int
dm_set_region(
dm_sessid_t sid,
void *hanp,
size_t hlen,
dm_token_t token,
u_int nelem,
dm_region_t *regbufp,
dm_boolean_t *exactflagp)
{
dm_fsys_vector_t *fsys_vector;
dm_tokdata_t *tdp;
int error;
error = dm_app_get_tdp(sid, hanp, hlen, token, DM_TDT_REG,
DM_RIGHT_EXCL, &tdp);
if (error != 0)
return(error);
VN_BHV_READ_LOCK(VN_BHV_HEAD(tdp->td_vp));
fsys_vector = dm_fsys_vector(tdp->td_vp);
error = fsys_vector->set_region(tdp->td_vp, tdp->td_right,
nelem, regbufp, exactflagp);
VN_BHV_READ_UNLOCK(VN_BHV_HEAD(tdp->td_vp));
dm_app_put_tdp(tdp);
return(error);
}
int
dm_get_config_events(
void *hanp,
size_t hlen,
u_int nelem,
dm_eventset_t *eventsetp,
u_int *nelemp)
{
dm_fsys_vector_t *fsys_vector;
dm_tokdata_t *tdp;
int error;
/* Convert the handle into a vnode. */
if ((error = dm_get_config_tdp(hanp, hlen, &tdp)) != 0)
return(error);
/* Now call the filesystem-specific routine to determine the
events supported by that filesystem.
*/
VN_BHV_READ_LOCK(VN_BHV_HEAD(tdp->td_vp));
fsys_vector = dm_fsys_vector(tdp->td_vp);
error = fsys_vector->get_config_events(tdp->td_vp, tdp->td_right,
nelem, eventsetp, nelemp);
VN_BHV_READ_UNLOCK(VN_BHV_HEAD(tdp->td_vp));
dm_app_put_tdp(tdp);
return(error);
}
int
dm_get_config(
void *hanp,
size_t hlen,
dm_config_t flagname,
dm_size_t *retvalp)
{
dm_fsys_vector_t *fsys_vector;
dm_tokdata_t *tdp;
dm_size_t retval;
int system = 1;
int error;
/* Trap and process configuration parameters which are system-wide. */
switch (flagname) {
case DM_CONFIG_LEGACY:
case DM_CONFIG_PENDING:
case DM_CONFIG_OBJ_REF:
retval = DM_TRUE;
break;
case DM_CONFIG_MAX_MESSAGE_DATA:
retval = DM_MAX_MSG_DATA;
break;
default:
system = 0;
break;
}
if (system) {
if (copy_to_user(retvalp, &retval, sizeof(retval)))
return(EFAULT);
return(0);
}
/* Must be filesystem-specific. Convert the handle into a vnode. */
if ((error = dm_get_config_tdp(hanp, hlen, &tdp)) != 0)
return(error);
/* Now call the filesystem-specific routine to determine the
value of the configuration option for that filesystem.
