static void
vfs_mountroot_wait(void)
{
struct root_hold_token *h;
struct timeval lastfail;
int curfail;
curfail = 0;
while (1) {
DROP_GIANT();
g_waitidle();
PICKUP_GIANT();
mtx_lock(&mountlist_mtx);
if (LIST_EMPTY(&root_holds)) {
mtx_unlock(&mountlist_mtx);
break;
}
if (ppsratecheck(&lastfail, &curfail, 1)) {
printf("Root mount waiting for:");
LIST_FOREACH(h, &root_holds, list)
printf(" %s", h->who);
printf("\n");
}
msleep(&root_holds, &mountlist_mtx, PZERO | PDROP, "roothold",
hz);
}
}
/*
* Unmount system call.
*/
static int
ext2_unmount(struct mount *mp, int mntflags)
{
struct ext2mount *ump;
struct m_ext2fs *fs;
struct csum *sump;
int error, flags, i, ronly;
flags = 0;
if (mntflags & MNT_FORCE) {
if (mp->mnt_flag & MNT_ROOTFS)
return (EINVAL);
flags |= FORCECLOSE;
}
if ((error = ext2_flushfiles(mp, flags, curthread)) != 0)
return (error);
ump = VFSTOEXT2(mp);
fs = ump->um_e2fs;
ronly = fs->e2fs_ronly;
if (ronly == 0 && ext2_cgupdate(ump, MNT_WAIT) == 0) {
if (fs->e2fs_wasvalid)
fs->e2fs->e2fs_state |= E2FS_ISCLEAN;
ext2_sbupdate(ump, MNT_WAIT);
}
DROP_GIANT();
g_topology_lock();
g_vfs_close(ump->um_cp);
g_topology_unlock();
PICKUP_GIANT();
vrele(ump->um_devvp);
sump = fs->e2fs_clustersum;
for (i = 0; i < fs->e2fs_gcount; i++, sump++)
free(sump->cs_sum, M_EXT2MNT);
free(fs->e2fs_clustersum, M_EXT2MNT);
free(fs->e2fs_maxcluster, M_EXT2MNT);
free(fs->e2fs_gd, M_EXT2MNT);
free(fs->e2fs_contigdirs, M_EXT2MNT);
free(fs->e2fs, M_EXT2MNT);
free(fs, M_EXT2MNT);
free(ump, M_EXT2MNT);
mp->mnt_data = NULL;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_LOCAL;
MNT_IUNLOCK(mp);
return (error);
}
void
uio_yield(void)
{
struct thread *td;
td = curthread;
DROP_GIANT();
#ifndef __rtems__
thread_lock(td);
sched_prio(td, td->td_user_pri);
mi_switch(SW_INVOL | SWT_RELINQUISH, NULL);
thread_unlock(td);
#else /* __rtems__ */
rtems_task_wake_after(RTEMS_YIELD_PROCESSOR);
#endif /* __rtems__ */
PICKUP_GIANT();
}
static int
udf_unmount(struct mount *mp, int mntflags)
{
struct udf_mnt *udfmp;
int error, flags = 0;
udfmp = VFSTOUDFFS(mp);
if (mntflags & MNT_FORCE)
flags |= FORCECLOSE;
if ((error = vflush(mp, 0, flags, curthread)))
return (error);
if (udfmp->im_flags & UDFMNT_KICONV && udf_iconv) {
if (udfmp->im_d2l)
udf_iconv->close(udfmp->im_d2l);
#if 0
if (udfmp->im_l2d)
udf_iconv->close(udfmp->im_l2d);
#endif
}
DROP_GIANT();
g_topology_lock();
g_vfs_close(udfmp->im_cp);
g_topology_unlock();
PICKUP_GIANT();
vrele(udfmp->im_devvp);
dev_rel(udfmp->im_dev);
if (udfmp->s_table != NULL)
free(udfmp->s_table, M_UDFMOUNT);
free(udfmp, M_UDFMOUNT);
mp->mnt_data = NULL;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_LOCAL;
MNT_IUNLOCK(mp);
return (0);
}
static int
g_mbr_ioctl(struct g_provider *pp, u_long cmd, void *data, int fflag, struct thread *td)
{
struct g_geom *gp;
struct g_mbr_softc *ms;
struct g_slicer *gsp;
struct g_consumer *cp;
int error, opened;
gp = pp->geom;
gsp = gp->softc;
ms = gsp->softc;
opened = 0;
error = 0;
switch(cmd) {
case DIOCSMBR: {
if (!(fflag & FWRITE))
return (EPERM);
DROP_GIANT();
g_topology_lock();
cp = LIST_FIRST(&gp->consumer);
if (cp->acw == 0) {
error = g_access(cp, 0, 1, 0);
if (error == 0)
opened = 1;
}
if (!error)
error = g_mbr_modify(gp, ms, data, 512);
if (!error)
error = g_write_data(cp, 0, data, 512);
if (opened)
g_access(cp, 0, -1 , 0);
g_topology_unlock();
PICKUP_GIANT();
return(error);
}
default:
return (ENOIOCTL);
}
}
static int
udf_mountfs(struct vnode *devvp, struct mount *mp)
{
struct buf *bp = NULL;
struct cdev *dev;
struct anchor_vdp avdp;
struct udf_mnt *udfmp = NULL;
struct part_desc *pd;
struct logvol_desc *lvd;
struct fileset_desc *fsd;
struct file_entry *root_fentry;
uint32_t sector, size, mvds_start, mvds_end;
uint32_t logical_secsize;
uint32_t fsd_offset = 0;
uint16_t part_num = 0, fsd_part = 0;
int error = EINVAL;
int logvol_found = 0, part_found = 0, fsd_found = 0;
int bsize;
struct g_consumer *cp;
struct bufobj *bo;
dev = devvp->v_rdev;
dev_ref(dev);
DROP_GIANT();
g_topology_lock();
error = g_vfs_open(devvp, &cp, "udf", 0);
g_topology_unlock();
PICKUP_GIANT();
VOP_UNLOCK(devvp, 0);
if (error)
goto bail;
bo = &devvp->v_bufobj;
if (devvp->v_rdev->si_iosize_max != 0)
mp->mnt_iosize_max = devvp->v_rdev->si_iosize_max;
if (mp->mnt_iosize_max > MAXPHYS)
mp->mnt_iosize_max = MAXPHYS;
/* XXX: should be M_WAITOK */
udfmp = malloc(sizeof(struct udf_mnt), M_UDFMOUNT,
M_NOWAIT | M_ZERO);
if (udfmp == NULL) {
printf("Cannot allocate UDF mount struct\n");
error = ENOMEM;
goto bail;
}
mp->mnt_data = udfmp;
mp->mnt_stat.f_fsid.val[0] = dev2udev(devvp->v_rdev);
mp->mnt_stat.f_fsid.val[1] = mp->mnt_vfc->vfc_typenum;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_LOCAL;
mp->mnt_kern_flag |= MNTK_LOOKUP_SHARED | MNTK_EXTENDED_SHARED;
MNT_IUNLOCK(mp);
udfmp->im_mountp = mp;
udfmp->im_dev = dev;
udfmp->im_devvp = devvp;
udfmp->im_d2l = NULL;
udfmp->im_cp = cp;
udfmp->im_bo = bo;
#if 0
udfmp->im_l2d = NULL;
#endif
/*
* The UDF specification defines a logical sectorsize of 2048
* for DVD media.
*/
logical_secsize = 2048;
if (((logical_secsize % cp->provider->sectorsize) != 0) ||
(logical_secsize < cp->provider->sectorsize)) {
error = EINVAL;
goto bail;
}
bsize = cp->provider->sectorsize;
/*
* Get the Anchor Volume Descriptor Pointer from sector 256.
* XXX Should also check sector n - 256, n, and 512.
*/
sector = 256;
if ((error = bread(devvp, sector * btodb(logical_secsize), bsize,
NOCRED, &bp)) != 0)
goto bail;
if ((error = udf_checktag((struct desc_tag *)bp->b_data, TAGID_ANCHOR)))
goto bail;
bcopy(bp->b_data, &avdp, sizeof(struct anchor_vdp));
brelse(bp);
bp = NULL;
/*
* Extract the Partition Descriptor and Logical Volume Descriptor
* from the Volume Descriptor Sequence.
* XXX Should we care about the partition type right now?
* XXX What about multiple partitions?
