/*
* Set up the correct mode on the file based on the supplied ACL. This
* makes sure that the mode on the file reflects the state of the
* u::,g::[m::], and o:: entries in the ACL. Since the mode is where
* the ACL is going to get the permissions for these entries, we must
* synchronize the mode whenever we set the ACL on a file.
*/
STATIC int
xfs_acl_setmode(
bhv_vnode_t *vp,
xfs_acl_t *acl,
int *basicperms)
{
bhv_vattr_t va;
xfs_acl_entry_t *ap;
xfs_acl_entry_t *gap = NULL;
int i, error, nomask = 1;
*basicperms = 1;
if (acl->acl_cnt == XFS_ACL_NOT_PRESENT)
return 0;
/*
* Copy the u::, g::, o::, and m:: bits from the ACL into the
* mode. The m:: bits take precedence over the g:: bits.
*/
va.va_mask = XFS_AT_MODE;
error = xfs_getattr(xfs_vtoi(vp), &va, 0);
if (error)
return error;
va.va_mask = XFS_AT_MODE;
va.va_mode &= ~(S_IRWXU|S_IRWXG|S_IRWXO);
ap = acl->acl_entry;
for (i = 0; i < acl->acl_cnt; ++i) {
switch (ap->ae_tag) {
case ACL_USER_OBJ:
va.va_mode |= ap->ae_perm << 6;
break;
case ACL_GROUP_OBJ:
gap = ap;
break;
case ACL_MASK: /* more than just standard modes */
nomask = 0;
va.va_mode |= ap->ae_perm << 3;
*basicperms = 0;
break;
case ACL_OTHER:
va.va_mode |= ap->ae_perm;
break;
default: /* more than just standard modes */
*basicperms = 0;
break;
}
ap++;
}
/* Set the group bits from ACL_GROUP_OBJ if there's no ACL_MASK */
if (gap && nomask)
va.va_mode |= gap->ae_perm << 3;
return xfs_setattr(xfs_vtoi(vp), &va, 0, sys_cred);
}
int
xfs_acl_vtoacl(
bhv_vnode_t *vp,
xfs_acl_t *access_acl,
xfs_acl_t *default_acl)
{
bhv_vattr_t va;
int error = 0;
if (access_acl) {
/*
* Get the Access ACL and the mode. If either cannot
* be obtained for some reason, invalidate the access ACL.
*/
xfs_acl_get_attr(vp, access_acl, _ACL_TYPE_ACCESS, 0, &error);
if (!error) {
/* Got the ACL, need the mode... */
va.va_mask = XFS_AT_MODE;
error = xfs_getattr(xfs_vtoi(vp), &va, 0);
}
if (error)
access_acl->acl_cnt = XFS_ACL_NOT_PRESENT;
else /* We have a good ACL and the file mode, synchronize. */
xfs_acl_sync_mode(va.va_mode, access_acl);
}
if (default_acl) {
xfs_acl_get_attr(vp, default_acl, _ACL_TYPE_DEFAULT, 0, &error);
if (error)
default_acl->acl_cnt = XFS_ACL_NOT_PRESENT;
}
return error;
}
/*
* Set the EA with the ACL and do endian conversion.
*/
STATIC void
xfs_acl_set_attr(
bhv_vnode_t *vp,
xfs_acl_t *aclp,
int kind,
int *error)
{
xfs_acl_entry_t *ace, *newace, *end;
xfs_acl_t *newacl;
int len;
if (!(_ACL_ALLOC(newacl))) {
*error = ENOMEM;
return;
}
len = sizeof(xfs_acl_t) -
(sizeof(xfs_acl_entry_t) * (XFS_ACL_MAX_ENTRIES - aclp->acl_cnt));
end = &aclp->acl_entry[0]+aclp->acl_cnt;
for (ace = &aclp->acl_entry[0], newace = &newacl->acl_entry[0];
ace < end;
ace++, newace++) {
INT_SET(newace->ae_tag, ARCH_CONVERT, ace->ae_tag);
INT_SET(newace->ae_id, ARCH_CONVERT, ace->ae_id);
INT_SET(newace->ae_perm, ARCH_CONVERT, ace->ae_perm);
}
INT_SET(newacl->acl_cnt, ARCH_CONVERT, aclp->acl_cnt);
*error = xfs_attr_set(xfs_vtoi(vp),
kind == _ACL_TYPE_ACCESS ?