*/
VN_BHV_READ_LOCK(VN_BHV_HEAD(tdp->td_vp));
fsys_vector = dm_fsys_vector(tdp->td_vp);
error = fsys_vector->get_config(tdp->td_vp, tdp->td_right,
flagname, retvalp);
VN_BHV_READ_UNLOCK(VN_BHV_HEAD(tdp->td_vp));
dm_app_put_tdp(tdp);
return(error);
}
struct vnode *
vn_initialize(
struct inode *inode)
{
struct vnode *vp = LINVFS_GET_VP(inode);
XFS_STATS_INC(vn_active);
XFS_STATS_INC(vn_alloc);
vp->v_flag = VMODIFIED;
spinlock_init(&vp->v_lock, "v_lock");
spin_lock(&vnumber_lock);
if (!++vn_generation) /* v_number shouldn't be zero */
vn_generation++;
vp->v_number = vn_generation;
spin_unlock(&vnumber_lock);
ASSERT(VN_CACHED(vp) == 0);
/* Initialize the first behavior and the behavior chain head. */
vn_bhv_head_init(VN_BHV_HEAD(vp), "vnode");
#ifdef XFS_VNODE_TRACE
vp->v_trace = ktrace_alloc(VNODE_TRACE_SIZE, KM_SLEEP);
#endif /* XFS_VNODE_TRACE */
vn_trace_exit(vp, "vn_initialize", (inst_t *)__return_address);
return vp;
}
void
xfs_initialize_vnode(
bhv_desc_t *bdp,
vnode_t *vp,
bhv_desc_t *inode_bhv,
int unlock)
{
xfs_inode_t *ip = XFS_BHVTOI(inode_bhv);
struct inode *inode = LINVFS_GET_IP(vp);
if (!inode_bhv->bd_vobj) {
vp->v_vfsp = bhvtovfs(bdp);
bhv_desc_init(inode_bhv, ip, vp, &xfs_vnodeops);
bhv_insert(VN_BHV_HEAD(vp), inode_bhv);
}
vp->v_type = IFTOVT(ip->i_d.di_mode);
/* Have we been called during the new inode create process,
* in which case we are too early to fill in the Linux inode.
*/
if (vp->v_type == VNON)
return;
xfs_revalidate_inode(XFS_BHVTOM(bdp), vp, ip);
/* For new inodes we need to set the ops vectors,
* and unlock the inode.
*/
if (unlock && (inode->i_state & I_NEW)) {
xfs_set_inodeops(inode);
unlock_new_inode(inode);
}
}
/*
* Get a XFS inode from a given vnode.
*/
xfs_inode_t *
xfs_vtoi(
struct vnode *vp)
{
bhv_desc_t *bdp;
bdp = bhv_lookup_range(VN_BHV_HEAD(vp),
VNODE_POSITION_XFS, VNODE_POSITION_XFS);
if (unlikely(bdp == NULL))
return NULL;
return XFS_BHVTOI(bdp);
}
static int
xfs_vn_allocate(xfs_mount_t *mp, xfs_inode_t *ip, struct xfs_vnode **vpp)
{
struct vnode *vp;
struct xfs_vnode *vdata;
int error;
/* Use zone allocator here? */
vdata = kmem_zalloc(sizeof(*vdata), KM_SLEEP);
error = getnewvnode("xfs", XVFSTOMNT(XFS_MTOVFS(mp)),
&xfs_vnops, &vp);
if (error) {
kmem_free(vdata, sizeof(*vdata));
return (error);
}
vp->v_vnlock->lk_flags |= LK_CANRECURSE;
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, curthread);
error = insmntque(vp, XVFSTOMNT(XFS_MTOVFS(mp)));
if (error != 0) {
kmem_free(vdata, sizeof(*vdata));
return (error);
}
vp->v_data = (void *)vdata;
vdata->v_number= 0;
vdata->v_inode = ip;
vdata->v_vfsp = XFS_MTOVFS(mp);
vdata->v_vnode = vp;
vn_bhv_head_init(VN_BHV_HEAD(vdata), "vnode");
#ifdef CONFIG_XFS_VNODE_TRACING
vp->v_trace = ktrace_alloc(VNODE_TRACE_SIZE, KM_SLEEP);
#endif /* CONFIG_XFS_VNODE_TRACING */
vn_trace_exit(vp, "vn_initialize", (inst_t *)__return_address);
if (error == 0)
*vpp = vdata;
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;
bhv_desc_t *bdp;
vp = VNAME_TO_VNODE(dentry);
bdp = vn_bhv_lookup_unlocked(VN_BHV_HEAD(vp), &xfs_vnodeops);
