/*
* Reclaim an inode so that it can be used for other purposes.
*/
int
ufs_reclaim(struct vnode *vp)
{
struct inode *ip = VTOI(vp);
if (prtactive && vp->v_usecount > 1)
vprint("ufs_reclaim: pushing active", vp);
if (!UFS_WAPBL_BEGIN(vp->v_mount)) {
UFS_UPDATE(vp, NULL, NULL, UPDATE_CLOSE);
UFS_WAPBL_END(vp->v_mount);
}
UFS_UPDATE(vp, NULL, NULL, UPDATE_CLOSE);
/*
* Remove the inode from its hash chain.
*/
ufs_ihashrem(ip);
/*
* Purge old data structures associated with the inode.
*/
cache_purge(vp);
if (ip->i_devvp) {
vrele(ip->i_devvp);
ip->i_devvp = 0;
}
#ifdef QUOTA
ufsquota_free(ip);
#endif
#ifdef UFS_DIRHASH
if (ip->i_dirhash != NULL)
ufsdirhash_free(ip);
#endif
return (0);
}
/*
* Reclaim an inode so that it can be used for other purposes.
*/
int
ufs_reclaim(struct vnode *vp, struct proc *p)
{
struct inode *ip;
#ifdef DIAGNOSTIC
extern int prtactive;
if (prtactive && vp->v_usecount != 0)
vprint("ufs_reclaim: pushing active", vp);
#endif
ip = VTOI(vp);
/*
* Stop deferring timestamp writes
*/
if (ip->i_flag & IN_LAZYMOD) {
int err = UFS_WAPBL_BEGIN(vp->v_mount);
if (err)
return (err);
ip->i_flag |= IN_MODIFIED;
UFS_UPDATE(ip, 0);
UFS_WAPBL_END(vp->v_mount);
}
/*
* Remove the inode from its hash chain.
*/
ufs_ihashrem(ip);
/*
* Purge old data structures associated with the inode.
*/
cache_purge(vp);
if (ip->i_devvp) {
vrele(ip->i_devvp);
}
#ifdef UFS_DIRHASH
if (ip->i_dirhash != NULL)
ufsdirhash_free(ip);
#endif
ufs_quota_delete(ip);
return (0);
}
// ffs文件系统的写入操作
int
ffs_write(void *v)
{
struct vop_write_args *ap = v;
struct vnode *vp;
struct uio *uio;
struct inode *ip;
struct fs *fs;
struct buf *bp;
daddr_t lbn;
off_t osize;
int blkoffset, error, extended, flags, ioflag, size, xfersize;
ssize_t resid, overrun;
extended = 0;
ioflag = ap->a_ioflag;
uio = ap->a_uio;
vp = ap->a_vp;
ip = VTOI(vp);
#ifdef DIAGNOSTIC
if (uio->uio_rw != UIO_WRITE)
panic("ffs_write: mode");
#endif
/*
* If writing 0 bytes, succeed and do not change
* update time or file offset (standards compliance)
*/
if (uio->uio_resid == 0)
return (0);
switch (vp->v_type) {
case VREG:
if (ioflag & IO_APPEND)
uio->uio_offset = DIP(ip, size);
if ((DIP(ip, flags) & APPEND) && uio->uio_offset != DIP(ip, size))
return (EPERM);
/* FALLTHROUGH */
case VLNK:
break;
case VDIR:
if ((ioflag & IO_SYNC) == 0)
panic("ffs_write: nonsync dir write");
break;
default:
panic("ffs_write: type");
}
fs = ip->i_fs;
if (uio->uio_offset < 0 ||
(u_int64_t)uio->uio_offset + uio->uio_resid > fs->fs_maxfilesize)
return (EFBIG);
/* do the filesize rlimit check */
if ((error = vn_fsizechk(vp, uio, ioflag, &overrun)))
return (error);
resid = uio->uio_resid;
osize = DIP(ip, size);
flags = ioflag & IO_SYNC ? B_SYNC : 0;
for (error = 0; uio->uio_resid > 0;) {
lbn = lblkno(fs, uio->uio_offset);
blkoffset = blkoff(fs, uio->uio_offset);
xfersize = fs->fs_bsize - blkoffset;
if (uio->uio_resid < xfersize)
xfersize = uio->uio_resid;
if (fs->fs_bsize > xfersize)
flags |= B_CLRBUF;
else
flags &= ~B_CLRBUF;
if ((error = UFS_BUF_ALLOC(ip, uio->uio_offset, xfersize,
ap->a_cred, flags, &bp)) != 0)
break;
if (uio->uio_offset + xfersize > DIP(ip, size)) {
DIP_ASSIGN(ip, size, uio->uio_offset + xfersize);
uvm_vnp_setsize(vp, DIP(ip, size));
extended = 1;
}
(void)uvm_vnp_uncache(vp);
size = blksize(fs, ip, lbn) - bp->b_resid;
if (size < xfersize)
xfersize = size;
error =
uiomovei(bp->b_data + blkoffset, xfersize, uio);
if (error != 0)
memset(bp->b_data + blkoffset, 0, xfersize);
#if 0
if (ioflag & IO_NOCACHE)
bp->b_flags |= B_NOCACHE;
#endif
if (ioflag & IO_SYNC)
(void)bwrite(bp);
else if (xfersize + blkoffset == fs->fs_bsize) {
if (doclusterwrite)
cluster_write(bp, &ip->i_ci, DIP(ip, size));
else
bawrite(bp);
} else
bdwrite(bp);
if (error || xfersize == 0)
break;
ip->i_flag |= IN_CHANGE | IN_UPDATE;
}
/*
* If we successfully wrote any data, and we are not the superuser
* we clear the setuid and setgid bits as a precaution against
* tampering.
*/
if (resid > uio->uio_resid && ap->a_cred && ap->a_cred->cr_uid != 0)
DIP_ASSIGN(ip, mode, DIP(ip, mode) & ~(ISUID | ISGID));
if (resid > uio->uio_resid)
VN_KNOTE(vp, NOTE_WRITE | (extended ? NOTE_EXTEND : 0));
if (error) {
if (ioflag & IO_UNIT) {
(void)UFS_TRUNCATE(ip, osize,
ioflag & IO_SYNC, ap->a_cred);
uio->uio_offset -= resid - uio->uio_resid;
uio->uio_resid = resid;
}
} else if (resid > uio->uio_resid && (ioflag & IO_SYNC)) {
error = UFS_UPDATE(ip, 1);
}
/* correct the result for writes clamped by vn_fsizechk() */
uio->uio_resid += overrun;
return (error);
}
/*
* Vnode op for writing.
