/*
* Clean a vnode of filesystem-specific data and prepare it for reuse.
*/
STATIC int
vn_reclaim(
struct vnode *vp)
{
int error;
XFS_STATS_INC(vn_reclaim);
vn_trace_entry(vp, "vn_reclaim", (inst_t *)__return_address);
/*
* Only make the VOP_RECLAIM call if there are behaviors
* to call.
*/
if (vp->v_fbhv) {
VOP_RECLAIM(vp, error);
if (error)
return -error;
}
ASSERT(vp->v_fbhv == NULL);
VN_LOCK(vp);
vp->v_flag &= (VRECLM|VWAIT);
VN_UNLOCK(vp, 0);
vp->v_type = VNON;
vp->v_fbhv = NULL;
#ifdef XFS_VNODE_TRACE
ktrace_free(vp->v_trace);
vp->v_trace = NULL;
#endif
return 0;
}
/*
* purge a vnode from the cache
* At this point the vnode is guaranteed to have no references (vn_count == 0)
* The caller has to make sure that there are no ways someone could
* get a handle (via vn_get) on the vnode (usually done via a mount/vfs lock).
*/
void
vn_purge(
struct vnode *vp,
vmap_t *vmap)
{
vn_trace_entry(vp, "vn_purge", (inst_t *)__return_address);
again:
/*
* Check whether vp has already been reclaimed since our caller
* sampled its version while holding a filesystem cache lock that
* its VOP_RECLAIM function acquires.
*/
VN_LOCK(vp);
if (vp->v_number != vmap->v_number) {
VN_UNLOCK(vp, 0);
return;
}
/*
* If vp is being reclaimed or inactivated, wait until it is inert,
* then proceed. Can't assume that vnode is actually reclaimed
* just because the reclaimed flag is asserted -- a vn_alloc
* reclaim can fail.
*/
if (vp->v_flag & (VINACT | VRECLM)) {
ASSERT(vn_count(vp) == 0);
vp->v_flag |= VWAIT;
sv_wait(vptosync(vp), PINOD, &vp->v_lock, 0);
goto again;
}
/*
* Another process could have raced in and gotten this vnode...
*/
if (vn_count(vp) > 0) {
VN_UNLOCK(vp, 0);
return;
}
XFS_STATS_DEC(vn_active);
vp->v_flag |= VRECLM;
VN_UNLOCK(vp, 0);
/*
* Call VOP_RECLAIM and clean vp. The FSYNC_INVAL flag tells
* vp's filesystem to flush and invalidate all cached resources.
* When vn_reclaim returns, vp should have no private data,
* either in a system cache or attached to v_data.
*/
if (vn_reclaim(vp) != 0)
panic("vn_purge: cannot reclaim");
/*
* Wakeup anyone waiting for vp to be reclaimed.
*/
vn_wakeup(vp);
}
ssize_t
xfs_sendfile(
bhv_desc_t *bdp,
struct file *filp,
loff_t *offset,
int ioflags,
size_t count,
read_actor_t actor,
void *target,
cred_t *credp)
{
ssize_t ret;
xfs_fsize_t n;
xfs_inode_t *ip;
xfs_mount_t *mp;
vnode_t *vp;
ip = XFS_BHVTOI(bdp);
vp = BHV_TO_VNODE(bdp);
mp = ip->i_mount;
vn_trace_entry(vp, "xfs_sendfile", (inst_t *)__return_address);
XFS_STATS_INC(xs_read_calls);
n = XFS_MAXIOFFSET(mp) - *offset;
if ((n <= 0) || (count == 0))
return 0;
if (n < count)
count = n;
if (XFS_FORCED_SHUTDOWN(ip->i_mount))
return -EIO;
xfs_ilock(ip, XFS_IOLOCK_SHARED);
if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ) &&
(!(ioflags & IO_INVIS))) {
vrwlock_t locktype = VRWLOCK_READ;
int error;
error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), *offset, count,
FILP_DELAY_FLAG(filp), &locktype);
if (error) {
xfs_iunlock(ip, XFS_IOLOCK_SHARED);
return -error;
}
}
ret = generic_file_sendfile(filp, offset, count, actor, target);
xfs_iunlock(ip, XFS_IOLOCK_SHARED);
XFS_STATS_ADD(xs_read_bytes, ret);
xfs_ichgtime(ip, XFS_ICHGTIME_ACC);
return ret;
}
STATIC void
linvfs_clear_inode(
struct inode *inode)
{
vnode_t *vp = LINVFS_GET_VP(inode);
if (vp) {
vn_rele(vp);
vn_trace_entry(vp, __FUNCTION__, (inst_t *)__return_address);
/*
* Do all our cleanup, and remove this vnode.