SGI_ACL_FILE: SGI_ACL_DEFAULT,
(char *)newacl, len, ATTR_ROOT);
_ACL_FREE(newacl);
}
int
xfs_acl_vget(
bhv_vnode_t *vp,
void *acl,
size_t size,
int kind)
{
int error;
xfs_acl_t *xfs_acl = NULL;
posix_acl_xattr_header *ext_acl = acl;
int flags = 0;
VN_HOLD(vp);
if(size) {
if (!(_ACL_ALLOC(xfs_acl))) {
error = ENOMEM;
goto out;
}
memset(xfs_acl, 0, sizeof(xfs_acl_t));
} else
flags = ATTR_KERNOVAL;
xfs_acl_get_attr(vp, xfs_acl, kind, flags, &error);
if (error)
goto out;
if (!size) {
error = -posix_acl_xattr_size(XFS_ACL_MAX_ENTRIES);
} else {
if (xfs_acl_invalid(xfs_acl)) {
error = EINVAL;
goto out;
}
if (kind == _ACL_TYPE_ACCESS) {
bhv_vattr_t va;
va.va_mask = XFS_AT_MODE;
error = xfs_getattr(xfs_vtoi(vp), &va, 0);
if (error)
goto out;
xfs_acl_sync_mode(va.va_mode, xfs_acl);
}
error = -posix_acl_xfs_to_xattr(xfs_acl, ext_acl, size);
}
out:
VN_RELE(vp);
if(xfs_acl)
_ACL_FREE(xfs_acl);
return -error;
}
/*
* xfs_get_dir_entry is used to get a reference to an inode given
* its parent directory inode and the name of the file. It does
* not lock the child inode, and it unlocks the directory before
* returning. The directory's generation number is returned for
* use by a later call to xfs_lock_dir_and_entry.
*/
int
xfs_get_dir_entry(
vname_t *dentry,
xfs_inode_t **ipp)
{
vnode_t *vp;
vp = VNAME_TO_VNODE(dentry);
*ipp = xfs_vtoi(vp);
if (!*ipp)
return XFS_ERROR(ENOENT);
VN_HOLD(vp);
return 0;
}
/* If the user is attempting to execute a file that is offline then
* we have to trigger a DMAPI READ event before the file is marked as busy
* otherwise the invisible I/O will not be able to write to the file to bring
* it back online.
*/
STATIC int
xfs_file_open_exec(
struct inode *inode)
{
bhv_vnode_t *vp = vn_from_inode(inode);
if (unlikely(vp->v_vfsp->vfs_flag & VFS_DMI)) {
xfs_mount_t *mp = XFS_VFSTOM(vp->v_vfsp);
xfs_inode_t *ip = xfs_vtoi(vp);
if (!ip)
return -EINVAL;
if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ))
return -XFS_SEND_DATA(mp, DM_EVENT_READ, vp,
0, 0, 0, NULL);
}
return 0;
}
/*
* Get the ACL from the EA and do endian conversion.
*/
STATIC void
xfs_acl_get_attr(
bhv_vnode_t *vp,
xfs_acl_t *aclp,
int kind,
int flags,
int *error)
{
int len = sizeof(xfs_acl_t);
ASSERT((flags & ATTR_KERNOVAL) ? (aclp == NULL) : 1);
flags |= ATTR_ROOT;
*error = xfs_attr_get(xfs_vtoi(vp),
kind == _ACL_TYPE_ACCESS ?
SGI_ACL_FILE : SGI_ACL_DEFAULT,
(char *)aclp, &len, flags, sys_cred);
if (*error || (flags & ATTR_KERNOVAL))
return;
xfs_acl_get_endian(aclp);
}
int
xfs_acl_vremove(
bhv_vnode_t *vp,
int kind)
{
int error;
VN_HOLD(vp);
error = xfs_acl_allow_set(vp, kind);
if (!error) {
error = xfs_attr_remove(xfs_vtoi(vp),
kind == _ACL_TYPE_DEFAULT?
SGI_ACL_DEFAULT: SGI_ACL_FILE,
ATTR_ROOT);
if (error == ENOATTR)
error = 0; /* 'scool */
}
VN_RELE(vp);
return -error;
}
/*
* Write wrapper for fifos.
*/
static int
_xfsfifo_write(
struct vop_write_args /* {
struct vnode *a_vp;
struct uio *a_uio;
int a_ioflag;
struct ucred *a_cred;
} */ *ap)
{
int error, resid;
struct uio *uio;
struct xfs_inode *ip;
uio = ap->a_uio;
resid = uio->uio_resid;
error = fifo_specops.vop_write(ap);
ip = xfs_vtoi(VPTOXFSVP(ap->a_vp));
if (ip != NULL && (uio->uio_resid != resid ||
(error == 0 && resid != 0)))
xfs_ichgtime(ip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
return (error);
}
/*
* Read wrapper for fifos.
*/
static int
_xfsfifo_read(
struct vop_read_args /* {
struct vnode *a_vp;
struct uio *a_uio;
int a_ioflag;
struct ucred *a_cred;
} */ *ap)
{
int error, resid;
struct xfs_inode *ip;
struct uio *uio;
uio = ap->a_uio;
resid = uio->uio_resid;
error = fifo_specops.vop_read(ap);
ip = xfs_vtoi(VPTOXFSVP(ap->a_vp));
if ((ap->a_vp->v_mount->mnt_flag & MNT_NOATIME) == 0 && ip != NULL &&
(uio->uio_resid != resid || (error == 0 && resid != 0)))
xfs_ichgtime(ip, XFS_ICHGTIME_ACC);
return (error);
}
/* If the user is attempting to execute a file that is offline then
* we have to trigger a DMAPI READ event before the file is marked as busy
* otherwise the invisible I/O will not be able to write to the file to bring
* it back online.