*/
int
ffs_write(void *v)
{
struct vop_write_args *ap = v;
struct vnode *vp;
struct uio *uio;
struct inode *ip;
struct fs *fs;
struct buf *bp;
struct proc *p;
daddr64_t lbn;
off_t osize;
int blkoffset, error, extended, flags, ioflag, resid, size, xfersize;
extended = 0;
ioflag = ap->a_ioflag;
uio = ap->a_uio;
vp = ap->a_vp;
ip = VTOI(vp);
#ifdef DIAGNOSTIC
if (uio->uio_rw != UIO_WRITE)
panic("ffs_write: mode");
#endif
/*
* If writing 0 bytes, succeed and do not change
* update time or file offset (standards compliance)
*/
if (uio->uio_resid == 0)
return (0);
switch (vp->v_type) {
case VREG:
if (ioflag & IO_APPEND)
uio->uio_offset = DIP(ip, size);
if ((DIP(ip, flags) & APPEND) && uio->uio_offset != DIP(ip, size))
return (EPERM);
/* FALLTHROUGH */
case VLNK:
break;
case VDIR:
if ((ioflag & IO_SYNC) == 0)
panic("ffs_write: nonsync dir write");
break;
default:
panic("ffs_write: type");
}
fs = ip->i_fs;
if (uio->uio_offset < 0 ||
(u_int64_t)uio->uio_offset + uio->uio_resid > fs->fs_maxfilesize)
return (EFBIG);
/*
* Maybe this should be above the vnode op call, but so long as
* file servers have no limits, I don't think it matters.
*/
p = uio->uio_procp;
if (vp->v_type == VREG && p && !(ioflag & IO_NOLIMIT) &&
uio->uio_offset + uio->uio_resid >
p->p_rlimit[RLIMIT_FSIZE].rlim_cur) {
psignal(p, SIGXFSZ);
return (EFBIG);
}
resid = uio->uio_resid;
osize = DIP(ip, size);
flags = ioflag & IO_SYNC ? B_SYNC : 0;
for (error = 0; uio->uio_resid > 0;) {
lbn = lblkno(fs, uio->uio_offset);
blkoffset = blkoff(fs, uio->uio_offset);
xfersize = fs->fs_bsize - blkoffset;
if (uio->uio_resid < xfersize)
xfersize = uio->uio_resid;
if (fs->fs_bsize > xfersize)
flags |= B_CLRBUF;
else
flags &= ~B_CLRBUF;
if ((error = UFS_BUF_ALLOC(ip, uio->uio_offset, xfersize,
ap->a_cred, flags, &bp)) != 0)
break;
if (uio->uio_offset + xfersize > DIP(ip, size)) {
DIP_ASSIGN(ip, size, uio->uio_offset + xfersize);
uvm_vnp_setsize(vp, DIP(ip, size));
extended = 1;
}
(void)uvm_vnp_uncache(vp);
size = blksize(fs, ip, lbn) - bp->b_resid;
if (size < xfersize)
xfersize = size;
error =
uiomove((char *)bp->b_data + blkoffset, xfersize, uio);
if (error != 0)
bzero((char *)bp->b_data + blkoffset, xfersize);
if (ioflag & IO_SYNC)
(void)bwrite(bp);
else if (xfersize + blkoffset == fs->fs_bsize) {
if (doclusterwrite)
cluster_write(bp, &ip->i_ci, DIP(ip, size));
else
bawrite(bp);
} else
bdwrite(bp);
if (error || xfersize == 0)
break;
ip->i_flag |= IN_CHANGE | IN_UPDATE;
}
/*
* If we successfully wrote any data, and we are not the superuser
* we clear the setuid and setgid bits as a precaution against
* tampering.
*/
if (resid > uio->uio_resid && ap->a_cred && ap->a_cred->cr_uid != 0)
DIP_ASSIGN(ip, mode, DIP(ip, mode) & ~(ISUID | ISGID));
if (resid > uio->uio_resid)
VN_KNOTE(vp, NOTE_WRITE | (extended ? NOTE_EXTEND : 0));
if (error) {
if (ioflag & IO_UNIT) {
(void)UFS_TRUNCATE(ip, osize,
ioflag & IO_SYNC, ap->a_cred);
uio->uio_offset -= resid - uio->uio_resid;
uio->uio_resid = resid;
}
} else if (resid > uio->uio_resid && (ioflag & IO_SYNC)) {
error = UFS_UPDATE(ip, MNT_WAIT);
}
return (error);
}
/*
* Truncate the inode oip to at most length size, freeing the
* disk blocks.
*/
int
ffs_truncate(struct inode *oip, off_t length, int flags, struct ucred *cred)
{
struct vnode *ovp;
daddr64_t lastblock;
daddr64_t bn, lbn, lastiblock[NIADDR], indir_lbn[NIADDR];
daddr64_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR];
struct fs *fs;
struct buf *bp;
int offset, size, level;
long count, nblocks, vflags, blocksreleased = 0;
int i, aflags, error, allerror, indirect = 0;
off_t osize;
extern int num_indirdep;
extern int max_indirdep;
if (length < 0)
return (EINVAL);
ovp = ITOV(oip);
if (ovp->v_type != VREG &&
ovp->v_type != VDIR &&
ovp->v_type != VLNK)
return (0);
if (DIP(oip, size) == length)
return (0);
if (ovp->v_type == VLNK &&
(DIP(oip, size) < ovp->v_mount->mnt_maxsymlinklen ||
(ovp->v_mount->mnt_maxsymlinklen == 0 &&
oip->i_din1->di_blocks == 0))) {
#ifdef DIAGNOSTIC
if (length != 0)
panic("ffs_truncate: partial truncate of symlink");
#endif
memset(SHORTLINK(oip), 0, (size_t) DIP(oip, size));
DIP_ASSIGN(oip, size, 0);
oip->i_flag |= IN_CHANGE | IN_UPDATE;
return (UFS_UPDATE(oip, MNT_WAIT));
}
if ((error = getinoquota(oip)) != 0)
return (error);
uvm_vnp_setsize(ovp, length);
oip->i_ci.ci_lasta = oip->i_ci.ci_clen
= oip->i_ci.ci_cstart = oip->i_ci.ci_lastw = 0;
if (DOINGSOFTDEP(ovp)) {
if (length > 0 || softdep_slowdown(ovp)) {
/*
* If a file is only partially truncated, then
* we have to clean up the data structures
* describing the allocation past the truncation
* point. Finding and deallocating those structures
* is a lot of work. Since partial truncation occurs
* rarely, we solve the problem by syncing the file
* so that it will have no data structures left.