*/
vn_remove(vp);
}
}
/*
* Attempt to flush the inode, this will actually fail
* if the inode is pinned, but we dirty the inode again
* at the point when it is unpinned after a log write,
* since this is when the inode itself becomes flushable.
*/
STATIC void
linvfs_write_inode(
struct inode *inode,
int sync)
{
vnode_t *vp = LINVFS_GET_VP(inode);
int error, flags = FLUSH_INODE;
if (vp) {
vn_trace_entry(vp, __FUNCTION__, (inst_t *)__return_address);
if (sync)
flags |= FLUSH_SYNC;
VOP_IFLUSH(vp, flags, error);
}
}
int
xfs_dir_lookup_int(
bhv_desc_t *dir_bdp,
uint lock_mode,
vname_t *dentry,
xfs_ino_t *inum,
xfs_inode_t **ipp)
{
vnode_t *dir_vp;
xfs_inode_t *dp;
int error;
dir_vp = BHV_TO_VNODE(dir_bdp);
vn_trace_entry(dir_vp, __FUNCTION__, (inst_t *)__return_address);
dp = XFS_BHVTOI(dir_bdp);
error = XFS_DIR_LOOKUP(dp->i_mount, NULL, dp,
VNAME(dentry), VNAMELEN(dentry), inum);
if (!error) {
/*
* Unlock the directory. We do this because we can't
* hold the directory lock while doing the vn_get()
* in xfs_iget(). Doing so could cause us to hold
* a lock while waiting for the inode to finish
* being inactive while it's waiting for a log
* reservation in the inactive routine.
*/
xfs_iunlock(dp, lock_mode);
error = xfs_iget(dp->i_mount, NULL, *inum, 0, 0, ipp, 0);
xfs_ilock(dp, lock_mode);
if (error) {
*ipp = NULL;
} else if ((*ipp)->i_d.di_mode == 0) {
/*
* The inode has been freed. Something is
* wrong so just get out of here.
*/
xfs_iunlock(dp, lock_mode);
xfs_iput_new(*ipp, 0);
*ipp = NULL;
xfs_ilock(dp, lock_mode);
error = XFS_ERROR(ENOENT);
}
}
return error;
}
/*
* Call VOP_INACTIVE on last reference.
*/
void
vn_rele(
struct vnode *vp)
{
int vcnt;
int cache;
XFS_STATS_INC(vn_rele);
VN_LOCK(vp);
vn_trace_entry(vp, "vn_rele", (inst_t *)__return_address);
vcnt = vn_count(vp);
/*
* Since we always get called from put_inode we know
* that i_count won't be decremented after we
* return.
*/
if (!vcnt) {
/*
* As soon as we turn this on, noone can find us in vn_get
* until we turn off VINACT or VRECLM
*/
vp->v_flag |= VINACT;
VN_UNLOCK(vp, 0);
/*
* Do not make the VOP_INACTIVE call if there
* are no behaviors attached to the vnode to call.
*/
if (vp->v_fbhv)
VOP_INACTIVE(vp, NULL, cache);
VN_LOCK(vp);
if (vp->v_flag & VWAIT)
sv_broadcast(vptosync(vp));
vp->v_flag &= ~(VINACT|VWAIT|VRECLM|VMODIFIED);
}
VN_UNLOCK(vp, 0);
vn_trace_exit(vp, "vn_rele", (inst_t *)__return_address);
}
/*
* Revalidate the Linux inode from the vnode.
*/
int
vn_revalidate(
struct vnode *vp)
{
vattr_t va;
int error;
vn_trace_entry(vp, "vn_revalidate", (inst_t *)__return_address);
ASSERT(vp->v_fbhv != NULL);
va.va_mask = XFS_AT_STAT|XFS_AT_XFLAGS;
VOP_GETATTR(vp, &va, 0, NULL, error);
if (!error) {
vn_revalidate_core(vp, &va);
VUNMODIFY(vp);
}
return -error;
}
/*
* Revalidate the Linux inode from the vnode.
