static int zfsfuse_lookup(fuse_req_t req, fuse_ino_t parent, const char *name)
{
if(strlen(name) >= MAXNAMELEN)
return ENAMETOOLONG;
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
ZFS_ENTER(zfsvfs);
znode_t *znode;
int error = zfs_zget(zfsvfs, parent, &znode, B_TRUE);
if(error) {
ZFS_EXIT(zfsvfs);
/* If the inode we are trying to get was recently deleted
dnode_hold_impl will return EEXIST instead of ENOENT */
return error == EEXIST ? ENOENT : error;
}
ASSERT(znode != NULL);
vnode_t *dvp = ZTOV(znode);
ASSERT(dvp != NULL);
vnode_t *vp = NULL;
cred_t cred;
zfsfuse_getcred(req, &cred);
error = VOP_LOOKUP(dvp, (char *) name, &vp, NULL, 0, NULL, &cred, NULL, NULL, NULL);
if(error)
goto out;
struct fuse_entry_param e = { 0 };
e.attr_timeout = 0.0;
e.entry_timeout = 0.0;
if(vp == NULL)
goto out;
e.ino = VTOZ(vp)->z_id;
if(e.ino == 3)
e.ino = 1;
e.generation = VTOZ(vp)->z_phys->zp_gen;
error = zfsfuse_stat(vp, &e.attr, &cred);
out:
if(vp != NULL)
VN_RELE(vp);
VN_RELE(dvp);
ZFS_EXIT(zfsvfs);
if(!error)
fuse_reply_entry(req, &e);
return error;
}
static int zfsfuse_release(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi)
{
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
ZFS_ENTER(zfsvfs);
file_info_t *info = (file_info_t *)(uintptr_t) fi->fh;
ASSERT(info->vp != NULL);
ASSERT(VTOZ(info->vp) != NULL);
ASSERT(VTOZ(info->vp)->z_id == ino);
cred_t cred;
zfsfuse_getcred(req, &cred);
int error = VOP_CLOSE(info->vp, info->flags, 1, (offset_t) 0, &cred, NULL);
VERIFY(error == 0);
VN_RELE(info->vp);
kmem_cache_free(file_info_cache, info);
ZFS_EXIT(zfsvfs);
return error;
}
void getfinderinfo(znode_t *zp, cred_t *cr, finderinfo_t *fip)
{
vnode_t *xdvp = NULLVP;
vnode_t *xvp = NULLVP;
struct uio *auio = NULL;
struct componentname cn;
int error;
uint64_t xattr = 0;
if (sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zp->z_zfsvfs),
&xattr, sizeof(xattr)) ||
(xattr == 0)) {
goto nodata;
}
auio = uio_create(1, 0, UIO_SYSSPACE, UIO_READ);
if (auio == NULL) {
goto nodata;
}
uio_addiov(auio, CAST_USER_ADDR_T(fip), sizeof (finderinfo_t));
/*
* Grab the hidden attribute directory vnode.
*
* XXX - switch to embedded Finder Info when it becomes available
*/
if ((error = zfs_get_xattrdir(zp, &xdvp, cr, 0))) {
goto out;
}
bzero(&cn, sizeof (cn));
cn.cn_nameiop = LOOKUP;
cn.cn_flags = ISLASTCN;
cn.cn_nameptr = XATTR_FINDERINFO_NAME;
cn.cn_namelen = strlen(cn.cn_nameptr);
if ((error = zfs_dirlook(VTOZ(xdvp), cn.cn_nameptr, &xvp, 0, NULL, &cn))) {
goto out;
}
error = dmu_read_uio(zp->z_zfsvfs->z_os, VTOZ(xvp)->z_id, auio,
sizeof (finderinfo_t));
out:
if (auio)
uio_free(auio);
if (xvp)
vnode_put(xvp);
if (xdvp)
vnode_put(xdvp);
if (error == 0)
return;
nodata:
bzero(fip, sizeof (finderinfo_t));
}
/*
* Create the '.zfs' directory. This directory is cached as part of the VFS
* structure. This results in a hold on the vfs_t. The code in zfs_umount()
* therefore checks against a vfs_count of 2 instead of 1. This reference
* is removed when the ctldir is destroyed in the unmount.
*/
void
zfsctl_create(zfsvfs_t *zfsvfs)
{
vnode_t *vp, *rvp;
zfsctl_node_t *zcp;
ASSERT(zfsvfs->z_ctldir == NULL);
vp = gfs_root_create(sizeof (zfsctl_node_t), zfsvfs->z_vfs,
zfsctl_ops_root, ZFSCTL_INO_ROOT, zfsctl_root_entries,
zfsctl_root_inode_cb, MAXNAMELEN, NULL, NULL);
zcp = vp->v_data;
zcp->zc_id = ZFSCTL_INO_ROOT;
VERIFY(VFS_ROOT(zfsvfs->z_vfs, &rvp) == 0);
ZFS_TIME_DECODE(&zcp->zc_cmtime, VTOZ(rvp)->z_phys->zp_crtime);
VN_RELE(rvp);
/*
* We're only faking the fact that we have a root of a filesystem for
* the sake of the GFS interfaces. Undo the flag manipulation it did
* for us.
*/
vp->v_flag &= ~(VROOT | VNOCACHE | VNOMAP | VNOSWAP | VNOMOUNT);
zfsvfs->z_ctldir = vp;
}
/*
* Create the '.zfs' directory. This directory is cached as part of the VFS
* structure. This results in a hold on the vfs_t. The code in zfs_umount()
* therefore checks against a vfs_count of 2 instead of 1. This reference
* is removed when the ctldir is destroyed in the unmount.
*/
void
zfsctl_create(zfsvfs_t *zfsvfs)
{
vnode_t *vp, *rvp;
zfsctl_node_t *zcp;
uint64_t crtime[2];
ASSERT(zfsvfs->z_ctldir == NULL);
vp = gfs_root_create(sizeof (zfsctl_node_t), zfsvfs->z_vfs,
&zfsctl_ops_root, ZFSCTL_INO_ROOT, zfsctl_root_entries,
zfsctl_root_inode_cb, MAXNAMELEN, NULL, NULL);
zcp = vp->v_data;
zcp->zc_id = ZFSCTL_INO_ROOT;
VERIFY(VFS_ROOT(zfsvfs->z_vfs, LK_EXCLUSIVE, &rvp) == 0);
VERIFY(0 == sa_lookup(VTOZ(rvp)->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs),
&crtime, sizeof (crtime)));
ZFS_TIME_DECODE(&zcp->zc_cmtime, crtime);
VN_URELE(rvp);
/*
* We're only faking the fact that we have a root of a filesystem for
* the sake of the GFS interfaces. Undo the flag manipulation it did
* for us.
*/
vp->v_vflag &= ~VV_ROOT;
zfsvfs->z_ctldir = vp;
VOP_UNLOCK(vp, 0);
}
static int zfsfuse_read(fuse_req_t req, fuse_ino_t ino, size_t size, off_t off, struct fuse_file_info *fi)
{
file_info_t *info = (file_info_t *)(uintptr_t) fi->fh;
vnode_t *vp = info->vp;
ASSERT(vp != NULL);
ASSERT(VTOZ(vp) != NULL);
ASSERT(VTOZ(vp)->z_id == ino);
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
char *outbuf = kmem_alloc(size, KM_NOSLEEP);
if(outbuf == NULL)
return ENOMEM;
ZFS_ENTER(zfsvfs);
iovec_t iovec;
uio_t uio;
uio.uio_iov = &iovec;
uio.uio_iovcnt = 1;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_fmode = 0;
uio.uio_llimit = RLIM64_INFINITY;
iovec.iov_base = outbuf;
iovec.iov_len = size;
uio.uio_resid = iovec.iov_len;
uio.uio_loffset = off;
cred_t cred;
zfsfuse_getcred(req, &cred);
int error = VOP_READ(vp, &uio, info->flags, &cred, NULL);
ZFS_EXIT(zfsvfs);
if(!error)
fuse_reply_buf(req, outbuf, uio.uio_loffset - off);
kmem_free(outbuf, size);
return error;
}
static int zfsfuse_write(fuse_req_t req, fuse_ino_t ino, const char *buf, size_t size, off_t off, struct fuse_file_info *fi)
{
file_info_t *info = (file_info_t *)(uintptr_t) fi->fh;
vnode_t *vp = info->vp;
ASSERT(vp != NULL);
ASSERT(VTOZ(vp) != NULL);
ASSERT(VTOZ(vp)->z_id == ino);
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
ZFS_ENTER(zfsvfs);
iovec_t iovec;
uio_t uio;
uio.uio_iov = &iovec;
uio.uio_iovcnt = 1;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_fmode = 0;
uio.uio_llimit = RLIM64_INFINITY;
iovec.iov_base = (void *) buf;
iovec.iov_len = size;
uio.uio_resid = iovec.iov_len;
uio.uio_loffset = off;
cred_t cred;
zfsfuse_getcred(req, &cred);
int error = VOP_WRITE(vp, &uio, info->flags, &cred, NULL);
ZFS_EXIT(zfsvfs);
if(!error) {
/* When not using direct_io, we must always write 'size' bytes */
VERIFY(uio.uio_resid == 0);
fuse_reply_write(req, size - uio.uio_resid);
}
return error;
}
/* ARGSUSED */
static ino64_t
zfsctl_root_inode_cb(vnode_t *vp, int index)
{
zfsvfs_t *zfsvfs = VTOZ(vp)->z_zfsvfs;
ASSERT(index <= 2);
if (index == 0)
return (ZFSCTL_INO_SNAPDIR);
return (zfsvfs->z_shares_dir);
}
static int
zfs_vfs_sync(struct mount *mp, __unused int waitfor, __unused vfs_context_t context)
{
zfsvfs_t *zfsvfs = vfs_fsprivate(mp);
ZFS_ENTER(zfsvfs);
/*
* Mac OS X needs a file system modify time
*
* We use the mtime of the "com.apple.system.mtime"
* extended attribute, which is associated with the
* file system root directory.