*/
mvds_start = le32toh(avdp.main_vds_ex.loc);
mvds_end = mvds_start + (le32toh(avdp.main_vds_ex.len) - 1) / bsize;
for (sector = mvds_start; sector < mvds_end; sector++) {
if ((error = bread(devvp, sector * btodb(logical_secsize),
bsize, NOCRED, &bp)) != 0) {
printf("Can't read sector %d of VDS\n", sector);
goto bail;
}
lvd = (struct logvol_desc *)bp->b_data;
if (!udf_checktag(&lvd->tag, TAGID_LOGVOL)) {
udfmp->bsize = le32toh(lvd->lb_size);
udfmp->bmask = udfmp->bsize - 1;
udfmp->bshift = ffs(udfmp->bsize) - 1;
fsd_part = le16toh(lvd->_lvd_use.fsd_loc.loc.part_num);
fsd_offset = le32toh(lvd->_lvd_use.fsd_loc.loc.lb_num);
if (udf_find_partmaps(udfmp, lvd))
break;
logvol_found = 1;
}
pd = (struct part_desc *)bp->b_data;
if (!udf_checktag(&pd->tag, TAGID_PARTITION)) {
part_found = 1;
part_num = le16toh(pd->part_num);
udfmp->part_len = le32toh(pd->part_len);
udfmp->part_start = le32toh(pd->start_loc);
}
brelse(bp);
bp = NULL;
if ((part_found) && (logvol_found))
break;
}
if (!part_found || !logvol_found) {
error = EINVAL;
goto bail;
}
if (fsd_part != part_num) {
printf("FSD does not lie within the partition!\n");
error = EINVAL;
goto bail;
}
/*
* Grab the Fileset Descriptor
* Thanks to Chuck McCrobie <*****@*****.**> for pointing
* me in the right direction here.
*/
sector = udfmp->part_start + fsd_offset;
if ((error = RDSECTOR(devvp, sector, udfmp->bsize, &bp)) != 0) {
printf("Cannot read sector %d of FSD\n", sector);
goto bail;
}
fsd = (struct fileset_desc *)bp->b_data;
if (!udf_checktag(&fsd->tag, TAGID_FSD)) {
fsd_found = 1;
bcopy(&fsd->rootdir_icb, &udfmp->root_icb,
sizeof(struct long_ad));
}
brelse(bp);
bp = NULL;
if (!fsd_found) {
printf("Couldn't find the fsd\n");
error = EINVAL;
goto bail;
}
/*
* Find the file entry for the root directory.
*/
sector = le32toh(udfmp->root_icb.loc.lb_num) + udfmp->part_start;
size = le32toh(udfmp->root_icb.len);
if ((error = udf_readdevblks(udfmp, sector, size, &bp)) != 0) {
printf("Cannot read sector %d\n", sector);
goto bail;
}
root_fentry = (struct file_entry *)bp->b_data;
if ((error = udf_checktag(&root_fentry->tag, TAGID_FENTRY))) {
printf("Invalid root file entry!\n");
goto bail;
}
brelse(bp);
bp = NULL;
return 0;
bail:
if (udfmp != NULL)
free(udfmp, M_UDFMOUNT);
if (bp != NULL)
brelse(bp);
if (cp != NULL) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
g_topology_unlock();
PICKUP_GIANT();
}
dev_rel(dev);
return error;
};
/*
* Unmount the filesystem described by mp.
*/
static int
msdosfs_unmount(struct mount *mp, int mntflags)
{
struct msdosfsmount *pmp;
int error, flags;
error = flags = 0;
pmp = VFSTOMSDOSFS(mp);
if ((pmp->pm_flags & MSDOSFSMNT_RONLY) == 0)
error = msdosfs_sync(mp, MNT_WAIT);
if ((mntflags & MNT_FORCE) != 0)
flags |= FORCECLOSE;
else if (error != 0)
return (error);
error = vflush(mp, 0, flags, curthread);
if (error != 0 && error != ENXIO)
return (error);
if ((pmp->pm_flags & MSDOSFSMNT_RONLY) == 0) {
error = markvoldirty(pmp, 0);
if (error && error != ENXIO) {
(void)markvoldirty(pmp, 1);
return (error);
}
}
if (pmp->pm_flags & MSDOSFSMNT_KICONV && msdosfs_iconv) {
if (pmp->pm_w2u)
msdosfs_iconv->close(pmp->pm_w2u);
if (pmp->pm_u2w)
msdosfs_iconv->close(pmp->pm_u2w);
if (pmp->pm_d2u)
msdosfs_iconv->close(pmp->pm_d2u);
if (pmp->pm_u2d)
msdosfs_iconv->close(pmp->pm_u2d);
}
#ifdef MSDOSFS_DEBUG
{
struct vnode *vp = pmp->pm_devvp;
struct bufobj *bo;
bo = &vp->v_bufobj;
BO_LOCK(bo);
VI_LOCK(vp);
vn_printf(vp,
"msdosfs_umount(): just before calling VOP_CLOSE()\n");
printf("freef %p, freeb %p, mount %p\n",
TAILQ_NEXT(vp, v_actfreelist), vp->v_actfreelist.tqe_prev,
vp->v_mount);
printf("cleanblkhd %p, dirtyblkhd %p, numoutput %ld, type %d\n",
TAILQ_FIRST(&vp->v_bufobj.bo_clean.bv_hd),
TAILQ_FIRST(&vp->v_bufobj.bo_dirty.bv_hd),
vp->v_bufobj.bo_numoutput, vp->v_type);
VI_UNLOCK(vp);
BO_UNLOCK(bo);
}
#endif
DROP_GIANT();
if (pmp->pm_devvp->v_type == VCHR && pmp->pm_devvp->v_rdev != NULL)
pmp->pm_devvp->v_rdev->si_mountpt = NULL;
g_topology_lock();
g_vfs_close(pmp->pm_cp);
g_topology_unlock();
PICKUP_GIANT();
vrele(pmp->pm_devvp);
dev_rel(pmp->pm_dev);
free(pmp->pm_inusemap, M_MSDOSFSFAT);
if (pmp->pm_flags & MSDOSFS_LARGEFS)
msdosfs_fileno_free(mp);
lockdestroy(&pmp->pm_fatlock);
free(pmp, M_MSDOSFSMNT);
mp->mnt_data = NULL;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_LOCAL;
MNT_IUNLOCK(mp);
return (error);
}
static int
mountmsdosfs(struct vnode *devvp, struct mount *mp)
{
struct msdosfsmount *pmp;
struct buf *bp;
struct cdev *dev;
union bootsector *bsp;
struct byte_bpb33 *b33;
struct byte_bpb50 *b50;
struct byte_bpb710 *b710;
u_int8_t SecPerClust;
u_long clusters;
int ronly, error;
struct g_consumer *cp;
struct bufobj *bo;
bp = NULL; /* This and pmp both used in error_exit. */
pmp = NULL;
ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
dev = devvp->v_rdev;
dev_ref(dev);
DROP_GIANT();
g_topology_lock();
error = g_vfs_open(devvp, &cp, "msdosfs", ronly ? 0 : 1);
g_topology_unlock();
PICKUP_GIANT();
VOP_UNLOCK(devvp, 0);
if (error)
goto error_exit;
bo = &devvp->v_bufobj;
/*
* Read the boot sector of the filesystem, and then check the
* boot signature. If not a dos boot sector then error out.
*
* NOTE: 8192 is a magic size that works for ffs.
*/
error = bread(devvp, 0, 8192, NOCRED, &bp);
if (error)
goto error_exit;
bp->b_flags |= B_AGE;
bsp = (union bootsector *)bp->b_data;
b33 = (struct byte_bpb33 *)bsp->bs33.bsBPB;
b50 = (struct byte_bpb50 *)bsp->bs50.bsBPB;
b710 = (struct byte_bpb710 *)bsp->bs710.bsBPB;
#ifndef MSDOSFS_NOCHECKSIG
if (bsp->bs50.bsBootSectSig0 != BOOTSIG0
|| bsp->bs50.bsBootSectSig1 != BOOTSIG1) {
error = EINVAL;
goto error_exit;
}
#endif
pmp = malloc(sizeof *pmp, M_MSDOSFSMNT, M_WAITOK | M_ZERO);
pmp->pm_mountp = mp;
pmp->pm_cp = cp;
pmp->pm_bo = bo;
lockinit(&pmp->pm_fatlock, 0, msdosfs_lock_msg, 0, 0);
/*
* Initialize ownerships and permissions, since nothing else will
* initialize them iff we are mounting root.
*/
pmp->pm_uid = UID_ROOT;
pmp->pm_gid = GID_WHEEL;
pmp->pm_mask = pmp->pm_dirmask = S_IXUSR | S_IXGRP | S_IXOTH |
S_IRUSR | S_IRGRP | S_IROTH | S_IWUSR;
/*
* Experimental support for large MS-DOS filesystems.