if (!bdp) {
*ipp = NULL;
return XFS_ERROR(ENOENT);
}
VN_HOLD(vp);
*ipp = XFS_BHVTOI(bdp);
return 0;
}
STATIC int
xfs_fssetdm_by_handle(
xfs_mount_t *mp,
void __user *arg,
struct file *parfilp,
struct inode *parinode)
{
int error;
struct fsdmidata fsd;
xfs_fsop_setdm_handlereq_t dmhreq;
struct inode *inode;
bhv_desc_t *bdp;
vnode_t *vp;
if (!capable(CAP_MKNOD))
return -XFS_ERROR(EPERM);
if (copy_from_user(&dmhreq, arg, sizeof(xfs_fsop_setdm_handlereq_t)))
return -XFS_ERROR(EFAULT);
error = xfs_vget_fsop_handlereq(mp, parinode, &dmhreq.hreq, &vp, &inode);
if (error)
return -error;
if (IS_IMMUTABLE(inode) || IS_APPEND(inode)) {
VN_RELE(vp);
return -XFS_ERROR(EPERM);
}
if (copy_from_user(&fsd, dmhreq.data, sizeof(fsd))) {
VN_RELE(vp);
return -XFS_ERROR(EFAULT);
}
bdp = bhv_base_unlocked(VN_BHV_HEAD(vp));
error = xfs_set_dmattrs(bdp, fsd.fsd_dmevmask, fsd.fsd_dmstate, NULL);
VN_RELE(vp);
if (error)
return -error;
return 0;
}
void
xfs_page_trace(
int tag,
struct inode *inode,
struct page *page,
int mask)
{
xfs_inode_t *ip;
bhv_desc_t *bdp;
vnode_t *vp = LINVFS_GET_VP(inode);
loff_t isize = i_size_read(inode);
loff_t offset = page->index << PAGE_CACHE_SHIFT;
int delalloc = -1, unmapped = -1, unwritten = -1;
if (page_has_buffers(page))
xfs_count_page_state(page, &delalloc, &unmapped, &unwritten);
bdp = vn_bhv_lookup(VN_BHV_HEAD(vp), &xfs_vnodeops);
ip = XFS_BHVTOI(bdp);
if (!ip->i_rwtrace)
return;
ktrace_enter(ip->i_rwtrace,
(void *)((unsigned long)tag),
(void *)ip,
(void *)inode,
(void *)page,
(void *)((unsigned long)mask),
(void *)((unsigned long)((ip->i_d.di_size >> 32) & 0xffffffff)),
(void *)((unsigned long)(ip->i_d.di_size & 0xffffffff)),
(void *)((unsigned long)((isize >> 32) & 0xffffffff)),
(void *)((unsigned long)(isize & 0xffffffff)),
(void *)((unsigned long)((offset >> 32) & 0xffffffff)),
(void *)((unsigned long)(offset & 0xffffffff)),
(void *)((unsigned long)delalloc),
(void *)((unsigned long)unmapped),
(void *)((unsigned long)unwritten),
(void *)NULL,
(void *)NULL);
}
/*
* 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,
unsigned long arg)
{
int hsize;
xfs_handle_t handle;
xfs_fsop_handlereq_t hreq;
struct inode *inode;
struct vnode *vp;
if (copy_from_user(&hreq, (xfs_fsop_handlereq_t *)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(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);
}
/* we need the vnode */
vp = LINVFS_GET_VP(inode);
if (vp->v_type != VREG && vp->v_type != VDIR && vp->v_type != VLNK) {
iput(inode);
return -XFS_ERROR(EBADF);
}
/* 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;
bhv_desc_t *bhv;
int lock_mode;
/* need to get access to the xfs_inode to read the generation */
bhv = vn_bhv_lookup_unlocked(VN_BHV_HEAD(vp), &xfs_vnodeops);
ASSERT(bhv);
ip = XFS_BHVTOI(bhv);
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((xfs_handle_t *)hreq.ohandle, &handle, hsize) ||
copy_to_user(hreq.ohandlen, &hsize, sizeof(__s32))) {
iput(inode);
return -XFS_ERROR(EFAULT);
}
iput(inode);
return 0;
}
/* Try to discard pages, in order to recycle a vcache entry.