*
* Here we sync any mtime changes to this attribute.
*/
if (zfsvfs->z_mtime_vp != NULL) {
timestruc_t mtime;
znode_t *zp;
top:
zp = VTOZ(zfsvfs->z_mtime_vp);
ZFS_TIME_DECODE(&mtime, zp->z_phys->zp_mtime);
if (zfsvfs->z_last_mtime_synced < mtime.tv_sec) {
dmu_tx_t *tx;
int error;
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_bonus(tx, zp->z_id);
error = dmu_tx_assign(tx, zfsvfs->z_assign);
if (error) {
if (error == ERESTART && zfsvfs->z_assign == TXG_NOWAIT) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
} else {
dmu_buf_will_dirty(zp->z_dbuf, tx);
dmu_tx_commit(tx);
zfsvfs->z_last_mtime_synced = mtime.tv_sec;
}
}
}
if (zfsvfs->z_log != NULL)
zil_commit(zfsvfs->z_log, UINT64_MAX, 0);
else
txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0);
ZFS_EXIT(zfsvfs);
return (0);
}
static int zfsfuse_fsync(fuse_req_t req, fuse_ino_t ino, int datasync, struct fuse_file_info *fi)
{
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
ZFS_ENTER(zfsvfs);
file_info_t *info = (file_info_t *)(uintptr_t) fi->fh;
ASSERT(info->vp != NULL);
ASSERT(VTOZ(info->vp) != NULL);
ASSERT(VTOZ(info->vp)->z_id == ino);
vnode_t *vp = info->vp;
cred_t cred;
zfsfuse_getcred(req, &cred);
int error = VOP_FSYNC(vp, datasync ? FDSYNC : FSYNC, &cred, NULL);
ZFS_EXIT(zfsvfs);
return error;
}
/*
* Create the '.zfs' directory.
*/
void
zfsctl_create(zfsvfs_t *zfsvfs)
{
zfsctl_root_t *dot_zfs;
sfs_node_t *snapdir;
vnode_t *rvp;
uint64_t crtime[2];
ASSERT(zfsvfs->z_ctldir == NULL);
snapdir = sfs_alloc_node(sizeof(*snapdir), "snapshot", ZFSCTL_INO_ROOT,
ZFSCTL_INO_SNAPDIR);
dot_zfs = (zfsctl_root_t *)sfs_alloc_node(sizeof(*dot_zfs), ".zfs", 0,
ZFSCTL_INO_ROOT);
dot_zfs->snapdir = snapdir;
VERIFY(VFS_ROOT(zfsvfs->z_vfs, LK_EXCLUSIVE, &rvp) == 0);
VERIFY(0 == sa_lookup(VTOZ(rvp)->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs),
&crtime, sizeof(crtime)));
ZFS_TIME_DECODE(&dot_zfs->cmtime, crtime);
vput(rvp);
zfsvfs->z_ctldir = dot_zfs;
}
/*
* Vnode pointer to File handle
*
* XXX Do we want to check the DSL sharenfs property?
*/
static int
zfs_vfs_vptofh(struct vnode *vp, int *fhlenp, unsigned char *fhp, __unused vfs_context_t context)
{
zfsvfs_t *zfsvfs = vfs_fsprivate(vnode_mount(vp));
zfs_zfid_t *zfid = (zfs_zfid_t *)fhp;
znode_t *zp = VTOZ(vp);
uint64_t obj_num;
uint64_t zp_gen;
int i;
int error;
if (*fhlenp < sizeof (zfs_zfid_t)) {
return (EOVERFLOW);
}
ZFS_ENTER(zfsvfs);
obj_num = zp->z_id;
zp_gen = zp->z_phys->zp_gen;
if (zp_gen == 0)
zp_gen = 1;
/*
* Store the object and gen numbers in an endian neutral manner
*/
for (i = 0; i < sizeof (zfid->zf_object); i++)
zfid->zf_object[i] = (uint8_t)(obj_num >> (8 * i));
for (i = 0; i < sizeof (zfid->zf_gen); i++)
zfid->zf_gen[i] = (uint8_t)(zp_gen >> (8 * i));
*fhlenp = sizeof (zfs_zfid_t);
ZFS_EXIT(zfsvfs);
return (0);
}
/*
* Lock a directory entry. A dirlock on <dzp, name> protects that name
* in dzp's directory zap object. As long as you hold a dirlock, you can
* assume two things: (1) dzp cannot be reaped, and (2) no other thread
* can change the zap entry for (i.e. link or unlink) this name.
*
* Input arguments:
* dzp - znode for directory
* name - name of entry to lock
* flag - ZNEW: if the entry already exists, fail with EEXIST.
* ZEXISTS: if the entry does not exist, fail with ENOENT.
* ZSHARED: allow concurrent access with other ZSHARED callers.
* ZXATTR: we want dzp's xattr directory
* ZCILOOK: On a mixed sensitivity file system,
* this lookup should be case-insensitive.
* ZCIEXACT: On a purely case-insensitive file system,
* this lookup should be case-sensitive.
* ZRENAMING: we are locking for renaming, force narrow locks
* ZHAVELOCK: Don't grab the z_name_lock for this call. The
* current thread already holds it.
*
* Output arguments:
* zpp - pointer to the znode for the entry (NULL if there isn't one)
* dlpp - pointer to the dirlock for this entry (NULL on error)
* direntflags - (case-insensitive lookup only)
* flags if multiple case-sensitive matches exist in directory
* realpnp - (case-insensitive lookup only)
* actual name matched within the directory
*
* Return value: 0 on success or errno on failure.
*
* NOTE: Always checks for, and rejects, '.' and '..'.
* NOTE: For case-insensitive file systems we take wide locks (see below),
* but return znode pointers to a single match.
*/
int
zfs_dirent_lock(zfs_dirlock_t **dlpp, znode_t *dzp, char *name, znode_t **zpp,
int flag, int *direntflags, pathname_t *realpnp)
{
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
zfs_dirlock_t *dl;
boolean_t update;
boolean_t exact;
uint64_t zoid;
vnode_t *vp = NULL;
int error = 0;
int cmpflags;
*zpp = NULL;
*dlpp = NULL;
/*
* Verify that we are not trying to lock '.', '..', or '.zfs'
*/
if ((name[0] == '.' &&
(name[1] == '\0' || (name[1] == '.' && name[2] == '\0'))) ||
(zfs_has_ctldir(dzp) && strcmp(name, ZFS_CTLDIR_NAME) == 0))
return (EEXIST);
/*
* Case sensitivity and normalization preferences are set when
* the file system is created. These are stored in the
* zfsvfs->z_case and zfsvfs->z_norm fields. These choices
* affect what vnodes can be cached in the DNLC, how we
* perform zap lookups, and the "width" of our dirlocks.
*
* A normal dirlock locks a single name. Note that with
* normalization a name can be composed multiple ways, but
* when normalized, these names all compare equal. A wide
* dirlock locks multiple names. We need these when the file
* system is supporting mixed-mode access. It is sometimes
* necessary to lock all case permutations of file name at
* once so that simultaneous case-insensitive/case-sensitive
* behaves as rationally as possible.
*/
/*
* Decide if exact matches should be requested when performing
* a zap lookup on file systems supporting case-insensitive
* access.
*/
exact =
((zfsvfs->z_case == ZFS_CASE_INSENSITIVE) && (flag & ZCIEXACT)) ||
((zfsvfs->z_case == ZFS_CASE_MIXED) && !(flag & ZCILOOK));
/*
* Only look in or update the DNLC if we are looking for the
* name on a file system that does not require normalization
* or case folding. We can also look there if we happen to be
* on a non-normalizing, mixed sensitivity file system IF we
* are looking for the exact name.
*
* Maybe can add TO-UPPERed version of name to dnlc in ci-only
* case for performance improvement?