*/
STATIC int
xfs_file_open_exec(
struct inode *inode)
{
vnode_t *vp = vn_from_inode(inode);
xfs_mount_t *mp = XFS_VFSTOM(vp->v_vfsp);
int error = 0;
xfs_inode_t *ip;
if (vp->v_vfsp->vfs_flag & VFS_DMI) {
ip = xfs_vtoi(vp);
if (!ip) {
error = -EINVAL;
goto open_exec_out;
}
if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ)) {
error = -XFS_SEND_DATA(mp, DM_EVENT_READ, vp,
0, 0, 0, NULL);
}
}
open_exec_out:
return error;
}
STATIC int
xfs_acl_allow_set(
bhv_vnode_t *vp,
int kind)
{
xfs_inode_t *ip = xfs_vtoi(vp);
bhv_vattr_t va;
int error;
if (vp->i_flags & (S_IMMUTABLE|S_APPEND))
return EPERM;
if (kind == _ACL_TYPE_DEFAULT && !VN_ISDIR(vp))
return ENOTDIR;
if (vp->i_sb->s_flags & MS_RDONLY)
return EROFS;
va.va_mask = XFS_AT_UID;
error = xfs_getattr(ip, &va, 0);
if (error)
return error;
if (va.va_uid != current->fsuid && !capable(CAP_FOWNER))
return EPERM;
return error;
}
/*
* xfs_rename
*/
int
xfs_rename(
bhv_desc_t *src_dir_bdp,
bhv_vname_t *src_vname,
bhv_vnode_t *target_dir_vp,
bhv_vname_t *target_vname,
cred_t *credp)
{
xfs_trans_t *tp;
xfs_inode_t *src_dp, *target_dp, *src_ip, *target_ip;
xfs_mount_t *mp;
int new_parent; /* moving to a new dir */
int src_is_directory; /* src_name is a directory */
int error;
xfs_bmap_free_t free_list;
xfs_fsblock_t first_block;
int cancel_flags;
int committed;
xfs_inode_t *inodes[4];
int target_ip_dropped = 0; /* dropped target_ip link? */
bhv_vnode_t *src_dir_vp;
int spaceres;
int target_link_zero = 0;
int num_inodes;
char *src_name = VNAME(src_vname);
char *target_name = VNAME(target_vname);
int src_namelen = VNAMELEN(src_vname);
int target_namelen = VNAMELEN(target_vname);
src_dir_vp = BHV_TO_VNODE(src_dir_bdp);
vn_trace_entry(src_dir_vp, "xfs_rename", (inst_t *)__return_address);
vn_trace_entry(target_dir_vp, "xfs_rename", (inst_t *)__return_address);
/*
* Find the XFS behavior descriptor for the target directory
* vnode since it was not handed to us.
*/
target_dp = xfs_vtoi(target_dir_vp);
if (target_dp == NULL) {
return XFS_ERROR(EXDEV);
}
src_dp = XFS_BHVTOI(src_dir_bdp);
mp = src_dp->i_mount;
if (DM_EVENT_ENABLED(src_dir_vp->v_vfsp, src_dp, DM_EVENT_RENAME) ||
DM_EVENT_ENABLED(target_dir_vp->v_vfsp,
target_dp, DM_EVENT_RENAME)) {
error = XFS_SEND_NAMESP(mp, DM_EVENT_RENAME,
src_dir_vp, DM_RIGHT_NULL,
target_dir_vp, DM_RIGHT_NULL,
src_name, target_name,
0, 0, 0);
if (error) {
return error;
}
}
/* Return through std_return after this point. */
/*
* Lock all the participating inodes. Depending upon whether
* the target_name exists in the target directory, and
* whether the target directory is the same as the source
* directory, we can lock from 2 to 4 inodes.
* xfs_lock_for_rename() will return ENOENT if src_name
* does not exist in the source directory.
*/
tp = NULL;
error = xfs_lock_for_rename(src_dp, target_dp, src_vname,
target_vname, &src_ip, &target_ip, inodes,
&num_inodes);
if (error) {
/*
* We have nothing locked, no inode references, and
* no transaction, so just get out.
*/
goto std_return;
}
ASSERT(src_ip != NULL);
if ((src_ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
/*
* Check for link count overflow on target_dp
*/
if (target_ip == NULL && (src_dp != target_dp) &&
target_dp->i_d.di_nlink >= XFS_MAXLINK) {
error = XFS_ERROR(EMLINK);
xfs_rename_unlock4(inodes, XFS_ILOCK_SHARED);
goto rele_return;
}
}
/*
* If we are using project inheritance, we only allow renames
* into our tree when the project IDs are the same; else the
* tree quota mechanism would be circumvented.