*/
int
vn_revalidate(
struct vnode *vp)
{
struct inode *inode;
vattr_t va;
int error;
vn_trace_entry(vp, "vn_revalidate", (inst_t *)__return_address);
ASSERT(vp->v_fbhv != NULL);
va.va_mask = XFS_AT_STAT|XFS_AT_XFLAGS;
VOP_GETATTR(vp, &va, 0, NULL, error);
if (!error) {
inode = LINVFS_GET_IP(vp);
inode->i_mode = VTTOIF(va.va_type) | va.va_mode;
inode->i_nlink = va.va_nlink;
inode->i_uid = va.va_uid;
inode->i_gid = va.va_gid;
inode->i_blocks = va.va_nblocks;
inode->i_mtime = va.va_mtime;
inode->i_ctime = va.va_ctime;
inode->i_atime = va.va_atime;
if (va.va_xflags & XFS_XFLAG_IMMUTABLE)
inode->i_flags |= S_IMMUTABLE;
else
inode->i_flags &= ~S_IMMUTABLE;
if (va.va_xflags & XFS_XFLAG_APPEND)
inode->i_flags |= S_APPEND;
else
inode->i_flags &= ~S_APPEND;
if (va.va_xflags & XFS_XFLAG_SYNC)
inode->i_flags |= S_SYNC;
else
inode->i_flags &= ~S_SYNC;
if (va.va_xflags & XFS_XFLAG_NOATIME)
inode->i_flags |= S_NOATIME;
else
inode->i_flags &= ~S_NOATIME;
VUNMODIFY(vp);
}
return -error;
}
/*
* Special iput for brand-new inodes that are still locked
*/
void
xfs_iput_new(xfs_inode_t *ip,
uint lock_flags)
{
xfs_vnode_t *vp = XFS_ITOV(ip);
vn_trace_entry(vp, "xfs_iput_new", (inst_t *)__return_address);
printf("xfs_iput_new: ip %p\n",ip);
if ((ip->i_d.di_mode == 0)) {
ASSERT(!(ip->i_flags & XFS_IRECLAIMABLE));
//vn_mark_bad(vp);
printf("xfs_iput_new: ip %p di_mode == 0\n",ip);
/* mabe call vgone here? RMC */
}
if (lock_flags)
xfs_iunlock(ip, lock_flags);
ASSERT_VOP_LOCKED(vp->v_vnode, "xfs_iput_new");
vput(vp->v_vnode);
}
int
xfs_ioctl(
bhv_desc_t *bdp,
struct inode *inode,
struct file *filp,
int ioflags,
unsigned int cmd,
unsigned long arg)
{
int error;
vnode_t *vp;
xfs_inode_t *ip;
xfs_mount_t *mp;
vp = LINVFS_GET_VP(inode);
vn_trace_entry(vp, "xfs_ioctl", (inst_t *)__return_address);
ip = XFS_BHVTOI(bdp);
mp = ip->i_mount;
switch (cmd) {
case XFS_IOC_ALLOCSP:
case XFS_IOC_FREESP:
case XFS_IOC_RESVSP:
case XFS_IOC_UNRESVSP:
case XFS_IOC_ALLOCSP64:
case XFS_IOC_FREESP64:
case XFS_IOC_RESVSP64:
case XFS_IOC_UNRESVSP64:
/*
* Only allow the sys admin to reserve space unless
* unwritten extents are enabled.
*/
if (!XFS_SB_VERSION_HASEXTFLGBIT(&mp->m_sb) &&
!capable(CAP_SYS_ADMIN))
return -EPERM;
return xfs_ioc_space(bdp, vp, filp, ioflags, cmd, arg);
case XFS_IOC_DIOINFO: {
struct dioattr da;
da.d_miniosz = mp->m_sb.sb_blocksize;
da.d_mem = mp->m_sb.sb_blocksize;
/*
* this only really needs to be BBSIZE.
* it is set to the file system block size to
* avoid having to do block zeroing on short writes.
