*
* CDDL HEADER END
*/
+
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 by Delphix. All rights reserved.
- * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
+ * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
* Copyright (c) 2015 by Chunwei Chen. All rights reserved.
+ * Copyright 2017 Nexenta Systems, Inc.
*/
/* Portions Copyright 2007 Jeremy Teo */
#include <sys/types.h>
#include <sys/param.h>
#include <sys/time.h>
-#include <sys/systm.h>
#include <sys/sysmacros.h>
-#include <sys/resource.h>
#include <sys/vfs.h>
-#include <sys/vfs_opreg.h>
#include <sys/file.h>
#include <sys/stat.h>
#include <sys/kmem.h>
#include <sys/uio.h>
#include <sys/vmsystm.h>
#include <sys/atomic.h>
-#include <vm/pvn.h>
#include <sys/pathname.h>
#include <sys/cmn_err.h>
#include <sys/errno.h>
-#include <sys/unistd.h>
#include <sys/zfs_dir.h>
#include <sys/zfs_acl.h>
#include <sys/zfs_ioctl.h>
#include <sys/dbuf.h>
#include <sys/zap.h>
#include <sys/sa.h>
-#include <sys/dirent.h>
#include <sys/policy.h>
#include <sys/sunddi.h>
#include <sys/sid.h>
#include <sys/mode.h>
-#include "fs/fs_subr.h"
#include <sys/zfs_ctldir.h>
#include <sys/zfs_fuid.h>
#include <sys/zfs_sa.h>
#include <sys/zfs_vnops.h>
-#include <sys/dnlc.h>
#include <sys/zfs_rlock.h>
-#include <sys/extdirent.h>
-#include <sys/kidmap.h>
#include <sys/cred.h>
-#include <sys/attr.h>
#include <sys/zpl.h>
+#include <sys/zil.h>
+#include <sys/sa_impl.h>
/*
* Programming rules.
* to freed memory. The example below illustrates the following Big Rules:
*
* (1) A check must be made in each zfs thread for a mounted file system.
- * This is done avoiding races using ZFS_ENTER(zsb).
- * A ZFS_EXIT(zsb) is needed before all returns. Any znodes
+ * This is done avoiding races using ZFS_ENTER(zfsvfs).
+ * A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes
* must be checked with ZFS_VERIFY_ZP(zp). Both of these macros
* can return EIO from the calling function.
*
* Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open()
* forever, because the previous txg can't quiesce until B's tx commits.
*
- * If dmu_tx_assign() returns ERESTART and zsb->z_assign is TXG_NOWAIT,
+ * If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT,
* then drop all locks, call dmu_tx_wait(), and try again. On subsequent
- * calls to dmu_tx_assign(), pass TXG_WAITED rather than TXG_NOWAIT,
+ * calls to dmu_tx_assign(), pass TXG_NOTHROTTLE in addition to TXG_NOWAIT,
* to indicate that this operation has already called dmu_tx_wait().
* This will ensure that we don't retry forever, waiting a short bit
* each time.
*
* In general, this is how things should be ordered in each vnode op:
*
- * ZFS_ENTER(zsb); // exit if unmounted
+ * ZFS_ENTER(zfsvfs); // exit if unmounted
* top:
* zfs_dirent_lock(&dl, ...) // lock directory entry (may igrab())
* rw_enter(...); // grab any other locks you need
* tx = dmu_tx_create(...); // get DMU tx
* dmu_tx_hold_*(); // hold each object you might modify
- * error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ * error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
* if (error) {
* rw_exit(...); // drop locks
* zfs_dirent_unlock(dl); // unlock directory entry
* goto top;
* }
* dmu_tx_abort(tx); // abort DMU tx
- * ZFS_EXIT(zsb); // finished in zfs
+ * ZFS_EXIT(zfsvfs); // finished in zfs
* return (error); // really out of space
* }
* error = do_real_work(); // do whatever this VOP does
* zfs_dirent_unlock(dl); // unlock directory entry
* iput(...); // release held vnodes
* zil_commit(zilog, foid); // synchronous when necessary
- * ZFS_EXIT(zsb); // finished in zfs
+ * ZFS_EXIT(zfsvfs); // finished in zfs
* return (error); // done, report error
*/
zfs_open(struct inode *ip, int mode, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
/* Honor ZFS_APPENDONLY file attribute */
if ((mode & FMODE_WRITE) && (zp->z_pflags & ZFS_APPENDONLY) &&
((flag & O_APPEND) == 0)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
/* Virus scan eligible files on open */
- if (!zfs_has_ctldir(zp) && zsb->z_vscan && S_ISREG(ip->i_mode) &&
+ if (!zfs_has_ctldir(zp) && zfsvfs->z_vscan && S_ISREG(ip->i_mode) &&
!(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) {
if (zfs_vscan(ip, cr, 0) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EACCES));
}
}
if (flag & O_SYNC)
atomic_inc_32(&zp->z_sync_cnt);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_open);
/* ARGSUSED */
int
zfs_close(struct inode *ip, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
/* Decrement the synchronous opens in the znode */
if (flag & O_SYNC)
atomic_dec_32(&zp->z_sync_cnt);
- if (!zfs_has_ctldir(zp) && zsb->z_vscan && S_ISREG(ip->i_mode) &&
+ if (!zfs_has_ctldir(zp) && zfsvfs->z_vscan && S_ISREG(ip->i_mode) &&
!(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0)
VERIFY(zfs_vscan(ip, cr, 1) == 0);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_close);
#if defined(SEEK_HOLE) && defined(SEEK_DATA)
/*
if (error == ESRCH)
return (SET_ERROR(ENXIO));
+ /* file was dirty, so fall back to using generic logic */
+ if (error == EBUSY) {
+ if (hole)
+ *off = file_sz;
+
+ return (0);
+ }
+
/*
* We could find a hole that begins after the logical end-of-file,
* because dmu_offset_next() only works on whole blocks. If the
zfs_holey(struct inode *ip, int cmd, loff_t *off)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int error;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
error = zfs_holey_common(ip, cmd, off);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_holey);
#endif /* SEEK_HOLE && SEEK_DATA */
#if defined(_KERNEL)
pp = find_lock_page(mp, start >> PAGE_SHIFT);
if (pp) {
ASSERT(PageUptodate(pp));
+ unlock_page(pp);
pb = kmap(pp);
error = uiomove(pb + off, bytes, UIO_READ, uio);
flush_dcache_page(pp);
mark_page_accessed(pp);
- unlock_page(pp);
put_page(pp);
} else {
error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
int
zfs_read(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
{
- znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
- ssize_t n, nbytes;
- int error = 0;
- rl_t *rl;
-#ifdef HAVE_UIO_ZEROCOPY
- xuio_t *xuio = NULL;
-#endif /* HAVE_UIO_ZEROCOPY */
+ int error = 0;
- ZFS_ENTER(zsb);
+ znode_t *zp = ITOZ(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if (zp->z_pflags & ZFS_AV_QUARANTINED) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EACCES));
}
* Validate file offset
*/
if (uio->uio_loffset < (offset_t)0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
* Fasttrack empty reads
*/
if (uio->uio_resid == 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
/*
* If we're in FRSYNC mode, sync out this znode before reading it.
+ * Only do this for non-snapshots.
*/
- if (ioflag & FRSYNC || zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
- zil_commit(zsb->z_log, zp->z_id);
+ if (zfsvfs->z_log &&
+ (ioflag & FRSYNC || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS))
+ zil_commit(zfsvfs->z_log, zp->z_id);
/*
* Lock the range against changes.