*/
if (unlikely((target_dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
(target_dp->i_d.di_projid != src_ip->i_d.di_projid))) {
error = XFS_ERROR(EXDEV);
xfs_rename_unlock4(inodes, XFS_ILOCK_SHARED);
goto rele_return;
}
new_parent = (src_dp != target_dp);
src_is_directory = ((src_ip->i_d.di_mode & S_IFMT) == S_IFDIR);
/*
* Drop the locks on our inodes so that we can start the transaction.
*/
xfs_rename_unlock4(inodes, XFS_ILOCK_SHARED);
XFS_BMAP_INIT(&free_list, &first_block);
tp = xfs_trans_alloc(mp, XFS_TRANS_RENAME);
cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
spaceres = XFS_RENAME_SPACE_RES(mp, target_namelen);
error = xfs_trans_reserve(tp, spaceres, XFS_RENAME_LOG_RES(mp), 0,
XFS_TRANS_PERM_LOG_RES, XFS_RENAME_LOG_COUNT);
if (error == ENOSPC) {
spaceres = 0;
error = xfs_trans_reserve(tp, 0, XFS_RENAME_LOG_RES(mp), 0,
XFS_TRANS_PERM_LOG_RES, XFS_RENAME_LOG_COUNT);
}
if (error) {
xfs_trans_cancel(tp, 0);
goto rele_return;
}
/*
* Attach the dquots to the inodes
*/
if ((error = XFS_QM_DQVOPRENAME(mp, inodes))) {
xfs_trans_cancel(tp, cancel_flags);
goto rele_return;
}
/*
* Reacquire the inode locks we dropped above.
*/
xfs_lock_inodes(inodes, num_inodes, 0, XFS_ILOCK_EXCL);
/*
* Join all the inodes to the transaction. From this point on,
* we can rely on either trans_commit or trans_cancel to unlock
* them. Note that we need to add a vnode reference to the
* directories since trans_commit & trans_cancel will decrement
* them when they unlock the inodes. Also, we need to be careful
* not to add an inode to the transaction more than once.
*/
VN_HOLD(src_dir_vp);
xfs_trans_ijoin(tp, src_dp, XFS_ILOCK_EXCL);
if (new_parent) {
VN_HOLD(target_dir_vp);
xfs_trans_ijoin(tp, target_dp, XFS_ILOCK_EXCL);
}
if ((src_ip != src_dp) && (src_ip != target_dp)) {
xfs_trans_ijoin(tp, src_ip, XFS_ILOCK_EXCL);
}
if ((target_ip != NULL) &&
(target_ip != src_ip) &&
(target_ip != src_dp) &&
(target_ip != target_dp)) {
xfs_trans_ijoin(tp, target_ip, XFS_ILOCK_EXCL);
}
/*
* Set up the target.
*/
if (target_ip == NULL) {
/*
* If there's no space reservation, check the entry will
* fit before actually inserting it.
*/
if (spaceres == 0 &&
(error = xfs_dir_canenter(tp, target_dp, target_name,
target_namelen)))
goto error_return;
/*
* If target does not exist and the rename crosses
* directories, adjust the target directory link count
* to account for the ".." reference from the new entry.
*/
error = xfs_dir_createname(tp, target_dp, target_name,
target_namelen, src_ip->i_ino,
&first_block, &free_list, spaceres);
if (error == ENOSPC)
goto error_return;
if (error)
goto abort_return;
xfs_ichgtime(target_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
if (new_parent && src_is_directory) {
error = xfs_bumplink(tp, target_dp);
if (error)
goto abort_return;
}
} else { /* target_ip != NULL */
/*
* If target exists and it's a directory, check that both
* target and source are directories and that target can be
* destroyed, or that neither is a directory.
*/
if ((target_ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
/*
* Make sure target dir is empty.
*/
if (!(xfs_dir_isempty(target_ip)) ||
(target_ip->i_d.di_nlink > 2)) {
error = XFS_ERROR(EEXIST);
goto error_return;
}
}
/*
* Link the source inode under the target name.
* If the source inode is a directory and we are moving
* it across directories, its ".." entry will be
* inconsistent until we replace that down below.
*
* In case there is already an entry with the same
* name at the destination directory, remove it first.
*/
error = xfs_dir_replace(tp, target_dp, target_name,
target_namelen, src_ip->i_ino,
&first_block, &free_list, spaceres);
if (error)
goto abort_return;
xfs_ichgtime(target_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
/*
* Decrement the link count on the target since the target
* dir no longer points to it.
*/
error = xfs_droplink(tp, target_ip);
if (error)
goto abort_return;
target_ip_dropped = 1;
if (src_is_directory) {
/*
* Drop the link from the old "." entry.
*/
error = xfs_droplink(tp, target_ip);
if (error)
goto abort_return;
}
/* Do this test while we still hold the locks */
target_link_zero = (target_ip)->i_d.di_nlink==0;
} /* target_ip != NULL */
/*
* Remove the source.
*/
if (new_parent && src_is_directory) {
/*
* Rewrite the ".." entry to point to the new
* directory.