*/
update = !zfsvfs->z_norm ||
((zfsvfs->z_case == ZFS_CASE_MIXED) &&
!(zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER) && !(flag & ZCILOOK));
/*
* ZRENAMING indicates we are in a situation where we should
* take narrow locks regardless of the file system's
* preferences for normalizing and case folding. This will
* prevent us deadlocking trying to grab the same wide lock
* twice if the two names happen to be case-insensitive
* matches.
*/
if (flag & ZRENAMING)
cmpflags = 0;
else
cmpflags = zfsvfs->z_norm;
/*
* Wait until there are no locks on this name.
*
* Don't grab the the lock if it is already held. However, cannot
* have both ZSHARED and ZHAVELOCK together.
*/
ASSERT(!(flag & ZSHARED) || !(flag & ZHAVELOCK));
if (!(flag & ZHAVELOCK))
rw_enter(&dzp->z_name_lock, RW_READER);
mutex_enter(&dzp->z_lock);
for (;;) {
if (dzp->z_unlinked) {
mutex_exit(&dzp->z_lock);
if (!(flag & ZHAVELOCK))
rw_exit(&dzp->z_name_lock);
return (ENOENT);
}
for (dl = dzp->z_dirlocks; dl != NULL; dl = dl->dl_next) {
if ((u8_strcmp(name, dl->dl_name, 0, cmpflags,
U8_UNICODE_LATEST, &error) == 0) || error != 0)
break;
}
if (error != 0) {
mutex_exit(&dzp->z_lock);
if (!(flag & ZHAVELOCK))
rw_exit(&dzp->z_name_lock);
return (ENOENT);
}
if (dl == NULL) {
/*
* Allocate a new dirlock and add it to the list.
*/
dl = kmem_alloc(sizeof (zfs_dirlock_t), KM_SLEEP);
cv_init(&dl->dl_cv, NULL, CV_DEFAULT, NULL);
dl->dl_name = name;
dl->dl_sharecnt = 0;
dl->dl_namelock = 0;
dl->dl_namesize = 0;
dl->dl_dzp = dzp;
dl->dl_next = dzp->z_dirlocks;
dzp->z_dirlocks = dl;
break;
}
if ((flag & ZSHARED) && dl->dl_sharecnt != 0)
break;
cv_wait(&dl->dl_cv, &dzp->z_lock);
}
/*
* If the z_name_lock was NOT held for this dirlock record it.
*/
if (flag & ZHAVELOCK)
dl->dl_namelock = 1;
if ((flag & ZSHARED) && ++dl->dl_sharecnt > 1 && dl->dl_namesize == 0) {
/*
* We're the second shared reference to dl. Make a copy of
* dl_name in case the first thread goes away before we do.
* Note that we initialize the new name before storing its
* pointer into dl_name, because the first thread may load
* dl->dl_name at any time. He'll either see the old value,
* which is his, or the new shared copy; either is OK.
*/
dl->dl_namesize = strlen(dl->dl_name) + 1;
name = kmem_alloc(dl->dl_namesize, KM_SLEEP);
bcopy(dl->dl_name, name, dl->dl_namesize);
dl->dl_name = name;
}
mutex_exit(&dzp->z_lock);
/*
* We have a dirlock on the name. (Note that it is the dirlock,
* not the dzp's z_lock, that protects the name in the zap object.)
* See if there's an object by this name; if so, put a hold on it.
*/
if (flag & ZXATTR) {
error = sa_lookup(dzp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &zoid,
sizeof (zoid));
if (error == 0)
error = (zoid == 0 ? ENOENT : 0);
} else {
if (update)
vp = dnlc_lookup(ZTOV(dzp), name);
if (vp == DNLC_NO_VNODE) {
VN_RELE(vp);
error = ENOENT;
} else if (vp) {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
VN_RELE(vp);
return (EEXIST);
}
*dlpp = dl;
*zpp = VTOZ(vp);
return (0);
} else {
error = zfs_match_find(zfsvfs, dzp, name, exact,
update, direntflags, realpnp, &zoid);
}
}
if (error) {
if (error != ENOENT || (flag & ZEXISTS)) {
zfs_dirent_unlock(dl);
return (error);
}
} else {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
return (EEXIST);
}
error = zfs_zget(zfsvfs, zoid, zpp);
if (error) {
zfs_dirent_unlock(dl);
return (error);
}
if (!(flag & ZXATTR) && update)
dnlc_update(ZTOV(dzp), name, ZTOV(*zpp));
}
*dlpp = dl;
return (0);
}
int
zfs_getattr_znode_unlocked(struct vnode *vp, vattr_t *vap)
{
znode_t *zp = VTOZ(vp);
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
int error = 0;
uint64_t parent;
//printf("getattr_osx\n");
ZFS_ENTER(zfsvfs);
/*
* On Mac OS X we always export the root directory id as 2
*/
vap->va_fileid = (zp->z_id == zfsvfs->z_root) ? 2 : zp->z_id;
//vap->va_fileid = (zp->z_id == zfsvfs->z_root) ? 2 : zp->z_vid;
vap->va_nlink = zp->z_links;
vap->va_data_size = zp->z_size;
vap->va_total_size = zp->z_size;
vap->va_gen = zp->z_gen;
/*
* For Carbon compatibility,pretend to support this legacy/unused attribute
*/
if (VATTR_IS_ACTIVE(vap, va_backup_time)) {
vap->va_backup_time.tv_sec = 0;
vap->va_backup_time.tv_nsec = 0;
VATTR_SET_SUPPORTED(vap, va_backup_time);
}
vap->va_flags = zfs_getbsdflags(zp);
/*
* On Mac OS X we always export the root directory id as 2
* and its parent as 1
*/
error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (parent));
if (!error) {
if (zp->z_id == zfsvfs->z_root)
vap->va_parentid = 1;
else if (parent == zfsvfs->z_root)
vap->va_parentid = 2;
else
vap->va_parentid = parent;
}
vap->va_iosize = zp->z_blksz ? zp->z_blksz : zfsvfs->z_max_blksz;
//vap->va_iosize = 512;
VATTR_SET_SUPPORTED(vap, va_iosize);
/* Don't include '.' and '..' in the number of entries */
if (VATTR_IS_ACTIVE(vap, va_nchildren) && vnode_isdir(vp)) {
VATTR_RETURN(vap, va_nchildren, vap->va_nlink - 2);
}
/*
* va_dirlinkcount is the count of directory hard links. When a file
* system does not support ATTR_DIR_LINKCOUNT, xnu will default to 1.
* Since we claim to support ATTR_DIR_LINKCOUNT both as valid and as
* native, we'll just return 1. We set 1 for this value in dirattrpack
* as well. If in the future ZFS actually supports directory hard links,
* we can return a real value.