*/
da.d_maxiosz = XFS_FSB_TO_B(mp,
XFS_B_TO_FSBT(mp, KIO_MAX_ATOMIC_IO << 10));
if (copy_to_user((struct dioattr *)arg, &da, sizeof(da)))
return -XFS_ERROR(EFAULT);
return 0;
}
case XFS_IOC_FSBULKSTAT_SINGLE:
case XFS_IOC_FSBULKSTAT:
case XFS_IOC_FSINUMBERS:
return xfs_ioc_bulkstat(mp, cmd, arg);
case XFS_IOC_FSGEOMETRY_V1:
return xfs_ioc_fsgeometry_v1(mp, arg);
case XFS_IOC_FSGEOMETRY:
return xfs_ioc_fsgeometry(mp, arg);
case XFS_IOC_GETVERSION:
case XFS_IOC_GETXFLAGS:
case XFS_IOC_SETXFLAGS:
case XFS_IOC_FSGETXATTR:
case XFS_IOC_FSSETXATTR:
case XFS_IOC_FSGETXATTRA:
return xfs_ioc_xattr(vp, ip, filp, cmd, arg);
case XFS_IOC_FSSETDM: {
struct fsdmidata dmi;
if (copy_from_user(&dmi, (struct fsdmidata *)arg, sizeof(dmi)))
return -XFS_ERROR(EFAULT);
error = xfs_set_dmattrs(bdp, dmi.fsd_dmevmask, dmi.fsd_dmstate,
NULL);
return -error;
}
case XFS_IOC_GETBMAP:
case XFS_IOC_GETBMAPA:
return xfs_ioc_getbmap(bdp, filp, ioflags, cmd, arg);
case XFS_IOC_GETBMAPX:
return xfs_ioc_getbmapx(bdp, arg);
case XFS_IOC_FD_TO_HANDLE:
case XFS_IOC_PATH_TO_HANDLE:
case XFS_IOC_PATH_TO_FSHANDLE:
return xfs_find_handle(cmd, arg);
case XFS_IOC_OPEN_BY_HANDLE:
return xfs_open_by_handle(mp, arg, filp, inode);
case XFS_IOC_FSSETDM_BY_HANDLE:
return xfs_fssetdm_by_handle(mp, arg, filp, inode);
case XFS_IOC_READLINK_BY_HANDLE:
return xfs_readlink_by_handle(mp, arg, filp, inode);
case XFS_IOC_ATTRLIST_BY_HANDLE:
return xfs_attrlist_by_handle(mp, arg, filp, inode);
case XFS_IOC_ATTRMULTI_BY_HANDLE:
return xfs_attrmulti_by_handle(mp, arg, filp, inode);
case XFS_IOC_SWAPEXT: {
error = xfs_swapext((struct xfs_swapext *)arg);
return -error;
}
case XFS_IOC_FSCOUNTS: {
xfs_fsop_counts_t out;
error = xfs_fs_counts(mp, &out);
if (error)
return -error;
if (copy_to_user((char *)arg, &out, sizeof(out)))
return -XFS_ERROR(EFAULT);
return 0;
}
case XFS_IOC_SET_RESBLKS: {
xfs_fsop_resblks_t inout;
__uint64_t in;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (copy_from_user(&inout, (char *)arg, sizeof(inout)))
return -XFS_ERROR(EFAULT);
/* input parameter is passed in resblks field of structure */
in = inout.resblks;
error = xfs_reserve_blocks(mp, &in, &inout);
if (error)
return -error;
if (copy_to_user((char *)arg, &inout, sizeof(inout)))
return -XFS_ERROR(EFAULT);
return 0;
}
case XFS_IOC_GET_RESBLKS: {
xfs_fsop_resblks_t out;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
error = xfs_reserve_blocks(mp, NULL, &out);
if (error)
return -error;
if (copy_to_user((char *)arg, &out, sizeof(out)))
return -XFS_ERROR(EFAULT);
return 0;
}
case XFS_IOC_FSGROWFSDATA: {
xfs_growfs_data_t in;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (copy_from_user(&in, (char *)arg, sizeof(in)))
return -XFS_ERROR(EFAULT);
error = xfs_growfs_data(mp, &in);
return -error;
}
case XFS_IOC_FSGROWFSLOG: {
xfs_growfs_log_t in;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (copy_from_user(&in, (char *)arg, sizeof(in)))
return -XFS_ERROR(EFAULT);
error = xfs_growfs_log(mp, &in);
return -error;
}
case XFS_IOC_FSGROWFSRT: {
xfs_growfs_rt_t in;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (copy_from_user(&in, (char *)arg, sizeof(in)))
return -XFS_ERROR(EFAULT);
error = xfs_growfs_rt(mp, &in);
return -error;
}
case XFS_IOC_FREEZE:
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (vp->v_vfsp->vfs_frozen == SB_UNFROZEN) {