*/
if ((error = VOP_FSYNC(ovp, cred, MNT_WAIT)) != 0)
return (error);
} else {
(void)ufs_quota_free_blocks(oip, DIP(oip, blocks),
NOCRED);
softdep_setup_freeblocks(oip, length);
(void) vinvalbuf(ovp, 0, cred, curproc, 0, 0);
oip->i_flag |= IN_CHANGE | IN_UPDATE;
return (UFS_UPDATE(oip, 0));
}
}
fs = oip->i_fs;
osize = DIP(oip, size);
/*
* Lengthen the size of the file. We must ensure that the
* last byte of the file is allocated. Since the smallest
* value of osize is 0, length will be at least 1.
*/
if (osize < length) {
if (length > fs->fs_maxfilesize)
return (EFBIG);
aflags = B_CLRBUF;
if (flags & IO_SYNC)
aflags |= B_SYNC;
error = UFS_BUF_ALLOC(oip, length - 1, 1,
cred, aflags, &bp);
if (error)
return (error);
if (bp->b_lblkno >= NDADDR)
indirect = 1;
DIP_ASSIGN(oip, size, length);
uvm_vnp_setsize(ovp, length);
(void) uvm_vnp_uncache(ovp);
if (aflags & B_SYNC)
bwrite(bp);
else
bawrite(bp);
oip->i_flag |= IN_CHANGE | IN_UPDATE;
error = UFS_UPDATE(oip, MNT_WAIT);
if (DOINGSOFTDEP(ovp) && num_indirdep > max_indirdep)
if (indirect) {
/*
* If the number of pending indirect block
* dependencies is sufficiently close to the
* maximum number of simultaneously mappable
* buffers force a sync on the vnode to prevent
* buffer cache exhaustion.
*/
VOP_FSYNC(ovp, curproc->p_ucred, MNT_WAIT);
}
return (error);
}
uvm_vnp_setsize(ovp, length);
/*
* Shorten the size of the file. If the file is not being
* truncated to a block boundary, the contents of the
* partial block following the end of the file must be
* zero'ed in case it ever becomes accessible again because
* of subsequent file growth. Directories however are not
* zero'ed as they should grow back initialized to empty.
*/
offset = blkoff(fs, length);
if (offset == 0) {
DIP_ASSIGN(oip, size, length);
} else {
lbn = lblkno(fs, length);
aflags = B_CLRBUF;
if (flags & IO_SYNC)
aflags |= B_SYNC;
error = UFS_BUF_ALLOC(oip, length - 1, 1,
cred, aflags, &bp);
if (error)
return (error);
/*
* When we are doing soft updates and the UFS_BALLOC
* above fills in a direct block hole with a full sized
* block that will be truncated down to a fragment below,
* we must flush out the block dependency with an FSYNC
* so that we do not get a soft updates inconsistency
* when we create the fragment below.
*/
if (DOINGSOFTDEP(ovp) && lbn < NDADDR &&
fragroundup(fs, blkoff(fs, length)) < fs->fs_bsize &&
(error = VOP_FSYNC(ovp, cred, MNT_WAIT)) != 0)
return (error);
DIP_ASSIGN(oip, size, length);
size = blksize(fs, oip, lbn);
(void) uvm_vnp_uncache(ovp);
if (ovp->v_type != VDIR)
bzero((char *)bp->b_data + offset,
(u_int)(size - offset));
bp->b_bcount = size;
if (aflags & B_SYNC)
bwrite(bp);
else
bawrite(bp);
}
/*
* Calculate index into inode's block list of
* last direct and indirect blocks (if any)
* which we want to keep. Lastblock is -1 when
* the file is truncated to 0.
*/
lastblock = lblkno(fs, length + fs->fs_bsize - 1) - 1;
lastiblock[SINGLE] = lastblock - NDADDR;
lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs);
lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs);
nblocks = btodb(fs->fs_bsize);
/*
* Update file and block pointers on disk before we start freeing
* blocks. If we crash before free'ing blocks below, the blocks
* will be returned to the free list. lastiblock values are also
* normalized to -1 for calls to ffs_indirtrunc below.
*/
for (level = TRIPLE; level >= SINGLE; level--) {
oldblks[NDADDR + level] = DIP(oip, ib[level]);
if (lastiblock[level] < 0) {
DIP_ASSIGN(oip, ib[level], 0);
lastiblock[level] = -1;
}
}
for (i = 0; i < NDADDR; i++) {
oldblks[i] = DIP(oip, db[i]);
if (i > lastblock)
DIP_ASSIGN(oip, db[i], 0);
}
oip->i_flag |= IN_CHANGE | IN_UPDATE;
if ((error = UFS_UPDATE(oip, MNT_WAIT)) != 0)
allerror = error;
/*
* Having written the new inode to disk, save its new configuration
* and put back the old block pointers long enough to process them.
* Note that we save the new block configuration so we can check it
* when we are done.
*/
for (i = 0; i < NDADDR; i++) {
newblks[i] = DIP(oip, db[i]);
DIP_ASSIGN(oip, db[i], oldblks[i]);
}
for (i = 0; i < NIADDR; i++) {
newblks[NDADDR + i] = DIP(oip, ib[i]);
DIP_ASSIGN(oip, ib[i], oldblks[NDADDR + i]);
}
DIP_ASSIGN(oip, size, osize);
vflags = ((length > 0) ? V_SAVE : 0) | V_SAVEMETA;
allerror = vinvalbuf(ovp, vflags, cred, curproc, 0, 0);
/*
* Indirect blocks first.
*/
indir_lbn[SINGLE] = -NDADDR;
indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) - 1;
indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1;
for (level = TRIPLE; level >= SINGLE; level--) {
bn = DIP(oip, ib[level]);
if (bn != 0) {
error = ffs_indirtrunc(oip, indir_lbn[level],
fsbtodb(fs, bn), lastiblock[level], level, &count);
if (error)
allerror = error;
blocksreleased += count;
if (lastiblock[level] < 0) {
DIP_ASSIGN(oip, ib[level], 0);
ffs_blkfree(oip, bn, fs->fs_bsize);
blocksreleased += nblocks;
}
}
if (lastiblock[level] >= 0)
goto done;
}
/*
* All whole direct blocks or frags.
*/
for (i = NDADDR - 1; i > lastblock; i--) {
long bsize;
bn = DIP(oip, db[i]);
if (bn == 0)
continue;
DIP_ASSIGN(oip, db[i], 0);
bsize = blksize(fs, oip, i);
ffs_blkfree(oip, bn, bsize);
blocksreleased += btodb(bsize);
}
if (lastblock < 0)
goto done;
/*
* Finally, look for a change in size of the
* last direct block; release any frags.