*/
- rl = zfs_range_lock(&zp->z_range_lock, uio->uio_loffset, uio->uio_resid,
- RL_READER);
+ locked_range_t *lr = rangelock_enter(&zp->z_rangelock,
+ uio->uio_loffset, uio->uio_resid, RL_READER);
/*
* If we are reading past end-of-file we can skip
}
ASSERT(uio->uio_loffset < zp->z_size);
- n = MIN(uio->uio_resid, zp->z_size - uio->uio_loffset);
+ ssize_t n = MIN(uio->uio_resid, zp->z_size - uio->uio_loffset);
+ ssize_t start_resid = n;
#ifdef HAVE_UIO_ZEROCOPY
+ xuio_t *xuio = NULL;
if ((uio->uio_extflg == UIO_XUIO) &&
(((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) {
int nblk;
#endif /* HAVE_UIO_ZEROCOPY */
while (n > 0) {
- nbytes = MIN(n, zfs_read_chunk_size -
+ ssize_t nbytes = MIN(n, zfs_read_chunk_size -
P2PHASE(uio->uio_loffset, zfs_read_chunk_size));
if (zp->z_is_mapped && !(ioflag & O_DIRECT)) {
n -= nbytes;
}
+
+ int64_t nread = start_resid - n;
+ dataset_kstats_update_read_kstats(&zfsvfs->z_kstat, nread);
+ task_io_account_read(nread);
out:
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_read);
/*
* Write the bytes to a file.
int
zfs_write(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
{
- znode_t *zp = ITOZ(ip);
- rlim64_t limit = uio->uio_limit;
- ssize_t start_resid = uio->uio_resid;
- ssize_t tx_bytes;
- uint64_t end_size;
- dmu_tx_t *tx;
- zfs_sb_t *zsb = ZTOZSB(zp);
- zilog_t *zilog;
- offset_t woff;
- ssize_t n, nbytes;
- rl_t *rl;
- int max_blksz = zsb->z_max_blksz;
- int error = 0;
- arc_buf_t *abuf;
- const iovec_t *aiov = NULL;
- xuio_t *xuio = NULL;
- int i_iov = 0;
- const iovec_t *iovp = uio->uio_iov;
- int write_eof;
- int count = 0;
- sa_bulk_attr_t bulk[4];
- uint64_t mtime[2], ctime[2];
- ASSERTV(int iovcnt = uio->uio_iovcnt);
+ int error = 0;
+ ssize_t start_resid = uio->uio_resid;
/*
* Fasttrack empty write
*/
- n = start_resid;
+ ssize_t n = start_resid;
if (n == 0)
return (0);
+ rlim64_t limit = uio->uio_limit;
if (limit == RLIM64_INFINITY || limit > MAXOFFSET_T)
limit = MAXOFFSET_T;
- ZFS_ENTER(zsb);
+ znode_t *zp = ITOZ(ip);
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL, &mtime, 16);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL, &ctime, 16);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, 8);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL,
+ sa_bulk_attr_t bulk[4];
+ int count = 0;
+ uint64_t mtime[2], ctime[2];
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL,
+ &zp->z_size, 8);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, 8);
/*
* Callers might not be able to detect properly that we are read-only,
* so check it explicitly here.
*/
- if (zfs_is_readonly(zsb)) {
- ZFS_EXIT(zsb);
+ if (zfs_is_readonly(zfsvfs)) {
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EROFS));
}
if ((zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY)) ||
((zp->z_pflags & ZFS_APPENDONLY) && !(ioflag & FAPPEND) &&
(uio->uio_loffset < zp->z_size))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
- zilog = zsb->z_log;
-
/*
* Validate file offset
*/
- woff = ioflag & FAPPEND ? zp->z_size : uio->uio_loffset;
+ offset_t woff = ioflag & FAPPEND ? zp->z_size : uio->uio_loffset;
if (woff < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
+ int max_blksz = zfsvfs->z_max_blksz;
+ xuio_t *xuio = NULL;
+
/*
* Pre-fault the pages to ensure slow (eg NFS) pages
* don't hold up txg.
xuio = (xuio_t *)uio;
else
#endif
- uio_prefaultpages(MIN(n, max_blksz), uio);
+ if (uio_prefaultpages(MIN(n, max_blksz), uio)) {
+ ZFS_EXIT(zfsvfs);
+ return (SET_ERROR(EFAULT));
+ }
/*
* If in append mode, set the io offset pointer to eof.
*/
+ locked_range_t *lr;
if (ioflag & FAPPEND) {
/*
* Obtain an appending range lock to guarantee file append
* semantics. We reset the write offset once we have the lock.
*/
- rl = zfs_range_lock(&zp->z_range_lock, 0, n, RL_APPEND);
- woff = rl->r_off;
- if (rl->r_len == UINT64_MAX) {
+ lr = rangelock_enter(&zp->z_rangelock, 0, n, RL_APPEND);
+ woff = lr->lr_offset;
+ if (lr->lr_length == UINT64_MAX) {
/*
* We overlocked the file because this write will cause
* the file block size to increase.
* this write, then this range lock will lock the entire file
* so that we can re-write the block safely.
*/
- rl = zfs_range_lock(&zp->z_range_lock, woff, n, RL_WRITER);
+ lr = rangelock_enter(&zp->z_rangelock, woff, n, RL_WRITER);
}
if (woff >= limit) {
- zfs_range_unlock(rl);
- ZFS_EXIT(zsb);
+ rangelock_exit(lr);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EFBIG));
}
n = limit - woff;
/* Will this write extend the file length? */
- write_eof = (woff + n > zp->z_size);
+ int write_eof = (woff + n > zp->z_size);
+
+ uint64_t end_size = MAX(zp->z_size, woff + n);
+ zilog_t *zilog = zfsvfs->z_log;
+#ifdef HAVE_UIO_ZEROCOPY
+ int i_iov = 0;
+ const iovec_t *iovp = uio->uio_iov;
+ ASSERTV(int iovcnt = uio->uio_iovcnt);
+#endif
- end_size = MAX(zp->z_size, woff + n);
/*
* Write the file in reasonable size chunks. Each chunk is written
* and allows us to do more fine-grained space accounting.
*/
while (n > 0) {
- abuf = NULL;
woff = uio->uio_loffset;
- if (zfs_owner_overquota(zsb, zp, B_FALSE) ||
- zfs_owner_overquota(zsb, zp, B_TRUE)) {
- if (abuf != NULL)
- dmu_return_arcbuf(abuf);
+
+ if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT,
+ KUID_TO_SUID(ip->i_uid)) ||
+ zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT,
+ KGID_TO_SGID(ip->i_gid)) ||
+ (zp->z_projid != ZFS_DEFAULT_PROJID &&
+ zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT,
+ zp->z_projid))) {
error = SET_ERROR(EDQUOT);
break;
}
- if (xuio && abuf == NULL) {
+ arc_buf_t *abuf = NULL;
+ const iovec_t *aiov = NULL;
+ if (xuio) {
+#ifdef HAVE_UIO_ZEROCOPY
ASSERT(i_iov < iovcnt);
ASSERT3U(uio->uio_segflg, !=, UIO_BVEC);
aiov = &iovp[i_iov];
((char *)aiov->iov_base - (char *)abuf->b_data +
aiov->iov_len == arc_buf_size(abuf)));
i_iov++;
- } else if (abuf == NULL && n >= max_blksz &&
- woff >= zp->z_size &&
+#endif
+ } else if (n >= max_blksz && woff >= zp->z_size &&
P2PHASE(woff, max_blksz) == 0 &&
zp->z_blksz == max_blksz) {
/*
/*
* Start a transaction.
*/
- tx = dmu_tx_create(zsb->z_os);
+ dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
dmu_tx_hold_write(tx, zp->z_id, woff, MIN(n, max_blksz));
zfs_sa_upgrade_txholds(tx, zp);
}
/*
- * If zfs_range_lock() over-locked we grow the blocksize
+ * If rangelock_enter() over-locked we grow the blocksize
* and then reduce the lock range. This will only happen
- * on the first iteration since zfs_range_reduce() will
- * shrink down r_len to the appropriate size.
+ * on the first iteration since rangelock_reduce() will
+ * shrink down lr_length to the appropriate size.
*/
- if (rl->r_len == UINT64_MAX) {
+ if (lr->lr_length == UINT64_MAX) {
uint64_t new_blksz;
if (zp->z_blksz > max_blksz) {
new_blksz = MIN(end_size, max_blksz);
}
zfs_grow_blocksize(zp, new_blksz, tx);
- zfs_range_reduce(rl, woff, n);
+ rangelock_reduce(lr, woff, n);
}
/*
* XXX - should we really limit each write to z_max_blksz?
* Perhaps we should use SPA_MAXBLOCKSIZE chunks?