*/
if (VATTR_IS_ACTIVE(vap, va_dirlinkcount) && vnode_isdir(vp)) {
VATTR_RETURN(vap, va_dirlinkcount, 1);
}
if (VATTR_IS_ACTIVE(vap, va_acl)) {
//printf("want acl\n");
#if 0
zfs_acl_phys_t acl;
if (sa_lookup(zp->z_sa_hdl, SA_ZPL_ZNODE_ACL(zfsvfs),
&acl, sizeof (zfs_acl_phys_t))) {
//if (zp->z_acl.z_acl_count == 0) {
vap->va_acl = (kauth_acl_t) KAUTH_FILESEC_NONE;
} else {
if ((error = zfs_getacl(zp, &vap->va_acl, B_TRUE, NULL))) {
dprintf("zfs_getacl returned error %d\n", error);
error = 0;
//ZFS_EXIT(zfsvfs);
//return (error);
}
}
#endif
//VATTR_SET_SUPPORTED(vap, va_acl);
VATTR_RETURN(vap, va_uuuid, kauth_null_guid);
VATTR_RETURN(vap, va_guuid, kauth_null_guid);
dprintf("Calling getacl\n");
if ((error = zfs_getacl(zp, &vap->va_acl, B_FALSE, NULL))) {
dprintf("zfs_getacl returned error %d\n", error);
error = 0;
} else {
VATTR_SET_SUPPORTED(vap, va_acl);
/* va_acl implies that va_uuuid and va_guuid are also supported. */
VATTR_RETURN(vap, va_uuuid, kauth_null_guid);
VATTR_RETURN(vap, va_guuid, kauth_null_guid);
}
}
if (VATTR_IS_ACTIVE(vap, va_data_alloc) || VATTR_IS_ACTIVE(vap, va_total_alloc)) {
uint32_t blksize;
u_longlong_t nblks;
sa_object_size(zp->z_sa_hdl, &blksize, &nblks);
vap->va_data_alloc = (uint64_t)512LL * (uint64_t)nblks;
vap->va_total_alloc = vap->va_data_alloc;
vap->va_supported |= VNODE_ATTR_va_data_alloc |
VNODE_ATTR_va_total_alloc;
}
if (VATTR_IS_ACTIVE(vap, va_name)) {
vap->va_name[0] = 0;
if (!vnode_isvroot(vp)) {
/* Lets not supply name as zap_cursor can cause panic */
#if 0
if (zap_value_search(zfsvfs->z_os, parent, zp->z_id,
ZFS_DIRENT_OBJ(-1ULL), vap->va_name) == 0)
VATTR_SET_SUPPORTED(vap, va_name);
#endif
} else {
/*
* The vroot objects must return a unique name for Finder to
* be able to distringuish between mounts. For this reason
* we simply return the fullname, from the statfs mountedfrom
*/
strlcpy(vap->va_name,
vfs_statfs(vnode_mount(vp))->f_mntfromname,
MAXPATHLEN);
VATTR_SET_SUPPORTED(vap, va_name);
}
}
if (VATTR_IS_ACTIVE(vap, va_filerev)) {
VATTR_RETURN(vap, va_filerev, 0);
}
if (VATTR_IS_ACTIVE(vap, va_linkid)) {
VATTR_RETURN(vap, va_linkid, vap->va_fileid);
}
if (VATTR_IS_ACTIVE(vap, va_fsid)) {
VATTR_RETURN(vap, va_fsid, vfs_statfs(zfsvfs->z_vfs)->f_fsid.val[0]);
}
if (VATTR_IS_ACTIVE(vap, va_type)) {
VATTR_RETURN(vap, va_type, vnode_vtype(ZTOV(zp)));
}
if (VATTR_IS_ACTIVE(vap, va_encoding)) {
VATTR_RETURN(vap, va_encoding, kTextEncodingMacUnicode);
}
#ifdef VNODE_ATTR_va_addedtime
if (VATTR_IS_ACTIVE(vap, va_addedtime)) {
VATTR_RETURN(vap, va_addedtime, vap->va_ctime);
}
#endif
if (VATTR_IS_ACTIVE(vap, va_uuuid)) {
kauth_cred_uid2guid(zp->z_uid, &vap->va_uuuid);
}
if (VATTR_IS_ACTIVE(vap, va_guuid)) {
kauth_cred_uid2guid(zp->z_gid, &vap->va_guuid);
}
vap->va_supported |= ZFS_SUPPORTED_VATTRS;
ZFS_EXIT(zfsvfs);
return (error);
}
static int
zfs_vfs_getattr(struct mount *mp, struct vfs_attr *fsap, __unused vfs_context_t context)
{
zfsvfs_t *zfsvfs = vfs_fsprivate(mp);
uint64_t refdbytes, availbytes, usedobjs, availobjs;
ZFS_ENTER(zfsvfs);
dmu_objset_space(zfsvfs->z_os,
&refdbytes, &availbytes, &usedobjs, &availobjs);
VFSATTR_RETURN(fsap, f_objcount, usedobjs);
VFSATTR_RETURN(fsap, f_maxobjcount, 0x7fffffffffffffff);
/*
* Carbon depends on f_filecount and f_dircount so
* make up some values based on total objects.
*/
VFSATTR_RETURN(fsap, f_filecount, usedobjs - (usedobjs / 4));
VFSATTR_RETURN(fsap, f_dircount, usedobjs / 4);
/*
* The underlying storage pool actually uses multiple block sizes.
* We report the fragsize as the smallest block size we support,
* and we report our blocksize as the filesystem's maximum blocksize.
*/
VFSATTR_RETURN(fsap, f_bsize, 1UL << SPA_MINBLOCKSHIFT);
VFSATTR_RETURN(fsap, f_iosize, zfsvfs->z_max_blksz);
/*
* The following report "total" blocks of various kinds in the
* file system, but reported in terms of f_frsize - the
* "fragment" size.
*/
VFSATTR_RETURN(fsap, f_blocks,
(u_int64_t)((refdbytes + availbytes) >> SPA_MINBLOCKSHIFT));
VFSATTR_RETURN(fsap, f_bfree, (u_int64_t)(availbytes >> SPA_MINBLOCKSHIFT));
VFSATTR_RETURN(fsap, f_bavail, fsap->f_bfree); /* no root reservation */
VFSATTR_RETURN(fsap, f_bused, fsap->f_blocks - fsap->f_bfree);
/*
* statvfs() should really be called statufs(), because it assumes
* static metadata. ZFS doesn't preallocate files, so the best
* we can do is report the max that could possibly fit in f_files,
* and that minus the number actually used in f_ffree.
* For f_ffree, report the smaller of the number of object available
* and the number of blocks (each object will take at least a block).
*/
VFSATTR_RETURN(fsap, f_ffree, (u_int64_t)MIN(availobjs, fsap->f_bfree));
VFSATTR_RETURN(fsap, f_files, fsap->f_ffree + usedobjs);
#if 0
statp->f_flag = vf_to_stf(vfsp->vfs_flag);
#endif
if (VFSATTR_IS_ACTIVE(fsap, f_fsid)) {
VFSATTR_RETURN(fsap, f_fsid, vfs_statfs(mp)->f_fsid);
}
if (VFSATTR_IS_ACTIVE(fsap, f_capabilities)) {
bcopy(&zfs_capabilities, &fsap->f_capabilities, sizeof (zfs_capabilities));
VFSATTR_SET_SUPPORTED(fsap, f_capabilities);
}
if (VFSATTR_IS_ACTIVE(fsap, f_attributes)) {
bcopy(&zfs_attributes, &fsap->f_attributes.validattr, sizeof (zfs_attributes));
bcopy(&zfs_attributes, &fsap->f_attributes.nativeattr, sizeof (zfs_attributes));
VFSATTR_SET_SUPPORTED(fsap, f_attributes);
}
if (VFSATTR_IS_ACTIVE(fsap, f_create_time)) {
dmu_objset_stats_t dmu_stat;
dmu_objset_fast_stat(zfsvfs->z_os, &dmu_stat);
fsap->f_create_time.tv_sec = dmu_stat.dds_creation_time;
fsap->f_create_time.tv_nsec = 0;
VFSATTR_SET_SUPPORTED(fsap, f_create_time);
}
if (VFSATTR_IS_ACTIVE(fsap, f_modify_time)) {
if (zfsvfs->z_mtime_vp != NULL) {
znode_t *mzp;
mzp = VTOZ(zfsvfs->z_mtime_vp);
ZFS_TIME_DECODE(&fsap->f_modify_time, mzp->z_phys->zp_mtime);
} else {
fsap->f_modify_time.tv_sec = 0;
fsap->f_modify_time.tv_nsec = 0;
}
VFSATTR_SET_SUPPORTED(fsap, f_modify_time);
}
/*
* For Carbon compatibility, pretend to support this legacy/unused
* attribute
*/
if (VFSATTR_IS_ACTIVE(fsap, f_backup_time)) {
fsap->f_backup_time.tv_sec = 0;
fsap->f_backup_time.tv_nsec = 0;
VFSATTR_SET_SUPPORTED(fsap, f_backup_time);
}
if (VFSATTR_IS_ACTIVE(fsap, f_vol_name)) {
spa_t *spa = dmu_objset_spa(zfsvfs->z_os);
spa_config_enter(spa, RW_READER, FTAG);
strlcpy(fsap->f_vol_name, spa_name(spa), MAXPATHLEN);
spa_config_exit(spa, FTAG);
VFSATTR_SET_SUPPORTED(fsap, f_vol_name);
}
VFSATTR_RETURN(fsap, f_fssubtype, 0);
VFSATTR_RETURN(fsap, f_signature, 0x5a21); /* 'Z!' */
VFSATTR_RETURN(fsap, f_carbon_fsid, 0);
ZFS_EXIT(zfsvfs);
return (0);
}
static int
zfs_vfs_mount(struct mount *mp, vnode_t devvp, user_addr_t data, vfs_context_t context)
{
char *osname = NULL;
size_t osnamelen = 0;
int error = 0;
int canwrite;
/*
* Get the objset name (the "special" mount argument).
* The filesystem that we mount as root is defined in the
* "zfs-bootfs" property.
*/
if (data) {
user_addr_t fspec = USER_ADDR_NULL;
#ifndef __APPLE__
if (ddi_prop_lookup_string(DDI_DEV_T_ANY, ddi_root_node(),
DDI_PROP_DONTPASS, "zfs-bootfs", &zfs_bootpath) !=
DDI_SUCCESS)
return (EIO);
error = parse_bootpath(zfs_bootpath, rootfs.bo_name);
ddi_prop_free(zfs_bootpath);
#endif
osname = kmem_alloc(MAXPATHLEN, KM_SLEEP);
if (vfs_context_is64bit(context)) {
if ( (error = copyin(data, (caddr_t)&fspec, sizeof(fspec))) )
goto out;
} else {
#ifdef ZFS_LEOPARD_ONLY
char *tmp;
#else
user32_addr_t tmp;
#endif
if ( (error = copyin(data, (caddr_t)&tmp, sizeof(tmp))) )
goto out;
/* munge into LP64 addr */
fspec = CAST_USER_ADDR_T(tmp);
}
if ( (error = copyinstr(fspec, osname, MAXPATHLEN, &osnamelen)) )
goto out;
}
#if 0
if (mvp->v_type != VDIR)
return (ENOTDIR);
mutex_enter(&mvp->v_lock);
if ((uap->flags & MS_REMOUNT) == 0 &&
(uap->flags & MS_OVERLAY) == 0 &&
(mvp->v_count != 1 || (mvp->v_flag & VROOT))) {
mutex_exit(&mvp->v_lock);
return (EBUSY);
}
mutex_exit(&mvp->v_lock);
/*
* ZFS does not support passing unparsed data in via MS_DATA.