freeze_bdev(mp->m_ddev_targp->pbr_bdev);
if (mp->m_rtdev_targp)
freeze_bdev(mp->m_rtdev_targp->pbr_bdev);
}
return 0;
case XFS_IOC_THAW:
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (vp->v_vfsp->vfs_frozen != SB_UNFROZEN) {
thaw_bdev(mp->m_ddev_targp->pbr_bdev, inode->i_sb);
if (mp->m_rtdev_targp)
thaw_bdev(mp->m_ddev_targp->pbr_bdev, NULL);
}
return 0;
case XFS_IOC_GOINGDOWN: {
__uint32_t in;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
if (get_user(in, (__uint32_t *)arg))
return -XFS_ERROR(EFAULT);
error = xfs_fs_goingdown(mp, in);
return -error;
}
case XFS_IOC_ERROR_INJECTION: {
xfs_error_injection_t in;
if (!capable(CAP_SYS_ADMIN))
return EPERM;
if (copy_from_user(&in, (char *)arg, sizeof(in)))
return -XFS_ERROR(EFAULT);
error = xfs_errortag_add(in.errtag, mp);
return -error;
}
case XFS_IOC_ERROR_CLEARALL:
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
error = xfs_errortag_clearall(mp);
return -error;
default:
return -ENOTTY;
}
}
/*
* 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;
}
ssize_t /* bytes written, or (-) error */
xfs_write(
bhv_desc_t *bdp,
struct kiocb *iocb,
const struct iovec *iovp,
unsigned int segs,
loff_t *offset,
int ioflags,
cred_t *credp)
{
struct file *file = iocb->ki_filp;
size_t size = 0;
xfs_inode_t *xip;
xfs_mount_t *mp;
ssize_t ret;
int error = 0;
xfs_fsize_t isize, new_size;
xfs_fsize_t n, limit;
xfs_iocore_t *io;
vnode_t *vp;
unsigned long seg;
int iolock;
int eventsent = 0;
vrwlock_t locktype;
XFS_STATS_INC(xs_write_calls);
vp = BHV_TO_VNODE(bdp);
vn_trace_entry(vp, "xfs_write", (inst_t *)__return_address);
xip = XFS_BHVTOI(bdp);
/* START copy & waste from filemap.c */
for (seg = 0; seg < segs; seg++) {
const struct iovec *iv = &iovp[seg];
/*
* If any segment has a negative length, or the cumulative
* length ever wraps negative then return -EINVAL.
*/
size += iv->iov_len;
if (unlikely((ssize_t)(size|iv->iov_len) < 0))
return XFS_ERROR(-EINVAL);
}
/* END copy & waste from filemap.c */
if (size == 0)
return 0;
io = &(xip->i_iocore);
mp = io->io_mount;
xfs_check_frozen(mp, bdp, XFS_FREEZE_WRITE);
if (XFS_FORCED_SHUTDOWN(mp)) {
return -EIO;
}
if (ioflags & IO_ISDIRECT) {
pb_target_t *target =
(xip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
mp->m_rtdev_targp : mp->m_ddev_targp;
if ((*offset & target->pbr_smask) ||
(size & target->pbr_smask)) {
return XFS_ERROR(-EINVAL);
}
iolock = XFS_IOLOCK_SHARED;
locktype = VRWLOCK_WRITE_DIRECT;
} else {
iolock = XFS_IOLOCK_EXCL;
locktype = VRWLOCK_WRITE;
}
xfs_ilock(xip, XFS_ILOCK_EXCL|iolock);
isize = xip->i_d.di_size;
limit = XFS_MAXIOFFSET(mp);
if (file->f_flags & O_APPEND)
*offset = isize;
start:
n = limit - *offset;
if (n <= 0) {
xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock);
return -EFBIG;
}
if (n < size)
size = n;
new_size = *offset + size;
if (new_size > isize) {
io->io_new_size = new_size;
}
if ((DM_EVENT_ENABLED(vp->v_vfsp, xip, DM_EVENT_WRITE) &&
!(ioflags & IO_INVIS) && !eventsent)) {
loff_t savedsize = *offset;
xfs_iunlock(xip, XFS_ILOCK_EXCL);
error = XFS_SEND_DATA(xip->i_mount, DM_EVENT_WRITE, vp,
*offset, size,
FILP_DELAY_FLAG(file), &locktype);
if (error) {
xfs_iunlock(xip, iolock);
return -error;
}
xfs_ilock(xip, XFS_ILOCK_EXCL);
eventsent = 1;
/*
* The iolock was dropped and reaquired in XFS_SEND_DATA
* so we have to recheck the size when appending.