*/
- nbytes = MIN(n, max_blksz - P2PHASE(woff, max_blksz));
+ ssize_t nbytes = MIN(n, max_blksz - P2PHASE(woff, max_blksz));
+ ssize_t tx_bytes;
if (abuf == NULL) {
tx_bytes = uio->uio_resid;
+ uio->uio_fault_disable = B_TRUE;
error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl),
uio, nbytes, tx);
+ if (error == EFAULT) {
+ dmu_tx_commit(tx);
+ if (uio_prefaultpages(MIN(n, max_blksz), uio)) {
+ break;
+ }
+ continue;
+ } else if (error != 0) {
+ dmu_tx_commit(tx);
+ break;
+ }
tx_bytes -= uio->uio_resid;
} else {
tx_bytes = nbytes;
if (tx_bytes < max_blksz && (!write_eof ||
aiov->iov_base != abuf->b_data)) {
ASSERT(xuio);
- dmu_write(zsb->z_os, zp->z_id, woff,
+ dmu_write(zfsvfs->z_os, zp->z_id, woff,
+ /* cppcheck-suppress nullPointer */
aiov->iov_len, aiov->iov_base, tx);
dmu_return_arcbuf(abuf);
xuio_stat_wbuf_copied();
} else {
ASSERT(xuio || tx_bytes == max_blksz);
- dmu_assign_arcbuf(sa_get_db(zp->z_sa_hdl),
- woff, abuf, tx);
+ error = dmu_assign_arcbuf_by_dbuf(
+ sa_get_db(zp->z_sa_hdl), woff, abuf, tx);
+ if (error != 0) {
+ dmu_return_arcbuf(abuf);
+ dmu_tx_commit(tx);
+ break;
+ }
}
ASSERT(tx_bytes <= uio->uio_resid);
uioskip(uio, tx_bytes);
}
-
- if (tx_bytes && zp->z_is_mapped && !(ioflag & O_DIRECT))
- update_pages(ip, woff, tx_bytes, zsb->z_os, zp->z_id);
+ if (tx_bytes && zp->z_is_mapped && !(ioflag & O_DIRECT)) {
+ update_pages(ip, woff,
+ tx_bytes, zfsvfs->z_os, zp->z_id);
+ }
/*
* If we made no progress, we're done. If we made even
* partial progress, update the znode and ZIL accordingly.
*/
if (tx_bytes == 0) {
- (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zsb),
+ (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
(void *)&zp->z_size, sizeof (uint64_t), tx);
dmu_tx_commit(tx);
ASSERT(error != 0);
/*
* Clear Set-UID/Set-GID bits on successful write if not
- * privileged and at least one of the excute bits is set.
+ * privileged and at least one of the execute bits is set.
*
* It would be nice to to this after all writes have
* been done, but that would still expose the ISUID/ISGID
* user 0 is not an ephemeral uid.
*/
mutex_enter(&zp->z_acl_lock);
+ uint32_t uid = KUID_TO_SUID(ip->i_uid);
if ((zp->z_mode & (S_IXUSR | (S_IXUSR >> 3) |
(S_IXUSR >> 6))) != 0 &&
(zp->z_mode & (S_ISUID | S_ISGID)) != 0 &&
secpolicy_vnode_setid_retain(cr,
- (zp->z_mode & S_ISUID) != 0 && zp->z_uid == 0) != 0) {
+ ((zp->z_mode & S_ISUID) != 0 && uid == 0)) != 0) {
uint64_t newmode;
zp->z_mode &= ~(S_ISUID | S_ISGID);
- newmode = zp->z_mode;
- (void) sa_update(zp->z_sa_hdl, SA_ZPL_MODE(zsb),
+ ip->i_mode = newmode = zp->z_mode;
+ (void) sa_update(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs),
(void *)&newmode, sizeof (uint64_t), tx);
}
mutex_exit(&zp->z_acl_lock);
* the file size to the specified eof. Note, there's no
* concurrency during replay.
*/
- if (zsb->z_replay && zsb->z_replay_eof != 0)
- zp->z_size = zsb->z_replay_eof;
+ if (zfsvfs->z_replay && zfsvfs->z_replay_eof != 0)
+ zp->z_size = zfsvfs->z_replay_eof;
error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
ASSERT(tx_bytes == nbytes);
n -= nbytes;
- if (!xuio && n > 0)
- uio_prefaultpages(MIN(n, max_blksz), uio);
+ if (!xuio && n > 0) {
+ if (uio_prefaultpages(MIN(n, max_blksz), uio)) {
+ error = EFAULT;
+ break;
+ }
+ }
}
zfs_inode_update(zp);
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
/*
* If we're in replay mode, or we made no progress, return error.
* Otherwise, it's at least a partial write, so it's successful.
*/
- if (zsb->z_replay || uio->uio_resid == start_resid) {
- ZFS_EXIT(zsb);
+ if (zfsvfs->z_replay || uio->uio_resid == start_resid) {
+ ZFS_EXIT(zfsvfs);
return (error);
}
if (ioflag & (FSYNC | FDSYNC) ||
- zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, zp->z_id);
- ZFS_EXIT(zsb);
+ int64_t nwritten = start_resid - uio->uio_resid;
+ dataset_kstats_update_write_kstats(&zfsvfs->z_kstat, nwritten);
+ task_io_account_write(nwritten);
+
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_write);
+/*
+ * Drop a reference on the passed inode asynchronously. This ensures
+ * that the caller will never drop the last reference on an inode in
+ * the current context. Doing so while holding open a tx could result
+ * in a deadlock if iput_final() re-enters the filesystem code.
+ */
void
zfs_iput_async(struct inode *ip)
{
ASSERT(os != NULL);
if (atomic_read(&ip->i_count) == 1)
- taskq_dispatch(dsl_pool_iput_taskq(dmu_objset_pool(os)),
- (task_func_t *)iput, ip, TQ_SLEEP);
+ VERIFY(taskq_dispatch(dsl_pool_iput_taskq(dmu_objset_pool(os)),
+ (task_func_t *)iput, ip, TQ_SLEEP) != TASKQID_INVALID);
else
iput(ip);
}
+/* ARGSUSED */
void
zfs_get_done(zgd_t *zgd, int error)
{
if (zgd->zgd_db)
dmu_buf_rele(zgd->zgd_db, zgd);
- zfs_range_unlock(zgd->zgd_rl);
+ rangelock_exit(zgd->zgd_lr);
/*
* Release the vnode asynchronously as we currently have the
*/
zfs_iput_async(ZTOI(zp));
- if (error == 0 && zgd->zgd_bp)
- zil_add_block(zgd->zgd_zilog, zgd->zgd_bp);
-
kmem_free(zgd, sizeof (zgd_t));
}
* Get data to generate a TX_WRITE intent log record.
*/
int
-zfs_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio)
+zfs_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio)
{
- zfs_sb_t *zsb = arg;
- objset_t *os = zsb->z_os;
+ zfsvfs_t *zfsvfs = arg;
+ objset_t *os = zfsvfs->z_os;
znode_t *zp;
uint64_t object = lr->lr_foid;
uint64_t offset = lr->lr_offset;
uint64_t size = lr->lr_length;
- blkptr_t *bp = &lr->lr_blkptr;
dmu_buf_t *db;
zgd_t *zgd;
int error = 0;
- ASSERT(zio != NULL);
- ASSERT(size != 0);
+ ASSERT3P(lwb, !=, NULL);
+ ASSERT3P(zio, !=, NULL);
+ ASSERT3U(size, !=, 0);
/*
* Nothing to do if the file has been removed
*/
- if (zfs_zget(zsb, object, &zp) != 0)
+ if (zfs_zget(zfsvfs, object, &zp) != 0)
return (SET_ERROR(ENOENT));
if (zp->z_unlinked) {
/*
}
zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP);
- zgd->zgd_zilog = zsb->z_log;
+ zgd->zgd_lwb = lwb;
zgd->zgd_private = zp;
/*
* we don't have to write the data twice.
*/
if (buf != NULL) { /* immediate write */
- zgd->zgd_rl = zfs_range_lock(&zp->z_range_lock, offset, size,
- RL_READER);
+ zgd->zgd_lr = rangelock_enter(&zp->z_rangelock,
+ offset, size, RL_READER);
/* test for truncation needs to be done while range locked */
if (offset >= zp->z_size) {
error = SET_ERROR(ENOENT);
} else { /* indirect write */
/*
* Have to lock the whole block to ensure when it's
- * written out and it's checksum is being calculated
+ * written out and its checksum is being calculated
* that no one can change the data. We need to re-check
* blocksize after we get the lock in case it's changed!