*/
bn = DIP(oip, db[lastblock]);
if (bn != 0) {
long oldspace, newspace;
/*
* Calculate amount of space we're giving
* back as old block size minus new block size.
*/
oldspace = blksize(fs, oip, lastblock);
DIP_ASSIGN(oip, size, length);
newspace = blksize(fs, oip, lastblock);
if (newspace == 0)
panic("ffs_truncate: newspace");
if (oldspace - newspace > 0) {
/*
* Block number of space to be free'd is
* the old block # plus the number of frags
* required for the storage we're keeping.
*/
bn += numfrags(fs, newspace);
ffs_blkfree(oip, bn, oldspace - newspace);
blocksreleased += btodb(oldspace - newspace);
}
}
done:
#ifdef DIAGNOSTIC
for (level = SINGLE; level <= TRIPLE; level++)
if (newblks[NDADDR + level] != DIP(oip, ib[level]))
panic("ffs_truncate1");
for (i = 0; i < NDADDR; i++)
if (newblks[i] != DIP(oip, db[i]))
panic("ffs_truncate2");
#endif /* DIAGNOSTIC */
/*
* Put back the real size.
*/
DIP_ASSIGN(oip, size, length);
DIP_ADD(oip, blocks, -blocksreleased);
oip->i_flag |= IN_CHANGE;
(void)ufs_quota_free_blocks(oip, blocksreleased, NOCRED);
return (allerror);
}
/*
* 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);
}
/*
* 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);
}
/*
* 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);
}
/*
* 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);
}
/*
* 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);
}
int
ufs_mkdir(void *v)
{
struct vop_mkdir_v3_args /* {
struct vnode *a_dvp;
struct vnode **a_vpp;
struct componentname *a_cnp;
struct vattr *a_vap;
} */ *ap = v;
struct vnode *dvp = ap->a_dvp, *tvp;
struct vattr *vap = ap->a_vap;
struct componentname *cnp = ap->a_cnp;
struct inode *ip, *dp = VTOI(dvp);
struct buf *bp;
struct dirtemplate dirtemplate;
struct direct *newdir;
int error, dmode;
struct ufsmount *ump = dp->i_ump;
int dirblksiz = ump->um_dirblksiz;
struct ufs_lookup_results *ulr;
fstrans_start(dvp->v_mount, FSTRANS_SHARED);
/* XXX should handle this material another way */
ulr = &dp->i_crap;
UFS_CHECK_CRAPCOUNTER(dp);
if ((nlink_t)dp->i_nlink >= LINK_MAX) {
error = EMLINK;
goto out;
}
dmode = vap->va_mode & ACCESSPERMS;
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_VALLOC(dvp, dmode, cnp->cn_cred, ap->a_vpp)) != 0)
goto out;
tvp = *ap->a_vpp;
ip = VTOI(tvp);
error = UFS_WAPBL_BEGIN(ap->a_dvp->v_mount);
if (error) {
UFS_VFREE(tvp, ip->i_number, dmode);
vput(tvp);
goto out;
}
ip->i_uid = kauth_cred_geteuid(cnp->cn_cred);
DIP_ASSIGN(ip, uid, ip->i_uid);
ip->i_gid = dp->i_gid;
DIP_ASSIGN(ip, gid, ip->i_gid);
#if defined(QUOTA) || defined(QUOTA2)
if ((error = chkiq(ip, 1, cnp->cn_cred, 0))) {
UFS_VFREE(tvp, ip->i_number, dmode);
UFS_WAPBL_END(dvp->v_mount);
fstrans_done(dvp->v_mount);
vput(tvp);
return (error);
}
#endif
ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
ip->i_mode = dmode;
DIP_ASSIGN(ip, mode, dmode);
tvp->v_type = VDIR; /* Rest init'd in getnewvnode(). */
ip->i_nlink = 2;
DIP_ASSIGN(ip, nlink, 2);
if (cnp->cn_flags & ISWHITEOUT) {
ip->i_flags |= UF_OPAQUE;
DIP_ASSIGN(ip, flags, ip->i_flags);
}
/*
* Bump link count in parent directory to reflect work done below.
* Should be done before reference is created so cleanup is
* possible if we crash.
*/
dp->i_nlink++;
DIP_ASSIGN(dp, nlink, dp->i_nlink);
dp->i_flag |= IN_CHANGE;
if ((error = UFS_UPDATE(dvp, NULL, NULL, UPDATE_DIROP)) != 0)
goto bad;
/*
* Initialize directory with "." and ".." from static template.
*/
dirtemplate = mastertemplate;
dirtemplate.dotdot_reclen = dirblksiz - dirtemplate.dot_reclen;
dirtemplate.dot_ino = ufs_rw32(ip->i_number, UFS_MPNEEDSWAP(ump));
dirtemplate.dotdot_ino = ufs_rw32(dp->i_number, UFS_MPNEEDSWAP(ump));
dirtemplate.dot_reclen = ufs_rw16(dirtemplate.dot_reclen,
UFS_MPNEEDSWAP(ump));
dirtemplate.dotdot_reclen = ufs_rw16(dirtemplate.dotdot_reclen,
UFS_MPNEEDSWAP(ump));
if (ump->um_maxsymlinklen <= 0) {
#if BYTE_ORDER == LITTLE_ENDIAN
if (UFS_MPNEEDSWAP(ump) == 0)
#else
if (UFS_MPNEEDSWAP(ump) != 0)
#endif
{
dirtemplate.dot_type = dirtemplate.dot_namlen;
dirtemplate.dotdot_type = dirtemplate.dotdot_namlen;
dirtemplate.dot_namlen = dirtemplate.dotdot_namlen = 0;
} else
dirtemplate.dot_type = dirtemplate.dotdot_type = 0;
}
if ((error = UFS_BALLOC(tvp, (off_t)0, dirblksiz, cnp->cn_cred,
B_CLRBUF, &bp)) != 0)
goto bad;
ip->i_size = dirblksiz;
DIP_ASSIGN(ip, size, dirblksiz);
ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
uvm_vnp_setsize(tvp, ip->i_size);
memcpy((void *)bp->b_data, (void *)&dirtemplate, sizeof dirtemplate);
/*
* Directory set up, now install its entry in the parent directory.
* We must write out the buffer containing the new directory body
* before entering the new name in the parent.