* We will only "goto start;" once, since having sent the
* event prevents another call to XFS_SEND_DATA, which is
* what allows the size to change in the first place.
*/
if ((file->f_flags & O_APPEND) &&
savedsize != xip->i_d.di_size) {
*offset = isize = xip->i_d.di_size;
goto start;
}
}
/*
* On Linux, generic_file_write updates the times even if
* no data is copied in so long as the write had a size.
*
* We must update xfs' times since revalidate will overcopy xfs.
*/
if (size && !(ioflags & IO_INVIS))
xfs_ichgtime(xip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
/*
* If the offset is beyond the size of the file, we have a couple
* of things to do. First, if there is already space allocated
* we need to either create holes or zero the disk or ...
*
* If there is a page where the previous size lands, we need
* to zero it out up to the new size.
*/
if (!(ioflags & IO_ISDIRECT) && (*offset > isize && isize)) {
error = xfs_zero_eof(BHV_TO_VNODE(bdp), io, *offset,
isize, *offset + size);
if (error) {
xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock);
return(-error);
}
}
xfs_iunlock(xip, XFS_ILOCK_EXCL);
/*
* If we're writing the file then make sure to clear the
* setuid and setgid bits if the process is not being run
* by root. This keeps people from modifying setuid and
* setgid binaries.
*/
if (((xip->i_d.di_mode & S_ISUID) ||
((xip->i_d.di_mode & (S_ISGID | S_IXGRP)) ==
(S_ISGID | S_IXGRP))) &&
!capable(CAP_FSETID)) {
error = xfs_write_clear_setuid(xip);
if (error) {
xfs_iunlock(xip, iolock);
return -error;
}
}
retry:
if (ioflags & IO_ISDIRECT) {
xfs_inval_cached_pages(vp, &xip->i_iocore, *offset, 1, 1);
}
ret = generic_file_aio_write_nolock(iocb, iovp, segs, offset);
if ((ret == -ENOSPC) &&
DM_EVENT_ENABLED(vp->v_vfsp, xip, DM_EVENT_NOSPACE) &&
!(ioflags & IO_INVIS)) {
xfs_rwunlock(bdp, locktype);
error = XFS_SEND_NAMESP(xip->i_mount, DM_EVENT_NOSPACE, vp,
DM_RIGHT_NULL, vp, DM_RIGHT_NULL, NULL, NULL,
0, 0, 0); /* Delay flag intentionally unused */
if (error)
return -error;
xfs_rwlock(bdp, locktype);
*offset = xip->i_d.di_size;
goto retry;
}
if (*offset > xip->i_d.di_size) {
xfs_ilock(xip, XFS_ILOCK_EXCL);
if (*offset > xip->i_d.di_size) {
struct inode *inode = LINVFS_GET_IP(vp);
xip->i_d.di_size = *offset;
i_size_write(inode, *offset);
xip->i_update_core = 1;
xip->i_update_size = 1;
}
xfs_iunlock(xip, XFS_ILOCK_EXCL);
}
if (ret <= 0) {
xfs_rwunlock(bdp, locktype);
return ret;
}
XFS_STATS_ADD(xs_write_bytes, ret);
/* Handle various SYNC-type writes */
if ((file->f_flags & O_SYNC) || IS_SYNC(file->f_dentry->d_inode)) {
/*
* If we're treating this as O_DSYNC and we have not updated the
* size, force the log.
*/
if (!(mp->m_flags & XFS_MOUNT_OSYNCISOSYNC)
&& !(xip->i_update_size)) {
/*
* If an allocation transaction occurred
* without extending the size, then we have to force
* the log up the proper point to ensure that the
* allocation is permanent. We can't count on
* the fact that buffered writes lock out direct I/O
* writes - the direct I/O write could have extended
* the size nontransactionally, then finished before
* we started. xfs_write_file will think that the file
* didn't grow but the update isn't safe unless the
* size change is logged.
*
* Force the log if we've committed a transaction
* against the inode or if someone else has and
* the commit record hasn't gone to disk (e.g.
* the inode is pinned). This guarantees that
* all changes affecting the inode are permanent
* when we return.