* Users should use the MS_OPTIONSTR interface; this means
* that all option parsing is already done and the options struct
* can be interrogated.
*/
if ((uap->flags & MS_DATA) && uap->datalen > 0)
return (EINVAL);
/*
* Get the objset name (the "special" mount argument).
*/
if (error = pn_get(uap->spec, fromspace, &spn))
return (error);
osname = spn.pn_path;
#endif
/*
* Check for mount privilege?
*
* If we don't have privilege then see if
* we have local permission to allow it
*/
#ifndef __APPLE__
error = secpolicy_fs_mount(cr, mvp, vfsp);
if (error) {
error = dsl_deleg_access(osname, ZFS_DELEG_PERM_MOUNT, cr);
if (error == 0) {
vattr_t vattr;
/*
* Make sure user is the owner of the mount point
* or has sufficient privileges.
*/
vattr.va_mask = AT_UID;
if (error = VOP_GETATTR(mvp, &vattr, 0, cr)) {
goto out;
}
if (error = secpolicy_vnode_owner(cr, vattr.va_uid)) {
goto out;
}
if (error = VOP_ACCESS(mvp, VWRITE, 0, cr)) {
goto out;
}
secpolicy_fs_mount_clearopts(cr, vfsp);
} else {
goto out;
}
}
#endif
error = zfs_domount(mp, 0, osname, context);
if (error)
printf("zfs_vfs_mount: error %d\n", error);
if (error == 0) {
zfsvfs_t *zfsvfs = NULL;
/* Make the Finder treat sub file systems just like a folder */
if (strpbrk(osname, "/"))
vfs_setflags(mp, (u_int64_t)((unsigned int)MNT_DONTBROWSE));
/* Indicate to VFS that we support ACLs. */
vfs_setextendedsecurity(mp);
/* Advisory locking should be handled at the VFS layer */
vfs_setlocklocal(mp);
/*
* Mac OS X needs a file system modify time
*
* We use the mtime of the "com.apple.system.mtime"
* extended attribute, which is associated with the
* file system root directory.
*
* Here we need to take a ref on z_mtime_vp to keep it around.
* If the attribute isn't there, attempt to create it.
*/
zfsvfs = vfs_fsprivate(mp);
if (zfsvfs->z_mtime_vp == NULL) {
struct vnode * rvp;
struct vnode *xdvp = NULLVP;
struct vnode *xvp = NULLVP;
znode_t *rootzp;
timestruc_t modify_time;
cred_t *cr;
timestruc_t now;
int flag;
int result;
if (zfs_zget(zfsvfs, zfsvfs->z_root, &rootzp) != 0) {
goto out;
}
rvp = ZTOV(rootzp);
cr = (cred_t *)vfs_context_ucred(context);
/* Grab the hidden attribute directory vnode. */
result = zfs_get_xattrdir(rootzp, &xdvp, cr, CREATE_XATTR_DIR);
vnode_put(rvp); /* all done with root vnode */
rvp = NULL;
if (result) {
goto out;
}
/*
* HACK - workaround missing vnode_setnoflush() KPI...
*
* We tag zfsvfs so that zfs_attach_vnode() can then set
* vnfs_marksystem when the vnode gets created.
*/
zfsvfs->z_last_unmount_time = 0xBADC0DE;
zfsvfs->z_last_mtime_synced = VTOZ(xdvp)->z_id;
flag = vfs_isrdonly(mp) ? 0 : ZEXISTS;
/* Lookup or create the named attribute. */
if ( zfs_obtain_xattr(VTOZ(xdvp), ZFS_MTIME_XATTR,
S_IRUSR | S_IWUSR, cr, &xvp,
flag) ) {
zfsvfs->z_last_unmount_time = 0;
zfsvfs->z_last_mtime_synced = 0;
vnode_put(xdvp);
goto out;
}
gethrestime(&now);
ZFS_TIME_ENCODE(&now, VTOZ(xvp)->z_phys->zp_mtime);
vnode_put(xdvp);
vnode_ref(xvp);
zfsvfs->z_mtime_vp = xvp;
ZFS_TIME_DECODE(&modify_time, VTOZ(xvp)->z_phys->zp_mtime);
zfsvfs->z_last_unmount_time = modify_time.tv_sec;
zfsvfs->z_last_mtime_synced = modify_time.tv_sec;
/*
* Keep this referenced vnode from impeding an unmount.
*
* XXX vnode_setnoflush() is MIA from KPI (see workaround above).
*/
#if 0
vnode_setnoflush(xvp);
#endif
vnode_put(xvp);
}
}
out:
if (osname) {
kmem_free(osname, MAXPATHLEN);
}
return (error);
}
/*
* Lock a directory entry. A dirlock on <dzp, name> protects that name
* in dzp's directory zap object. As long as you hold a dirlock, you can
* assume two things: (1) dzp cannot be reaped, and (2) no other thread
* can change the zap entry for (i.e. link or unlink) this name.
*
* Input arguments:
* dzp - znode for directory
* name - name of entry to lock
* flag - ZNEW: if the entry already exists, fail with EEXIST.
* ZEXISTS: if the entry does not exist, fail with ENOENT.
* ZSHARED: allow concurrent access with other ZSHARED callers.
* ZXATTR: we want dzp's xattr directory
*
* Output arguments:
* zpp - pointer to the znode for the entry (NULL if there isn't one)
* dlpp - pointer to the dirlock for this entry (NULL on error)
*
* Return value: 0 on success or errno on failure.
*
* NOTE: Always checks for, and rejects, '.' and '..'.
*/
int
zfs_dirent_lock(zfs_dirlock_t **dlpp, znode_t *dzp, char *name, znode_t **zpp,
int flag)
{
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
zfs_dirlock_t *dl;
uint64_t zoid;
int error;
vnode_t *vp;
*zpp = NULL;
*dlpp = NULL;
/*
* Verify that we are not trying to lock '.', '..', or '.zfs'
*/
if (name[0] == '.' &&
(name[1] == '\0' || (name[1] == '.' && name[2] == '\0')) ||
zfs_has_ctldir(dzp) && strcmp(name, ZFS_CTLDIR_NAME) == 0)
return (EEXIST);
/*
* Wait until there are no locks on this name.
*/
mutex_enter(&dzp->z_lock);
for (;;) {
if (dzp->z_reap) {
mutex_exit(&dzp->z_lock);
return (ENOENT);
}
for (dl = dzp->z_dirlocks; dl != NULL; dl = dl->dl_next)
if (strcmp(name, dl->dl_name) == 0)
break;
if (dl == NULL) {
/*
* Allocate a new dirlock and add it to the list.
*/
dl = kmem_alloc(sizeof (zfs_dirlock_t), KM_SLEEP);
cv_init(&dl->dl_cv, NULL, CV_DEFAULT, NULL);
dl->dl_name = name;
dl->dl_sharecnt = 0;
dl->dl_namesize = 0;
dl->dl_dzp = dzp;
dl->dl_next = dzp->z_dirlocks;
dzp->z_dirlocks = dl;
break;
}
if ((flag & ZSHARED) && dl->dl_sharecnt != 0)
break;
cv_wait(&dl->dl_cv, &dzp->z_lock);
}
if ((flag & ZSHARED) && ++dl->dl_sharecnt > 1 && dl->dl_namesize == 0) {
/*
* We're the second shared reference to dl. Make a copy of
* dl_name in case the first thread goes away before we do.
* Note that we initialize the new name before storing its
* pointer into dl_name, because the first thread may load
* dl->dl_name at any time. He'll either see the old value,
* which is his, or the new shared copy; either is OK.
*/
dl->dl_namesize = strlen(dl->dl_name) + 1;
name = kmem_alloc(dl->dl_namesize, KM_SLEEP);
bcopy(dl->dl_name, name, dl->dl_namesize);
dl->dl_name = name;
}
mutex_exit(&dzp->z_lock);
/*
* We have a dirlock on the name. (Note that it is the dirlock,
* not the dzp's z_lock, that protects the name in the zap object.)
* See if there's an object by this name; if so, put a hold on it.