*/
if ((error = VOP_BWRITE(bp->b_vp, bp)) != 0)
goto bad;
if ((error = UFS_UPDATE(tvp, NULL, NULL, UPDATE_DIROP)) != 0) {
goto bad;
}
newdir = pool_cache_get(ufs_direct_cache, PR_WAITOK);
ufs_makedirentry(ip, cnp, newdir);
error = ufs_direnter(dvp, ulr, tvp, newdir, cnp, bp);
pool_cache_put(ufs_direct_cache, newdir);
bad:
if (error == 0) {
VN_KNOTE(dvp, NOTE_WRITE | NOTE_LINK);
VOP_UNLOCK(tvp);
UFS_WAPBL_END(dvp->v_mount);
} else {
dp->i_nlink--;
DIP_ASSIGN(dp, nlink, dp->i_nlink);
dp->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(dvp, NULL, NULL, UPDATE_DIROP);
/*
* No need to do an explicit UFS_TRUNCATE here, vrele will
* do this for us because we set the link count to 0.
*/
ip->i_nlink = 0;
DIP_ASSIGN(ip, nlink, 0);
ip->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(tvp, NULL, NULL, UPDATE_DIROP);
UFS_WAPBL_END(dvp->v_mount);
vput(tvp);
}
out:
fstrans_done(dvp->v_mount);
return (error);
}
/*
* ufs_link: create hard link.
*/
int
ufs_link(void *v)
{
struct vop_link_args /* {
struct vnode *a_dvp;
struct vnode *a_vp;
struct componentname *a_cnp;
} */ *ap = v;
struct vnode *dvp = ap->a_dvp;
struct vnode *vp = ap->a_vp;
struct componentname *cnp = ap->a_cnp;
struct inode *ip;
struct direct *newdir;
int error;
struct ufs_lookup_results *ulr;
KASSERT(dvp != vp);
KASSERT(vp->v_type != VDIR);
KASSERT(dvp->v_mount == vp->v_mount);
/* XXX should handle this material another way */
ulr = &VTOI(dvp)->i_crap;
UFS_CHECK_CRAPCOUNTER(VTOI(dvp));
fstrans_start(dvp->v_mount, FSTRANS_SHARED);
error = vn_lock(vp, LK_EXCLUSIVE);
if (error) {
VOP_ABORTOP(dvp, cnp);
goto out2;
}
ip = VTOI(vp);
if ((nlink_t)ip->i_nlink >= LINK_MAX) {
VOP_ABORTOP(dvp, cnp);
error = EMLINK;
goto out1;
}
if (ip->i_flags & (IMMUTABLE | APPEND)) {
VOP_ABORTOP(dvp, cnp);
error = EPERM;
goto out1;
}
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error) {
VOP_ABORTOP(dvp, cnp);
goto out1;
}
ip->i_nlink++;
DIP_ASSIGN(ip, nlink, ip->i_nlink);
ip->i_flag |= IN_CHANGE;
error = UFS_UPDATE(vp, NULL, NULL, UPDATE_DIROP);
if (!error) {
newdir = pool_cache_get(ufs_direct_cache, PR_WAITOK);
ufs_makedirentry(ip, cnp, newdir);
error = ufs_direnter(dvp, ulr, vp, newdir, cnp, NULL);
pool_cache_put(ufs_direct_cache, newdir);
}
if (error) {
ip->i_nlink--;
DIP_ASSIGN(ip, nlink, ip->i_nlink);
ip->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(vp, NULL, NULL, UPDATE_DIROP);
}
UFS_WAPBL_END(vp->v_mount);
out1:
VOP_UNLOCK(vp);
out2:
VN_KNOTE(vp, NOTE_LINK);
VN_KNOTE(dvp, NOTE_WRITE);
vput(dvp);
fstrans_done(dvp->v_mount);
return (error);
}
/*
* Set attribute vnode op. called from several syscalls
*/
int
ufs_setattr(void *v)
{
struct vop_setattr_args /* {
struct vnode *a_vp;
struct vattr *a_vap;
kauth_cred_t a_cred;
} */ *ap = v;
struct vattr *vap;
struct vnode *vp;
struct inode *ip;
kauth_cred_t cred;
struct lwp *l;
int error;
kauth_action_t action;
bool changing_sysflags;
vap = ap->a_vap;
vp = ap->a_vp;
ip = VTOI(vp);
cred = ap->a_cred;
l = curlwp;
action = KAUTH_VNODE_WRITE_FLAGS;
changing_sysflags = false;
/*
* Check for unsettable attributes.
*/
if ((vap->va_type != VNON) || (vap->va_nlink != VNOVAL) ||
(vap->va_fsid != VNOVAL) || (vap->va_fileid != VNOVAL) ||
(vap->va_blocksize != VNOVAL) || (vap->va_rdev != VNOVAL) ||
((int)vap->va_bytes != VNOVAL) || (vap->va_gen != VNOVAL)) {
return (EINVAL);
}
fstrans_start(vp->v_mount, FSTRANS_SHARED);
if (vap->va_flags != VNOVAL) {
if (vp->v_mount->mnt_flag & MNT_RDONLY) {
error = EROFS;
goto out;
}
/* Snapshot flag cannot be set or cleared */
if ((vap->va_flags & (SF_SNAPSHOT | SF_SNAPINVAL)) !=
(ip->i_flags & (SF_SNAPSHOT | SF_SNAPINVAL))) {
error = EPERM;
goto out;
}
if (ip->i_flags & (SF_IMMUTABLE | SF_APPEND)) {
action |= KAUTH_VNODE_HAS_SYSFLAGS;
}
if ((vap->va_flags & SF_SETTABLE) !=
(ip->i_flags & SF_SETTABLE)) {
action |= KAUTH_VNODE_WRITE_SYSFLAGS;
changing_sysflags = true;
}
error = kauth_authorize_vnode(cred, action, vp, NULL,
genfs_can_chflags(cred, vp->v_type, ip->i_uid,
changing_sysflags));
if (error)
goto out;
if (changing_sysflags) {
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
ip->i_flags = vap->va_flags;
DIP_ASSIGN(ip, flags, ip->i_flags);
} else {
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
ip->i_flags &= SF_SETTABLE;
ip->i_flags |= (vap->va_flags & UF_SETTABLE);
DIP_ASSIGN(ip, flags, ip->i_flags);
}
ip->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(vp, NULL, NULL, 0);
UFS_WAPBL_END(vp->v_mount);
if (vap->va_flags & (IMMUTABLE | APPEND)) {
error = 0;
goto out;
}
}
if (ip->i_flags & (IMMUTABLE | APPEND)) {
error = EPERM;
goto out;
}
/*
* Go through the fields and update iff not VNOVAL.