*/
size = zp->z_blksz;
blkoff = ISP2(size) ? P2PHASE(offset, size) : offset;
offset -= blkoff;
- zgd->zgd_rl = zfs_range_lock(&zp->z_range_lock, offset,
- size, RL_READER);
+ zgd->zgd_lr = rangelock_enter(&zp->z_rangelock,
+ offset, size, RL_READER);
if (zp->z_blksz == size)
break;
offset += blkoff;
- zfs_range_unlock(zgd->zgd_rl);
+ rangelock_exit(zgd->zgd_lr);
}
/* test for truncation needs to be done while range locked */
if (lr->lr_offset >= zp->z_size)
DMU_READ_NO_PREFETCH);
if (error == 0) {
- blkptr_t *obp = dmu_buf_get_blkptr(db);
- if (obp) {
- ASSERT(BP_IS_HOLE(bp));
- *bp = *obp;
- }
+ blkptr_t *bp = &lr->lr_blkptr;
zgd->zgd_db = db;
zgd->zgd_bp = bp;
error = dmu_sync(zio, lr->lr_common.lrc_txg,
zfs_get_done, zgd);
- ASSERT(error || lr->lr_length <= zp->z_blksz);
+ ASSERT(error || lr->lr_length <= size);
/*
* On success, we need to wait for the write I/O
if (error == EALREADY) {
lr->lr_common.lrc_txtype = TX_WRITE2;
+ /*
+ * TX_WRITE2 relies on the data previously
+ * written by the TX_WRITE that caused
+ * EALREADY. We zero out the BP because
+ * it is the old, currently-on-disk BP.
+ */
+ zgd->zgd_bp = NULL;
+ BP_ZERO(bp);
error = 0;
}
}
zfs_access(struct inode *ip, int mode, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int error;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if (flag & V_ACE_MASK)
else
error = zfs_zaccess_rwx(zp, mode, flag, cr);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_access);
/*
* Lookup an entry in a directory, or an extended attribute directory.
cred_t *cr, int *direntflags, pathname_t *realpnp)
{
znode_t *zdp = ITOZ(dip);
- zfs_sb_t *zsb = ITOZSB(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
int error = 0;
- /* fast path */
+ /*
+ * Fast path lookup, however we must skip DNLC lookup
+ * for case folding or normalizing lookups because the
+ * DNLC code only stores the passed in name. This means
+ * creating 'a' and removing 'A' on a case insensitive
+ * file system would work, but DNLC still thinks 'a'
+ * exists and won't let you create it again on the next
+ * pass through fast path.
+ */
if (!(flags & (LOOKUP_XATTR | FIGNORECASE))) {
if (!S_ISDIR(dip->i_mode)) {
return (0);
}
return (error);
-#ifdef HAVE_DNLC
- } else {
- vnode_t *tvp = dnlc_lookup(dvp, nm);
-
- if (tvp) {
- error = zfs_fastaccesschk_execute(zdp, cr);
- if (error) {
- iput(tvp);
- return (error);
- }
- if (tvp == DNLC_NO_VNODE) {
- iput(tvp);
- return (SET_ERROR(ENOENT));
- } else {
- *vpp = tvp;
- return (specvp_check(vpp, cr));
- }
- }
-#endif /* HAVE_DNLC */
}
}
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zdp);
*ipp = NULL;
* Maybe someday we will.
*/
if (zdp->z_pflags & ZFS_XATTR) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
if ((error = zfs_get_xattrdir(zdp, ipp, cr, flags))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
*ipp = NULL;
}
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
if (!S_ISDIR(dip->i_mode)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOTDIR));
}
*/
if ((error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
- if (zsb->z_utf8 && u8_validate(nm, strlen(nm),
+ if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if ((error == 0) && (*ipp))
zfs_inode_update(ITOZ(*ipp));
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_lookup);
/*
* Attempt to create a new entry in a directory. If the entry
int mode, struct inode **ipp, cred_t *cr, int flag, vsecattr_t *vsecp)
{
znode_t *zp, *dzp = ITOZ(dip);
- zfs_sb_t *zsb = ITOZSB(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
zilog_t *zilog;
objset_t *os;
zfs_dirlock_t *dl;
gid = crgetgid(cr);
uid = crgetuid(cr);
- if (zsb->z_use_fuids == B_FALSE &&
+ if (zfsvfs->z_use_fuids == B_FALSE &&
(vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (SET_ERROR(EINVAL));
- ZFS_ENTER(zsb);
+ if (name == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
- os = zsb->z_os;
- zilog = zsb->z_log;
+ os = zfsvfs->z_os;
+ zilog = zfsvfs->z_log;
- if (zsb->z_utf8 && u8_validate(name, strlen(name),
+ if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
crgetuid(cr), cr, vap->va_mode)) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
}
zfs_acl_ids_free(&acl_ids);
if (strcmp(name, "..") == 0)
error = SET_ERROR(EISDIR);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
}
if (zp == NULL) {
uint64_t txtype;
+ uint64_t projid = ZFS_DEFAULT_PROJID;
/*
* Create a new file object and update the directory
goto out;
have_acl = B_TRUE;
- if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
+ if (S_ISREG(vap->va_mode) || S_ISDIR(vap->va_mode))
+ projid = zfs_inherit_projid(dzp);
+ if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
zfs_acl_ids_free(&acl_ids);
error = SET_ERROR(EDQUOT);
goto out;
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
- fuid_dirtied = zsb->z_fuid_dirty;
+ fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
- zfs_fuid_txhold(zsb, tx);
+ zfs_fuid_txhold(zfsvfs, tx);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
- if (!zsb->z_use_sa &&
+ if (!zfsvfs->z_use_sa &&
acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, acl_ids.z_aclp->z_acl_bytes);
}
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+
+ error = dmu_tx_assign(tx,
+ (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
+ error = zfs_link_create(dl, zp, tx, ZNEW);
+ if (error != 0) {
+ /*
+ * Since, we failed to add the directory entry for it,
+ * delete the newly created dnode.
+ */
+ zfs_znode_delete(zp, tx);
+ remove_inode_hash(ZTOI(zp));
+ zfs_acl_ids_free(&acl_ids);
+ dmu_tx_commit(tx);
+ goto out;
+ }
+
if (fuid_dirtied)
- zfs_fuid_sync(zsb, tx);
+ zfs_fuid_sync(zfsvfs, tx);
- (void) zfs_link_create(dl, zp, tx, ZNEW);
txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap);
if (flag & FIGNORECASE)
txtype |= TX_CI;
if (S_ISREG(ZTOI(zp)->i_mode) &&
(vap->va_mask & ATTR_SIZE) && (vap->va_size == 0)) {
/* we can't hold any locks when calling zfs_freesp() */
- zfs_dirent_unlock(dl);
- dl = NULL;
+ if (dl) {
+ zfs_dirent_unlock(dl);
+ dl = NULL;
+ }
error = zfs_freesp(zp, 0, 0, mode, TRUE);
}
}
*ipp = ZTOI(zp);
}
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
+ return (error);
+}
+
+/* ARGSUSED */
+int
+zfs_tmpfile(struct inode *dip, vattr_t *vap, int excl,
+ int mode, struct inode **ipp, cred_t *cr, int flag, vsecattr_t *vsecp)
+{
+ znode_t *zp = NULL, *dzp = ITOZ(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
+ objset_t *os;
+ dmu_tx_t *tx;
+ int error;
+ uid_t uid;
+ gid_t gid;
+ zfs_acl_ids_t acl_ids;
+ uint64_t projid = ZFS_DEFAULT_PROJID;
+ boolean_t fuid_dirtied;
+ boolean_t have_acl = B_FALSE;
+ boolean_t waited = B_FALSE;
+
+ /*
+ * If we have an ephemeral id, ACL, or XVATTR then
+ * make sure file system is at proper version
+ */
+
+ gid = crgetgid(cr);
+ uid = crgetuid(cr);
+
+ if (zfsvfs->z_use_fuids == B_FALSE &&
+ (vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
+ ZFS_VERIFY_ZP(dzp);
+ os = zfsvfs->z_os;
+
+ if (vap->va_mask & ATTR_XVATTR) {
+ if ((error = secpolicy_xvattr((xvattr_t *)vap,
+ crgetuid(cr), cr, vap->va_mode)) != 0) {
+ ZFS_EXIT(zfsvfs);
+ return (error);
+ }
+ }
+
+top:
+ *ipp = NULL;
+
+ /*
+ * Create a new file object and update the directory
+ * to reference it.
