuf.st_mode) && is_selinux_enabled() > 0) {
security_context_t context_string;
security_context_t oldcontext;
context_t newcontext;
if (fgetfilecon(DEV, &oldcontext) < 0) {
if (errno != ENODATA) {
fprintf(stderr, _("%s: %s: unable to obtain selinux file label: %s\n"),
program_name, device_name,
strerror(errno));
exit(1);
}
if (matchpathcon(device_name, statbuf.st_mode, &oldcontext))
die(_("unable to matchpathcon()"));
}
if (!(newcontext = context_new(oldcontext)))
die(_("unable to create new selinux context"));
if (context_type_set(newcontext, SELINUX_SWAPFILE_TYPE))
die(_("couldn't compute selinux context"));
context_string = context_str(newcontext);
if (strcmp(context_string, oldcontext)!=0) {
if (fsetfilecon(DEV, context_string)) {
fprintf(stderr, _("%s: unable to relabel %s to %s: %s\n"),
program_name, device_name,
context_string,
strerror(errno));
exit(1);
}
}
context_free(newcontext);
freecon(oldcontext);
}
#endif
return 0;
}
int mnt_context_setup_loopdev(struct libmnt_context *cxt)
{
const char *backing_file;
char *loopdev = NULL;
size_t len;
struct loopdev_cxt lc;
int rc, lo_flags = 0;
assert(cxt);
assert(cxt->fs);
assert((cxt->flags & MNT_FL_MOUNTFLAGS_MERGED));
backing_file = mnt_fs_get_srcpath(cxt->fs);
if (!backing_file)
return -EINVAL;
DBG(CXT, mnt_debug_h(cxt, "trying to setup loopdev for %s", backing_file));
if (cxt->mountflags & MS_RDONLY) {
DBG(CXT, mnt_debug_h(cxt, "enabling READ-ONLY flag"));
lo_flags |= LO_FLAGS_READ_ONLY;
}
loopcxt_init(&lc, 0);
if ((cxt->user_mountflags & MNT_MS_LOOP) &&
mnt_fs_get_option(cxt->fs, "loop", &loopdev, &len) == 0 && loopdev) {
char *tmp = strndup(loopdev, len);
if (!tmp)
rc = -ENOMEM;
else {
rc = loopcxt_set_device(&lc, tmp);
free(tmp);
}
}
/* since 2.6.37 we don't have to store backing filename to mtab
* because kernel provides the name in /sys.
*/
if (get_linux_version() >= KERNEL_VERSION(2, 6, 37) ||
!cxt->mtab_writable) {
DBG(CXT, mnt_debug_h(cxt, "enabling AUTOCLEAR flag"));
lo_flags |= LO_FLAGS_AUTOCLEAR;
}
do {
/* found free device */
if (!loopdev) {
rc = loopcxt_find_unused(&lc);
if (rc)
goto done;
DBG(CXT, mnt_debug_h(cxt, "trying to use %s",
loopcxt_get_device(&lc)));
}
/* set device attributes */
rc = loopcxt_set_backing_file(&lc, backing_file);
if (rc)
goto done;
loopcxt_set_flags(&lc, lo_flags);
/* setup the device */
rc = loopcxt_setup_device(&lc);
if (!rc)
break; /* success */
if (loopdev || rc != -EBUSY) {
DBG(CXT, mnt_debug_h(cxt, "failed to setup device"));
break;
}
DBG(CXT, mnt_debug_h(cxt, "loopdev stolen...trying again"));
} while (1);
if (!rc)
rc = mnt_fs_set_source(cxt->fs, loopcxt_get_device(&lc));
if (!rc) {
/* success */
cxt->flags |= MNT_FL_LOOPDEV_READY;
if ((cxt->user_mountflags & MNT_MS_LOOP) &&
loopcxt_is_autoclear(&lc))
/*
* autoclear flag accepted by kernel, don't store
* the "loop=" option to mtab.
*/
cxt->user_mountflags &= ~MNT_MS_LOOP;
if (!(cxt->mountflags & MS_RDONLY) &&
loopcxt_is_readonly(&lc))
/*
* mount planned read-write, but loopdev is read-only,
* let's fix mount options...