*/
if (flag & ZXATTR) {
zoid = dzp->z_phys->zp_xattr;
error = (zoid == 0 ? ENOENT : 0);
} else {
vp = dnlc_lookup(ZTOV(dzp), name);
if (vp == DNLC_NO_VNODE) {
VN_RELE(vp);
error = ENOENT;
} else if (vp) {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
VN_RELE(vp);
return (EEXIST);
}
*dlpp = dl;
*zpp = VTOZ(vp);
return (0);
} else {
error = zap_lookup(zfsvfs->z_os, dzp->z_id, name,
8, 1, &zoid);
if (error == ENOENT)
dnlc_update(ZTOV(dzp), name, DNLC_NO_VNODE);
}
}
if (error) {
if (error != ENOENT || (flag & ZEXISTS)) {
zfs_dirent_unlock(dl);
return (error);
}
} else {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
return (EEXIST);
}
error = zfs_zget(zfsvfs, zoid, zpp);
if (error) {
zfs_dirent_unlock(dl);
return (error);
}
if (!(flag & ZXATTR))
dnlc_update(ZTOV(dzp), name, ZTOV(*zpp));
}
*dlpp = dl;
return (0);
}
/* ARGSUSED */
static int
zfsctl_snapdir_lookup(vnode_t *dvp, char *nm, vnode_t **vpp, pathname_t *pnp,
int flags, vnode_t *rdir, cred_t *cr, caller_context_t *ct,
int *direntflags, pathname_t *realpnp)
{
zfsctl_snapdir_t *sdp = dvp->v_data;
objset_t *snap;
char snapname[MAXNAMELEN];
char real[MAXNAMELEN];
char *mountpoint;
zfs_snapentry_t *sep, search;
struct mounta margs;
vfs_t *vfsp;
size_t mountpoint_len;
avl_index_t where;
zfsvfs_t *zfsvfs = dvp->v_vfsp->vfs_data;
int err;
/*
* No extended attributes allowed under .zfs
*/
if (flags & LOOKUP_XATTR)
return (EINVAL);
ASSERT(dvp->v_type == VDIR);
/*
* If we get a recursive call, that means we got called
* from the domount() code while it was trying to look up the
* spec (which looks like a local path for zfs). We need to
* add some flag to domount() to tell it not to do this lookup.
*/
if (MUTEX_HELD(&sdp->sd_lock))
return (ENOENT);
ZFS_ENTER(zfsvfs);
if (gfs_lookup_dot(vpp, dvp, zfsvfs->z_ctldir, nm) == 0) {
ZFS_EXIT(zfsvfs);
return (0);
}
if (flags & FIGNORECASE) {
boolean_t conflict = B_FALSE;
err = dmu_snapshot_realname(zfsvfs->z_os, nm, real,
MAXNAMELEN, &conflict);
if (err == 0) {
nm = real;
} else if (err != ENOTSUP) {
ZFS_EXIT(zfsvfs);
return (err);
}
if (realpnp)
(void) strlcpy(realpnp->pn_buf, nm,
realpnp->pn_bufsize);
if (conflict && direntflags)
*direntflags = ED_CASE_CONFLICT;
}
mutex_enter(&sdp->sd_lock);
search.se_name = (char *)nm;
if ((sep = avl_find(&sdp->sd_snaps, &search, &where)) != NULL) {
*vpp = sep->se_root;
VN_HOLD(*vpp);
err = traverse(vpp);
if (err) {
VN_RELE(*vpp);
*vpp = NULL;
} else if (*vpp == sep->se_root) {
/*
* The snapshot was unmounted behind our backs,
* try to remount it.
*/
goto domount;
} else {
/*
* VROOT was set during the traverse call. We need
* to clear it since we're pretending to be part
* of our parent's vfs.
*/
(*vpp)->v_flag &= ~VROOT;
}
mutex_exit(&sdp->sd_lock);
ZFS_EXIT(zfsvfs);
return (err);
}
/*
* The requested snapshot is not currently mounted, look it up.
*/
err = zfsctl_snapshot_zname(dvp, nm, MAXNAMELEN, snapname);
if (err) {
mutex_exit(&sdp->sd_lock);
ZFS_EXIT(zfsvfs);
/*
* handle "ls *" or "?" in a graceful manner,
* forcing EILSEQ to ENOENT.
* Since shell ultimately passes "*" or "?" as name to lookup
*/
return (err == EILSEQ ? ENOENT : err);
}
if (dmu_objset_hold(snapname, FTAG, &snap) != 0) {
mutex_exit(&sdp->sd_lock);
ZFS_EXIT(zfsvfs);
return (ENOENT);
}
sep = kmem_alloc(sizeof (zfs_snapentry_t), KM_SLEEP);
sep->se_name = kmem_alloc(strlen(nm) + 1, KM_SLEEP);
(void) strcpy(sep->se_name, nm);
*vpp = sep->se_root = zfsctl_snapshot_mknode(dvp, dmu_objset_id(snap));
avl_insert(&sdp->sd_snaps, sep, where);
dmu_objset_rele(snap, FTAG);
domount:
mountpoint_len = strlen(refstr_value(dvp->v_vfsp->vfs_mntpt)) +
strlen("/.zfs/snapshot/") + strlen(nm) + 1;
mountpoint = kmem_alloc(mountpoint_len, KM_SLEEP);
(void) snprintf(mountpoint, mountpoint_len, "%s/.zfs/snapshot/%s",
refstr_value(dvp->v_vfsp->vfs_mntpt), nm);
margs.spec = snapname;
margs.dir = mountpoint;
margs.flags = MS_SYSSPACE | MS_NOMNTTAB;
margs.fstype = "zfs";
margs.dataptr = NULL;
margs.datalen = 0;
margs.optptr = NULL;
margs.optlen = 0;
err = domount("zfs", &margs, *vpp, kcred, &vfsp);
kmem_free(mountpoint, mountpoint_len);
if (err == 0) {
/*
* Return the mounted root rather than the covered mount point.
* Takes the GFS vnode at .zfs/snapshot/<snapname> and returns
* the ZFS vnode mounted on top of the GFS node. This ZFS
* vnode is the root of the newly created vfsp.
*/
VFS_RELE(vfsp);
err = traverse(vpp);
}
if (err == 0) {
/*
* Fix up the root vnode mounted on .zfs/snapshot/<snapname>.
*
* This is where we lie about our v_vfsp in order to
* make .zfs/snapshot/<snapname> accessible over NFS
* without requiring manual mounts of <snapname>.
*/
ASSERT(VTOZ(*vpp)->z_zfsvfs != zfsvfs);
VTOZ(*vpp)->z_zfsvfs->z_parent = zfsvfs;
(*vpp)->v_vfsp = zfsvfs->z_vfs;
(*vpp)->v_flag &= ~VROOT;
}
mutex_exit(&sdp->sd_lock);
ZFS_EXIT(zfsvfs);
/*
* If we had an error, drop our hold on the vnode and
* zfsctl_snapshot_inactive() will clean up.