*/
if (vap->va_uid != (uid_t)VNOVAL || vap->va_gid != (gid_t)VNOVAL) {
if (vp->v_mount->mnt_flag & MNT_RDONLY) {
error = EROFS;
goto out;
}
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
error = ufs_chown(vp, vap->va_uid, vap->va_gid, cred, l);
UFS_WAPBL_END(vp->v_mount);
if (error)
goto out;
}
if (vap->va_size != VNOVAL) {
/*
* Disallow write attempts on read-only file systems;
* unless the file is a socket, fifo, or a block or
* character device resident on the file system.
*/
switch (vp->v_type) {
case VDIR:
error = EISDIR;
goto out;
case VCHR:
case VBLK:
case VFIFO:
break;
case VREG:
if (vp->v_mount->mnt_flag & MNT_RDONLY) {
error = EROFS;
goto out;
}
if ((ip->i_flags & SF_SNAPSHOT) != 0) {
error = EPERM;
goto out;
}
error = ufs_truncate(vp, vap->va_size, cred);
if (error)
goto out;
break;
default:
error = EOPNOTSUPP;
goto out;
}
}
ip = VTOI(vp);
if (vap->va_atime.tv_sec != VNOVAL || vap->va_mtime.tv_sec != VNOVAL ||
vap->va_birthtime.tv_sec != VNOVAL) {
if (vp->v_mount->mnt_flag & MNT_RDONLY) {
error = EROFS;
goto out;
}
if ((ip->i_flags & SF_SNAPSHOT) != 0) {
error = EPERM;
goto out;
}
error = kauth_authorize_vnode(cred, KAUTH_VNODE_WRITE_TIMES, vp,
NULL, genfs_can_chtimes(vp, vap->va_vaflags, ip->i_uid, cred));
if (error)
goto out;
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
if (vap->va_atime.tv_sec != VNOVAL)
if (!(vp->v_mount->mnt_flag & MNT_NOATIME))
ip->i_flag |= IN_ACCESS;
if (vap->va_mtime.tv_sec != VNOVAL) {
ip->i_flag |= IN_CHANGE | IN_UPDATE;
if (vp->v_mount->mnt_flag & MNT_RELATIME)
ip->i_flag |= IN_ACCESS;
}
if (vap->va_birthtime.tv_sec != VNOVAL &&
ip->i_ump->um_fstype == UFS2) {
ip->i_ffs2_birthtime = vap->va_birthtime.tv_sec;
ip->i_ffs2_birthnsec = vap->va_birthtime.tv_nsec;
}
error = UFS_UPDATE(vp, &vap->va_atime, &vap->va_mtime, 0);
UFS_WAPBL_END(vp->v_mount);
if (error)
goto out;
}
error = 0;
if (vap->va_mode != (mode_t)VNOVAL) {
if (vp->v_mount->mnt_flag & MNT_RDONLY) {
error = EROFS;
goto out;
}
if ((ip->i_flags & SF_SNAPSHOT) != 0 &&
(vap->va_mode & (S_IXUSR | S_IWUSR | S_IXGRP | S_IWGRP |
S_IXOTH | S_IWOTH))) {
error = EPERM;
goto out;
}
error = UFS_WAPBL_BEGIN(vp->v_mount);
if (error)
goto out;
error = ufs_chmod(vp, (int)vap->va_mode, cred, l);
UFS_WAPBL_END(vp->v_mount);
}
VN_KNOTE(vp, NOTE_ATTRIB);
out:
fstrans_done(vp->v_mount);
return (error);
}
/*
* Allocate a new inode.
*/
int
ufs_makeinode(int mode, struct vnode *dvp, const struct ufs_lookup_results *ulr,
struct vnode **vpp, struct componentname *cnp)
{
struct inode *ip, *pdir;
struct direct *newdir;
struct vnode *tvp;
int error;
UFS_WAPBL_JUNLOCK_ASSERT(dvp->v_mount);
pdir = VTOI(dvp);
if ((mode & IFMT) == 0)
mode |= IFREG;
if ((error = UFS_VALLOC(dvp, mode, cnp->cn_cred, vpp)) != 0) {
return (error);
}
tvp = *vpp;
ip = VTOI(tvp);
ip->i_gid = pdir->i_gid;
DIP_ASSIGN(ip, gid, ip->i_gid);
ip->i_uid = kauth_cred_geteuid(cnp->cn_cred);
DIP_ASSIGN(ip, uid, ip->i_uid);
error = UFS_WAPBL_BEGIN1(dvp->v_mount, dvp);
if (error) {
/*
* Note, we can't VOP_VFREE(tvp) here like we should
* because we can't write to the disk. Instead, we leave
* the vnode dangling from the journal.
*/
vput(tvp);
return (error);
}
#if defined(QUOTA) || defined(QUOTA2)
if ((error = chkiq(ip, 1, cnp->cn_cred, 0))) {
UFS_VFREE(tvp, ip->i_number, mode);
UFS_WAPBL_END1(dvp->v_mount, dvp);
vput(tvp);
return (error);
}
#endif
ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
ip->i_mode = mode;
DIP_ASSIGN(ip, mode, mode);
tvp->v_type = IFTOVT(mode); /* Rest init'd in getnewvnode(). */
ip->i_nlink = 1;
DIP_ASSIGN(ip, nlink, 1);
/* Authorize setting SGID if needed. */
if (ip->i_mode & ISGID) {
error = kauth_authorize_vnode(cnp->cn_cred, KAUTH_VNODE_WRITE_SECURITY,
tvp, NULL, genfs_can_chmod(tvp->v_type, cnp->cn_cred, ip->i_uid,
ip->i_gid, mode));
if (error) {
ip->i_mode &= ~ISGID;
DIP_ASSIGN(ip, mode, ip->i_mode);
}
}
if (cnp->cn_flags & ISWHITEOUT) {
ip->i_flags |= UF_OPAQUE;
DIP_ASSIGN(ip, flags, ip->i_flags);
}
/*
* Make sure inode goes to disk before directory entry.
*/
if ((error = UFS_UPDATE(tvp, NULL, NULL, UPDATE_DIROP)) != 0)
goto bad;
newdir = pool_cache_get(ufs_direct_cache, PR_WAITOK);
ufs_makedirentry(ip, cnp, newdir);
error = ufs_direnter(dvp, ulr, tvp, newdir, cnp, NULL);
pool_cache_put(ufs_direct_cache, newdir);
if (error)
goto bad;
*vpp = tvp;
return (0);
bad:
/*
* Write error occurred trying to update the inode
* or the directory so must deallocate the inode.