* WARNING: This uses at least 32 bytes of kernel memory (which is not
* reclaimed until the FS is unmounted) for each file on disk to map
* between the 32-bit inode numbers used by VFS and the 64-bit
* pseudo-inode numbers used internally by msdosfs. This is only
* safe to use in certain controlled situations (e.g. read-only FS
* with less than 1 million files).
* Since the mappings do not persist across unmounts (or reboots), these
* filesystems are not suitable for exporting through NFS, or any other
* application that requires fixed inode numbers.
*/
vfs_flagopt(mp->mnt_optnew, "large", &pmp->pm_flags, MSDOSFS_LARGEFS);
/*
* Compute several useful quantities from the bpb in the
* bootsector. Copy in the dos 5 variant of the bpb then fix up
* the fields that are different between dos 5 and dos 3.3.
*/
SecPerClust = b50->bpbSecPerClust;
pmp->pm_BytesPerSec = getushort(b50->bpbBytesPerSec);
if (pmp->pm_BytesPerSec < DEV_BSIZE) {
error = EINVAL;
goto error_exit;
}
pmp->pm_ResSectors = getushort(b50->bpbResSectors);
pmp->pm_FATs = b50->bpbFATs;
pmp->pm_RootDirEnts = getushort(b50->bpbRootDirEnts);
pmp->pm_Sectors = getushort(b50->bpbSectors);
pmp->pm_FATsecs = getushort(b50->bpbFATsecs);
pmp->pm_SecPerTrack = getushort(b50->bpbSecPerTrack);
pmp->pm_Heads = getushort(b50->bpbHeads);
pmp->pm_Media = b50->bpbMedia;
/* calculate the ratio of sector size to DEV_BSIZE */
pmp->pm_BlkPerSec = pmp->pm_BytesPerSec / DEV_BSIZE;
/*
* We don't check pm_Heads nor pm_SecPerTrack, because
* these may not be set for EFI file systems. We don't
* use these anyway, so we're unaffected if they are
* invalid.
*/
if (!pmp->pm_BytesPerSec || !SecPerClust) {
error = EINVAL;
goto error_exit;
}
if (pmp->pm_Sectors == 0) {
pmp->pm_HiddenSects = getulong(b50->bpbHiddenSecs);
pmp->pm_HugeSectors = getulong(b50->bpbHugeSectors);
} else {
pmp->pm_HiddenSects = getushort(b33->bpbHiddenSecs);
pmp->pm_HugeSectors = pmp->pm_Sectors;
}
if (!(pmp->pm_flags & MSDOSFS_LARGEFS)) {
if (pmp->pm_HugeSectors > 0xffffffff /
(pmp->pm_BytesPerSec / sizeof(struct direntry)) + 1) {
/*
* We cannot deal currently with this size of disk
* due to fileid limitations (see msdosfs_getattr and
* msdosfs_readdir)
*/
error = EINVAL;
vfs_mount_error(mp,
"Disk too big, try '-o large' mount option");
goto error_exit;
}
}
if (pmp->pm_RootDirEnts == 0) {
if (pmp->pm_FATsecs
|| getushort(b710->bpbFSVers)) {
error = EINVAL;
#ifdef MSDOSFS_DEBUG
printf("mountmsdosfs(): bad FAT32 filesystem\n");
#endif
goto error_exit;
}
pmp->pm_fatmask = FAT32_MASK;
pmp->pm_fatmult = 4;
pmp->pm_fatdiv = 1;
pmp->pm_FATsecs = getulong(b710->bpbBigFATsecs);
if (getushort(b710->bpbExtFlags) & FATMIRROR)
pmp->pm_curfat = getushort(b710->bpbExtFlags) & FATNUM;
else
pmp->pm_flags |= MSDOSFS_FATMIRROR;
} else
pmp->pm_flags |= MSDOSFS_FATMIRROR;
/*
* Check a few values (could do some more):
* - logical sector size: power of 2, >= block size
* - sectors per cluster: power of 2, >= 1
* - number of sectors: >= 1, <= size of partition
* - number of FAT sectors: >= 1
*/
if ( (SecPerClust == 0)
|| (SecPerClust & (SecPerClust - 1))
|| (pmp->pm_BytesPerSec < DEV_BSIZE)
|| (pmp->pm_BytesPerSec & (pmp->pm_BytesPerSec - 1))
|| (pmp->pm_HugeSectors == 0)
|| (pmp->pm_FATsecs == 0)
|| (SecPerClust * pmp->pm_BlkPerSec > MAXBSIZE / DEV_BSIZE)
) {
error = EINVAL;
goto error_exit;
}
pmp->pm_HugeSectors *= pmp->pm_BlkPerSec;
pmp->pm_HiddenSects *= pmp->pm_BlkPerSec; /* XXX not used? */
pmp->pm_FATsecs *= pmp->pm_BlkPerSec;
SecPerClust *= pmp->pm_BlkPerSec;
pmp->pm_fatblk = pmp->pm_ResSectors * pmp->pm_BlkPerSec;
if (FAT32(pmp)) {
pmp->pm_rootdirblk = getulong(b710->bpbRootClust);
pmp->pm_firstcluster = pmp->pm_fatblk
+ (pmp->pm_FATs * pmp->pm_FATsecs);
pmp->pm_fsinfo = getushort(b710->bpbFSInfo) * pmp->pm_BlkPerSec;
} else {
pmp->pm_rootdirblk = pmp->pm_fatblk +
(pmp->pm_FATs * pmp->pm_FATsecs);
pmp->pm_rootdirsize = howmany(pmp->pm_RootDirEnts *
sizeof(struct direntry), DEV_BSIZE); /* in blocks */
pmp->pm_firstcluster = pmp->pm_rootdirblk + pmp->pm_rootdirsize;
}
pmp->pm_maxcluster = (pmp->pm_HugeSectors - pmp->pm_firstcluster) /
SecPerClust + 1;
pmp->pm_fatsize = pmp->pm_FATsecs * DEV_BSIZE; /* XXX not used? */
if (pmp->pm_fatmask == 0) {
if (pmp->pm_maxcluster
<= ((CLUST_RSRVD - CLUST_FIRST) & FAT12_MASK)) {
/*
* This will usually be a floppy disk. This size makes
* sure that one fat entry will not be split across
* multiple blocks.
*/
pmp->pm_fatmask = FAT12_MASK;
pmp->pm_fatmult = 3;
pmp->pm_fatdiv = 2;
} else {
pmp->pm_fatmask = FAT16_MASK;
pmp->pm_fatmult = 2;
pmp->pm_fatdiv = 1;
}
}
clusters = (pmp->pm_fatsize / pmp->pm_fatmult) * pmp->pm_fatdiv;
if (pmp->pm_maxcluster >= clusters) {
#ifdef MSDOSFS_DEBUG
printf("Warning: number of clusters (%ld) exceeds FAT "
"capacity (%ld)\n", pmp->pm_maxcluster + 1, clusters);
#endif
pmp->pm_maxcluster = clusters - 1;
}
if (FAT12(pmp))
pmp->pm_fatblocksize = 3 * 512;
else
pmp->pm_fatblocksize = PAGE_SIZE;
pmp->pm_fatblocksize = roundup(pmp->pm_fatblocksize,
pmp->pm_BytesPerSec);
pmp->pm_fatblocksec = pmp->pm_fatblocksize / DEV_BSIZE;
pmp->pm_bnshift = ffs(DEV_BSIZE) - 1;
/*
* Compute mask and shift value for isolating cluster relative byte
* offsets and cluster numbers from a file offset.
*/
pmp->pm_bpcluster = SecPerClust * DEV_BSIZE;
pmp->pm_crbomask = pmp->pm_bpcluster - 1;
pmp->pm_cnshift = ffs(pmp->pm_bpcluster) - 1;
/*
* Check for valid cluster size
* must be a power of 2
*/
if (pmp->pm_bpcluster ^ (1 << pmp->pm_cnshift)) {
error = EINVAL;
goto error_exit;
}
/*
* Release the bootsector buffer.
*/
brelse(bp);
bp = NULL;
/*
* Check the fsinfo sector if we have one. Silently fix up our
* in-core copy of fp->fsinxtfree if it is unknown (0xffffffff)
* or too large. Ignore fp->fsinfree for now, since we need to
* read the entire FAT anyway to fill the inuse map.