*
* We also make some sanity checks: ref count, open count, held locks.
*
* We also do some non-VM-related chores, such as releasing the cred pointer
* (for AIX and Solaris) and releasing the gnode (for AIX).
*
* Locking: afs_xvcache lock is held. It must not be dropped.
*/
int
osi_VM_FlushVCache(struct vcache *avc)
{
int s, code;
vnode_t *vp = &avc->v;
if (avc->vrefCount != 0)
return EBUSY;
if (avc->opens != 0)
return EBUSY;
/*
* Just in case someone is still referring to the vnode we give up
* trying to get rid of this guy.
*/
if (CheckLock(&avc->lock) || LockWaiters(&avc->lock))
return EBUSY;
s = VN_LOCK(vp);
/*
* we just need to avoid the race
* in vn_rele between the ref count going to 0 and VOP_INACTIVE
* finishing up.
* Note that although we checked vcount above, we didn't have the lock
*/
if (vp->v_count > 0 || (vp->v_flag & VINACT)) {
VN_UNLOCK(vp, s);
return EBUSY;
}
VN_UNLOCK(vp, s);
/*
* Since we store on last close and on VOP_INACTIVE
* there should be NO dirty pages
* Note that we hold the xvcache lock the entire time.
*/
AFS_GUNLOCK();
PTOSSVP(vp, (off_t) 0, (off_t) MAXLONG);
AFS_GLOCK();
/* afs_chkpgoob will drop and re-acquire the global lock. */
afs_chkpgoob(vp, 0);
osi_Assert(!VN_GET_PGCNT(vp));
osi_Assert(!AFS_VN_MAPPED(vp));
osi_Assert(!AFS_VN_DIRTY(&avc->v));
#if defined(AFS_SGI65_ENV)
if (vp->v_filocks)
cleanlocks(vp, IGN_PID, 0);
mutex_destroy(&vp->v_filocksem);
#else /* AFS_SGI65_ENV */
if (vp->v_filocksem) {
if (vp->v_filocks)
#ifdef AFS_SGI64_ENV
cleanlocks(vp, &curprocp->p_flid);
#else
cleanlocks(vp, IGN_PID, 0);
#endif
osi_Assert(vp->v_filocks == NULL);
mutex_destroy(vp->v_filocksem);
kmem_free(vp->v_filocksem, sizeof *vp->v_filocksem);
vp->v_filocksem = NULL;
}
#endif /* AFS_SGI65_ENV */
if (avc->vrefCount)
osi_Panic("flushVcache: vm race");
#ifdef AFS_SGI64_ENV
AFS_GUNLOCK();
vnode_pcache_reclaim(vp); /* this can sleep */
vnode_pcache_free(vp);
if (vp->v_op != &Afs_vnodeops) {
VOP_RECLAIM(vp, FSYNC_WAIT, code);
}
AFS_GLOCK();
#ifdef AFS_SGI65_ENV
#ifdef VNODE_TRACING
ktrace_free(vp->v_trace);
#endif /* VNODE_TRACING */
vn_bhv_remove(VN_BHV_HEAD(vp), &(avc->vc_bhv_desc));
vn_bhv_head_destroy(&(vp->v_bh));
destroy_bitlock(&vp->v_pcacheflag);
mutex_destroy(&vp->v_buf_lock);
#else
bhv_remove(VN_BHV_HEAD(vp), &(avc->vc_bhv_desc));
bhv_head_destroy(&(vp->v_bh));
#endif
vp->v_flag = 0; /* debug */
#if defined(DEBUG) && defined(VNODE_INIT_BITLOCK)
destroy_bitlock(&vp->v_flag);
#endif
#ifdef INTR_KTHREADS
AFS_VN_DESTROY_BUF_LOCK(vp);
#endif
#endif /* AFS_SGI64_ENV */
return 0;
}
/*
* Syssgi interface for swapext
*/
int
xfs_swapext(
xfs_swapext_t __user *sxp)
{
xfs_swapext_t sx;
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;
bhv_desc_t *bdp, *tbdp;
vn_bhv_head_t *bhp, *tbhp;
uint lock_flags=0;
int ilf_fields, tilf_fields;
int error = 0;
xfs_ifork_t tempif, *ifp, *tifp;
__uint64_t tmp;
int aforkblks = 0;
int taforkblks = 0;
int locked = 0;
if (copy_from_user(&sx, sxp, sizeof(sx)))
return XFS_ERROR(EFAULT);
/* Pull information for the target fd */
if (((fp = fget((int)sx.sx_fdtarget)) == NULL) ||