*/
ip->i_nlink = 0;
DIP_ASSIGN(ip, nlink, 0);
ip->i_flag |= IN_CHANGE;
UFS_WAPBL_UPDATE(tvp, NULL, NULL, 0);
tvp->v_type = VNON; /* explodes later if VBLK */
UFS_WAPBL_END1(dvp->v_mount, dvp);
vput(tvp);
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);
}
/*
* Synch an open file.
*/
int
ffs_fsync(void *v)
{
struct vop_fsync_args *ap = v;
struct vnode *vp = ap->a_vp;
struct buf *bp, *nbp;
int s, error, passes, skipmeta;
if (vp->v_type == VBLK &&
vp->v_specmountpoint != NULL &&
(vp->v_specmountpoint->mnt_flag & MNT_SOFTDEP))
softdep_fsync_mountdev(vp, ap->a_waitfor);
/*
* Flush all dirty buffers associated with a vnode.
*/
passes = NIADDR + 1;
skipmeta = 0;
if (ap->a_waitfor == MNT_WAIT)
skipmeta = 1;
s = splbio();
loop:
for (bp = LIST_FIRST(&vp->v_dirtyblkhd); bp;
bp = LIST_NEXT(bp, b_vnbufs))
bp->b_flags &= ~B_SCANNED;
for (bp = LIST_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) {
nbp = LIST_NEXT(bp, b_vnbufs);
/*
* Reasons to skip this buffer: it has already been considered
* on this pass, this pass is the first time through on a
* synchronous flush request and the buffer being considered
* is metadata, the buffer has dependencies that will cause
* it to be redirtied and it has not already been deferred,
* or it is already being written.
*/
if (bp->b_flags & (B_BUSY | B_SCANNED))
continue;
if ((bp->b_flags & B_DELWRI) == 0)
panic("ffs_fsync: not dirty");
if (skipmeta && bp->b_lblkno < 0)
continue;
if (ap->a_waitfor != MNT_WAIT &&
LIST_FIRST(&bp->b_dep) != NULL &&
(bp->b_flags & B_DEFERRED) == 0 &&
buf_countdeps(bp, 0, 1)) {
bp->b_flags |= B_DEFERRED;
continue;
}
bremfree(bp);
buf_acquire(bp);
bp->b_flags |= B_SCANNED;
splx(s);
/*
* On our final pass through, do all I/O synchronously
* so that we can find out if our flush is failing
* because of write errors.
*/
if (passes > 0 || ap->a_waitfor != MNT_WAIT)
(void) bawrite(bp);
else if ((error = bwrite(bp)) != 0)
return (error);
s = splbio();
/*
* Since we may have slept during the I/O, we need
* to start from a known point.
*/
nbp = LIST_FIRST(&vp->v_dirtyblkhd);
}
if (skipmeta) {
skipmeta = 0;
goto loop;
}
if (ap->a_waitfor == MNT_WAIT) {
vwaitforio(vp, 0, "ffs_fsync", 0);
/*
* Ensure that any filesystem metadata associated
* with the vnode has been written.
*/
splx(s);
if ((error = softdep_sync_metadata(ap)) != 0)
return (error);
s = splbio();
if (!LIST_EMPTY(&vp->v_dirtyblkhd)) {
/*
* Block devices associated with filesystems may
* have new I/O requests posted for them even if
* the vnode is locked, so no amount of trying will
* get them clean. Thus we give block devices a
* good effort, then just give up. For all other file
* types, go around and try again until it is clean.
*/
if (passes > 0) {
passes -= 1;
goto loop;
}
#ifdef DIAGNOSTIC
if (vp->v_type != VBLK)
vprint("ffs_fsync: dirty", vp);
#endif
}
}
splx(s);
return (UFS_UPDATE(VTOI(vp), ap->a_waitfor == MNT_WAIT));
}
/*
* Write a directory entry after a call to namei, using the parameters
* that it left in nameidata. The argument dirp is the new directory
* entry contents. Dvp is a pointer to the directory to be written,
* which was left locked by namei. Remaining parameters (dp->i_offset,
* dp->i_count) indicate how the space for the new entry is to be obtained.
* Non-null bp indicates that a directory is being created (for the
* soft dependency code).
*/
int
ufs_direnter(struct vnode *dvp, struct vnode *tvp, struct direct *dirp,
struct componentname *cnp, struct buf *newdirbp)
{
struct ucred *cr;
struct proc *p;
int newentrysize;
struct inode *dp;
struct buf *bp;
u_int dsize;
struct direct *ep, *nep;
int error, ret, blkoff, loc, spacefree, flags;
char *dirbuf;
UFS_WAPBL_JLOCK_ASSERT(dvp->v_mount);
error = 0;
cr = cnp->cn_cred;
p = cnp->cn_proc;
dp = VTOI(dvp);
newentrysize = DIRSIZ(FSFMT(dvp), dirp);
if (dp->i_count == 0) {
/*
* If dp->i_count is 0, then namei could find no
* space in the directory. Here, dp->i_offset will
* be on a directory block boundary and we will write the
* new entry into a fresh block.
*/
if (dp->i_offset & (DIRBLKSIZ - 1))
panic("ufs_direnter: newblk");
flags = B_CLRBUF;
if (!DOINGSOFTDEP(dvp))
flags |= B_SYNC;
if ((error = UFS_BUF_ALLOC(dp, (off_t)dp->i_offset, DIRBLKSIZ,
cr, flags, &bp)) != 0) {
if (DOINGSOFTDEP(dvp) && newdirbp != NULL)
bdwrite(newdirbp);
return (error);
}
DIP_ASSIGN(dp, size, dp->i_offset + DIRBLKSIZ);
dp->i_flag |= IN_CHANGE | IN_UPDATE;
uvm_vnp_setsize(dvp, DIP(dp, size));
dirp->d_reclen = DIRBLKSIZ;
blkoff = dp->i_offset &
(VFSTOUFS(dvp->v_mount)->um_mountp->mnt_stat.f_iosize - 1);
memcpy(bp->b_data + blkoff, dirp, newentrysize);
#ifdef UFS_DIRHASH
if (dp->i_dirhash != NULL) {
ufsdirhash_newblk(dp, dp->i_offset);
ufsdirhash_add(dp, dirp, dp->i_offset);
ufsdirhash_checkblock(dp, (char *)bp->b_data + blkoff,
dp->i_offset);
}
#endif
if (DOINGSOFTDEP(dvp)) {
/*
* 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 (softdep_setup_directory_add(bp, dp, dp->i_offset,
dirp->d_ino, newdirbp, 1) == 0) {
bdwrite(bp);
return (UFS_UPDATE(dp, 0));
}
/* We have just allocated a directory block in an
* indirect block. Rather than tracking when it gets
* claimed by the inode, we simply do a VOP_FSYNC
* now to ensure that it is there (in case the user
* does a future fsync). Note that we have to unlock
* the inode for the entry that we just entered, as
* the VOP_FSYNC may need to lock other inodes which
* can lead to deadlock if we also hold a lock on
* the newly entered node.