+ */
+ if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
+ if (have_acl)
+ zfs_acl_ids_free(&acl_ids);
+ goto out;
+ }
+
+ if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
+ cr, vsecp, &acl_ids)) != 0)
+ goto out;
+ have_acl = B_TRUE;
+
+ if (S_ISREG(vap->va_mode) || S_ISDIR(vap->va_mode))
+ projid = zfs_inherit_projid(dzp);
+ if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
+ zfs_acl_ids_free(&acl_ids);
+ error = SET_ERROR(EDQUOT);
+ goto out;
+ }
+
+ tx = dmu_tx_create(os);
+
+ dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
+ ZFS_SA_BASE_ATTR_SIZE);
+ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
+
+ fuid_dirtied = zfsvfs->z_fuid_dirty;
+ if (fuid_dirtied)
+ zfs_fuid_txhold(zfsvfs, tx);
+ if (!zfsvfs->z_use_sa &&
+ acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
+ dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
+ 0, acl_ids.z_aclp->z_acl_bytes);
+ }
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
+ if (error) {
+ if (error == ERESTART) {
+ waited = B_TRUE;
+ dmu_tx_wait(tx);
+ dmu_tx_abort(tx);
+ goto top;
+ }
+ zfs_acl_ids_free(&acl_ids);
+ dmu_tx_abort(tx);
+ ZFS_EXIT(zfsvfs);
+ return (error);
+ }
+ zfs_mknode(dzp, vap, tx, cr, IS_TMPFILE, &zp, &acl_ids);
+
+ if (fuid_dirtied)
+ zfs_fuid_sync(zfsvfs, tx);
+
+ /* Add to unlinked set */
+ zp->z_unlinked = 1;
+ zfs_unlinked_add(zp, tx);
+ zfs_acl_ids_free(&acl_ids);
+ dmu_tx_commit(tx);
+out:
+
+ if (error) {
+ if (zp)
+ iput(ZTOI(zp));
+ } else {
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ *ipp = ZTOI(zp);
+ }
+
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_create);
/*
* Remove an entry from a directory.
znode_t *zp, *dzp = ITOZ(dip);
znode_t *xzp;
struct inode *ip;
- zfs_sb_t *zsb = ITOZSB(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
zilog_t *zilog;
uint64_t acl_obj, xattr_obj;
uint64_t xattr_obj_unlinked = 0;
uint64_t obj = 0;
+ uint64_t links;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
boolean_t may_delete_now, delete_now = FALSE;
int zflg = ZEXISTS;
boolean_t waited = B_FALSE;
- ZFS_ENTER(zsb);
+ if (name == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
- zilog = zsb->z_log;
+ zilog = zfsvfs->z_log;
if (flags & FIGNORECASE) {
zflg |= ZCILOOK;
NULL, realnmp))) {
if (realnmp)
pn_free(realnmp);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
goto out;
}
-#ifdef HAVE_DNLC
- if (realnmp)
- dnlc_remove(dvp, realnmp->pn_buf);
- else
- dnlc_remove(dvp, name);
-#endif /* HAVE_DNLC */
-
mutex_enter(&zp->z_lock);
may_delete_now = atomic_read(&ip->i_count) == 1 && !(zp->z_is_mapped);
mutex_exit(&zp->z_lock);
* allow for either case.
*/
obj = zp->z_id;
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
}
/* are there any extended attributes? */
- error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
+ error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj, sizeof (xattr_obj));
if (error == 0 && xattr_obj) {
- error = zfs_zget(zsb, xattr_obj, &xzp);
+ error = zfs_zget(zfsvfs, xattr_obj, &xzp);
ASSERT0(error);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
mutex_exit(&zp->z_lock);
/* charge as an update -- would be nice not to charge at all */
- dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, FALSE, NULL);
+ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
/*
* Mark this transaction as typically resulting in a net free of space
*/
dmu_tx_mark_netfree(tx);
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
- iput(ip);
- if (xzp)
- iput(ZTOI(xzp));
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
+ iput(ip);
+ if (xzp)
+ iput(ZTOI(xzp));
goto top;
}
if (realnmp)
pn_free(realnmp);
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ iput(ip);
+ if (xzp)
+ iput(ZTOI(xzp));
+ ZFS_EXIT(zfsvfs);
return (error);
}
* zfs_sa_upgrade().
*/
mutex_enter(&zp->z_lock);
- (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
+ (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj_unlinked, sizeof (xattr_obj_unlinked));
delete_now = may_delete_now && !toobig &&
atomic_read(&ip->i_count) == 1 && !(zp->z_is_mapped) &&
if (delete_now) {
if (xattr_obj_unlinked) {
- ASSERT3U(xzp->z_links, ==, 2);
+ ASSERT3U(ZTOI(xzp)->i_nlink, ==, 2);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = 1;
- xzp->z_links = 0;
- error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zsb),
- &xzp->z_links, sizeof (xzp->z_links), tx);
+ clear_nlink(ZTOI(xzp));
+ links = 0;
+ error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
+ &links, sizeof (links), tx);
ASSERT3U(error, ==, 0);
mutex_exit(&xzp->z_lock);
zfs_unlinked_add(xzp, tx);
if (zp->z_is_sa)
error = sa_remove(zp->z_sa_hdl,
- SA_ZPL_XATTR(zsb), tx);
+ SA_ZPL_XATTR(zfsvfs), tx);
else
error = sa_update(zp->z_sa_hdl,
- SA_ZPL_XATTR(zsb), &null_xattr,
+ SA_ZPL_XATTR(zfsvfs), &null_xattr,
sizeof (uint64_t), tx);
ASSERT0(error);
}
*/
zfs_unlinked_add(zp, tx);
mutex_exit(&zp->z_lock);
- zfs_inode_update(zp);
- iput(ip);
} else if (unlinked) {
mutex_exit(&zp->z_lock);
zfs_unlinked_add(zp, tx);
zfs_dirent_unlock(dl);
zfs_inode_update(dzp);
+ zfs_inode_update(zp);
- if (!delete_now) {
- zfs_inode_update(zp);
+ if (delete_now)
+ iput(ip);
+ else
zfs_iput_async(ip);
- }
if (xzp) {
zfs_inode_update(xzp);
zfs_iput_async(ZTOI(xzp));
}
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_remove);
/*
* Create a new directory and insert it into dip using the name
cred_t *cr, int flags, vsecattr_t *vsecp)
{
znode_t *zp, *dzp = ITOZ(dip);
- zfs_sb_t *zsb = ITOZSB(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
zilog_t *zilog;
zfs_dirlock_t *dl;
uint64_t txtype;
*/
uid = crgetuid(cr);
- if (zsb->z_use_fuids == B_FALSE &&
+ if (zfsvfs->z_use_fuids == B_FALSE &&
(vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (SET_ERROR(EINVAL));
- ZFS_ENTER(zsb);
+ if (dirname == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
- zilog = zsb->z_log;
+ zilog = zfsvfs->z_log;
if (dzp->z_pflags & ZFS_XATTR) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
- if (zsb->z_utf8 && u8_validate(dirname,
+ if (zfsvfs->z_utf8 && u8_validate(dirname,
strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
crgetuid(cr), cr, vap->va_mode)) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
}
if ((error = zfs_acl_ids_create(dzp, 0, vap, cr,
vsecp, &acl_ids)) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
/*
if ((error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf,
NULL, NULL))) {
zfs_acl_ids_free(&acl_ids);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
if ((error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
- if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
+ if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, zfs_inherit_projid(dzp))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EDQUOT));
}
/*
* Add a new entry to the directory.
*/
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
- fuid_dirtied = zsb->z_fuid_dirty;
+ fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
- zfs_fuid_txhold(zsb, tx);
- if (!zsb->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
+ zfs_fuid_txhold(zfsvfs, tx);
+ if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
acl_ids.z_aclp->z_acl_bytes);
}
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
*/
zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
- if (fuid_dirtied)
- zfs_fuid_sync(zsb, tx);
-
/*
* Now put new name in parent dir.
