/*
* Rewrite an existing directory entry to point at the inode
* supplied. The parameters describing the directory entry are
* set up by a call to namei.
*/
int
ufs_dirrewrite(struct inode *dp, struct inode *oip, ufsino_t newinum,
int newtype, int isrmdir)
{
struct buf *bp;
struct direct *ep;
struct vnode *vdp = ITOV(dp);
int error;
error = UFS_BUFATOFF(dp, (off_t)dp->i_offset, (char **)&ep, &bp);
if (error)
return (error);
ep->d_ino = newinum;
if (vdp->v_mount->mnt_maxsymlinklen > 0)
ep->d_type = newtype;
oip->i_effnlink--;
if (DOINGSOFTDEP(vdp)) {
softdep_change_linkcnt(oip, 0);
softdep_setup_directory_change(bp, dp, oip, newinum, isrmdir);
bdwrite(bp);
} else {
DIP_ADD(oip, nlink, -1);
oip->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(oip, MNT_WAIT);
if (DOINGASYNC(vdp)) {
bdwrite(bp);
error = 0;
} else {
error = VOP_BWRITE(bp);
}
}
dp->i_flag |= IN_CHANGE | IN_UPDATE;
UFS_WAPBL_UPDATE(dp, MNT_WAIT);
return (error);
}
/*
* Remove a directory entry after a call to namei, using
* the parameters which it left in nameidata. The entry
* dp->i_offset contains the offset into the directory of the
* entry to be eliminated. The dp->i_count field contains the
* size of the previous record in the directory. If this
* is 0, the first entry is being deleted, so we need only
* zero the inode number to mark the entry as free. If the
* entry is not the first in the directory, we must reclaim
* the space of the now empty record by adding the record size
* to the size of the previous entry.
*/
int
ufs_dirremove(struct vnode *dvp, struct inode *ip, int flags, int isrmdir)
{
struct inode *dp;
struct direct *ep;
struct buf *bp;
int error;
UFS_WAPBL_JLOCK_ASSERT(dvp->v_mount);
dp = VTOI(dvp);
if ((error = UFS_BUFATOFF(dp,
(off_t)(dp->i_offset - dp->i_count), (char **)&ep, &bp)) != 0)
return (error);
#ifdef UFS_DIRHASH
/*
* Remove the dirhash entry. This is complicated by the fact
* that `ep' is the previous entry when dp->i_count != 0.
*/
if (dp->i_dirhash != NULL)
ufsdirhash_remove(dp, (dp->i_count == 0) ? ep :
(struct direct *)((char *)ep + ep->d_reclen), dp->i_offset);
#endif
if (dp->i_count == 0) {
/*
* First entry in block: set d_ino to zero.
*/
ep->d_ino = 0;
} else {
/*
* Collapse new free space into previous entry.
*/
ep->d_reclen += dp->i_reclen;
}
#ifdef UFS_DIRHASH
if (dp->i_dirhash != NULL)
ufsdirhash_checkblock(dp, (char *)ep -
((dp->i_offset - dp->i_count) & (DIRBLKSIZ - 1)),
dp->i_offset & ~(DIRBLKSIZ - 1));
#endif
if (DOINGSOFTDEP(dvp)) {
if (ip) {
ip->i_effnlink--;
softdep_change_linkcnt(ip, 0);
softdep_setup_remove(bp, dp, ip, isrmdir);
}
if (softdep_slowdown(dvp)) {
error = bwrite(bp);
} else {
bdwrite(bp);
error = 0;
}
} else {
if (ip) {
ip->i_effnlink--;
DIP_ADD(ip, nlink, -1);
ip->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(ip, 0);
}
if (DOINGASYNC(dvp) && dp->i_count != 0) {
bdwrite(bp);
error = 0;
} else
error = bwrite(bp);
}
dp->i_flag |= IN_CHANGE | IN_UPDATE;
UFS_WAPBL_UPDATE(dp, 0);
return (error);
}
/*
* Rename system call.
* rename("foo", "bar");
* is essentially
* unlink("bar");
* link("foo", "bar");
* unlink("foo");
* but ``atomically''. Can't do full commit without saving state in the
* inode on disk which isn't feasible at this time. Best we can do is
* always guarantee the target exists.
*
* Basic algorithm is:
*
* 1) Bump link count on source while we're linking it to the
* target. This also ensure the inode won't be deleted out
* from underneath us while we work (it may be truncated by
* a concurrent `trunc' or `open' for creation).
* 2) Link source to destination. If destination already exists,
* delete it first.
* 3) Unlink source reference to inode if still around. If a
* directory was moved and the parent of the destination
* is different from the source, patch the ".." entry in the
* directory.