*/
if (pmp->pm_fsinfo) {
struct fsinfo *fp;
if ((error = bread(devvp, pmp->pm_fsinfo, pmp->pm_BytesPerSec,
NOCRED, &bp)) != 0)
goto error_exit;
fp = (struct fsinfo *)bp->b_data;
if (!bcmp(fp->fsisig1, "RRaA", 4)
&& !bcmp(fp->fsisig2, "rrAa", 4)
&& !bcmp(fp->fsisig3, "\0\0\125\252", 4)) {
pmp->pm_nxtfree = getulong(fp->fsinxtfree);
if (pmp->pm_nxtfree > pmp->pm_maxcluster)
pmp->pm_nxtfree = CLUST_FIRST;
} else
pmp->pm_fsinfo = 0;
brelse(bp);
bp = NULL;
}
/*
* Finish initializing pmp->pm_nxtfree (just in case the first few
* sectors aren't properly reserved in the FAT). This completes
* the fixup for fp->fsinxtfree, and fixes up the zero-initialized
* value if there is no fsinfo. We will use pmp->pm_nxtfree
* internally even if there is no fsinfo.
*/
if (pmp->pm_nxtfree < CLUST_FIRST)
pmp->pm_nxtfree = CLUST_FIRST;
/*
* Allocate memory for the bitmap of allocated clusters, and then
* fill it in.
*/
pmp->pm_inusemap = malloc(howmany(pmp->pm_maxcluster + 1, N_INUSEBITS)
* sizeof(*pmp->pm_inusemap),
M_MSDOSFSFAT, M_WAITOK);
/*
* fillinusemap() needs pm_devvp.
*/
pmp->pm_devvp = devvp;
pmp->pm_dev = dev;
/*
* Have the inuse map filled in.
*/
MSDOSFS_LOCK_MP(pmp);
error = fillinusemap(pmp);
MSDOSFS_UNLOCK_MP(pmp);
if (error != 0)
goto error_exit;
/*
* If they want fat updates to be synchronous then let them suffer
* the performance degradation in exchange for the on disk copy of
* the fat being correct just about all the time. I suppose this
* would be a good thing to turn on if the kernel is still flakey.
*/
if (mp->mnt_flag & MNT_SYNCHRONOUS)
pmp->pm_flags |= MSDOSFSMNT_WAITONFAT;
/*
* Finish up.
*/
if (ronly)
pmp->pm_flags |= MSDOSFSMNT_RONLY;
else {
if ((error = markvoldirty(pmp, 1)) != 0) {
(void)markvoldirty(pmp, 0);
goto error_exit;
}
pmp->pm_fmod = 1;
}
mp->mnt_data = pmp;
mp->mnt_stat.f_fsid.val[0] = dev2udev(dev);
mp->mnt_stat.f_fsid.val[1] = mp->mnt_vfc->vfc_typenum;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_LOCAL;
mp->mnt_kern_flag |= MNTK_USES_BCACHE;
MNT_IUNLOCK(mp);
if (pmp->pm_flags & MSDOSFS_LARGEFS)
msdosfs_fileno_init(mp);
return 0;
error_exit:
if (bp)
brelse(bp);
if (cp != NULL) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
g_topology_unlock();
PICKUP_GIANT();
}
if (pmp) {
lockdestroy(&pmp->pm_fatlock);
if (pmp->pm_inusemap)
free(pmp->pm_inusemap, M_MSDOSFSFAT);
free(pmp, M_MSDOSFSMNT);
mp->mnt_data = NULL;
}
dev_rel(dev);
return (error);
}
/*
* mp - path - addr in user space of mount point (ie /usr or whatever)
* data - addr in user space of mount params including the name of the block
* special file to treat as a filesystem.
*/
static int
msdosfs_mount(struct mount *mp)
{
struct vnode *devvp; /* vnode for blk device to mount */
struct thread *td;
/* msdosfs specific mount control block */
struct msdosfsmount *pmp = NULL;
struct nameidata ndp;
int error, flags;
accmode_t accmode;
char *from;
td = curthread;
if (vfs_filteropt(mp->mnt_optnew, msdosfs_opts))
return (EINVAL);
/*
* If updating, check whether changing from read-only to
* read/write; if there is no device name, that's all we do.
*/
if (mp->mnt_flag & MNT_UPDATE) {
pmp = VFSTOMSDOSFS(mp);
if (vfs_flagopt(mp->mnt_optnew, "export", NULL, 0)) {
/*
* Forbid export requests if filesystem has
* MSDOSFS_LARGEFS flag set.
*/
if ((pmp->pm_flags & MSDOSFS_LARGEFS) != 0) {
vfs_mount_error(mp,
"MSDOSFS_LARGEFS flag set, cannot export");
return (EOPNOTSUPP);
}
}
if (!(pmp->pm_flags & MSDOSFSMNT_RONLY) &&
vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0)) {
error = VFS_SYNC(mp, MNT_WAIT);
if (error)
return (error);
flags = WRITECLOSE;
if (mp->mnt_flag & MNT_FORCE)
flags |= FORCECLOSE;
error = vflush(mp, 0, flags, td);
if (error)
return (error);
/*
* Now the volume is clean. Mark it so while the
* device is still rw.
*/
error = markvoldirty(pmp, 0);
if (error) {
(void)markvoldirty(pmp, 1);
return (error);
}
/* Downgrade the device from rw to ro. */
DROP_GIANT();
g_topology_lock();
error = g_access(pmp->pm_cp, 0, -1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error) {
(void)markvoldirty(pmp, 1);
return (error);
}
/*
* Backing out after an error was painful in the
* above. Now we are committed to succeeding.
*/
pmp->pm_fmod = 0;
pmp->pm_flags |= MSDOSFSMNT_RONLY;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_RDONLY;
MNT_IUNLOCK(mp);
} else if ((pmp->pm_flags & MSDOSFSMNT_RONLY) &&
!vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0)) {
/*
* If upgrade to read-write by non-root, then verify
* that user has necessary permissions on the device.
*/
devvp = pmp->pm_devvp;
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
error = VOP_ACCESS(devvp, VREAD | VWRITE,
td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
VOP_UNLOCK(devvp, 0);
return (error);
}
VOP_UNLOCK(devvp, 0);
DROP_GIANT();
g_topology_lock();
error = g_access(pmp->pm_cp, 0, 1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error)
return (error);
pmp->pm_fmod = 1;
pmp->pm_flags &= ~MSDOSFSMNT_RONLY;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_RDONLY;
MNT_IUNLOCK(mp);
/* Now that the volume is modifiable, mark it dirty. */
error = markvoldirty(pmp, 1);
if (error)
return (error);
}
}
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible disk device.
*/
if (vfs_getopt(mp->mnt_optnew, "from", (void **)&from, NULL))
return (EINVAL);
NDINIT(&ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, from, td);
error = namei(&ndp);
if (error)
return (error);
devvp = ndp.ni_vp;
NDFREE(&ndp, NDF_ONLY_PNBUF);
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/*
* If mount by non-root, then verify that user has necessary
* permissions on the device.
*/
accmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accmode |= VWRITE;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = mountmsdosfs(devvp, mp);
#ifdef MSDOSFS_DEBUG /* only needed for the printf below */
pmp = VFSTOMSDOSFS(mp);
#endif
} else {
vput(devvp);
if (devvp != pmp->pm_devvp)
return (EINVAL); /* XXX needs translation */
}
if (error) {
vrele(devvp);
return (error);
}
error = update_mp(mp, td);
if (error) {
if ((mp->mnt_flag & MNT_UPDATE) == 0)
msdosfs_unmount(mp, MNT_FORCE);
return error;
}
if (devvp->v_type == VCHR && devvp->v_rdev != NULL)
devvp->v_rdev->si_mountpt = mp;
vfs_mountedfrom(mp, from);
#ifdef MSDOSFS_DEBUG
printf("msdosfs_mount(): mp %p, pmp %p, inusemap %p\n", mp, pmp, pmp->pm_inusemap);
#endif
return (0);
}
/*
* Common code for mount and mountroot.