*/
if ((error = VOP_BWRITE(bp)))
return (error);
if (tvp != NULL)
VOP_UNLOCK(tvp, 0);
error = VOP_FSYNC(dvp, p->p_ucred, MNT_WAIT);
if (tvp != NULL)
vn_lock(tvp, LK_EXCLUSIVE | LK_RETRY, p);
return (error);
}
error = VOP_BWRITE(bp);
ret = UFS_UPDATE(dp, !DOINGSOFTDEP(dvp));
if (error == 0)
return (ret);
return (error);
}
/*
* If dp->i_count is non-zero, then namei found space for the new
* entry in the range dp->i_offset to dp->i_offset + dp->i_count
* in the directory. To use this space, we may have to compact
* the entries located there, by copying them together towards the
* beginning of the block, leaving the free space in one usable
* chunk at the end.
*/
/*
* Increase size of directory if entry eats into new space.
* This should never push the size past a new multiple of
* DIRBLKSIZE.
*
* N.B. - THIS IS AN ARTIFACT OF 4.2 AND SHOULD NEVER HAPPEN.
*/
if (dp->i_offset + dp->i_count > DIP(dp, size)) {
DIP_ASSIGN(dp, size, dp->i_offset + dp->i_count);
dp->i_flag |= IN_CHANGE | IN_UPDATE;
UFS_WAPBL_UPDATE(dp, MNT_WAIT);
}
/*
* Get the block containing the space for the new directory entry.
*/
if ((error = UFS_BUFATOFF(dp, (off_t)dp->i_offset, &dirbuf, &bp))
!= 0) {
if (DOINGSOFTDEP(dvp) && newdirbp != NULL)
bdwrite(newdirbp);
return (error);
}
/*
* Find space for the new entry. In the simple case, the entry at
* offset base will have the space. If it does not, then namei
* arranged that compacting the region dp->i_offset to
* dp->i_offset + dp->i_count would yield the space.
*/
ep = (struct direct *)dirbuf;
dsize = ep->d_ino ? DIRSIZ(FSFMT(dvp), ep) : 0;
spacefree = ep->d_reclen - dsize;
for (loc = ep->d_reclen; loc < dp->i_count; ) {
nep = (struct direct *)(dirbuf + loc);
/* Trim the existing slot (NB: dsize may be zero). */
ep->d_reclen = dsize;
ep = (struct direct *)((char *)ep + dsize);
/* Read nep->d_reclen now as the memmove() may clobber it. */
loc += nep->d_reclen;
if (nep->d_ino == 0) {
/*
* A mid-block unused entry. Such entries are
* never created by the kernel, but fsck_ffs
* can create them (and it doesn't fix them).
*
* Add up the free space, and initialise the
* relocated entry since we don't memmove it.
*/
spacefree += nep->d_reclen;
ep->d_ino = 0;
dsize = 0;
continue;
}
dsize = DIRSIZ(FSFMT(dvp), nep);
spacefree += nep->d_reclen - dsize;
#ifdef UFS_DIRHASH
if (dp->i_dirhash != NULL)
ufsdirhash_move(dp, nep,
dp->i_offset + ((char *)nep - dirbuf),
dp->i_offset + ((char *)ep - dirbuf));
#endif
if (DOINGSOFTDEP(dvp))
softdep_change_directoryentry_offset(dp, dirbuf,
(caddr_t)nep, (caddr_t)ep, dsize);
else
memmove(ep, nep, dsize);
}
/*
* Here, `ep' points to a directory entry containing `dsize' in-use
* bytes followed by `spacefree' unused bytes. If ep->d_ino == 0,
* then the entry is completely unused (dsize == 0). The value
* of ep->d_reclen is always indeterminate.
*
* Update the pointer fields in the previous entry (if any),
* copy in the new entry, and write out the block.
*/
if (ep->d_ino == 0) {
if (spacefree + dsize < newentrysize)
panic("ufs_direnter: compact1");
dirp->d_reclen = spacefree + dsize;
} else {
if (spacefree < newentrysize)
panic("ufs_direnter: compact2");
dirp->d_reclen = spacefree;
ep->d_reclen = dsize;
ep = (struct direct *)((char *)ep + dsize);
}
#ifdef UFS_DIRHASH
if (dp->i_dirhash != NULL && (ep->d_ino == 0 ||
dirp->d_reclen == spacefree))
ufsdirhash_add(dp, dirp, dp->i_offset + ((char *)ep - dirbuf));
#endif
memcpy(ep, dirp, newentrysize);
#ifdef UFS_DIRHASH
if (dp->i_dirhash != NULL)
ufsdirhash_checkblock(dp, dirbuf -
(dp->i_offset & (DIRBLKSIZ - 1)),
dp->i_offset & ~(DIRBLKSIZ - 1));
#endif
if (DOINGSOFTDEP(dvp)) {
(void)softdep_setup_directory_add(bp, dp,
dp->i_offset + (caddr_t)ep - dirbuf,
dirp->d_ino, newdirbp, 0);
bdwrite(bp);
} else {
error = VOP_BWRITE(bp);
}
dp->i_flag |= IN_CHANGE | IN_UPDATE;
/*
* If all went well, and the directory can be shortened, proceed
* with the truncation. Note that we have to unlock the inode for
* the entry that we just entered, as the truncation may need to
* lock other inodes which can lead to deadlock if we also hold a
* lock on the newly entered node.
*/
if (error == 0 && dp->i_endoff && dp->i_endoff < DIP(dp, size)) {
if (tvp != NULL)
VOP_UNLOCK(tvp, 0);
#ifdef UFS_DIRHASH
if (dp->i_dirhash != NULL)
ufsdirhash_dirtrunc(dp, dp->i_endoff);
#endif
error = UFS_TRUNCATE(dp, (off_t)dp->i_endoff, IO_SYNC, cr);
if (tvp != NULL)
vn_lock(tvp, LK_EXCLUSIVE | LK_RETRY, p);
}
UFS_WAPBL_UPDATE(dp, MNT_WAIT);
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);
}