*/
cxt->mountflags |= MS_RDONLY;
/* we have to keep the device open until mount(1),
* otherwise it will auto-cleared by kernel
*/
cxt->loopdev_fd = loopcxt_get_fd(&lc);
loopcxt_set_fd(&lc, -1, 0);
}
done:
loopcxt_deinit(&lc);
return rc;
}
/*
* this has to be called after mnt_context_evaluate_permissions()
*/
static int fix_optstr(struct libmnt_context *cxt)
{
int rc = 0;
char *next;
char *name, *val;
size_t namesz, valsz;
struct libmnt_fs *fs;
#ifdef HAVE_LIBSELINUX
int se_fix = 0, se_rem = 0;
#endif
assert(cxt);
assert(cxt->fs);
assert((cxt->flags & MNT_FL_MOUNTFLAGS_MERGED));
if (!cxt)
return -EINVAL;
if (!cxt->fs || (cxt->flags & MNT_FL_MOUNTOPTS_FIXED))
return 0;
DBG(CXT, mnt_debug_h(cxt, "mount: fixing optstr"));
fs = cxt->fs;
/* The propagation flags should not be used together with any other
* flags (except MS_REC and MS_SILENT) */
if (cxt->mountflags & MS_PROPAGATION)
cxt->mountflags &= (MS_PROPAGATION | MS_REC | MS_SILENT);
if (!mnt_optstr_get_option(fs->user_optstr, "user", &val, &valsz)) {
if (val) {
cxt->orig_user = strndup(val, valsz);
if (!cxt->orig_user) {
rc = -ENOMEM;
goto done;
}
}
cxt->flags |= MNT_FL_SAVED_USER;
}
/*
* Sync mount options with mount flags
*/
rc = mnt_optstr_apply_flags(&fs->vfs_optstr, cxt->mountflags,
mnt_get_builtin_optmap(MNT_LINUX_MAP));
if (rc)
goto done;
rc = mnt_optstr_apply_flags(&fs->user_optstr, cxt->user_mountflags,
mnt_get_builtin_optmap(MNT_USERSPACE_MAP));
if (rc)
goto done;
next = fs->fs_optstr;
#ifdef HAVE_LIBSELINUX
if (!is_selinux_enabled())
/* Always remove SELinux garbage if SELinux disabled */
se_rem = 1;
else if (cxt->mountflags & MS_REMOUNT)
/*
* Linux kernel < 2.6.39 does not allow to remount with any
* selinux specific mount options.
*
* Kernel 2.6.39 commits: ff36fe2c845cab2102e4826c1ffa0a6ebf487c65
* 026eb167ae77244458fa4b4b9fc171209c079ba7
* fix this odd behavior, so we don't have to care about it in
* userspace.
*/
se_rem = get_linux_version() < KERNEL_VERSION(2, 6, 39);
else
/* For normal mount we have translate the contexts */
se_fix = 1;
#endif
while (!mnt_optstr_next_option(&next, &name, &namesz, &val, &valsz)) {
if (namesz == 3 && !strncmp(name, "uid", 3))
rc = mnt_optstr_fix_uid(&fs->fs_optstr, val, valsz, &next);
else if (namesz == 3 && !strncmp(name, "gid", 3))
rc = mnt_optstr_fix_gid(&fs->fs_optstr, val, valsz, &next);
#ifdef HAVE_LIBSELINUX
else if ((se_rem || se_fix) &&
namesz >= 7 && (!strncmp(name, "context", 7) ||
!strncmp(name, "fscontext", 9) ||
!strncmp(name, "defcontext", 10) ||
!strncmp(name, "rootcontext", 11) ||
!strncmp(name, "seclabel", 8))) {
if (se_rem) {
/* remove context= option */
next = name;
rc = mnt_optstr_remove_option_at(&fs->fs_optstr,
name,
val ? val + valsz :
name + namesz);
} else if (se_fix && val && valsz)
/* translate selinux contexts */
rc = mnt_optstr_fix_secontext(&fs->fs_optstr,
val, valsz, &next);
}
#endif
if (rc)
goto done;
}
if (!rc && cxt->user_mountflags & MNT_MS_USER)
rc = mnt_optstr_fix_user(&fs->user_optstr);
/* refresh merged optstr */
free(fs->optstr);
fs->optstr = NULL;
fs->optstr = mnt_fs_strdup_options(fs);
done:
cxt->flags |= MNT_FL_MOUNTOPTS_FIXED;
DBG(CXT, mnt_debug_h(cxt, "fixed options [rc=%d]: "
"vfs: '%s' fs: '%s' user: '******', optstr: '%s'", rc,
fs->vfs_optstr, fs->fs_optstr, fs->user_optstr, fs->optstr));
if (rc)
rc = -MNT_ERR_MOUNTOPT;
return rc;
}
int
main(int argc, char **argv) {
struct stat statbuf;
struct swap_header_v1_2 *hdr;
int c;
unsigned long long maxpages;
unsigned long long goodpages;
unsigned long long sz;
off_t offset;
int force = 0;
int version = 1;
char *block_count = 0;