*/
- (void) zfs_link_create(dl, zp, tx, ZNEW);
+ error = zfs_link_create(dl, zp, tx, ZNEW);
+ if (error != 0) {
+ zfs_znode_delete(zp, tx);
+ remove_inode_hash(ZTOI(zp));
+ goto out;
+ }
+
+ if (fuid_dirtied)
+ zfs_fuid_sync(zfsvfs, tx);
*ipp = ZTOI(zp);
zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, vsecp,
acl_ids.z_fuidp, vap);
+out:
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
zfs_dirent_unlock(dl);
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
- zfs_inode_update(dzp);
- zfs_inode_update(zp);
- ZFS_EXIT(zsb);
- return (0);
+ if (error != 0) {
+ iput(ZTOI(zp));
+ } else {
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ }
+ ZFS_EXIT(zfsvfs);
+ return (error);
}
-EXPORT_SYMBOL(zfs_mkdir);
/*
* Remove a directory subdir entry. If the current working
znode_t *dzp = ITOZ(dip);
znode_t *zp;
struct inode *ip;
- zfs_sb_t *zsb = ITOZSB(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
int zflg = ZEXISTS;
boolean_t waited = B_FALSE;
- ZFS_ENTER(zsb);
+ if (name == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
- zilog = zsb->z_log;
+ zilog = zfsvfs->z_log;
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
*/
if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
NULL, NULL))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
}
/*
- * Grab a lock on the directory to make sure that noone is
+ * Grab a lock on the directory to make sure that no one is
* trying to add (or lookup) entries while we are removing it.
*/
rw_enter(&zp->z_name_lock, RW_WRITER);
*/
rw_enter(&zp->z_parent_lock, RW_WRITER);
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
- dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, FALSE, NULL);
+ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
zfs_sa_upgrade_txholds(tx, zp);
zfs_sa_upgrade_txholds(tx, dzp);
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ dmu_tx_mark_netfree(tx);
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
rw_exit(&zp->z_parent_lock);
rw_exit(&zp->z_name_lock);
zfs_dirent_unlock(dl);
- iput(ip);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
+ iput(ip);
goto top;
}
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ iput(ip);
+ ZFS_EXIT(zfsvfs);
return (error);
}
zfs_inode_update(zp);
iput(ip);
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_rmdir);
/*
* Read as many directory entries as will fit into the provided
*/
/* ARGSUSED */
int
-zfs_readdir(struct inode *ip, struct dir_context *ctx, cred_t *cr)
+zfs_readdir(struct inode *ip, zpl_dir_context_t *ctx, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
objset_t *os;
zap_cursor_t zc;
zap_attribute_t zap;
uint64_t parent;
uint64_t offset; /* must be unsigned; checks for < 1 */
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
- if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zsb),
+ if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (parent))) != 0)
goto out;
goto out;
error = 0;
- os = zsb->z_os;
+ os = zfsvfs->z_os;
offset = ctx->pos;
prefetch = zp->z_zn_prefetch;
type = ZFS_DIRENT_TYPE(zap.za_first_integer);
}
- done = !dir_emit(ctx, zap.za_name, strlen(zap.za_name),
+ done = !zpl_dir_emit(ctx, zap.za_name, strlen(zap.za_name),
objnum, type);
if (done)
break;
if (error == ENOENT)
error = 0;
out:
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_readdir);
ulong_t zfs_fsync_sync_cnt = 4;
zfs_fsync(struct inode *ip, int syncflag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
(void) tsd_set(zfs_fsyncer_key, (void *)zfs_fsync_sync_cnt);
- if (zsb->z_os->os_sync != ZFS_SYNC_DISABLED) {
- ZFS_ENTER(zsb);
+ if (zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED) {
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
- zil_commit(zsb->z_log, zp->z_id);
- ZFS_EXIT(zsb);
+ zil_commit(zfsvfs->z_log, zp->z_id);
+ ZFS_EXIT(zfsvfs);
}
tsd_set(zfs_fsyncer_key, NULL);
return (0);
}
-EXPORT_SYMBOL(zfs_fsync);
/*
zfs_getattr(struct inode *ip, vattr_t *vap, int flags, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int error = 0;
uint64_t links;
uint64_t atime[2], mtime[2], ctime[2];
sa_bulk_attr_t bulk[3];
int count = 0;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL, &atime, 16);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL, &mtime, 16);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL, &ctime, 16);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
if ((error = sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
(vap->va_uid != crgetuid(cr))) {
if ((error = zfs_zaccess(zp, ACE_READ_ATTRIBUTES, 0,
skipaclchk, cr))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
}
vap->va_mode = zp->z_mode;
vap->va_fsid = ZTOI(zp)->i_sb->s_dev;
vap->va_nodeid = zp->z_id;
- if ((zp->z_id == zsb->z_root) && zfs_show_ctldir(zp))
- links = zp->z_links + 1;
+ if ((zp->z_id == zfsvfs->z_root) && zfs_show_ctldir(zp))
+ links = ZTOI(zp)->i_nlink + 1;
else
- links = zp->z_links;
+ links = ZTOI(zp)->i_nlink;
vap->va_nlink = MIN(links, ZFS_LINK_MAX);
vap->va_size = i_size_read(ip);
vap->va_rdev = ip->i_rdev;
* Add in any requested optional attributes and the create time.
* Also set the corresponding bits in the returned attribute bitmap.
*/
- if ((xoap = xva_getxoptattr(xvap)) != NULL && zsb->z_use_fuids) {
+ if ((xoap = xva_getxoptattr(xvap)) != NULL && zfsvfs->z_use_fuids) {
if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) {
xoap->xoa_archive =
((zp->z_pflags & ZFS_ARCHIVE) != 0);
if (XVA_ISSET_REQ(xvap, XAT_CREATETIME)) {
uint64_t times[2];
- (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(zsb),
+ (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs),
times, sizeof (times));
ZFS_TIME_DECODE(&xoap->xoa_createtime, times);
XVA_SET_RTN(xvap, XAT_CREATETIME);
((zp->z_pflags & ZFS_SPARSE) != 0);
XVA_SET_RTN(xvap, XAT_SPARSE);
}
+
+ if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
+ xoap->xoa_projinherit =
+ ((zp->z_pflags & ZFS_PROJINHERIT) != 0);
+ XVA_SET_RTN(xvap, XAT_PROJINHERIT);
+ }
+
+ if (XVA_ISSET_REQ(xvap, XAT_PROJID)) {
+ xoap->xoa_projid = zp->z_projid;
+ XVA_SET_RTN(xvap, XAT_PROJID);
+ }
}
ZFS_TIME_DECODE(&vap->va_atime, atime);
/*
* Block size hasn't been set; suggest maximal I/O transfers.
*/
- vap->va_blksize = zsb->z_max_blksz;
+ vap->va_blksize = zfsvfs->z_max_blksz;
}
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_getattr);
/*
* Get the basic file attributes and place them in the provided kstat
zfs_getattr_fast(struct inode *ip, struct kstat *sp)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
uint32_t blksize;
u_longlong_t nblocks;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
mutex_enter(&zp->z_lock);
/*
* Block size hasn't been set; suggest maximal I/O transfers.
*/
- sp->blksize = zsb->z_max_blksz;
+ sp->blksize = zfsvfs->z_max_blksz;
}
mutex_exit(&zp->z_lock);
* Required to prevent NFS client from detecting different inode
* numbers of snapshot root dentry before and after snapshot mount.
*/
- if (zsb->z_issnap) {
+ if (zfsvfs->z_issnap) {
if (ip->i_sb->s_root->d_inode == ip)
sp->ino = ZFSCTL_INO_SNAPDIRS -
- dmu_objset_id(zsb->z_os);
+ dmu_objset_id(zfsvfs->z_os);
}
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_getattr_fast);
+
+/*
+ * For the operation of changing file's user/group/project, we need to
+ * handle not only the main object that is assigned to the file directly,
+ * but also the ones that are used by the file via hidden xattr directory.
+ *
+ * Because the xattr directory may contains many EA entries, as to it may
+ * be impossible to change all of them via the transaction of changing the
+ * main object's user/group/project attributes. Then we have to change them
+ * via other multiple independent transactions one by one. It may be not good
+ * solution, but we have no better idea yet.