*/
int
ufs_rename(void *v)
{
struct vop_rename_args *ap = v;
struct vnode *tvp = ap->a_tvp;
struct vnode *tdvp = ap->a_tdvp;
struct vnode *fvp = ap->a_fvp;
struct vnode *fdvp = ap->a_fdvp;
struct componentname *tcnp = ap->a_tcnp;
struct componentname *fcnp = ap->a_fcnp;
struct proc *p = fcnp->cn_proc;
struct inode *ip, *xp, *dp;
struct direct newdir;
int doingdirectory = 0, oldparent = 0, newparent = 0;
int error = 0;
#ifdef DIAGNOSTIC
if ((tcnp->cn_flags & HASBUF) == 0 ||
(fcnp->cn_flags & HASBUF) == 0)
panic("ufs_rename: no name");
#endif
/*
* Check for cross-device rename.
*/
if ((fvp->v_mount != tdvp->v_mount) ||
(tvp && (fvp->v_mount != tvp->v_mount))) {
error = EXDEV;
abortit:
VOP_ABORTOP(tdvp, tcnp);
if (tdvp == tvp)
vrele(tdvp);
else
vput(tdvp);
if (tvp)
vput(tvp);
VOP_ABORTOP(fdvp, fcnp);
vrele(fdvp);
vrele(fvp);
return (error);
}
if (tvp && ((DIP(VTOI(tvp), flags) & (IMMUTABLE | APPEND)) ||
(DIP(VTOI(tdvp), flags) & APPEND))) {
error = EPERM;
goto abortit;
}
/*
* Check if just deleting a link name or if we've lost a race.
* If another process completes the same rename after we've looked
* up the source and have blocked looking up the target, then the
* source and target inodes may be identical now although the
* names were never linked.
*/
if (fvp == tvp) {
if (fvp->v_type == VDIR) {
/*
* Linked directories are impossible, so we must
* have lost the race. Pretend that the rename
* completed before the lookup.
*/
error = ENOENT;
goto abortit;
}
/* Release destination completely. */
VOP_ABORTOP(tdvp, tcnp);
vput(tdvp);
vput(tvp);
/*
* Delete source. There is another race now that everything
* is unlocked, but this doesn't cause any new complications.
* relookup() may find a file that is unrelated to the
* original one, or it may fail. Too bad.
*/
vrele(fvp);
fcnp->cn_flags &= ~MODMASK;
fcnp->cn_flags |= LOCKPARENT | LOCKLEAF;
if ((fcnp->cn_flags & SAVESTART) == 0)
panic("ufs_rename: lost from startdir");
fcnp->cn_nameiop = DELETE;
if ((error = vfs_relookup(fdvp, &fvp, fcnp)) != 0)
return (error); /* relookup did vrele() */
vrele(fdvp);
return (VOP_REMOVE(fdvp, fvp, fcnp));
}
if ((error = vn_lock(fvp, LK_EXCLUSIVE, p)) != 0)
goto abortit;
/* fvp, tdvp, tvp now locked */
dp = VTOI(fdvp);
ip = VTOI(fvp);
if ((nlink_t) DIP(ip, nlink) >= LINK_MAX) {
VOP_UNLOCK(fvp, 0);
error = EMLINK;
goto abortit;
}
if ((DIP(ip, flags) & (IMMUTABLE | APPEND)) ||
(DIP(dp, flags) & APPEND)) {
VOP_UNLOCK(fvp, 0);
error = EPERM;
goto abortit;
}
if ((DIP(ip, mode) & IFMT) == IFDIR) {
error = VOP_ACCESS(fvp, VWRITE, tcnp->cn_cred);
if (!error && tvp)
error = VOP_ACCESS(tvp, VWRITE, tcnp->cn_cred);
if (error) {
VOP_UNLOCK(fvp, 0);
error = EACCES;
goto abortit;
}
/*
* Avoid ".", "..", and aliases of "." for obvious reasons.
*/
if ((fcnp->cn_namelen == 1 && fcnp->cn_nameptr[0] == '.') ||
dp == ip ||
(fcnp->cn_flags & ISDOTDOT) ||
(tcnp->cn_flags & ISDOTDOT) ||
(ip->i_flag & IN_RENAME)) {
VOP_UNLOCK(fvp, 0);
error = EINVAL;
goto abortit;
}
ip->i_flag |= IN_RENAME;
oldparent = dp->i_number;
doingdirectory = 1;
}
VN_KNOTE(fdvp, NOTE_WRITE); /* XXX right place? */
/*
* When the target exists, both the directory
* and target vnodes are returned locked.
*/
dp = VTOI(tdvp);
xp = NULL;
if (tvp)
xp = VTOI(tvp);
/*
* 1) Bump link count while we're moving stuff
* around. If we crash somewhere before
* completing our work, the link count
* may be wrong, but correctable.
*/
ip->i_effnlink++;
DIP_ADD(ip, nlink, 1);
ip->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(fvp))
softdep_change_linkcnt(ip, 0);
if ((error = UFS_UPDATE(ip, !DOINGSOFTDEP(fvp))) != 0) {
VOP_UNLOCK(fvp, 0);
goto bad;
}
/*
* If ".." must be changed (ie the directory gets a new
* parent) then the source directory must not be in the
* directory hierarchy above the target, as this would
* orphan everything below the source directory. Also
* the user must have write permission in the source so
* as to be able to change "..". We must repeat the call
* to namei, as the parent directory is unlocked by the
* call to checkpath().