*/
static int
ext2_mountfs(struct vnode *devvp, struct mount *mp)
{
struct ext2mount *ump;
struct buf *bp;
struct m_ext2fs *fs;
struct ext2fs *es;
struct cdev *dev = devvp->v_rdev;
struct g_consumer *cp;
struct bufobj *bo;
struct csum *sump;
int error;
int ronly;
int i, size;
int32_t *lp;
ronly = vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0);
/* XXX: use VOP_ACESS to check FS perms */
DROP_GIANT();
g_topology_lock();
error = g_vfs_open(devvp, &cp, "ext2fs", ronly ? 0 : 1);
g_topology_unlock();
PICKUP_GIANT();
VOP_UNLOCK(devvp, 0);
if (error)
return (error);
/* XXX: should we check for some sectorsize or 512 instead? */
if (((SBSIZE % cp->provider->sectorsize) != 0) ||
(SBSIZE < cp->provider->sectorsize)) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
g_topology_unlock();
PICKUP_GIANT();
return (EINVAL);
}
bo = &devvp->v_bufobj;
bo->bo_private = cp;
bo->bo_ops = g_vfs_bufops;
if (devvp->v_rdev->si_iosize_max != 0)
mp->mnt_iosize_max = devvp->v_rdev->si_iosize_max;
if (mp->mnt_iosize_max > MAXPHYS)
mp->mnt_iosize_max = MAXPHYS;
bp = NULL;
ump = NULL;
if ((error = bread(devvp, SBLOCK, SBSIZE, NOCRED, &bp)) != 0)
goto out;
es = (struct ext2fs *)bp->b_data;
if (ext2_check_sb_compat(es, dev, ronly) != 0) {
error = EINVAL; /* XXX needs translation */
goto out;
}
if ((es->e2fs_state & E2FS_ISCLEAN) == 0 ||
(es->e2fs_state & E2FS_ERRORS)) {
if (ronly || (mp->mnt_flag & MNT_FORCE)) {
printf(
"WARNING: Filesystem was not properly dismounted\n");
} else {
printf(
"WARNING: R/W mount denied. Filesystem is not clean - run fsck\n");
error = EPERM;
goto out;
}
}
ump = malloc(sizeof(*ump), M_EXT2MNT, M_WAITOK | M_ZERO);
/*
* I don't know whether this is the right strategy. Note that
* we dynamically allocate both an ext2_sb_info and an ext2_super_block
* while Linux keeps the super block in a locked buffer.
*/
ump->um_e2fs = malloc(sizeof(struct m_ext2fs),
M_EXT2MNT, M_WAITOK);
ump->um_e2fs->e2fs = malloc(sizeof(struct ext2fs),
M_EXT2MNT, M_WAITOK);
mtx_init(EXT2_MTX(ump), "EXT2FS", "EXT2FS Lock", MTX_DEF);
bcopy(es, ump->um_e2fs->e2fs, (u_int)sizeof(struct ext2fs));
if ((error = compute_sb_data(devvp, ump->um_e2fs->e2fs, ump->um_e2fs)))
goto out;
/*
* Calculate the maximum contiguous blocks and size of cluster summary
* array. In FFS this is done by newfs; however, the superblock
* in ext2fs doesn't have these variables, so we can calculate
* them here.
*/
ump->um_e2fs->e2fs_maxcontig = MAX(1, MAXPHYS / ump->um_e2fs->e2fs_bsize);
if (ump->um_e2fs->e2fs_maxcontig > 0)
ump->um_e2fs->e2fs_contigsumsize =
MIN(ump->um_e2fs->e2fs_maxcontig, EXT2_MAXCONTIG);
else
ump->um_e2fs->e2fs_contigsumsize = 0;
if (ump->um_e2fs->e2fs_contigsumsize > 0) {
size = ump->um_e2fs->e2fs_gcount * sizeof(int32_t);
ump->um_e2fs->e2fs_maxcluster = malloc(size, M_EXT2MNT, M_WAITOK);
size = ump->um_e2fs->e2fs_gcount * sizeof(struct csum);
ump->um_e2fs->e2fs_clustersum = malloc(size, M_EXT2MNT, M_WAITOK);
lp = ump->um_e2fs->e2fs_maxcluster;
sump = ump->um_e2fs->e2fs_clustersum;
for (i = 0; i < ump->um_e2fs->e2fs_gcount; i++, sump++) {
*lp++ = ump->um_e2fs->e2fs_contigsumsize;
sump->cs_init = 0;
sump->cs_sum = malloc((ump->um_e2fs->e2fs_contigsumsize + 1) *
sizeof(int32_t), M_EXT2MNT, M_WAITOK | M_ZERO);
}
}
brelse(bp);
bp = NULL;
fs = ump->um_e2fs;
fs->e2fs_ronly = ronly; /* ronly is set according to mnt_flags */
/*
* If the fs is not mounted read-only, make sure the super block is
* always written back on a sync().
*/
fs->e2fs_wasvalid = fs->e2fs->e2fs_state & E2FS_ISCLEAN ? 1 : 0;
if (ronly == 0) {
fs->e2fs_fmod = 1; /* mark it modified */
fs->e2fs->e2fs_state &= ~E2FS_ISCLEAN; /* set fs invalid */
}
mp->mnt_data = ump;
mp->mnt_stat.f_fsid.val[0] = dev2udev(dev);
mp->mnt_stat.f_fsid.val[1] = mp->mnt_vfc->vfc_typenum;
mp->mnt_maxsymlinklen = EXT2_MAXSYMLINKLEN;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_LOCAL;
MNT_IUNLOCK(mp);
ump->um_mountp = mp;
ump->um_dev = dev;
ump->um_devvp = devvp;
ump->um_bo = &devvp->v_bufobj;
ump->um_cp = cp;
/*
* Setting those two parameters allowed us to use
* ufs_bmap w/o changse!
*/
ump->um_nindir = EXT2_ADDR_PER_BLOCK(fs);
ump->um_bptrtodb = fs->e2fs->e2fs_log_bsize + 1;
ump->um_seqinc = EXT2_FRAGS_PER_BLOCK(fs);
if (ronly == 0)
ext2_sbupdate(ump, MNT_WAIT);
/*
* Initialize filesystem stat information in mount struct.
*/
MNT_ILOCK(mp);
mp->mnt_kern_flag |= MNTK_MPSAFE | MNTK_LOOKUP_SHARED |
MNTK_EXTENDED_SHARED;
MNT_IUNLOCK(mp);
return (0);
out:
if (bp)
brelse(bp);
if (cp != NULL) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
g_topology_unlock();
PICKUP_GIANT();
}
if (ump) {
mtx_destroy(EXT2_MTX(ump));
free(ump->um_e2fs->e2fs_gd, M_EXT2MNT);
free(ump->um_e2fs->e2fs_contigdirs, M_EXT2MNT);
free(ump->um_e2fs->e2fs, M_EXT2MNT);
free(ump->um_e2fs, M_EXT2MNT);
free(ump, M_EXT2MNT);
mp->mnt_data = NULL;
}
return (error);
}
/*-
* This start routine is only called for non-trivial requests, all the
* trivial ones are handled autonomously by the slice code.
* For requests we handle here, we must call the g_io_deliver() on the
* bio, and return non-zero to indicate to the slice code that we did so.
* This code executes in the "DOWN" I/O path, this means:
* * No sleeping.
* * Don't grab the topology lock.