+ */
+static int
+zfs_setattr_dir(znode_t *dzp)
+{
+ struct inode *dxip = ZTOI(dzp);
+ struct inode *xip = NULL;
+ zfsvfs_t *zfsvfs = ITOZSB(dxip);
+ objset_t *os = zfsvfs->z_os;
+ zap_cursor_t zc;
+ zap_attribute_t zap;
+ zfs_dirlock_t *dl;
+ znode_t *zp;
+ dmu_tx_t *tx = NULL;
+ uint64_t uid, gid;
+ sa_bulk_attr_t bulk[4];
+ int count = 0;
+ int err;
+
+ zap_cursor_init(&zc, os, dzp->z_id);
+ while ((err = zap_cursor_retrieve(&zc, &zap)) == 0) {
+ if (zap.za_integer_length != 8 || zap.za_num_integers != 1) {
+ err = ENXIO;
+ break;
+ }
+
+ err = zfs_dirent_lock(&dl, dzp, (char *)zap.za_name, &zp,
+ ZEXISTS, NULL, NULL);
+ if (err == ENOENT)
+ goto next;
+ if (err)
+ break;
+
+ xip = ZTOI(zp);
+ if (KUID_TO_SUID(xip->i_uid) == KUID_TO_SUID(dxip->i_uid) &&
+ KGID_TO_SGID(xip->i_gid) == KGID_TO_SGID(dxip->i_gid) &&
+ zp->z_projid == dzp->z_projid)
+ goto next;
+
+ tx = dmu_tx_create(os);
+ if (!(zp->z_pflags & ZFS_PROJID))
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
+ else
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
+
+ err = dmu_tx_assign(tx, TXG_WAIT);
+ if (err)
+ break;
+
+ mutex_enter(&dzp->z_lock);
+
+ if (KUID_TO_SUID(xip->i_uid) != KUID_TO_SUID(dxip->i_uid)) {
+ xip->i_uid = dxip->i_uid;
+ uid = zfs_uid_read(dxip);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
+ &uid, sizeof (uid));
+ }
+
+ if (KGID_TO_SGID(xip->i_gid) != KGID_TO_SGID(dxip->i_gid)) {
+ xip->i_gid = dxip->i_gid;
+ gid = zfs_gid_read(dxip);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL,
+ &gid, sizeof (gid));
+ }
+
+ if (zp->z_projid != dzp->z_projid) {
+ if (!(zp->z_pflags & ZFS_PROJID)) {
+ zp->z_pflags |= ZFS_PROJID;
+ SA_ADD_BULK_ATTR(bulk, count,
+ SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags,
+ sizeof (zp->z_pflags));
+ }
+
+ zp->z_projid = dzp->z_projid;
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PROJID(zfsvfs),
+ NULL, &zp->z_projid, sizeof (zp->z_projid));
+ }
+
+ mutex_exit(&dzp->z_lock);
+
+ if (likely(count > 0)) {
+ err = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
+ dmu_tx_commit(tx);
+ } else {
+ dmu_tx_abort(tx);
+ }
+ tx = NULL;
+ if (err != 0 && err != ENOENT)
+ break;
+
+next:
+ if (xip) {
+ iput(xip);
+ xip = NULL;
+ zfs_dirent_unlock(dl);
+ }
+ zap_cursor_advance(&zc);
+ }
+
+ if (tx)
+ dmu_tx_abort(tx);
+ if (xip) {
+ iput(xip);
+ zfs_dirent_unlock(dl);
+ }
+ zap_cursor_fini(&zc);
+
+ return (err == ENOENT ? 0 : err);
+}
/*
* Set the file attributes to the values contained in the
zfs_setattr(struct inode *ip, vattr_t *vap, int flags, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
+ objset_t *os = zfsvfs->z_os;
zilog_t *zilog;
dmu_tx_t *tx;
vattr_t oldva;
uint_t saved_mask = 0;
int trim_mask = 0;
uint64_t new_mode;
- uint64_t new_uid, new_gid;
+ uint64_t new_kuid = 0, new_kgid = 0, new_uid, new_gid;
uint64_t xattr_obj;
uint64_t mtime[2], ctime[2], atime[2];
+ uint64_t projid = ZFS_INVALID_PROJID;
znode_t *attrzp;
int need_policy = FALSE;
- int err, err2;
+ int err, err2 = 0;
zfs_fuid_info_t *fuidp = NULL;
xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */
xoptattr_t *xoap;
zfs_acl_t *aclp;
boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
boolean_t fuid_dirtied = B_FALSE;
+ boolean_t handle_eadir = B_FALSE;
sa_bulk_attr_t *bulk, *xattr_bulk;
- int count = 0, xattr_count = 0;
+ int count = 0, xattr_count = 0, bulks = 8;
if (mask == 0)
return (0);
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
- zilog = zsb->z_log;
+ /*
+ * If this is a xvattr_t, then get a pointer to the structure of
+ * optional attributes. If this is NULL, then we have a vattr_t.
+ */
+ xoap = xva_getxoptattr(xvap);
+ if (xoap != NULL && (mask & ATTR_XVATTR)) {
+ if (XVA_ISSET_REQ(xvap, XAT_PROJID)) {
+ if (!dmu_objset_projectquota_enabled(os) ||
+ (!S_ISREG(ip->i_mode) && !S_ISDIR(ip->i_mode))) {
+ ZFS_EXIT(zfsvfs);
+ return (SET_ERROR(ENOTSUP));
+ }
+
+ projid = xoap->xoa_projid;
+ if (unlikely(projid == ZFS_INVALID_PROJID)) {
+ ZFS_EXIT(zfsvfs);
+ return (SET_ERROR(EINVAL));
+ }
+
+ if (projid == zp->z_projid && zp->z_pflags & ZFS_PROJID)
+ projid = ZFS_INVALID_PROJID;
+ else
+ need_policy = TRUE;
+ }
+
+ if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT) &&
+ (xoap->xoa_projinherit !=
+ ((zp->z_pflags & ZFS_PROJINHERIT) != 0)) &&
+ (!dmu_objset_projectquota_enabled(os) ||
+ (!S_ISREG(ip->i_mode) && !S_ISDIR(ip->i_mode)))) {
+ ZFS_EXIT(zfsvfs);
+ return (SET_ERROR(ENOTSUP));
+ }
+ }
+
+ zilog = zfsvfs->z_log;
/*
* Make sure that if we have ephemeral uid/gid or xvattr specified
* that file system is at proper version level
*/
- if (zsb->z_use_fuids == B_FALSE &&
+ if (zfsvfs->z_use_fuids == B_FALSE &&
(((mask & ATTR_UID) && IS_EPHEMERAL(vap->va_uid)) ||
((mask & ATTR_GID) && IS_EPHEMERAL(vap->va_gid)) ||
(mask & ATTR_XVATTR))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
if (mask & ATTR_SIZE && S_ISDIR(ip->i_mode)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EISDIR));
}
if (mask & ATTR_SIZE && !S_ISREG(ip->i_mode) && !S_ISFIFO(ip->i_mode)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
- /*
- * If this is an xvattr_t, then get a pointer to the structure of
- * optional attributes. If this is NULL, then we have a vattr_t.