char *opt_label = NULL;
unsigned char *uuid = NULL;
#ifdef HAVE_LIBUUID
const char *opt_uuid = NULL;
uuid_t uuid_dat;
#endif
static const struct option longopts[] = {
{ "check", no_argument, 0, 'c' },
{ "force", no_argument, 0, 'f' },
{ "pagesize", required_argument, 0, 'p' },
{ "label", required_argument, 0, 'L' },
{ "swapversion", required_argument, 0, 'v' },
{ "uuid", required_argument, 0, 'U' },
{ "version", no_argument, 0, 'V' },
{ "help", no_argument, 0, 'h' },
{ NULL, 0, 0, 0 }
};
setlocale(LC_ALL, "");
bindtextdomain(PACKAGE, LOCALEDIR);
textdomain(PACKAGE);
atexit(close_stdout);
while((c = getopt_long(argc, argv, "cfp:L:v:U:Vh", longopts, NULL)) != -1) {
switch (c) {
case 'c':
check=1;
break;
case 'f':
force=1;
break;
case 'p':
user_pagesize = strtou32_or_err(optarg, _("parsing page size failed"));
break;
case 'L':
opt_label = optarg;
break;
case 'v':
version = strtos32_or_err(optarg, _("parsing version number failed"));
break;
case 'U':
#ifdef HAVE_LIBUUID
opt_uuid = optarg;
#else
warnx(_("warning: ignoring -U (UUIDs are unsupported by %s)"),
program_invocation_short_name);
#endif
break;
case 'V':
printf(UTIL_LINUX_VERSION);
exit(EXIT_SUCCESS);
case 'h':
usage(stdout);
default:
usage(stderr);
}
}
if (optind < argc)
device_name = argv[optind++];
if (optind < argc)
block_count = argv[optind++];
if (optind != argc) {
warnx(_("only one device argument is currently supported"));
usage(stderr);
}
if (version != 1)
errx(EXIT_FAILURE,
_("swapspace version %d is not supported"), version);
#ifdef HAVE_LIBUUID
if(opt_uuid) {
if (uuid_parse(opt_uuid, uuid_dat) != 0)
errx(EXIT_FAILURE, _("error: parsing UUID failed"));
} else
uuid_generate(uuid_dat);
uuid = uuid_dat;
#endif
init_signature_page(); /* get pagesize */
if (!device_name) {
warnx(_("error: Nowhere to set up swap on?"));
usage(stderr);
}
if (block_count) {
/* this silly user specified the number of blocks explicitly */
uint64_t blks = strtou64_or_err(block_count,
_("invalid block count argument"));
PAGES = blks / (pagesize / 1024);
}
sz = get_size(device_name);
if (!PAGES)
PAGES = sz;
else if (PAGES > sz && !force) {
errx(EXIT_FAILURE,
_("error: "
"size %llu KiB is larger than device size %llu KiB"),
PAGES*(pagesize/1024), sz*(pagesize/1024));
}
if (PAGES < MIN_GOODPAGES) {
warnx(_("error: swap area needs to be at least %ld KiB"),
(long)(MIN_GOODPAGES * pagesize/1024));
usage(stderr);
}
#ifdef __linux__
if (get_linux_version() >= KERNEL_VERSION(2,3,4))
maxpages = UINT_MAX + 1ULL;
else if (get_linux_version() >= KERNEL_VERSION(2,2,1))
maxpages = V1_MAX_PAGES;
else
#endif
maxpages = V1_OLD_MAX_PAGES;
if (PAGES > maxpages) {
PAGES = maxpages;
warnx(_("warning: truncating swap area to %llu KiB"),
PAGES * pagesize / 1024);
}
if (is_mounted(device_name))
errx(EXIT_FAILURE, _("error: "
"%s is mounted; will not make swapspace"),
device_name);
if (stat(device_name, &statbuf) < 0) {
perror(device_name);
exit(EXIT_FAILURE);
}
if (S_ISBLK(statbuf.st_mode))
DEV = open(device_name, O_RDWR | O_EXCL);
else
DEV = open(device_name, O_RDWR);
if (DEV < 0) {
perror(device_name);
exit(EXIT_FAILURE);
}
if (!S_ISBLK(statbuf.st_mode))
check=0;
else if (blkdev_is_misaligned(DEV))
warnx(_("warning: %s is misaligned"), device_name);
if (check)
check_blocks();
wipe_device(DEV, device_name, force);
hdr = (struct swap_header_v1_2 *) signature_page;
hdr->version = 1;
hdr->last_page = PAGES - 1;
hdr->nr_badpages = badpages;
if (badpages > PAGES - MIN_GOODPAGES)
errx(EXIT_FAILURE, _("Unable to set up swap-space: unreadable"));
goodpages = PAGES - badpages - 1;
printf(_("Setting up swapspace version 1, size = %llu KiB\n"),