* * Don't call biowait, g_getattr(), g_setattr() or g_read_data()
*/
static int
g_bsd_ioctl(struct g_provider *pp, u_long cmd, void *data, int fflag, struct thread *td)
{
struct g_geom *gp;
struct g_bsd_softc *ms;
struct g_slicer *gsp;
u_char *label;
int error;
gp = pp->geom;
gsp = gp->softc;
ms = gsp->softc;
switch(cmd) {
case DIOCGDINFO:
/* Return a copy of the disklabel to userland. */
bsd_disklabel_le_dec(ms->label, data, MAXPARTITIONS);
return(0);
case DIOCBSDBB: {
struct g_consumer *cp;
u_char *buf;
void *p;
int error, i;
uint64_t sum;
if (!(fflag & FWRITE))
return (EPERM);
/* The disklabel to set is the ioctl argument. */
buf = g_malloc(BBSIZE, M_WAITOK);
p = *(void **)data;
error = copyin(p, buf, BBSIZE);
if (!error) {
/* XXX: Rude, but supposedly safe */
DROP_GIANT();
g_topology_lock();
/* Validate and modify our slice instance to match. */
error = g_bsd_modify(gp, buf + ms->labeloffset);
if (!error) {
cp = LIST_FIRST(&gp->consumer);
if (ms->labeloffset == ALPHA_LABEL_OFFSET) {
sum = 0;
for (i = 0; i < 63; i++)
sum += le64dec(buf + i * 8);
le64enc(buf + 504, sum);
}
error = g_write_data(cp, 0, buf, BBSIZE);
}
g_topology_unlock();
PICKUP_GIANT();
}
g_free(buf);
return (error);
}
case DIOCSDINFO:
case DIOCWDINFO: {
if (!(fflag & FWRITE))
return (EPERM);
label = g_malloc(LABELSIZE, M_WAITOK);
/* The disklabel to set is the ioctl argument. */
bsd_disklabel_le_enc(label, data);
DROP_GIANT();
g_topology_lock();
/* Validate and modify our slice instance to match. */
error = g_bsd_modify(gp, label);
if (error == 0 && cmd == DIOCWDINFO)
error = g_bsd_writelabel(gp, NULL);
g_topology_unlock();
PICKUP_GIANT();
g_free(label);
return(error);
}
default:
return (ENOIOCTL);
}
}
static int
parse_dir_md(char **conf)
{
struct stat sb;
struct thread *td;
struct md_ioctl *mdio;
char *path, *tok;
int error, fd, len;
td = curthread;
error = parse_token(conf, &tok);
if (error)
return (error);
len = strlen(tok);
mdio = malloc(sizeof(*mdio) + len + 1, M_TEMP, M_WAITOK | M_ZERO);
path = (void *)(mdio + 1);
bcopy(tok, path, len);
free(tok, M_TEMP);
/* Get file status. */
error = kern_stat(td, path, UIO_SYSSPACE, &sb);
if (error)
goto out;
/* Open /dev/mdctl so that we can attach/detach. */
error = kern_open(td, "/dev/" MDCTL_NAME, UIO_SYSSPACE, O_RDWR, 0);
if (error)
goto out;
fd = td->td_retval[0];
mdio->md_version = MDIOVERSION;
mdio->md_type = MD_VNODE;
if (root_mount_mddev != -1) {
mdio->md_unit = root_mount_mddev;
DROP_GIANT();
error = kern_ioctl(td, fd, MDIOCDETACH, (void *)mdio);
PICKUP_GIANT();
/* Ignore errors. We don't care. */
root_mount_mddev = -1;
}
mdio->md_file = (void *)(mdio + 1);
mdio->md_options = MD_AUTOUNIT | MD_READONLY;
mdio->md_mediasize = sb.st_size;
mdio->md_unit = 0;
DROP_GIANT();
error = kern_ioctl(td, fd, MDIOCATTACH, (void *)mdio);
PICKUP_GIANT();
if (error)
goto out;
if (mdio->md_unit > 9) {
printf("rootmount: too many md units\n");
mdio->md_file = NULL;
mdio->md_options = 0;
mdio->md_mediasize = 0;
DROP_GIANT();
error = kern_ioctl(td, fd, MDIOCDETACH, (void *)mdio);
PICKUP_GIANT();
/* Ignore errors. We don't care. */
error = ERANGE;
goto out;
}
root_mount_mddev = mdio->md_unit;
printf(MD_NAME "%u attached to %s\n", root_mount_mddev, mdio->md_file);
error = kern_close(td, fd);
out:
free(mdio, M_TEMP);
return (error);
}
/*
* Shutdown the system cleanly to prepare for reboot, halt, or power off.
*/
void
kern_reboot(int howto)
{
static int first_buf_printf = 1;
#if defined(SMP)
/*
* Bind us to CPU 0 so that all shutdown code runs there. Some
* systems don't shutdown properly (i.e., ACPI power off) if we
* run on another processor.
*/
if (!SCHEDULER_STOPPED()) {
thread_lock(curthread);
sched_bind(curthread, 0);
thread_unlock(curthread);
KASSERT(PCPU_GET(cpuid) == 0, ("boot: not running on cpu 0"));
}
#endif
/* We're in the process of rebooting. */
rebooting = 1;
/* collect extra flags that shutdown_nice might have set */
howto |= shutdown_howto;
/* We are out of the debugger now. */
kdb_active = 0;
/*
* Do any callouts that should be done BEFORE syncing the filesystems.
*/
EVENTHANDLER_INVOKE(shutdown_pre_sync, howto);
/*
* Now sync filesystems
*/
if (!cold && (howto & RB_NOSYNC) == 0 && waittime < 0) {
register struct buf *bp;
int iter, nbusy, pbusy;
#ifndef PREEMPTION
int subiter;
#endif
waittime = 0;
wdog_kern_pat(WD_LASTVAL);
sys_sync(curthread, NULL);
/*
* With soft updates, some buffers that are
* written will be remarked as dirty until other
* buffers are written.
*/
for (iter = pbusy = 0; iter < 20; iter++) {
nbusy = 0;
for (bp = &buf[nbuf]; --bp >= buf; )
if (isbufbusy(bp))
nbusy++;
if (nbusy == 0) {
if (first_buf_printf)
printf("All buffers synced.");
break;
}
if (first_buf_printf) {
printf("Syncing disks, buffers remaining... ");
first_buf_printf = 0;
}
printf("%d ", nbusy);
if (nbusy < pbusy)
iter = 0;
pbusy = nbusy;
wdog_kern_pat(WD_LASTVAL);
sys_sync(curthread, NULL);
#ifdef PREEMPTION
/*
* Drop Giant and spin for a while to allow
* interrupt threads to run.
*/
DROP_GIANT();
DELAY(50000 * iter);
PICKUP_GIANT();
#else
/*
* Drop Giant and context switch several times to
* allow interrupt threads to run.
*/
DROP_GIANT();
for (subiter = 0; subiter < 50 * iter; subiter++) {
thread_lock(curthread);
mi_switch(SW_VOL, NULL);
thread_unlock(curthread);
DELAY(1000);
}
PICKUP_GIANT();
#endif
}
printf("\n");
/*
* Count only busy local buffers to prevent forcing
* a fsck if we're just a client of a wedged NFS server
*/
nbusy = 0;
for (bp = &buf[nbuf]; --bp >= buf; ) {
if (isbufbusy(bp)) {
#if 0
/* XXX: This is bogus. We should probably have a BO_REMOTE flag instead */
if (bp->b_dev == NULL) {
TAILQ_REMOVE(&mountlist,
bp->b_vp->v_mount, mnt_list);
continue;
}
#endif
nbusy++;
if (show_busybufs > 0) {
printf(
"%d: buf:%p, vnode:%p, flags:%0x, blkno:%jd, lblkno:%jd, buflock:",
nbusy, bp, bp->b_vp, bp->b_flags,
(intmax_t)bp->b_blkno,
(intmax_t)bp->b_lblkno);
BUF_LOCKPRINTINFO(bp);
if (show_busybufs > 1)
vn_printf(bp->b_vp,
"vnode content: ");
}
}
}
if (nbusy) {
/*
* Failed to sync all blocks. Indicate this and don't
* unmount filesystems (thus forcing an fsck on reboot).
*/
printf("Giving up on %d buffers\n", nbusy);
DELAY(5000000); /* 5 seconds */
} else {
if (!first_buf_printf)
printf("Final sync complete\n");
/*
* Unmount filesystems
*/
if (panicstr == 0)
vfs_unmountall();
}
swapoff_all();
DELAY(100000); /* wait for console output to finish */
}
print_uptime();
cngrab();
/*
* Ok, now do things that assume all filesystem activity has
* been completed.
*/
EVENTHANDLER_INVOKE(shutdown_post_sync, howto);
if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold && !dumping)
doadump(TRUE);
/* Now that we're going to really halt the system... */
EVENTHANDLER_INVOKE(shutdown_final, howto);
for(;;) ; /* safety against shutdown_reset not working */
/* NOTREACHED */
}
/*
* Common code for mount and mountroot
*/
static int
reiserfs_mountfs(struct vnode *devvp, struct mount *mp, struct thread *td)
{
int error, old_format = 0;
struct reiserfs_mount *rmp;
struct reiserfs_sb_info *sbi;
struct reiserfs_super_block *rs;
struct cdev *dev;
struct g_consumer *cp;
struct bufobj *bo;
//ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
dev = devvp->v_rdev;
dev_ref(dev);
DROP_GIANT();
g_topology_lock();
error = g_vfs_open(devvp, &cp, "reiserfs", /* read-only */ 0);
g_topology_unlock();
PICKUP_GIANT();
VOP_UNLOCK(devvp, 0);
if (error) {
dev_rel(dev);
return (error);
}
bo = &devvp->v_bufobj;
bo->bo_private = cp;
bo->bo_ops = g_vfs_bufops;
if (devvp->v_rdev->si_iosize_max != 0)
mp->mnt_iosize_max = devvp->v_rdev->si_iosize_max;
if (mp->mnt_iosize_max > MAXPHYS)
mp->mnt_iosize_max = MAXPHYS;
rmp = NULL;
sbi = NULL;
/* rmp contains any information about this specific mount */
rmp = malloc(sizeof *rmp, M_REISERFSMNT, M_WAITOK | M_ZERO);
if (!rmp) {
error = (ENOMEM);
goto out;
}
sbi = malloc(sizeof *sbi, M_REISERFSMNT, M_WAITOK | M_ZERO);
if (!sbi) {
error = (ENOMEM);
goto out;
}
rmp->rm_reiserfs = sbi;
rmp->rm_mountp = mp;
rmp->rm_devvp = devvp;
rmp->rm_dev = dev;
rmp->rm_bo = &devvp->v_bufobj;
rmp->rm_cp = cp;
/* Set default values for options: non-aggressive tails */
REISERFS_SB(sbi)->s_mount_opt = (1 << REISERFS_SMALLTAIL);
REISERFS_SB(sbi)->s_rd_only = 1;
REISERFS_SB(sbi)->s_devvp = devvp;
/* Read the super block */
if ((error = read_super_block(rmp, REISERFS_OLD_DISK_OFFSET)) == 0) {
/* The read process succeeded, it's an old format */
old_format = 1;
} else if ((error = read_super_block(rmp, REISERFS_DISK_OFFSET)) != 0) {
reiserfs_log(LOG_ERR, "can not find a ReiserFS filesystem\n");
goto out;
}
rs = SB_DISK_SUPER_BLOCK(sbi);
/*
* Let's do basic sanity check to verify that underlying device is
* not smaller than the filesystem. If the check fails then abort and
* scream, because bad stuff will happen otherwise.