*/
error = xfs_dir_replace(tp, src_ip, "..", 2, target_dp->i_ino,
&first_block, &free_list, spaceres);
ASSERT(error != EEXIST);
if (error)
goto abort_return;
xfs_ichgtime(src_ip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
} else {
/*
* We always want to hit the ctime on the source inode.
* We do it in the if clause above for the 'new_parent &&
* src_is_directory' case, and here we get all the other
* cases. This isn't strictly required by the standards
* since the source inode isn't really being changed,
* but old unix file systems did it and some incremental
* backup programs won't work without it.
*/
xfs_ichgtime(src_ip, XFS_ICHGTIME_CHG);
}
/*
* Adjust the link count on src_dp. This is necessary when
* renaming a directory, either within one parent when
* the target existed, or across two parent directories.
*/
if (src_is_directory && (new_parent || target_ip != NULL)) {
/*
* Decrement link count on src_directory since the
* entry that's moved no longer points to it.
*/
error = xfs_droplink(tp, src_dp);
if (error)
goto abort_return;
}
error = xfs_dir_removename(tp, src_dp, src_name, src_namelen,
src_ip->i_ino, &first_block, &free_list, spaceres);
if (error)
goto abort_return;
xfs_ichgtime(src_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
/*
* Update the generation counts on all the directory inodes
* that we're modifying.
*/
src_dp->i_gen++;
xfs_trans_log_inode(tp, src_dp, XFS_ILOG_CORE);
if (new_parent) {
target_dp->i_gen++;
xfs_trans_log_inode(tp, target_dp, XFS_ILOG_CORE);
}
/*
* If there was a target inode, take an extra reference on
* it here so that it doesn't go to xfs_inactive() from
* within the commit.
*/
if (target_ip != NULL) {
IHOLD(target_ip);
}
/*
* If this is a synchronous mount, make sure that the
* rename transaction goes to disk before returning to
* the user.
*/
if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC)) {
xfs_trans_set_sync(tp);
}
/*
* Take refs. for vop_link_removed calls below. No need to worry
* about directory refs. because the caller holds them.
*
* Do holds before the xfs_bmap_finish since it might rele them down
* to zero.
*/
if (target_ip_dropped)
IHOLD(target_ip);
IHOLD(src_ip);
error = xfs_bmap_finish(&tp, &free_list, &committed);
if (error) {
xfs_bmap_cancel(&free_list);
xfs_trans_cancel(tp, (XFS_TRANS_RELEASE_LOG_RES |
XFS_TRANS_ABORT));
if (target_ip != NULL) {
IRELE(target_ip);
}
if (target_ip_dropped) {
IRELE(target_ip);
}
IRELE(src_ip);
goto std_return;
}
/*
* trans_commit will unlock src_ip, target_ip & decrement
* the vnode references.
*/
error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
if (target_ip != NULL) {
xfs_refcache_purge_ip(target_ip);
IRELE(target_ip);
}
/*
* Let interposed file systems know about removed links.
*/
if (target_ip_dropped) {
bhv_vop_link_removed(XFS_ITOV(target_ip), target_dir_vp,
target_link_zero);
IRELE(target_ip);
}
IRELE(src_ip);
/* Fall through to std_return with error = 0 or errno from
* xfs_trans_commit */
std_return:
if (DM_EVENT_ENABLED(src_dir_vp->v_vfsp, src_dp, DM_EVENT_POSTRENAME) ||
DM_EVENT_ENABLED(target_dir_vp->v_vfsp,
target_dp, DM_EVENT_POSTRENAME)) {
(void) XFS_SEND_NAMESP (mp, DM_EVENT_POSTRENAME,
src_dir_vp, DM_RIGHT_NULL,
target_dir_vp, DM_RIGHT_NULL,
src_name, target_name,
0, error, 0);
}
return error;
abort_return:
cancel_flags |= XFS_TRANS_ABORT;
/* FALLTHROUGH */
error_return:
xfs_bmap_cancel(&free_list);
xfs_trans_cancel(tp, cancel_flags);
goto std_return;
rele_return:
IRELE(src_ip);
if (target_ip != NULL) {
IRELE(target_ip);
}
goto std_return;
}
/*
* xfs_find_handle maps from userspace xfs_fsop_handlereq structure to
* a file or fs handle.