*/
if (err) {
VN_RELE(*vpp);
*vpp = NULL;
}
return (err);
}
void fileattrpack(attrinfo_t *aip, zfsvfs_t *zfsvfs, znode_t *zp)
{
attrgroup_t fileattr = aip->ai_attrlist->fileattr;
void *attrbufptr = *aip->ai_attrbufpp;
void *varbufptr = *aip->ai_varbufpp;
uint64_t allocsize = 0;
cred_t *cr = (cred_t *)vfs_context_ucred(aip->ai_context);
if ((ATTR_FILE_ALLOCSIZE | ATTR_FILE_DATAALLOCSIZE) & fileattr && zp) {
uint32_t blksize;
u_longlong_t nblks;
sa_object_size(zp->z_sa_hdl, &blksize, &nblks);
allocsize = (uint64_t)512LL * (uint64_t)nblks;
}
if (ATTR_FILE_LINKCOUNT & fileattr) {
uint64_t val;
VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
&val, sizeof(val)) == 0);
*((u_int32_t *)attrbufptr) = val;
attrbufptr = ((u_int32_t *)attrbufptr) + 1;
}
if (ATTR_FILE_TOTALSIZE & fileattr) {
uint64_t val;
VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
&val, sizeof(val)) == 0);
*((off_t *)attrbufptr) = val;
attrbufptr = ((off_t *)attrbufptr) + 1;
}
if (ATTR_FILE_ALLOCSIZE & fileattr) {
*((off_t *)attrbufptr) = allocsize;
attrbufptr = ((off_t *)attrbufptr) + 1;
}
if (ATTR_FILE_IOBLOCKSIZE & fileattr && zp) {
*((u_int32_t *)attrbufptr) =
zp->z_blksz ? zp->z_blksz : zfsvfs->z_max_blksz;
attrbufptr = ((u_int32_t *)attrbufptr) + 1;
}
if (ATTR_FILE_DEVTYPE & fileattr) {
uint64_t mode, val=0;
VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs),
&mode, sizeof(mode)) == 0);
sa_lookup(zp->z_sa_hdl, SA_ZPL_RDEV(zfsvfs),
&val, sizeof(val));
if (S_ISBLK(mode) || S_ISCHR(mode))
*((u_int32_t *)attrbufptr) = (u_int32_t)val;
else
*((u_int32_t *)attrbufptr) = 0;
attrbufptr = ((u_int32_t *)attrbufptr) + 1;
}
if (ATTR_FILE_DATALENGTH & fileattr) {
uint64_t val;
VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
&val, sizeof(val)) == 0);
*((off_t *)attrbufptr) = val;
attrbufptr = ((off_t *)attrbufptr) + 1;
}
if (ATTR_FILE_DATAALLOCSIZE & fileattr) {
*((off_t *)attrbufptr) = allocsize;
attrbufptr = ((off_t *)attrbufptr) + 1;
}
if ((ATTR_FILE_RSRCLENGTH | ATTR_FILE_RSRCALLOCSIZE) & fileattr) {
uint64_t rsrcsize = 0;
uint64_t xattr;
if (!sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr, sizeof(xattr)) &&
xattr) {
vnode_t *xdvp = NULLVP;
vnode_t *xvp = NULLVP;
struct componentname cn;
bzero(&cn, sizeof (cn));
cn.cn_nameiop = LOOKUP;
cn.cn_flags = ISLASTCN;
cn.cn_nameptr = XATTR_RESOURCEFORK_NAME;
cn.cn_namelen = strlen(cn.cn_nameptr);
/* Grab the hidden attribute directory vnode. */
if (zfs_get_xattrdir(zp, &xdvp, cr, 0) == 0 &&
zfs_dirlook(VTOZ(xdvp), cn.cn_nameptr, &xvp, 0, NULL,
&cn) == 0) {
rsrcsize = VTOZ(xvp)->z_size;
}
if (xvp)
vnode_put(xvp);
if (xdvp)
vnode_put(xdvp);
}
if (ATTR_FILE_RSRCLENGTH & fileattr) {
*((off_t *)attrbufptr) = rsrcsize;
attrbufptr = ((off_t *)attrbufptr) + 1;
}
if (ATTR_FILE_RSRCALLOCSIZE & fileattr) {
*((off_t *)attrbufptr) = roundup(rsrcsize, 512);
attrbufptr = ((off_t *)attrbufptr) + 1;
}
}
*aip->ai_attrbufpp = attrbufptr;
*aip->ai_varbufpp = varbufptr;
}
zfs_dirent_lock(zfs_dirlock_t **dlpp, znode_t *dzp, char *name, znode_t **zpp,
int flag)
#endif
{
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
zfs_dirlock_t *dl;
uint64_t zoid;
int error;
vnode_t *vp;
#ifdef __APPLE__
char *name;
u_int8_t *nfc_name = NULL; /* NFC form of name */
int nfc_namesize = 0;
#endif
*zpp = NULL;
*dlpp = NULL;
#ifdef __APPLE__
/* Note: cnp will be NULL for ZXATTR case */
name = cnp ? cnp->cn_nameptr : "";
if (cnp)
ASSERT(name[cnp->cn_namelen] == '\0');
#endif
/*
* Verify that we are not trying to lock '.', '..', or '.zfs'
*/
if ((name[0] == '.') &&
((name[1] == '\0') || ((name[1] == '.') && (name[2] == '\0'))) ||
zfs_has_ctldir(dzp) && strcmp(name, ZFS_CTLDIR_NAME) == 0)
return (EEXIST);
#ifdef __APPLE__
/*
* Mac OS X: store non-ascii names in UTF-8 NFC (pre-composed) on disk.
*
* The NFC name ptr is stored in dl->dl_name (allocated here)
* and its freed by zfs_dirent_unlock (since dl_namesize != 0).
*
* Since NFC size will not expand, we can allocate the same sized buffer.
*/
if (!is_ascii_str(name)) {
size_t outlen;
nfc_namesize = strlen(name) + 1;
nfc_name = kmem_alloc(nfc_namesize, KM_SLEEP);
if (utf8_normalizestr((const u_int8_t *)name, nfc_namesize, nfc_name,
&outlen, nfc_namesize, UTF_PRECOMPOSED) == 0) {
/* Normalization succeeded, switch to NFC name. */
name = (char *)nfc_name;
} else {
/* Normalization failed, just use input name as-is. */
kmem_free(nfc_name, nfc_namesize);
nfc_name = NULL;
}
}
#endif
/*
* Wait until there are no locks on this name.
*/
rw_enter(&dzp->z_name_lock, RW_READER);
mutex_enter(&dzp->z_lock);
for (;;) {
if (dzp->z_unlinked) {
mutex_exit(&dzp->z_lock);
rw_exit(&dzp->z_name_lock);
#ifdef __APPLE__
/* Release any unused NFC name before returning */
if (nfc_name) {
kmem_free(nfc_name, nfc_namesize);
}
#endif
return (ENOENT);
}
for (dl = dzp->z_dirlocks; dl != NULL; dl = dl->dl_next)
if (strcmp(name, dl->dl_name) == 0)
break;
if (dl == NULL) {
/*
* Allocate a new dirlock and add it to the list.
*/
dl = kmem_alloc(sizeof (zfs_dirlock_t), KM_SLEEP);
cv_init(&dl->dl_cv, NULL, CV_DEFAULT, NULL);
dl->dl_name = name;
dl->dl_sharecnt = 0;
dl->dl_namesize = 0;
dl->dl_dzp = dzp;
dl->dl_next = dzp->z_dirlocks;
dzp->z_dirlocks = dl;
#ifdef __APPLE__
/*
* Keep the NFC name around in dir lock by tagging it
* (setting nfc_namesize).
*/
if (nfc_name) {
dl->dl_namesize = nfc_namesize;
nfc_name = NULL; /* its now part of the dir lock */
}
#endif
break;
}
if ((flag & ZSHARED) && dl->dl_sharecnt != 0)
break;
cv_wait(&dl->dl_cv, &dzp->z_lock);
dl=NULL;
}
#ifdef __APPLE__
/*
* Release any unused NFC name (ie if we found a pre-existing lock entry)
*/
if (nfc_name) {
kmem_free(nfc_name, nfc_namesize);
nfc_name = NULL;
}
#endif
if ((flag & ZSHARED) && ++dl->dl_sharecnt > 1 && dl->dl_namesize == 0) {
/*
* We're the second shared reference to dl. Make a copy of
* dl_name in case the first thread goes away before we do.
* Note that we initialize the new name before storing its
* pointer into dl_name, because the first thread may load
* dl->dl_name at any time. He'll either see the old value,
* which is his, or the new shared copy; either is OK.
*/
dl->dl_namesize = strlen(dl->dl_name) + 1;
name = kmem_alloc(dl->dl_namesize, KM_SLEEP);
bcopy(dl->dl_name, name, dl->dl_namesize);
dl->dl_name = name;
}
mutex_exit(&dzp->z_lock);
/*
* We have a dirlock on the name. (Note that it is the dirlock,
* not the dzp's z_lock, that protects the name in the zap object.)
* See if there's an object by this name; if so, put a hold on it.
*/
if (flag & ZXATTR) {
error = sa_lookup(dzp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &zoid,
sizeof (zoid));
if (error == 0)
error = (zoid == 0 ? ENOENT : 0);
} else {
#ifdef __APPLE__
/*
* Lookup an entry in the vnode name cache
*
* If the lookup succeeds, the vnode is returned in *vpp,
* and a status of -1 is returned.
*
* If the lookup determines that the name does not exist
* (negative caching), a status of ENOENT is returned.
*
* If the lookup fails, a status of zero is returned.
*/
switch ( cache_lookup(ZTOV(dzp), &vp, cnp) ) {
case -1:
break;
case ENOENT:
vp = DNLC_NO_VNODE;
break;
default:
vp = NULLVP;
}
#else
vp = dnlc_lookup(ZTOV(dzp), name);
#endif /* __APPLE__ */
if (vp == DNLC_NO_VNODE) {
VN_RELE(vp);
error = ENOENT;
} else if (vp) {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
VN_RELE(vp);
return (EEXIST);
}
*dlpp = dl;
*zpp = VTOZ(vp);
return (0);
} else {
error = zap_lookup(zfsvfs->z_os, dzp->z_id, name,
8, 1, &zoid);
zoid = ZFS_DIRENT_OBJ(zoid);
if (error == ENOENT)
#ifdef __APPLE__
/*
* Add a negative entry into the VFS name cache
*/
if ((flag & ZNEW) == 0 &&
(dzp->z_pflags & ZFS_XATTR) == 0 &&
(cnp) &&
(cnp->cn_flags & MAKEENTRY) &&
(cnp->cn_nameiop != CREATE) &&
(cnp->cn_nameiop != RENAME)) {
cache_enter(ZTOV(dzp), NULLVP, cnp);
}
#else
dnlc_update(ZTOV(dzp), name, DNLC_NO_VNODE);
#endif /* __APPLE__ */
}
}
if (error) {
if (error != ENOENT || (flag & ZEXISTS)) {
zfs_dirent_unlock(dl);
return (error);
}
} else {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
return (EEXIST);
}
//error = zfs_zget_sans_vnode(zfsvfs, zoid, zpp);
error = zfs_zget(zfsvfs, zoid, zpp);
if (error) {
zfs_dirent_unlock(dl);
return (error);
} else {
// Should this be here?