*/
error = VOP_ACCESS(fvp, VWRITE, tcnp->cn_cred);
VOP_UNLOCK(fvp, 0);
/* tdvp and tvp locked */
if (oldparent != dp->i_number)
newparent = dp->i_number;
if (doingdirectory && newparent) {
if (error) /* write access check above */
goto bad;
if (xp != NULL)
vput(tvp);
/*
* Compensate for the reference ufs_checkpath() loses.
*/
vref(tdvp);
/* Only tdvp is locked */
if ((error = ufs_checkpath(ip, dp, tcnp->cn_cred)) != 0) {
vrele(tdvp);
goto out;
}
if ((tcnp->cn_flags & SAVESTART) == 0)
panic("ufs_rename: lost to startdir");
if ((error = vfs_relookup(tdvp, &tvp, tcnp)) != 0)
goto out;
vrele(tdvp); /* relookup() acquired a reference */
dp = VTOI(tdvp);
xp = NULL;
if (tvp)
xp = VTOI(tvp);
}
/*
* 2) If target doesn't exist, link the target
* to the source and unlink the source.
* Otherwise, rewrite the target directory
* entry to reference the source inode and
* expunge the original entry's existence.
*/
if (xp == NULL) {
if (dp->i_dev != ip->i_dev)
panic("rename: EXDEV");
/*
* Account for ".." in new directory.
* When source and destination have the same
* parent we don't fool with the link count.
*/
if (doingdirectory && newparent) {
if ((nlink_t) DIP(dp, nlink) >= LINK_MAX) {
error = EMLINK;
goto bad;
}
dp->i_effnlink++;
DIP_ADD(dp, nlink, 1);
dp->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(tdvp))
softdep_change_linkcnt(dp, 0);
if ((error = UFS_UPDATE(dp, !DOINGSOFTDEP(tdvp)))
!= 0) {
dp->i_effnlink--;
DIP_ADD(dp, nlink, -1);
dp->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(tdvp))
softdep_change_linkcnt(dp, 0);
goto bad;
}
}
ufs_makedirentry(ip, tcnp, &newdir);
if ((error = ufs_direnter(tdvp, NULL, &newdir, tcnp, NULL)) != 0) {
if (doingdirectory && newparent) {
dp->i_effnlink--;
DIP_ADD(dp, nlink, -1);
dp->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(tdvp))
softdep_change_linkcnt(dp, 0);
(void)UFS_UPDATE(dp, 1);
}
goto bad;
}
VN_KNOTE(tdvp, NOTE_WRITE);
vput(tdvp);
} else {
if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev)
panic("rename: EXDEV");
/*
* Short circuit rename(foo, foo).
*/
if (xp->i_number == ip->i_number)
panic("ufs_rename: same file");
/*
* If the parent directory is "sticky", then the user must
* own the parent directory, or the destination of the rename,
* otherwise the destination may not be changed (except by
* root). This implements append-only directories.
*/
if ((DIP(dp, mode) & S_ISTXT) && tcnp->cn_cred->cr_uid != 0 &&
tcnp->cn_cred->cr_uid != DIP(dp, uid) &&
DIP(xp, uid )!= tcnp->cn_cred->cr_uid) {
error = EPERM;
goto bad;
}
/*
* Target must be empty if a directory and have no links
* to it. Also, ensure source and target are compatible
* (both directories, or both not directories).
*/
if ((DIP(xp, mode) & IFMT) == IFDIR) {
if (xp->i_effnlink > 2 ||
!ufs_dirempty(xp, dp->i_number, tcnp->cn_cred)) {
error = ENOTEMPTY;
goto bad;
}
if (!doingdirectory) {
error = ENOTDIR;
goto bad;
}
cache_purge(tdvp);
} else if (doingdirectory) {
error = EISDIR;
goto bad;
}
if ((error = ufs_dirrewrite(dp, xp, ip->i_number,
IFTODT(DIP(ip, mode)), (doingdirectory && newparent) ?
newparent : doingdirectory)) != 0)
goto bad;
if (doingdirectory) {
if (!newparent) {
dp->i_effnlink--;
if (DOINGSOFTDEP(tdvp))
softdep_change_linkcnt(dp, 0);
}
xp->i_effnlink--;
if (DOINGSOFTDEP(tvp))
softdep_change_linkcnt(xp, 0);
}
if (doingdirectory && !DOINGSOFTDEP(tvp)) {
/*
* Truncate inode. The only stuff left in the directory
* is "." and "..". The "." reference is inconsequential
* since we are quashing it. We have removed the "."
* reference and the reference in the parent directory,
* but there may be other hard links. The soft
* dependency code will arrange to do these operations
* after the parent directory entry has been deleted on
* disk, so when running with that code we avoid doing
* them now.