*
* XFS_IOC_PATH_TO_FSHANDLE
* returns fs handle for a mount point or path within that mount point
* XFS_IOC_FD_TO_HANDLE
* returns full handle for a FD opened in user space
* XFS_IOC_PATH_TO_HANDLE
* returns full handle for a path
*/
STATIC int
xfs_find_handle(
unsigned int cmd,
void __user *arg)
{
int hsize;
xfs_handle_t handle;
xfs_fsop_handlereq_t hreq;
struct inode *inode;
struct vnode *vp;
if (copy_from_user(&hreq, arg, sizeof(hreq)))
return -XFS_ERROR(EFAULT);
memset((char *)&handle, 0, sizeof(handle));
switch (cmd) {
case XFS_IOC_PATH_TO_FSHANDLE:
case XFS_IOC_PATH_TO_HANDLE: {
struct nameidata nd;
int error;
error = user_path_walk_link((const char __user *)hreq.path, &nd);
if (error)
return error;
ASSERT(nd.dentry);
ASSERT(nd.dentry->d_inode);
inode = igrab(nd.dentry->d_inode);
path_release(&nd);
break;
}
case XFS_IOC_FD_TO_HANDLE: {
struct file *file;
file = fget(hreq.fd);
if (!file)
return -EBADF;
ASSERT(file->f_dentry);
ASSERT(file->f_dentry->d_inode);
inode = igrab(file->f_dentry->d_inode);
fput(file);
break;
}
default:
ASSERT(0);
return -XFS_ERROR(EINVAL);
}
if (inode->i_sb->s_magic != XFS_SB_MAGIC) {
/* we're not in XFS anymore, Toto */
iput(inode);
return -XFS_ERROR(EINVAL);
}
switch (inode->i_mode & S_IFMT) {
case S_IFREG:
case S_IFDIR:
case S_IFLNK:
break;
default:
iput(inode);
return -XFS_ERROR(EBADF);
}
/* we need the vnode */
vp = vn_from_inode(inode);
/* now we can grab the fsid */
memcpy(&handle.ha_fsid, vp->v_vfsp->vfs_altfsid, sizeof(xfs_fsid_t));
hsize = sizeof(xfs_fsid_t);
if (cmd != XFS_IOC_PATH_TO_FSHANDLE) {
xfs_inode_t *ip;
int lock_mode;
/* need to get access to the xfs_inode to read the generation */
ip = xfs_vtoi(vp);
ASSERT(ip);
lock_mode = xfs_ilock_map_shared(ip);
/* fill in fid section of handle from inode */
handle.ha_fid.xfs_fid_len = sizeof(xfs_fid_t) -
sizeof(handle.ha_fid.xfs_fid_len);
handle.ha_fid.xfs_fid_pad = 0;
handle.ha_fid.xfs_fid_gen = ip->i_d.di_gen;
handle.ha_fid.xfs_fid_ino = ip->i_ino;
xfs_iunlock_map_shared(ip, lock_mode);
hsize = XFS_HSIZE(handle);
}
/* now copy our handle into the user buffer & write out the size */
if (copy_to_user(hreq.ohandle, &handle, hsize) ||
copy_to_user(hreq.ohandlen, &hsize, sizeof(__s32))) {
iput(inode);
return -XFS_ERROR(EFAULT);
}
iput(inode);
return 0;
}
/*
* Syssgi interface for swapext
*/
int
xfs_swapext(
xfs_swapext_t __user *sxu)
{
xfs_swapext_t *sxp;
xfs_inode_t *ip=NULL, *tip=NULL, *ips[2];
xfs_trans_t *tp;
xfs_mount_t *mp;
xfs_bstat_t *sbp;
struct file *fp = NULL, *tfp = NULL;
vnode_t *vp, *tvp;
static uint lock_flags = XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL;
int ilf_fields, tilf_fields;
int error = 0;
xfs_ifork_t *tempifp, *ifp, *tifp;
__uint64_t tmp;
int aforkblks = 0;
int taforkblks = 0;
char locked = 0;
sxp = kmem_alloc(sizeof(xfs_swapext_t), KM_MAYFAIL);
tempifp = kmem_alloc(sizeof(xfs_ifork_t), KM_MAYFAIL);
if (!sxp || !tempifp) {
error = XFS_ERROR(ENOMEM);
goto error0;
}
if (copy_from_user(sxp, sxu, sizeof(xfs_swapext_t))) {
error = XFS_ERROR(EFAULT);
goto error0;
}
/* Pull information for the target fd */
if (((fp = fget((int)sxp->sx_fdtarget)) == NULL) ||
((vp = vn_from_inode(fp->f_dentry->d_inode)) == NULL)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
ip = xfs_vtoi(vp);
if (ip == NULL) {
error = XFS_ERROR(EBADF);
goto error0;
}
if (((tfp = fget((int)sxp->sx_fdtmp)) == NULL) ||
((tvp = vn_from_inode(tfp->f_dentry->d_inode)) == NULL)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
tip = xfs_vtoi(tvp);
if (tip == NULL) {
error = XFS_ERROR(EBADF);
goto error0;
}
if (ip->i_mount != tip->i_mount) {
error = XFS_ERROR(EINVAL);
goto error0;
}
if (ip->i_ino == tip->i_ino) {
error = XFS_ERROR(EINVAL);
goto error0;
}
mp = ip->i_mount;
sbp = &sxp->sx_stat;
if (XFS_FORCED_SHUTDOWN(mp)) {
error = XFS_ERROR(EIO);
goto error0;
}
locked = 1;
/* Lock in i_ino order */
if (ip->i_ino < tip->i_ino) {
ips[0] = ip;
ips[1] = tip;
} else {
ips[0] = tip;
ips[1] = ip;
}
xfs_lock_inodes(ips, 2, 0, lock_flags);
/* Check permissions */
error = xfs_iaccess(ip, S_IWUSR, NULL);
if (error)