*/
#if 0
if (s->s_bdev && s->s_bdev->bd_inode &&
i_size_read(s->s_bdev->bd_inode) <
sb_block_count(rs) * sb_blocksize(rs)) {
reiserfs_log(LOG_ERR,
"reiserfs: filesystem cannot be mounted because it is "
"bigger than the device.\n");
reiserfs_log(LOG_ERR, "reiserfs: you may need to run fsck "
"rr may be you forgot to reboot after fdisk when it "
"told you to.\n");
goto out;
}
#endif
/*
* XXX This is from the original Linux code, but why affecting 2 values
* to the same variable?
*/
sbi->s_mount_state = SB_REISERFS_STATE(sbi);
sbi->s_mount_state = REISERFS_VALID_FS;
if ((error = (old_format ?
read_old_bitmaps(rmp) : read_bitmaps(rmp)))) {
reiserfs_log(LOG_ERR, "unable to read bitmap\n");
goto out;
}
/* Make data=ordered the default */
if (!reiserfs_data_log(sbi) && !reiserfs_data_ordered(sbi) &&
!reiserfs_data_writeback(sbi)) {
REISERFS_SB(sbi)->s_mount_opt |= (1 << REISERFS_DATA_ORDERED);
}
if (reiserfs_data_log(sbi)) {
reiserfs_log(LOG_INFO, "using journaled data mode\n");
} else if (reiserfs_data_ordered(sbi)) {
reiserfs_log(LOG_INFO, "using ordered data mode\n");
} else {
reiserfs_log(LOG_INFO, "using writeback data mode\n");
}
/* TODO Not yet supported */
#if 0
if(journal_init(sbi, jdev_name, old_format, commit_max_age)) {
reiserfs_log(LOG_ERR, "unable to initialize journal space\n");
goto out;
} else {
jinit_done = 1 ; /* once this is set, journal_release must
be called if we error out of the mount */
}
if (reread_meta_blocks(sbi)) {
reiserfs_log(LOG_ERR,
"unable to reread meta blocks after journal init\n");
goto out;
}
#endif
/* Define and initialize hash function */
sbi->s_hash_function = hash_function(rmp);
if (sbi->s_hash_function == NULL) {
reiserfs_log(LOG_ERR, "couldn't determined hash function\n");
error = (EINVAL);
goto out;
}
if (is_reiserfs_3_5(rs) ||
(is_reiserfs_jr(rs) && SB_VERSION(sbi) == REISERFS_VERSION_1))
bit_set(&(sbi->s_properties), REISERFS_3_5);
else
bit_set(&(sbi->s_properties), REISERFS_3_6);
mp->mnt_data = rmp;
mp->mnt_stat.f_fsid.val[0] = dev2udev(dev);
mp->mnt_stat.f_fsid.val[1] = mp->mnt_vfc->vfc_typenum;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_LOCAL;
mp->mnt_kern_flag |= MNTK_MPSAFE;
MNT_IUNLOCK(mp);
#if defined(si_mountpoint)
devvp->v_rdev->si_mountpoint = mp;
#endif
return (0);
out:
reiserfs_log(LOG_INFO, "*** error during mount ***\n");
if (sbi) {
if (SB_AP_BITMAP(sbi)) {
int i;
for (i = 0; i < SB_BMAP_NR(sbi); i++) {
if (!SB_AP_BITMAP(sbi)[i].bp_data)
break;
free(SB_AP_BITMAP(sbi)[i].bp_data,
M_REISERFSMNT);
}
free(SB_AP_BITMAP(sbi), M_REISERFSMNT);
}
if (sbi->s_rs) {
free(sbi->s_rs, M_REISERFSMNT);
sbi->s_rs = NULL;
}
}
if (cp != NULL) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
g_topology_unlock();
PICKUP_GIANT();
}
if (sbi)
free(sbi, M_REISERFSMNT);
if (rmp)
free(rmp, M_REISERFSMNT);
dev_rel(dev);
return (error);
}
/*
* Unmount system call
*/
static int
reiserfs_unmount(struct mount *mp, int mntflags)
{
int error, flags = 0;
struct reiserfs_mount *rmp;
struct reiserfs_sb_info *sbi;
reiserfs_log(LOG_DEBUG, "get private data\n");
rmp = VFSTOREISERFS(mp);
sbi = rmp->rm_reiserfs;
/* Flangs handling */
reiserfs_log(LOG_DEBUG, "handle mntflags\n");
if (mntflags & MNT_FORCE)
flags |= FORCECLOSE;
/* Flush files -> vflush */
reiserfs_log(LOG_DEBUG, "flush vnodes\n");
if ((error = vflush(mp, 0, flags, curthread)))
return (error);
/* XXX Super block update */
if (sbi) {
if (SB_AP_BITMAP(sbi)) {
int i;
reiserfs_log(LOG_DEBUG,
"release bitmap buffers (total: %d)\n",
SB_BMAP_NR(sbi));
for (i = 0; i < SB_BMAP_NR(sbi); i++) {
if (SB_AP_BITMAP(sbi)[i].bp_data) {
free(SB_AP_BITMAP(sbi)[i].bp_data,
M_REISERFSMNT);
SB_AP_BITMAP(sbi)[i].bp_data = NULL;
}
}
reiserfs_log(LOG_DEBUG, "free bitmaps structure\n");
free(SB_AP_BITMAP(sbi), M_REISERFSMNT);
SB_AP_BITMAP(sbi) = NULL;
}
if (sbi->s_rs) {
reiserfs_log(LOG_DEBUG, "free super block data\n");
free(sbi->s_rs, M_REISERFSMNT);
sbi->s_rs = NULL;
}
}
reiserfs_log(LOG_DEBUG, "close device\n");
#if defined(si_mountpoint)
rmp->rm_devvp->v_rdev->si_mountpoint = NULL;
#endif
DROP_GIANT();
g_topology_lock();
g_vfs_close(rmp->rm_cp);
g_topology_unlock();
PICKUP_GIANT();
vrele(rmp->rm_devvp);
dev_rel(rmp->rm_dev);
if (sbi) {
reiserfs_log(LOG_DEBUG, "free sbi\n");
free(sbi, M_REISERFSMNT);
sbi = rmp->rm_reiserfs = NULL;
}
if (rmp) {
reiserfs_log(LOG_DEBUG, "free rmp\n");
free(rmp, M_REISERFSMNT);
rmp = NULL;
}
mp->mnt_data = 0;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_LOCAL;
MNT_IUNLOCK(mp);
reiserfs_log(LOG_DEBUG, "done\n");
return (error);
}
static int
_xfs_mount(struct mount *mp)
{
struct xfsmount *xmp;
struct xfs_vnode *rootvp;
struct ucred *curcred;
struct vnode *rvp, *devvp;
struct cdev *ddev;
struct g_consumer *cp;
struct thread *td;
int error;
td = curthread;
ddev = NULL;
cp = NULL;
if (vfs_filteropt(mp->mnt_optnew, xfs_opts))
return (EINVAL);
if (mp->mnt_flag & MNT_UPDATE)
return (0);
if ((mp->mnt_flag & MNT_RDONLY) == 0)
return (EPERM);
xmp = xfsmount_allocate(mp);
if (xmp == NULL)
return (ENOMEM);
if((error = _xfs_param_copyin(mp, td)) != 0)
goto fail;
curcred = td->td_ucred;
XVFS_MOUNT(XFSTOVFS(xmp), &xmp->m_args, curcred, error);
if (error)
goto fail;
XVFS_ROOT(XFSTOVFS(xmp), &rootvp, error);
ddev = XFS_VFSTOM(XFSTOVFS(xmp))->m_ddev_targp->dev;
devvp = XFS_VFSTOM(XFSTOVFS(xmp))->m_ddev_targp->specvp;
if (error)
goto fail_unmount;
if (ddev->si_iosize_max != 0)
mp->mnt_iosize_max = ddev->si_iosize_max;
if (mp->mnt_iosize_max > MAXPHYS)
mp->mnt_iosize_max = MAXPHYS;
mp->mnt_flag |= MNT_LOCAL;
mp->mnt_stat.f_fsid.val[0] = dev2udev(ddev);
mp->mnt_stat.f_fsid.val[1] = mp->mnt_vfc->vfc_typenum;
if ((error = VFS_STATFS(mp, &mp->mnt_stat)) != 0)
goto fail_unmount;
rvp = rootvp->v_vnode;
rvp->v_vflag |= VV_ROOT;
VN_RELE(rootvp);
return (0);
fail_unmount:
XVFS_UNMOUNT(XFSTOVFS(xmp), 0, curcred, error);
if (devvp != NULL) {
cp = devvp->v_bufobj.bo_private;
if (cp != NULL) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
g_topology_unlock();
PICKUP_GIANT();
}
}
fail:
if (xmp != NULL)
xfsmount_deallocate(xmp);
return (error);
}
/*
* VFS Operations.