*/
if (!newparent) {
DIP_ADD(dp, nlink, -1);
dp->i_flag |= IN_CHANGE;
}
DIP_ADD(xp, nlink, -1);
xp->i_flag |= IN_CHANGE;
if ((error = UFS_TRUNCATE(VTOI(tvp), (off_t)0, IO_SYNC,
tcnp->cn_cred)) != 0)
goto bad;
}
VN_KNOTE(tdvp, NOTE_WRITE);
vput(tdvp);
VN_KNOTE(tvp, NOTE_DELETE);
vput(tvp);
xp = NULL;
}
/*
* 3) Unlink the source.
*/
fcnp->cn_flags &= ~MODMASK;
fcnp->cn_flags |= LOCKPARENT | LOCKLEAF;
if ((fcnp->cn_flags & SAVESTART) == 0)
panic("ufs_rename: lost from startdir");
if ((error = vfs_relookup(fdvp, &fvp, fcnp)) != 0) {
vrele(ap->a_fvp);
return (error);
}
vrele(fdvp);
if (fvp == NULL) {
/*
* From name has disappeared.
*/
if (doingdirectory)
panic("ufs_rename: lost dir entry");
vrele(ap->a_fvp);
return (0);
}
xp = VTOI(fvp);
dp = VTOI(fdvp);
/*
* Ensure that the directory entry still exists and has not
* changed while the new name has been entered. If the source is
* a file then the entry may have been unlinked or renamed. In
* either case there is no further work to be done. If the source
* is a directory then it cannot have been rmdir'ed; the IN_RENAME
* flag ensures that it cannot be moved by another rename or removed
* by a rmdir.
*/
if (xp != ip) {
if (doingdirectory)
panic("ufs_rename: lost dir entry");
} else {
/*
* If the source is a directory with a
* new parent, the link count of the old
* parent directory must be decremented
* and ".." set to point to the new parent.
*/
if (doingdirectory && newparent) {
xp->i_offset = mastertemplate.dot_reclen;
ufs_dirrewrite(xp, dp, newparent, DT_DIR, 0);
cache_purge(fdvp);
}
error = ufs_dirremove(fdvp, xp, fcnp->cn_flags, 0);
xp->i_flag &= ~IN_RENAME;
}
VN_KNOTE(fvp, NOTE_RENAME);
if (dp)
vput(fdvp);
if (xp)
vput(fvp);
vrele(ap->a_fvp);
return (error);
bad:
if (xp)
vput(ITOV(xp));
vput(ITOV(dp));
out:
vrele(fdvp);
if (doingdirectory)
ip->i_flag &= ~IN_RENAME;
if (vn_lock(fvp, LK_EXCLUSIVE, p) == 0) {
ip->i_effnlink--;
DIP_ADD(ip, nlink, -1);
ip->i_flag |= IN_CHANGE;
ip->i_flag &= ~IN_RENAME;
if (DOINGSOFTDEP(fvp))
softdep_change_linkcnt(ip, 0);
vput(fvp);
} else
vrele(fvp);
return (error);
}
/*
* Allocate a new inode.
*/
int
ufs_makeinode(int mode, struct vnode *dvp, struct vnode **vpp,
struct componentname *cnp)
{
struct inode *ip, *pdir;
struct direct newdir;
struct vnode *tvp;
int error;
pdir = VTOI(dvp);
#ifdef DIAGNOSTIC
if ((cnp->cn_flags & HASBUF) == 0)
panic("ufs_makeinode: no name");
#endif
*vpp = NULL;
if ((mode & IFMT) == 0)
mode |= IFREG;
if ((error = UFS_INODE_ALLOC(pdir, mode, cnp->cn_cred, &tvp)) != 0) {
pool_put(&namei_pool, cnp->cn_pnbuf);
vput(dvp);
return (error);
}
ip = VTOI(tvp);
DIP_ASSIGN(ip, gid, DIP(pdir, gid));
DIP_ASSIGN(ip, uid, cnp->cn_cred->cr_uid);
if ((error = getinoquota(ip)) ||
(error = ufs_quota_alloc_inode(ip, cnp->cn_cred))) {
pool_put(&namei_pool, cnp->cn_pnbuf);
UFS_INODE_FREE(ip, ip->i_number, mode);
vput(tvp);
vput(dvp);
return (error);
}
ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
DIP_ASSIGN(ip, mode, mode);
tvp->v_type = IFTOVT(mode); /* Rest init'd in getnewvnode(). */
ip->i_effnlink = 1;
DIP_ASSIGN(ip, nlink, 1);
if (DOINGSOFTDEP(tvp))
softdep_change_linkcnt(ip, 0);
if ((DIP(ip, mode) & ISGID) &&
!groupmember(DIP(ip, gid), cnp->cn_cred) &&
suser_ucred(cnp->cn_cred))
DIP_AND(ip, mode, ~ISGID);
/*
* Make sure inode goes to disk before directory entry.
*/
if ((error = UFS_UPDATE(ip, !DOINGSOFTDEP(tvp))) != 0)
goto bad;
ufs_makedirentry(ip, cnp, &newdir);
if ((error = ufs_direnter(dvp, tvp, &newdir, cnp, NULL)) != 0)
goto bad;
if ((cnp->cn_flags & SAVESTART) == 0)
pool_put(&namei_pool, cnp->cn_pnbuf);
vput(dvp);
*vpp = tvp;
return (0);
bad:
/*
* Write error occurred trying to update the inode
* or the directory so must deallocate the inode.