goto error0;
error = xfs_iaccess(tip, S_IWUSR, NULL);
if (error)
goto error0;
/* Verify that both files have the same format */
if ((ip->i_d.di_mode & S_IFMT) != (tip->i_d.di_mode & S_IFMT)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
/* Verify both files are either real-time or non-realtime */
if ((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) !=
(tip->i_d.di_flags & XFS_DIFLAG_REALTIME)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
/* Should never get a local format */
if (ip->i_d.di_format == XFS_DINODE_FMT_LOCAL ||
tip->i_d.di_format == XFS_DINODE_FMT_LOCAL) {
error = XFS_ERROR(EINVAL);
goto error0;
}
if (VN_CACHED(tvp) != 0) {
xfs_inval_cached_trace(&tip->i_iocore, 0, -1, 0, -1);
VOP_FLUSHINVAL_PAGES(tvp, 0, -1, FI_REMAPF_LOCKED);
}
/* Verify O_DIRECT for ftmp */
if (VN_CACHED(tvp) != 0) {
error = XFS_ERROR(EINVAL);
goto error0;
}
/* Verify all data are being swapped */
if (sxp->sx_offset != 0 ||
sxp->sx_length != ip->i_d.di_size ||
sxp->sx_length != tip->i_d.di_size) {
error = XFS_ERROR(EFAULT);
goto error0;
}
/*
* If the target has extended attributes, the tmp file
* must also in order to ensure the correct data fork
* format.
*/
if ( XFS_IFORK_Q(ip) != XFS_IFORK_Q(tip) ) {
error = XFS_ERROR(EINVAL);
goto error0;
}
/*
* Compare the current change & modify times with that
* passed in. If they differ, we abort this swap.
* This is the mechanism used to ensure the calling
* process that the file was not changed out from
* under it.
*/
if ((sbp->bs_ctime.tv_sec != ip->i_d.di_ctime.t_sec) ||
(sbp->bs_ctime.tv_nsec != ip->i_d.di_ctime.t_nsec) ||
(sbp->bs_mtime.tv_sec != ip->i_d.di_mtime.t_sec) ||
(sbp->bs_mtime.tv_nsec != ip->i_d.di_mtime.t_nsec)) {
error = XFS_ERROR(EBUSY);
goto error0;
}
/* We need to fail if the file is memory mapped. Once we have tossed
* all existing pages, the page fault will have no option
* but to go to the filesystem for pages. By making the page fault call
* VOP_READ (or write in the case of autogrow) they block on the iolock
* until we have switched the extents.
*/
if (VN_MAPPED(vp)) {
error = XFS_ERROR(EBUSY);
goto error0;
}
xfs_iunlock(ip, XFS_ILOCK_EXCL);
xfs_iunlock(tip, XFS_ILOCK_EXCL);
/*
* There is a race condition here since we gave up the
* ilock. However, the data fork will not change since
* we have the iolock (locked for truncation too) so we
* are safe. We don't really care if non-io related
* fields change.
*/
VOP_TOSS_PAGES(vp, 0, -1, FI_REMAPF);
tp = xfs_trans_alloc(mp, XFS_TRANS_SWAPEXT);
if ((error = xfs_trans_reserve(tp, 0,
XFS_ICHANGE_LOG_RES(mp), 0,
0, 0))) {
xfs_iunlock(ip, XFS_IOLOCK_EXCL);
xfs_iunlock(tip, XFS_IOLOCK_EXCL);
xfs_trans_cancel(tp, 0);
locked = 0;
goto error0;
}
xfs_lock_inodes(ips, 2, 0, XFS_ILOCK_EXCL);
/*
* Count the number of extended attribute blocks
*/
if ( ((XFS_IFORK_Q(ip) != 0) && (ip->i_d.di_anextents > 0)) &&
(ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
error = xfs_bmap_count_blocks(tp, ip, XFS_ATTR_FORK, &aforkblks);
if (error) {
xfs_trans_cancel(tp, 0);
goto error0;
}
}
if ( ((XFS_IFORK_Q(tip) != 0) && (tip->i_d.di_anextents > 0)) &&
(tip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
error = xfs_bmap_count_blocks(tp, tip, XFS_ATTR_FORK,
&taforkblks);
if (error) {
xfs_trans_cancel(tp, 0);
goto error0;
}
}
/*
* Swap the data forks of the inodes
*/
ifp = &ip->i_df;
tifp = &tip->i_df;
*tempifp = *ifp; /* struct copy */
*ifp = *tifp; /* struct copy */
*tifp = *tempifp; /* struct copy */
/*
* Fix the on-disk inode values
*/
tmp = (__uint64_t)ip->i_d.di_nblocks;
ip->i_d.di_nblocks = tip->i_d.di_nblocks - taforkblks + aforkblks;
tip->i_d.di_nblocks = tmp + taforkblks - aforkblks;
tmp = (__uint64_t) ip->i_d.di_nextents;
ip->i_d.di_nextents = tip->i_d.di_nextents;
tip->i_d.di_nextents = tmp;
tmp = (__uint64_t) ip->i_d.di_format;
ip->i_d.di_format = tip->i_d.di_format;
tip->i_d.di_format = tmp;
ilf_fields = XFS_ILOG_CORE;
switch(ip->i_d.di_format) {
case XFS_DINODE_FMT_EXTENTS:
/* If the extents fit in the inode, fix the
* pointer. Otherwise it's already NULL or
* pointing to the extent.