*
* mount system call
*/
static int
ext2_mount(struct mount *mp)
{
struct vfsoptlist *opts;
struct vnode *devvp;
struct thread *td;
struct ext2mount *ump = NULL;
struct m_ext2fs *fs;
struct nameidata nd, *ndp = &nd;
accmode_t accmode;
char *path, *fspec;
int error, flags, len;
td = curthread;
opts = mp->mnt_optnew;
if (vfs_filteropt(opts, ext2_opts))
return (EINVAL);
vfs_getopt(opts, "fspath", (void **)&path, NULL);
/* Double-check the length of path.. */
if (strlen(path) >= MAXMNTLEN - 1)
return (ENAMETOOLONG);
fspec = NULL;
error = vfs_getopt(opts, "from", (void **)&fspec, &len);
if (!error && fspec[len - 1] != '\0')
return (EINVAL);
/*
* If updating, check whether changing from read-only to
* read/write; if there is no device name, that's all we do.
*/
if (mp->mnt_flag & MNT_UPDATE) {
ump = VFSTOEXT2(mp);
fs = ump->um_e2fs;
error = 0;
if (fs->e2fs_ronly == 0 &&
vfs_flagopt(opts, "ro", NULL, 0)) {
error = VFS_SYNC(mp, MNT_WAIT);
if (error)
return (error);
flags = WRITECLOSE;
if (mp->mnt_flag & MNT_FORCE)
flags |= FORCECLOSE;
error = ext2_flushfiles(mp, flags, td);
if ( error == 0 && fs->e2fs_wasvalid && ext2_cgupdate(ump, MNT_WAIT) == 0) {
fs->e2fs->e2fs_state |= E2FS_ISCLEAN;
ext2_sbupdate(ump, MNT_WAIT);
}
fs->e2fs_ronly = 1;
vfs_flagopt(opts, "ro", &mp->mnt_flag, MNT_RDONLY);
DROP_GIANT();
g_topology_lock();
g_access(ump->um_cp, 0, -1, 0);
g_topology_unlock();
PICKUP_GIANT();
}
if (!error && (mp->mnt_flag & MNT_RELOAD))
error = ext2_reload(mp, td);
if (error)
return (error);
devvp = ump->um_devvp;
if (fs->e2fs_ronly && !vfs_flagopt(opts, "ro", NULL, 0)) {
if (ext2_check_sb_compat(fs->e2fs, devvp->v_rdev, 0))
return (EPERM);
/*
* If upgrade to read-write by non-root, then verify
* that user has necessary permissions on the device.
*/
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
error = VOP_ACCESS(devvp, VREAD | VWRITE,
td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
VOP_UNLOCK(devvp, 0);
return (error);
}
VOP_UNLOCK(devvp, 0);
DROP_GIANT();
g_topology_lock();
error = g_access(ump->um_cp, 0, 1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error)
return (error);
if ((fs->e2fs->e2fs_state & E2FS_ISCLEAN) == 0 ||
(fs->e2fs->e2fs_state & E2FS_ERRORS)) {
if (mp->mnt_flag & MNT_FORCE) {
printf(
"WARNING: %s was not properly dismounted\n", fs->e2fs_fsmnt);
} else {
printf(
"WARNING: R/W mount of %s denied. Filesystem is not clean - run fsck\n",
fs->e2fs_fsmnt);
return (EPERM);
}
}
fs->e2fs->e2fs_state &= ~E2FS_ISCLEAN;
(void)ext2_cgupdate(ump, MNT_WAIT);
fs->e2fs_ronly = 0;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_RDONLY;
MNT_IUNLOCK(mp);
}
if (vfs_flagopt(opts, "export", NULL, 0)) {
/* Process export requests in vfs_mount.c. */
return (error);
}
}
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible disk device.
*/
if (fspec == NULL)
return (EINVAL);
NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspec, td);
if ((error = namei(ndp)) != 0)
return (error);
NDFREE(ndp, NDF_ONLY_PNBUF);
devvp = ndp->ni_vp;
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/*
* If mount by non-root, then verify that user has necessary
* permissions on the device.
*
* XXXRW: VOP_ACCESS() enough?
*/
accmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accmode |= VWRITE;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = ext2_mountfs(devvp, mp);
} else {
if (devvp != ump->um_devvp) {
vput(devvp);
return (EINVAL); /* needs translation */
} else
vput(devvp);
}
if (error) {
vrele(devvp);
return (error);
}
ump = VFSTOEXT2(mp);
fs = ump->um_e2fs;
/*
* Note that this strncpy() is ok because of a check at the start
* of ext2_mount().
*/
strncpy(fs->e2fs_fsmnt, path, MAXMNTLEN);
fs->e2fs_fsmnt[MAXMNTLEN - 1] = '\0';
vfs_mountedfrom(mp, fspec);
return (0);
}
/*ARGSUSED*/
static int
zfs_mount(vfs_t *vfsp)
{
kthread_t *td = curthread;
vnode_t *mvp = vfsp->mnt_vnodecovered;
cred_t *cr = td->td_ucred;
char *osname;
int error = 0;
int canwrite;
if (vfs_getopt(vfsp->mnt_optnew, "from", (void **)&osname, NULL))
return (EINVAL);
/*
* If full-owner-access is enabled and delegated administration is
* turned on, we must set nosuid.
*/
if (zfs_super_owner &&
dsl_deleg_access(osname, ZFS_DELEG_PERM_MOUNT, cr) != ECANCELED) {
secpolicy_fs_mount_clearopts(cr, vfsp);
}
/*
* Check for mount privilege?
*
* If we don't have privilege then see if
* we have local permission to allow it
*/
error = secpolicy_fs_mount(cr, mvp, vfsp);
if (error) {
error = dsl_deleg_access(osname, ZFS_DELEG_PERM_MOUNT, cr);
if (error != 0)
goto out;
if (!(vfsp->vfs_flag & MS_REMOUNT)) {
vattr_t vattr;
/*
* Make sure user is the owner of the mount point
* or has sufficient privileges.
*/
vattr.va_mask = AT_UID;
vn_lock(mvp, LK_SHARED | LK_RETRY);
if (error = VOP_GETATTR(mvp, &vattr, cr)) {
VOP_UNLOCK(mvp, 0);
goto out;
}
#if 0 /* CHECK THIS! Is probably needed for zfs_suser. */
if (secpolicy_vnode_owner(mvp, cr, vattr.va_uid) != 0 &&
VOP_ACCESS(mvp, VWRITE, cr, td) != 0) {
error = EPERM;
goto out;
}
#else
if (error = secpolicy_vnode_owner(mvp, cr, vattr.va_uid)) {
VOP_UNLOCK(mvp, 0);
goto out;
}
if (error = VOP_ACCESS(mvp, VWRITE, cr, td)) {
VOP_UNLOCK(mvp, 0);
goto out;
}
VOP_UNLOCK(mvp, 0);
#endif
}
secpolicy_fs_mount_clearopts(cr, vfsp);
}
/*
* Refuse to mount a filesystem if we are in a local zone and the
* dataset is not visible.
*/
if (!INGLOBALZONE(curthread) &&
(!zone_dataset_visible(osname, &canwrite) || !canwrite)) {
error = EPERM;
goto out;
}
/*
* When doing a remount, we simply refresh our temporary properties
* according to those options set in the current VFS options.
*/
if (vfsp->vfs_flag & MS_REMOUNT) {
/* refresh mount options */
zfs_unregister_callbacks(vfsp->vfs_data);
error = zfs_register_callbacks(vfsp);
goto out;
}
DROP_GIANT();
error = zfs_domount(vfsp, osname);
PICKUP_GIANT();
out:
return (error);
}