*/
xfs_inode_log_item_t *iip;
xfs_lsn_t lsn;
iip = xip->i_itemp;
if (iip && iip->ili_last_lsn) {
lsn = iip->ili_last_lsn;
xfs_log_force(mp, lsn,
XFS_LOG_FORCE | XFS_LOG_SYNC);
} else if (xfs_ipincount(xip) > 0) {
xfs_log_force(mp, (xfs_lsn_t)0,
XFS_LOG_FORCE | XFS_LOG_SYNC);
}
} else {
xfs_trans_t *tp;
/*
* O_SYNC or O_DSYNC _with_ a size update are handled
* the same way.
*
* If the write was synchronous then we need to make
* sure that the inode modification time is permanent.
* We'll have updated the timestamp above, so here
* we use a synchronous transaction to log the inode.
* It's not fast, but it's necessary.
*
* If this a dsync write and the size got changed
* non-transactionally, then we need to ensure that
* the size change gets logged in a synchronous
* transaction.
*/
tp = xfs_trans_alloc(mp, XFS_TRANS_WRITE_SYNC);
if ((error = xfs_trans_reserve(tp, 0,
XFS_SWRITE_LOG_RES(mp),
0, 0, 0))) {
/* Transaction reserve failed */
xfs_trans_cancel(tp, 0);
} else {
/* Transaction reserve successful */
xfs_ilock(xip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, xip, XFS_ILOCK_EXCL);
xfs_trans_ihold(tp, xip);
xfs_trans_log_inode(tp, xip, XFS_ILOG_CORE);
xfs_trans_set_sync(tp);
error = xfs_trans_commit(tp, 0, (xfs_lsn_t)0);
xfs_iunlock(xip, XFS_ILOCK_EXCL);
}
}
} /* (ioflags & O_SYNC) */
xfs_rwunlock(bdp, locktype);
return(ret);
}
ssize_t /* bytes read, or (-) error */
xfs_read(
bhv_desc_t *bdp,
struct kiocb *iocb,
const struct iovec *iovp,
unsigned int segs,
loff_t *offset,
int ioflags,
cred_t *credp)
{
struct file *file = iocb->ki_filp;
size_t size = 0;
ssize_t ret;
xfs_fsize_t n;
xfs_inode_t *ip;
xfs_mount_t *mp;
vnode_t *vp;
unsigned long seg;
ip = XFS_BHVTOI(bdp);
vp = BHV_TO_VNODE(bdp);
mp = ip->i_mount;
vn_trace_entry(vp, "xfs_read", (inst_t *)__return_address);
XFS_STATS_INC(xs_read_calls);
/* START copy & waste from filemap.c */
for (seg = 0; seg < segs; seg++) {
const struct iovec *iv = &iovp[seg];
/*
* If any segment has a negative length, or the cumulative
* length ever wraps negative then return -EINVAL.
*/
size += iv->iov_len;
if (unlikely((ssize_t)(size|iv->iov_len) < 0))
return XFS_ERROR(-EINVAL);
}
/* END copy & waste from filemap.c */
if (ioflags & IO_ISDIRECT) {
pb_target_t *target =
(ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
mp->m_rtdev_targp : mp->m_ddev_targp;
if ((*offset & target->pbr_smask) ||
(size & target->pbr_smask)) {
if (*offset == ip->i_d.di_size) {
return (0);
}
return -XFS_ERROR(EINVAL);
}
}
n = XFS_MAXIOFFSET(mp) - *offset;
if ((n <= 0) || (size == 0))
return 0;
if (n < size)
size = n;
if (XFS_FORCED_SHUTDOWN(mp)) {
return -EIO;
}
/* OK so we are holding the I/O lock for the duration
* of the submission, then what happens if the I/O
* does not really happen here, but is scheduled
* later?
*/
xfs_ilock(ip, XFS_IOLOCK_SHARED);
if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ) &&
!(ioflags & IO_INVIS)) {
int error;
vrwlock_t locktype = VRWLOCK_READ;
error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), *offset, size,
FILP_DELAY_FLAG(file), &locktype);
if (error) {
xfs_iunlock(ip, XFS_IOLOCK_SHARED);
return -error;
}
}
ret = __generic_file_aio_read(iocb, iovp, segs, offset);
xfs_iunlock(ip, XFS_IOLOCK_SHARED);
if (ret > 0)
XFS_STATS_ADD(xs_read_bytes, ret);
if (likely(!(ioflags & IO_INVIS)))
xfs_ichgtime(ip, XFS_ICHGTIME_ACC);
return ret;
}