*/
pool_put(&namei_pool, cnp->cn_pnbuf);
vput(dvp);
ip->i_effnlink = 0;
DIP_ASSIGN(ip, nlink, 0);
ip->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(tvp))
softdep_change_linkcnt(ip, 0);
tvp->v_type = VNON;
vput(tvp);
return (error);
}
/*
* link vnode call
*/
int
ufs_link(void *v)
{
struct vop_link_args *ap = v;
struct vnode *dvp = ap->a_dvp;
struct vnode *vp = ap->a_vp;
struct componentname *cnp = ap->a_cnp;
struct proc *p = cnp->cn_proc;
struct inode *ip;
struct direct newdir;
int error;
#ifdef DIAGNOSTIC
if ((cnp->cn_flags & HASBUF) == 0)
panic("ufs_link: no name");
#endif
if (vp->v_type == VDIR) {
VOP_ABORTOP(dvp, cnp);
error = EPERM;
goto out2;
}
if (dvp->v_mount != vp->v_mount) {
VOP_ABORTOP(dvp, cnp);
error = EXDEV;
goto out2;
}
if (dvp != vp && (error = vn_lock(vp, LK_EXCLUSIVE, p))) {
VOP_ABORTOP(dvp, cnp);
goto out2;
}
ip = VTOI(vp);
if ((nlink_t) DIP(ip, nlink) >= LINK_MAX) {
VOP_ABORTOP(dvp, cnp);
error = EMLINK;
goto out1;
}
if (DIP(ip, flags) & (IMMUTABLE | APPEND)) {
VOP_ABORTOP(dvp, cnp);
error = EPERM;
goto out1;
}
ip->i_effnlink++;
DIP_ADD(ip, nlink, 1);
ip->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(vp))
softdep_change_linkcnt(ip, 0);
if ((error = UFS_UPDATE(ip, !DOINGSOFTDEP(vp))) == 0) {
ufs_makedirentry(ip, cnp, &newdir);
error = ufs_direnter(dvp, vp, &newdir, cnp, NULL);
}
if (error) {
ip->i_effnlink--;
DIP_ADD(ip, nlink, -1);
ip->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(vp))
softdep_change_linkcnt(ip, 0);
}
pool_put(&namei_pool, cnp->cn_pnbuf);
VN_KNOTE(vp, NOTE_LINK);
VN_KNOTE(dvp, NOTE_WRITE);
out1:
if (dvp != vp)
VOP_UNLOCK(vp, 0);
out2:
vput(dvp);
return (error);
}
/*
* Rmdir system call.
*/
int
ufs_rmdir(void *v)
{
struct vop_rmdir_args *ap = v;
struct vnode *vp = ap->a_vp;
struct vnode *dvp = ap->a_dvp;
struct componentname *cnp = ap->a_cnp;
struct inode *ip, *dp;
int error;
ip = VTOI(vp);
dp = VTOI(dvp);
/*
* No rmdir "." or of mounted on directories.
*/
if (dp == ip || vp->v_mountedhere != NULL) {
if (dp == ip)
vrele(dvp);
else
vput(dvp);
vput(vp);
return (EINVAL);
}
/*
* Do not remove a directory that is in the process of being renamed.
* Verify the directory is empty (and valid). Rmdir ".." will not be
* valid since ".." will contain a reference to the current directory
* and thus be non-empty.
*/
error = 0;
if (ip->i_flag & IN_RENAME) {
error = EINVAL;
goto out;
}
if (ip->i_effnlink != 2 ||
!ufs_dirempty(ip, dp->i_number, cnp->cn_cred)) {
error = ENOTEMPTY;
goto out;
}
if ((DIP(dp, flags) & APPEND) ||
(DIP(ip, flags) & (IMMUTABLE | APPEND))) {
error = EPERM;
goto out;
}
/*
* Delete reference to directory before purging
* inode. If we crash in between, the directory
* will be reattached to lost+found,
*/
dp->i_effnlink--;
ip->i_effnlink--;
if (DOINGSOFTDEP(vp)) {
softdep_change_linkcnt(dp, 0);
softdep_change_linkcnt(ip, 0);
}
if ((error = ufs_dirremove(dvp, ip, cnp->cn_flags, 1)) != 0) {
dp->i_effnlink++;
ip->i_effnlink++;
if (DOINGSOFTDEP(vp)) {
softdep_change_linkcnt(dp, 0);
softdep_change_linkcnt(ip, 0);
}
goto out;
}
VN_KNOTE(dvp, NOTE_WRITE | NOTE_LINK);
cache_purge(dvp);
/*
* Truncate inode. The only stuff left in the directory is "." and
* "..". The "." reference is inconsequential since we are quashing
* it. The soft dependency code will arrange to do these operations
* after the parent directory entry has been deleted on disk, so
* when running with that code we avoid doing them now.