goodpages * pagesize / 1024);
write_signature("SWAPSPACE2");
write_uuid_and_label(uuid, opt_label);
offset = 1024;
if (lseek(DEV, offset, SEEK_SET) != offset)
errx(EXIT_FAILURE, _("unable to rewind swap-device"));
if (write_all(DEV, (char *) signature_page + offset,
pagesize - offset) == -1)
err(EXIT_FAILURE,
_("%s: unable to write signature page"),
device_name);
#ifdef HAVE_LIBSELINUX
if (S_ISREG(statbuf.st_mode) && is_selinux_enabled() > 0) {
security_context_t context_string;
security_context_t oldcontext;
context_t newcontext;
if (fgetfilecon(DEV, &oldcontext) < 0) {
if (errno != ENODATA)
err(EXIT_FAILURE,
_("%s: unable to obtain selinux file label"),
device_name);
if (matchpathcon(device_name, statbuf.st_mode, &oldcontext))
errx(EXIT_FAILURE, _("unable to matchpathcon()"));
}
if (!(newcontext = context_new(oldcontext)))
errx(EXIT_FAILURE, _("unable to create new selinux context"));
if (context_type_set(newcontext, SELINUX_SWAPFILE_TYPE))
errx(EXIT_FAILURE, _("couldn't compute selinux context"));
context_string = context_str(newcontext);
if (strcmp(context_string, oldcontext)!=0) {
if (fsetfilecon(DEV, context_string))
err(EXIT_FAILURE, _("unable to relabel %s to %s"),
device_name, context_string);
}
context_free(newcontext);
freecon(oldcontext);
}
#endif
/*
* A subsequent swapon() will fail if the signature
* is not actually on disk. (This is a kernel bug.)
* The fsync() in close_fd() will take care of writing.
*/
if (close_fd(DEV) != 0)
err(EXIT_FAILURE, _("write failed"));
return EXIT_SUCCESS;
}
int
main(int argc, char * argv[]) {
int read_value = 0xbeefbeef, write_value = 0xbeefbeef, bomb_value = 0xbeefbeef;
int read_set, write_set, bomb_set, set;
char * devname;
int fd;
blkelv_ioctl_arg_t elevator;
read_set = write_set = bomb_set = set = 0;
setlocale(LC_MESSAGES, "");
bindtextdomain(PACKAGE, LOCALEDIR);
textdomain(PACKAGE);
for (;;) {
int opt;
opt = getopt(argc, argv, "r:w:b:hv");
if (opt == -1)
break;
switch (opt) {
case 'r':
read_value = atoi(optarg);
read_set = set = 1;
break;
case 'w':
write_value = atoi(optarg);
write_set = set = 1;
break;
case 'b':
bomb_value = atoi(optarg);
bomb_set = set = 1;
break;
case 'h':
usage(), exit(0);
case 'v':
version(), exit(0);
default:
usage(), exit(1);
}
}
if (optind >= argc)
fprintf(stderr, _("missing blockdevice, use -h for help\n")), exit(1);
while (optind < argc) {
devname = argv[optind++];
fd = open(devname, O_RDONLY|O_NONBLOCK);
if (fd < 0) {
perror("open");
break;
}
/* mmj: If we get EINVAL it's not a 2.4 kernel, so warn about
that and exit. It should return ENOTTY however, so check for
that as well in case it gets corrected in the future */
if (ioctl(fd, BLKELVGET, &elevator) < 0) {
int errsv = errno;
perror("ioctl get");
if ((errsv == EINVAL || errsv == ENOTTY) &&
get_linux_version() >= KERNEL_VERSION(2,5,58)) {
fprintf(stderr,
_("\nelvtune is only useful on older "
"kernels;\nfor 2.6 use IO scheduler "
"sysfs tunables instead..\n"));
}
break;
}
if (set) {
if (read_set)
elevator.read_latency = read_value;
if (write_set)
elevator.write_latency = write_value;
if (bomb_set)
elevator.max_bomb_segments = bomb_value;
if (ioctl(fd, BLKELVSET, &elevator) < 0) {
perror("ioctl set");
break;
}
if (ioctl(fd, BLKELVGET, &elevator) < 0) {
perror("ioctl reget");
break;
}
}
printf("\n%s elevator ID\t\t%d\n", devname, elevator.queue_ID);
printf("\tread_latency:\t\t%d\n", elevator.read_latency);
printf("\twrite_latency:\t\t%d\n", elevator.write_latency);
printf("\tmax_bomb_segments:\t%d\n\n", elevator.max_bomb_segments);
if (close(fd) < 0) {
perror("close");
break;
}
}
return 0;
}