*/
if (ip->i_d.di_nextents <= XFS_INLINE_EXTS) {
ifp->if_u1.if_extents =
ifp->if_u2.if_inline_ext;
}
ilf_fields |= XFS_ILOG_DEXT;
break;
case XFS_DINODE_FMT_BTREE:
ilf_fields |= XFS_ILOG_DBROOT;
break;
}
tilf_fields = XFS_ILOG_CORE;
switch(tip->i_d.di_format) {
case XFS_DINODE_FMT_EXTENTS:
/* If the extents fit in the inode, fix the
* pointer. Otherwise it's already NULL or
* pointing to the extent.
*/
if (tip->i_d.di_nextents <= XFS_INLINE_EXTS) {
tifp->if_u1.if_extents =
tifp->if_u2.if_inline_ext;
}
tilf_fields |= XFS_ILOG_DEXT;
break;
case XFS_DINODE_FMT_BTREE:
tilf_fields |= XFS_ILOG_DBROOT;
break;
}
/*
* Increment vnode ref counts since xfs_trans_commit &
* xfs_trans_cancel will both unlock the inodes and
* decrement the associated ref counts.
*/
VN_HOLD(vp);
VN_HOLD(tvp);
xfs_trans_ijoin(tp, ip, lock_flags);
xfs_trans_ijoin(tp, tip, lock_flags);
xfs_trans_log_inode(tp, ip, ilf_fields);
xfs_trans_log_inode(tp, tip, tilf_fields);
/*
* If this is a synchronous mount, make sure that the
* transaction goes to disk before returning to the user.
*/
if (mp->m_flags & XFS_MOUNT_WSYNC) {
xfs_trans_set_sync(tp);
}
error = xfs_trans_commit(tp, XFS_TRANS_SWAPEXT, NULL);
locked = 0;
error0:
if (locked) {
xfs_iunlock(ip, lock_flags);
xfs_iunlock(tip, lock_flags);
}
if (fp != NULL)
fput(fp);
if (tfp != NULL)
fput(tfp);
if (sxp != NULL)
kmem_free(sxp, sizeof(xfs_swapext_t));
if (tempifp != NULL)
kmem_free(tempifp, sizeof(xfs_ifork_t));
return error;
}
/*
* Syssgi interface for swapext
*/
int
xfs_swapext(
xfs_swapext_t __user *sxu)
{
xfs_swapext_t *sxp;
xfs_inode_t *ip=NULL, *tip=NULL;
xfs_mount_t *mp;
struct file *fp = NULL, *tfp = NULL;
bhv_vnode_t *vp, *tvp;
int error = 0;
sxp = kmem_alloc(sizeof(xfs_swapext_t), KM_MAYFAIL);
if (!sxp) {
error = XFS_ERROR(ENOMEM);
goto error0;
}
if (copy_from_user(sxp, sxu, sizeof(xfs_swapext_t))) {
error = XFS_ERROR(EFAULT);
goto error0;
}
/* Pull information for the target fd */
if (((fp = fget((int)sxp->sx_fdtarget)) == NULL) ||
((vp = vn_from_inode(fp->f_path.dentry->d_inode)) == NULL)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
ip = xfs_vtoi(vp);
if (ip == NULL) {
error = XFS_ERROR(EBADF);
goto error0;
}
if (((tfp = fget((int)sxp->sx_fdtmp)) == NULL) ||
((tvp = vn_from_inode(tfp->f_path.dentry->d_inode)) == NULL)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
tip = xfs_vtoi(tvp);
if (tip == NULL) {
error = XFS_ERROR(EBADF);
goto error0;
}
if (ip->i_mount != tip->i_mount) {
error = XFS_ERROR(EINVAL);
goto error0;
}
if (ip->i_ino == tip->i_ino) {
error = XFS_ERROR(EINVAL);
goto error0;
}
mp = ip->i_mount;
if (XFS_FORCED_SHUTDOWN(mp)) {
error = XFS_ERROR(EIO);
goto error0;
}
error = XFS_SWAP_EXTENTS(mp, &ip->i_iocore, &tip->i_iocore, sxp);
error0:
if (fp != NULL)
fput(fp);
if (tfp != NULL)
fput(tfp);
if (sxp != NULL)
kmem_free(sxp, sizeof(xfs_swapext_t));
return error;
}