*/
if (!DOINGSOFTDEP(vp)) {
int ioflag;
DIP_ADD(dp, nlink, -1);
dp->i_flag |= IN_CHANGE;
DIP_ADD(ip, nlink, -1);
ip->i_flag |= IN_CHANGE;
ioflag = DOINGASYNC(vp) ? 0 : IO_SYNC;
error = UFS_TRUNCATE(ip, (off_t)0, ioflag, cnp->cn_cred);
}
cache_purge(vp);
#ifdef UFS_DIRHASH
/* Kill any active hash; i_effnlink == 0, so it will not come back. */
if (ip->i_dirhash != NULL)
ufsdirhash_free(ip);
#endif
out:
VN_KNOTE(vp, NOTE_DELETE);
vput(dvp);
vput(vp);
return (error);
}
/*
* Mkdir system call
*/
int
ufs_mkdir(void *v)
{
struct vop_mkdir_args *ap = v;
struct vnode *dvp = ap->a_dvp;
struct vattr *vap = ap->a_vap;
struct componentname *cnp = ap->a_cnp;
struct inode *ip, *dp;
struct vnode *tvp;
struct buf *bp;
struct direct newdir;
struct dirtemplate dirtemplate, *dtp;
int error, dmode, blkoff;
#ifdef DIAGNOSTIC
if ((cnp->cn_flags & HASBUF) == 0)
panic("ufs_mkdir: no name");
#endif
dp = VTOI(dvp);
if ((nlink_t) DIP(dp, nlink) >= LINK_MAX) {
error = EMLINK;
goto out;
}
dmode = vap->va_mode & 0777;
dmode |= IFDIR;
/*
* Must simulate part of ufs_makeinode here to acquire the inode,
* but not have it entered in the parent directory. The entry is
* made later after writing "." and ".." entries.
*/
if ((error = UFS_INODE_ALLOC(dp, dmode, cnp->cn_cred, &tvp)) != 0)
goto out;
ip = VTOI(tvp);
DIP_ASSIGN(ip, uid, cnp->cn_cred->cr_uid);
DIP_ASSIGN(ip, gid, DIP(dp, gid));
if ((error = getinoquota(ip)) ||
(error = ufs_quota_alloc_inode(ip, cnp->cn_cred))) {
pool_put(&namei_pool, cnp->cn_pnbuf);
UFS_INODE_FREE(ip, ip->i_number, dmode);
vput(tvp);
vput(dvp);
return (error);
}
ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
DIP_ASSIGN(ip, mode, dmode);
tvp->v_type = VDIR; /* Rest init'd in getnewvnode(). */
ip->i_effnlink = 2;
DIP_ASSIGN(ip, nlink, 2);
if (DOINGSOFTDEP(tvp))
softdep_change_linkcnt(ip, 0);
/*
* Bump link count in parent directory to reflect work done below.
* Should be done before reference is create so cleanup is
* possible if we crash.
*/
dp->i_effnlink++;
DIP_ADD(dp, nlink, 1);
dp->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(dvp))
softdep_change_linkcnt(dp, 0);
if ((error = UFS_UPDATE(dp, !DOINGSOFTDEP(dvp))) != 0)
goto bad;
/*
* Initialize directory with "." and ".." from static template.
*/
if (dvp->v_mount->mnt_maxsymlinklen > 0)
dtp = &mastertemplate;
else
dtp = (struct dirtemplate *)&omastertemplate;
dirtemplate = *dtp;
dirtemplate.dot_ino = ip->i_number;
dirtemplate.dotdot_ino = dp->i_number;
if ((error = UFS_BUF_ALLOC(ip, (off_t)0, DIRBLKSIZ, cnp->cn_cred,
B_CLRBUF, &bp)) != 0)
goto bad;
DIP_ASSIGN(ip, size, DIRBLKSIZ);
ip->i_flag |= IN_CHANGE | IN_UPDATE;
uvm_vnp_setsize(tvp, DIP(ip, size));
bcopy((caddr_t)&dirtemplate, (caddr_t)bp->b_data, sizeof dirtemplate);
if (DOINGSOFTDEP(tvp)) {
/*
* Ensure that the entire newly allocated block is a
* valid directory so that future growth within the
* block does not have to ensure that the block is
* written before the inode
*/
blkoff = DIRBLKSIZ;
while (blkoff < bp->b_bcount) {
((struct direct *)
(bp->b_data + blkoff))->d_reclen = DIRBLKSIZ;
blkoff += DIRBLKSIZ;
}
}
if ((error = UFS_UPDATE(ip, !DOINGSOFTDEP(tvp))) != 0) {
(void)VOP_BWRITE(bp);
goto bad;
}
/*
* Directory set up, now install its entry in the parent directory.
*
* If we are not doing soft dependencies, then we must write out the
* buffer containing the new directory body before entering the new
* name in the parent. If we are doing soft dependencies, then the
* buffer containing the new directory body will be passed to and
* released in the soft dependency code after the code has attached
* an appropriate ordering dependency to the buffer which ensures that
* the buffer is written before the new name is written in the parent.