//printf("zfs_dir attach 1\n");
//zfs_attach_vnode(*zpp);
}
if (!(flag & ZXATTR))
#ifdef __APPLE__
if (cnp && cnp->cn_flags & MAKEENTRY)
cache_enter(ZTOV(dzp), ZTOV(*zpp), cnp);
#else
dnlc_update(ZTOV(dzp), name, ZTOV(*zpp));
#endif /* __APPLE__ */
}
*dlpp = dl;
return (0);
}
static int zfsfuse_mknod(fuse_req_t req, fuse_ino_t parent, const char *name, mode_t mode, dev_t rdev)
{
if(strlen(name) >= MAXNAMELEN)
return ENAMETOOLONG;
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
ZFS_ENTER(zfsvfs);
znode_t *znode;
int error = zfs_zget(zfsvfs, parent, &znode, B_FALSE);
if(error) {
ZFS_EXIT(zfsvfs);
/* If the inode we are trying to get was recently deleted
dnode_hold_impl will return EEXIST instead of ENOENT */
return error == EEXIST ? ENOENT : error;
}
ASSERT(znode != NULL);
vnode_t *dvp = ZTOV(znode);
ASSERT(dvp != NULL);
cred_t cred;
zfsfuse_getcred(req, &cred);
vattr_t vattr;
vattr.va_type = IFTOVT(mode);
vattr.va_mode = mode & PERMMASK;
vattr.va_mask = AT_TYPE | AT_MODE;
if(mode & (S_IFCHR | S_IFBLK)) {
vattr.va_rdev = rdev;
vattr.va_mask |= AT_RDEV;
}
vnode_t *vp = NULL;
/* FIXME: check filesystem boundaries */
error = VOP_CREATE(dvp, (char *) name, &vattr, EXCL, 0, &vp, &cred, 0, NULL, NULL);
VN_RELE(dvp);
if(error)
goto out;
ASSERT(vp != NULL);
struct fuse_entry_param e = { 0 };
e.attr_timeout = 0.0;
e.entry_timeout = 0.0;
e.ino = VTOZ(vp)->z_id;
if(e.ino == 3)
e.ino = 1;
e.generation = VTOZ(vp)->z_phys->zp_gen;
error = zfsfuse_stat(vp, &e.attr, &cred);
out:
if(vp != NULL)
VN_RELE(vp);
ZFS_EXIT(zfsvfs);
if(!error)
fuse_reply_entry(req, &e);
return error;
}
static int zfsfuse_opencreate(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, int fflags, mode_t createmode, const char *name)
{
if(name && strlen(name) >= MAXNAMELEN)
return ENAMETOOLONG;
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
ZFS_ENTER(zfsvfs);
cred_t cred;
zfsfuse_getcred(req, &cred);
/* Map flags */
int mode, flags;
if(fflags & O_WRONLY) {
mode = VWRITE;
flags = FWRITE;
} else if(fflags & O_RDWR) {
mode = VREAD | VWRITE;
flags = FREAD | FWRITE;
} else {
mode = VREAD;
flags = FREAD;
}
if(fflags & O_CREAT)
flags |= FCREAT;
if(fflags & O_SYNC)
flags |= FSYNC;
if(fflags & O_DSYNC)
flags |= FDSYNC;
if(fflags & O_RSYNC)
flags |= FRSYNC;
if(fflags & O_APPEND)
flags |= FAPPEND;
if(fflags & O_LARGEFILE)
flags |= FOFFMAX;
if(fflags & O_NOFOLLOW)
flags |= FNOFOLLOW;
if(fflags & O_TRUNC)
flags |= FTRUNC;
if(fflags & O_EXCL)
flags |= FEXCL;
znode_t *znode;
int error = zfs_zget(zfsvfs, ino, &znode, B_FALSE);
if(error) {
ZFS_EXIT(zfsvfs);
/* If the inode we are trying to get was recently deleted
dnode_hold_impl will return EEXIST instead of ENOENT */
return error == EEXIST ? ENOENT : error;
}
ASSERT(znode != NULL);
vnode_t *vp = ZTOV(znode);
ASSERT(vp != NULL);
if (flags & FCREAT) {
enum vcexcl excl;
/*
* Wish to create a file.
*/
vattr_t vattr;
vattr.va_type = VREG;
vattr.va_mode = createmode;
vattr.va_mask = AT_TYPE|AT_MODE;
if (flags & FTRUNC) {
vattr.va_size = 0;
vattr.va_mask |= AT_SIZE;
}
if (flags & FEXCL)
excl = EXCL;
else
excl = NONEXCL;
vnode_t *new_vp;
/* FIXME: check filesystem boundaries */
error = VOP_CREATE(vp, (char *) name, &vattr, excl, mode, &new_vp, &cred, 0, NULL, NULL);
if(error)
goto out;
VN_RELE(vp);
vp = new_vp;
} else {
/*
* Get the attributes to check whether file is large.
* We do this only if the O_LARGEFILE flag is not set and
* only for regular files.
*/
if (!(flags & FOFFMAX) && (vp->v_type == VREG)) {
vattr_t vattr;
vattr.va_mask = AT_SIZE;
if ((error = VOP_GETATTR(vp, &vattr, 0, &cred, NULL)))
goto out;
if (vattr.va_size > (u_offset_t) MAXOFF32_T) {
/*
* Large File API - regular open fails
* if FOFFMAX flag is set in file mode
*/
error = EOVERFLOW;
goto out;
}
}
/*
* Check permissions.
*/
if (error = VOP_ACCESS(vp, mode, 0, &cred, NULL))
goto out;
}
if ((flags & FNOFOLLOW) && vp->v_type == VLNK) {
error = ELOOP;
goto out;
}
vnode_t *old_vp = vp;
error = VOP_OPEN(&vp, flags, &cred, NULL);
ASSERT(old_vp == vp);
if(error)
goto out;
struct fuse_entry_param e = { 0 };
if(flags & FCREAT) {
error = zfsfuse_stat(vp, &e.attr, &cred);
if(error)
goto out;
}
file_info_t *info = kmem_cache_alloc(file_info_cache, KM_NOSLEEP);
if(info == NULL) {
error = ENOMEM;
goto out;
}
info->vp = vp;
info->flags = flags;
fi->fh = (uint64_t) (uintptr_t) info;
fi->keep_cache = 1;
if(flags & FCREAT) {
e.attr_timeout = 0.0;
e.entry_timeout = 0.0;
e.ino = VTOZ(vp)->z_id;
if(e.ino == 3)
e.ino = 1;
e.generation = VTOZ(vp)->z_phys->zp_gen;
}
out:
if(error) {
ASSERT(vp->v_count > 0);
VN_RELE(vp);
}
ZFS_EXIT(zfsvfs);
if(!error) {
if(!(flags & FCREAT))
fuse_reply_open(req, fi);
else
fuse_reply_create(req, &e, fi);
}
return error;
}
static int zfsfuse_readdir(fuse_req_t req, fuse_ino_t ino, size_t size, off_t off, struct fuse_file_info *fi)
{
vnode_t *vp = ((file_info_t *)(uintptr_t) fi->fh)->vp;
ASSERT(vp != NULL);
ASSERT(VTOZ(vp) != NULL);
ASSERT(VTOZ(vp)->z_id == ino);
if(vp->v_type != VDIR)
return ENOTDIR;
vfs_t *vfs = (vfs_t *) fuse_req_userdata(req);
zfsvfs_t *zfsvfs = vfs->vfs_data;
char *outbuf = kmem_alloc(size, KM_NOSLEEP);
if(outbuf == NULL)
return ENOMEM;
ZFS_ENTER(zfsvfs);
cred_t cred;
zfsfuse_getcred(req, &cred);
union {
char buf[DIRENT64_RECLEN(MAXNAMELEN)];
struct dirent64 dirent;
} entry;
struct stat fstat = { 0 };
iovec_t iovec;
uio_t uio;
uio.uio_iov = &iovec;
uio.uio_iovcnt = 1;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_fmode = 0;
uio.uio_llimit = RLIM64_INFINITY;
int eofp = 0;
int outbuf_off = 0;
int outbuf_resid = size;
off_t next = off;
int error;
for(;;) {
iovec.iov_base = entry.buf;
iovec.iov_len = sizeof(entry.buf);
uio.uio_resid = iovec.iov_len;
uio.uio_loffset = next;
error = VOP_READDIR(vp, &uio, &cred, &eofp, NULL, 0);
if(error)
goto out;
/* No more directory entries */
if(iovec.iov_base == entry.buf)
break;
fstat.st_ino = entry.dirent.d_ino;
fstat.st_mode = 0;
int dsize = fuse_dirent_size(strlen(entry.dirent.d_name));
if(dsize > outbuf_resid)
break;
fuse_add_dirent(outbuf + outbuf_off, entry.dirent.d_name, &fstat, entry.dirent.d_off);
outbuf_off += dsize;
outbuf_resid -= dsize;
next = entry.dirent.d_off;
}
out:
ZFS_EXIT(zfsvfs);
if(!error)
fuse_reply_buf(req, outbuf, outbuf_off);
kmem_free(outbuf, size);
return error;
}