((vp = LINVFS_GET_VP(fp->f_dentry->d_inode)) == NULL)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
bhp = VN_BHV_HEAD(vp);
bdp = vn_bhv_lookup(bhp, &xfs_vnodeops);
if (bdp == NULL) {
error = XFS_ERROR(EBADF);
goto error0;
} else {
ip = XFS_BHVTOI(bdp);
}
if (((tfp = fget((int)sx.sx_fdtmp)) == NULL) ||
((tvp = LINVFS_GET_VP(tfp->f_dentry->d_inode)) == NULL)) {
error = XFS_ERROR(EINVAL);
goto error0;
}
tbhp = VN_BHV_HEAD(tvp);
tbdp = vn_bhv_lookup(tbhp, &xfs_vnodeops);
if (tbdp == NULL) {
error = XFS_ERROR(EBADF);
goto error0;
} else {
tip = XFS_BHVTOI(tbdp);
}
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 = &sx.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;
}
lock_flags = XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL;
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_pages(XFS_ITOV(tip), &(tip->i_iocore),
(xfs_off_t)0, 0, 0);
/* Verify O_DIRECT for ftmp */
if (VN_CACHED(tvp) != 0) {
error = XFS_ERROR(EINVAL);
goto error0;
}
/* Verify all data are being swapped */
if (sx.sx_offset != 0 ||
sx.sx_length != ip->i_d.di_size ||
sx.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);
return error;
}
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_iunlock(ip, lock_flags);
xfs_iunlock(tip, lock_flags);
xfs_trans_cancel(tp, 0);
return error;
}
}
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_iunlock(ip, lock_flags);
xfs_iunlock(tip, lock_flags);
xfs_trans_cancel(tp, 0);
return error;
}
}
/*
* Swap the data forks of the inodes
*/
ifp = &ip->i_df;
tifp = &tip->i_df;
tempif = *ifp; /* struct copy */
*ifp = *tifp; /* struct copy */
*tifp = tempif; /* 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);
fput(fp);
fput(tfp);
return error;
error0:
if (locked) {
xfs_iunlock(ip, lock_flags);
xfs_iunlock(tip, lock_flags);
}
if (fp != NULL) fput(fp);
if (tfp != NULL) fput(tfp);
return error;
}
/*
* xfs_rename
*/
int
xfs_rename(
bhv_desc_t *src_dir_bdp,
vname_t *src_vname,
vnode_t *target_dir_vp,
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? */
vnode_t *src_dir_vp;
bhv_desc_t *target_dir_bdp;
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_dir_bdp = vn_bhv_lookup_unlocked(VN_BHV_HEAD(target_dir_vp),
&xfs_vnodeops);
if (target_dir_bdp == NULL) {
return XFS_ERROR(EXDEV);
}
src_dp = XFS_BHVTOI(src_dir_bdp);
target_dp = XFS_BHVTOI(target_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;
}
}
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(mp, 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(mp, 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->i_mount, 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(mp, 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(mp, 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(mp, 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, first_block, &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, NULL);
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) {
VOP_LINK_REMOVED(XFS_ITOV(target_ip), target_dir_vp,
target_link_zero);
IRELE(target_ip);
}
FSC_NOTIFY_NAME_CHANGED(XFS_ITOV(src_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;
}