*/
if (!DOINGSOFTDEP(dvp) && ((error = VOP_BWRITE(bp)) != 0))
goto bad;
ufs_makedirentry(ip, cnp, &newdir);
error = ufs_direnter(dvp, tvp, &newdir, cnp, bp);
bad:
if (error == 0) {
VN_KNOTE(dvp, NOTE_WRITE | NOTE_LINK);
*ap->a_vpp = tvp;
} else {
dp->i_effnlink--;
DIP_ADD(dp, nlink, -1);
dp->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(dvp))
softdep_change_linkcnt(dp, 0);
/*
* No need to do an explicit VOP_TRUNCATE here, vrele will
* do this for us because we set the link count to 0.
*/
ip->i_effnlink = 0;
DIP_ASSIGN(ip, nlink, 0);
ip->i_flag |= IN_CHANGE;
if (DOINGSOFTDEP(tvp))
softdep_change_linkcnt(ip, 0);
vput(tvp);
}
out:
pool_put(&namei_pool, cnp->cn_pnbuf);
vput(dvp);
return (error);
}
/*
* Last reference to an inode. If necessary, write or delete it.
*/
int
ufs_inactive(void *v)
{
struct vop_inactive_args *ap = v;
struct vnode *vp = ap->a_vp;
struct inode *ip = VTOI(vp);
struct proc *p = ap->a_p;
mode_t mode;
int error = 0;
#ifdef DIAGNOSTIC
extern int prtactive;
if (prtactive && vp->v_usecount != 0)
vprint("ffs_inactive: pushing active", vp);
#endif
/*
* Ignore inodes related to stale file handles.
*/
if (ip->i_din1 == NULL || DIP(ip, mode) == 0)
goto out;
if (DIP(ip, nlink) <= 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) {
if (getinoquota(ip) == 0)
(void)ufs_quota_free_inode(ip, NOCRED);
error = UFS_TRUNCATE(ip, (off_t)0, IO_EXT | IO_NORMAL, NOCRED);
DIP_ASSIGN(ip, rdev, 0);
mode = DIP(ip, mode);
DIP_ASSIGN(ip, mode, 0);
ip->i_flag |= IN_CHANGE | IN_UPDATE;
/*
* Setting the mode to zero needs to wait for the inode to be
* written just as does a change to the link count. So, rather
* than creating a new entry point to do the same thing, we
* just use softdep_change_linkcnt(). Also, we can't let
* softdep co-opt us to help on its worklist, as we may end up
* trying to recycle vnodes and getting to this same point a
* couple of times, blowing the kernel stack. However, this
* could be optimized by checking if we are coming from
* vrele(), vput() or vclean() (by checking for VXLOCK) and
* just avoiding the co-opt to happen in the last case.
*/
if (DOINGSOFTDEP(vp))
softdep_change_linkcnt(ip, 1);
UFS_INODE_FREE(ip, ip->i_number, mode);
}
if (ip->i_flag & (IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE)) {
UFS_UPDATE(ip, 0);
}
out:
VOP_UNLOCK(vp, 0, p);
/*
* If we are done with the inode, reclaim it
* so that it can be reused immediately.
*/
if (ip->i_din1 == NULL || DIP(ip, mode) == 0)
vrecycle(vp, p);
return (error);
}
/*
* Last reference to an inode. If necessary, write or delete it.
*/
int
ufs_inactive(void *v)
{
struct vop_inactive_args /* {
struct vnode *a_vp;
struct bool *a_recycle;
} */ *ap = v;
struct vnode *vp = ap->a_vp;
struct inode *ip = VTOI(vp);
struct mount *transmp;
mode_t mode;
int error = 0;
int logged = 0;
UFS_WAPBL_JUNLOCK_ASSERT(vp->v_mount);
transmp = vp->v_mount;
fstrans_start(transmp, FSTRANS_SHARED);
/*
* Ignore inodes related to stale file handles.
*/
if (ip->i_mode == 0)
goto out;
if (ip->i_ffs_effnlink == 0 && DOINGSOFTDEP(vp))
softdep_releasefile(ip);
if (ip->i_nlink <= 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) {
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
logged = 1;
#ifdef QUOTA
(void)chkiq(ip, -1, NOCRED, 0);
#endif
#ifdef UFS_EXTATTR
ufs_extattr_vnode_inactive(vp, curlwp);
#endif
if (ip->i_size != 0) {
/*
* When journaling, only truncate one indirect block
* at a time
*/
if (vp->v_mount->mnt_wapbl) {
uint64_t incr = MNINDIR(ip->i_ump) <<
vp->v_mount->mnt_fs_bshift; /* Power of 2 */
uint64_t base = NDADDR <<
vp->v_mount->mnt_fs_bshift;
while (!error && ip->i_size > base + incr) {
/*
* round down to next full indirect
* block boundary.
*/
uint64_t nsize = base +
((ip->i_size - base - 1) &
~(incr - 1));
error = UFS_TRUNCATE(vp, nsize, 0,
NOCRED);
if (error)
break;
UFS_WAPBL_END(vp->v_mount);
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
}
}
if (!error)
error = UFS_TRUNCATE(vp, (off_t)0, 0, NOCRED);
}
/*
* Setting the mode to zero needs to wait for the inode
* to be written just as does a change to the link count.
* So, rather than creating a new entry point to do the
* same thing, we just use softdep_change_linkcnt().
*/
DIP_ASSIGN(ip, rdev, 0);
mode = ip->i_mode;
ip->i_mode = 0;
DIP_ASSIGN(ip, mode, 0);
ip->i_flag |= IN_CHANGE | IN_UPDATE;
mutex_enter(&vp->v_interlock);
vp->v_iflag |= VI_FREEING;
mutex_exit(&vp->v_interlock);
if (DOINGSOFTDEP(vp))
softdep_change_linkcnt(ip);
UFS_VFREE(vp, ip->i_number, mode);
}
if (ip->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) {
if (!logged++) {
int err;
err = UFS_WAPBL_BEGIN(vp->v_mount);
if (err)
goto out;
}
UFS_UPDATE(vp, NULL, NULL, 0);
}
if (logged)
UFS_WAPBL_END(vp->v_mount);
out:
/*
* If we are done with the inode, reclaim it
* so that it can be reused immediately.
*/
*ap->a_recycle = (ip->i_mode == 0);
VOP_UNLOCK(vp, 0);
fstrans_done(transmp);
return (error);
}
/*
* Last reference to an inode. If necessary, write or delete it.
*/
int
ufs_inactive(void *v)
{
struct vop_inactive_args *ap = v;
struct vnode *vp = ap->a_vp;
struct inode *ip = VTOI(vp);
struct fs *fs = ip->i_fs;
struct proc *p = curproc;
mode_t mode;
int error = 0, logged = 0, truncate_error = 0;
#ifdef DIAGNOSTIC
extern int prtactive;
if (prtactive && vp->v_usecount != 0)
vprint("ufs_inactive: pushing active", vp);
#endif
UFS_WAPBL_JUNLOCK_ASSERT(vp->v_mount);
/*
* Ignore inodes related to stale file handles.
*/
if (ip->i_din1 == NULL || DIP(ip, mode) == 0)
goto out;
if (DIP(ip, nlink) <= 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) {
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
logged = 1;
if (getinoquota(ip) == 0)
(void)ufs_quota_free_inode(ip, NOCRED);
if (DIP(ip, size) != 0 && vp->v_mount->mnt_wapbl) {
/*
* When journaling, only truncate one indirect block at
* a time.
*/
uint64_t incr = MNINDIR(ip->i_ump) << fs->fs_bshift;
uint64_t base = NDADDR << fs->fs_bshift;
while (!error && DIP(ip, size) > base + incr) {
/*
* round down to next full indirect block
* boundary.
*/
uint64_t nsize = base +
((DIP(ip, size) - base - 1) &
~(incr - 1));
error = UFS_TRUNCATE(ip, nsize, 0, NOCRED);
if (error)
break;
UFS_WAPBL_END(vp->v_mount);
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
}
}
if (error == 0) {
truncate_error = UFS_TRUNCATE(ip, (off_t)0, 0, NOCRED);
/* XXX pedro: remove me */
if (truncate_error)
printf("UFS_TRUNCATE()=%d\n", truncate_error);
}
DIP_ASSIGN(ip, rdev, 0);
mode = DIP(ip, mode);
DIP_ASSIGN(ip, mode, 0);
ip->i_flag |= IN_CHANGE | IN_UPDATE;
/*
* Setting the mode to zero needs to wait for the inode to be
* written just as does a change to the link count. So, rather
* than creating a new entry point to do the same thing, we
* just use softdep_change_linkcnt(). Also, we can't let
* softdep co-opt us to help on its worklist, as we may end up
* trying to recycle vnodes and getting to this same point a
* couple of times, blowing the kernel stack. However, this
* could be optimized by checking if we are coming from
* vrele(), vput() or vclean() (by checking for VXLOCK) and
* just avoiding the co-opt to happen in the last case.
*/
if (DOINGSOFTDEP(vp))
softdep_change_linkcnt(ip, 1);
UFS_INODE_FREE(ip, ip->i_number, mode);
}
if (ip->i_flag & (IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE)) {
if (!logged++) {
int err;
err = UFS_WAPBL_BEGIN(vp->v_mount);
if (err) {
error = err;
goto out;
}
}
UFS_UPDATE(ip, 0);
}
if (logged)
UFS_WAPBL_END(vp->v_mount);
out:
VOP_UNLOCK(vp, 0);
/*
* If we are done with the inode, reclaim it
* so that it can be reused immediately.
*/
if (error == 0 && truncate_error == 0 &&
(ip->i_din1 == NULL || DIP(ip, mode) == 0))
vrecycle(vp, p);
return (truncate_error ? truncate_error : error);
}