- */
- xoap = xva_getxoptattr(xvap);
-
tmpxvattr = kmem_alloc(sizeof (xvattr_t), KM_SLEEP);
xva_init(tmpxvattr);
- bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * 7, KM_SLEEP);
- xattr_bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * 7, KM_SLEEP);
+ bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * bulks, KM_SLEEP);
+ xattr_bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * bulks, KM_SLEEP);
/*
* Immutable files can only alter immutable bit and atime
if ((zp->z_pflags & ZFS_IMMUTABLE) &&
((mask & (ATTR_SIZE|ATTR_UID|ATTR_GID|ATTR_MTIME|ATTR_MODE)) ||
((mask & ATTR_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) {
- err = EPERM;
+ err = SET_ERROR(EPERM);
goto out3;
}
if ((mask & ATTR_SIZE) && (zp->z_pflags & ZFS_READONLY)) {
- err = EPERM;
+ err = SET_ERROR(EPERM);
goto out3;
}
TIMESPEC_OVERFLOW(&vap->va_atime)) ||
((mask & ATTR_MTIME) &&
TIMESPEC_OVERFLOW(&vap->va_mtime))) {
- err = EOVERFLOW;
+ err = SET_ERROR(EOVERFLOW);
goto out3;
}
}
aclp = NULL;
/* Can this be moved to before the top label? */
- if (zfs_is_readonly(zsb)) {
- err = EROFS;
+ if (zfs_is_readonly(zfsvfs)) {
+ err = SET_ERROR(EROFS);
goto out3;
}
take_owner = (mask & ATTR_UID) && (vap->va_uid == crgetuid(cr));
take_group = (mask & ATTR_GID) &&
- zfs_groupmember(zsb, vap->va_gid, cr);
+ zfs_groupmember(zfsvfs, vap->va_gid, cr);
/*
* If both ATTR_UID and ATTR_GID are set then take_owner and
}
}
+ if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
+ if (xoap->xoa_projinherit !=
+ ((zp->z_pflags & ZFS_PROJINHERIT) != 0)) {
+ need_policy = TRUE;
+ } else {
+ XVA_CLR_REQ(xvap, XAT_PROJINHERIT);
+ XVA_SET_REQ(tmpxvattr, XAT_PROJINHERIT);
+ }
+ }
+
if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
if (xoap->xoa_nounlink !=
((zp->z_pflags & ZFS_NOUNLINK) != 0)) {
if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
mutex_exit(&zp->z_lock);
- err = EPERM;
+ err = SET_ERROR(EPERM);
goto out3;
}
*/
mask = vap->va_mask;
- if ((mask & (ATTR_UID | ATTR_GID))) {
- err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
+ if ((mask & (ATTR_UID | ATTR_GID)) || projid != ZFS_INVALID_PROJID) {
+ handle_eadir = B_TRUE;
+ err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj, sizeof (xattr_obj));
if (err == 0 && xattr_obj) {
goto out2;
}
if (mask & ATTR_UID) {
- new_uid = zfs_fuid_create(zsb,
+ new_kuid = zfs_fuid_create(zfsvfs,
(uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
- if (new_uid != zp->z_uid &&
- zfs_fuid_overquota(zsb, B_FALSE, new_uid)) {
+ if (new_kuid != KUID_TO_SUID(ZTOI(zp)->i_uid) &&
+ zfs_id_overquota(zfsvfs, DMU_USERUSED_OBJECT,
+ new_kuid)) {
if (attrzp)
iput(ZTOI(attrzp));
- err = EDQUOT;
+ err = SET_ERROR(EDQUOT);
goto out2;
}
}
if (mask & ATTR_GID) {
- new_gid = zfs_fuid_create(zsb, (uint64_t)vap->va_gid,
- cr, ZFS_GROUP, &fuidp);
- if (new_gid != zp->z_gid &&
- zfs_fuid_overquota(zsb, B_TRUE, new_gid)) {
+ new_kgid = zfs_fuid_create(zfsvfs,
+ (uint64_t)vap->va_gid, cr, ZFS_GROUP, &fuidp);
+ if (new_kgid != KGID_TO_SGID(ZTOI(zp)->i_gid) &&
+ zfs_id_overquota(zfsvfs, DMU_GROUPUSED_OBJECT,
+ new_kgid)) {
if (attrzp)
iput(ZTOI(attrzp));
- err = EDQUOT;
+ err = SET_ERROR(EDQUOT);
goto out2;
}
}
+
+ if (projid != ZFS_INVALID_PROJID &&
+ zfs_id_overquota(zfsvfs, DMU_PROJECTUSED_OBJECT, projid)) {
+ if (attrzp)
+ iput(ZTOI(attrzp));
+ err = EDQUOT;
+ goto out2;
+ }
}
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(os);
if (mask & ATTR_MODE) {
uint64_t pmode = zp->z_mode;
* Are we upgrading ACL from old V0 format
* to V1 format?
*/
- if (zsb->z_version >= ZPL_VERSION_FUID &&
+ if (zfsvfs->z_version >= ZPL_VERSION_FUID &&
zfs_znode_acl_version(zp) ==
ZFS_ACL_VERSION_INITIAL) {
dmu_tx_hold_free(tx, acl_obj, 0,
mutex_exit(&zp->z_lock);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
} else {
- if ((mask & ATTR_XVATTR) &&
- XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
+ if (((mask & ATTR_XVATTR) &&
+ XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) ||
+ (projid != ZFS_INVALID_PROJID &&
+ !(zp->z_pflags & ZFS_PROJID)))
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
else
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE);
}
- fuid_dirtied = zsb->z_fuid_dirty;
+ fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
- zfs_fuid_txhold(zsb, tx);
+ zfs_fuid_txhold(zfsvfs, tx);
zfs_sa_upgrade_txholds(tx, zp);
* updated as a side-effect of calling this function.
*/
+ if (projid != ZFS_INVALID_PROJID && !(zp->z_pflags & ZFS_PROJID)) {
+ /*
+ * For the existed object that is upgraded from old system,
+ * its on-disk layout has no slot for the project ID attribute.
+ * But quota accounting logic needs to access related slots by
+ * offset directly. So we need to adjust old objects' layout
+ * to make the project ID to some unified and fixed offset.
+ */
+ if (attrzp)
+ err = sa_add_projid(attrzp->z_sa_hdl, tx, projid);
+ if (err == 0)
+ err = sa_add_projid(zp->z_sa_hdl, tx, projid);
+
+ if (unlikely(err == EEXIST))
+ err = 0;
+ else if (err != 0)
+ goto out;
+ else
+ projid = ZFS_INVALID_PROJID;
+ }
if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
mutex_enter(&zp->z_acl_lock);
mutex_enter(&zp->z_lock);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, sizeof (zp->z_pflags));
if (attrzp) {
mutex_enter(&attrzp->z_acl_lock);
mutex_enter(&attrzp->z_lock);
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
- SA_ZPL_FLAGS(zsb), NULL, &attrzp->z_pflags,
+ SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags,
sizeof (attrzp->z_pflags));
+ if (projid != ZFS_INVALID_PROJID) {
+ attrzp->z_projid = projid;
+ SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
+ SA_ZPL_PROJID(zfsvfs), NULL, &attrzp->z_projid,
+ sizeof (attrzp->z_projid));
+ }
}
if (mask & (ATTR_UID|ATTR_GID)) {
if (mask & ATTR_UID) {
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), NULL,
+ ZTOI(zp)->i_uid = SUID_TO_KUID(new_kuid);
+ new_uid = zfs_uid_read(ZTOI(zp));
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
&new_uid, sizeof (new_uid));
- zp->z_uid = new_uid;
if (attrzp) {
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
- SA_ZPL_UID(zsb), NULL, &new_uid,
+ SA_ZPL_UID(zfsvfs), NULL, &new_uid,
sizeof (new_uid));
- attrzp->z_uid = new_uid;
+ ZTOI(attrzp)->i_uid = SUID_TO_KUID(new_uid);
}
}
if (mask & ATTR_GID) {
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb),
+ ZTOI(zp)->i_gid = SGID_TO_KGID(new_kgid);
+ new_gid = zfs_gid_read(ZTOI(zp));
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs),
NULL, &new_gid, sizeof (new_gid));
- zp->z_gid = new_gid;
if (attrzp) {
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
- SA_ZPL_GID(zsb), NULL, &new_gid,
+ SA_ZPL_GID(zfsvfs), NULL, &new_gid,
sizeof (new_gid));
- attrzp->z_gid = new_gid;
+ ZTOI(attrzp)->i_gid = SGID_TO_KGID(new_kgid);
}
}
if (!(mask & ATTR_MODE)) {
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb),
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs),
NULL, &new_mode, sizeof (new_mode));
new_mode = zp->z_mode;
}
}
if (mask & ATTR_MODE) {
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
&new_mode, sizeof (new_mode));
- zp->z_mode = new_mode;
+ zp->z_mode = ZTOI(zp)->i_mode = new_mode;
ASSERT3P(aclp, !=, NULL);
err = zfs_aclset_common(zp, aclp, cr, tx);
ASSERT0(err);
aclp = NULL;
}
-
if ((mask & ATTR_ATIME) || zp->z_atime_dirty) {
zp->z_atime_dirty = 0;
ZFS_TIME_ENCODE(&ip->i_atime, atime);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
&atime, sizeof (atime));
}
- if (mask & ATTR_MTIME) {
+ if (mask & (ATTR_MTIME | ATTR_SIZE)) {
ZFS_TIME_ENCODE(&vap->va_mtime, mtime);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL,
+ ZTOI(zp)->i_mtime = zpl_inode_timespec_trunc(vap->va_mtime,
+ ZTOI(zp)->i_sb->s_time_gran);
+
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
mtime, sizeof (mtime));
}
- /* XXX - shouldn't this be done *before* the ATIME/MTIME checks? */
- if (mask & ATTR_SIZE && !(mask & ATTR_MTIME)) {
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb),
- NULL, mtime, sizeof (mtime));
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL,
- &ctime, sizeof (ctime));
- zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime);
- } else if (mask != 0) {
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL,
- &ctime, sizeof (ctime));
- zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime, ctime);
- if (attrzp) {
- SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
- SA_ZPL_CTIME(zsb), NULL,
- &ctime, sizeof (ctime));
- zfs_tstamp_update_setup(attrzp, STATE_CHANGED,
- mtime, ctime);
- }
+ if (mask & (ATTR_CTIME | ATTR_SIZE)) {
+ ZFS_TIME_ENCODE(&vap->va_ctime, ctime);
+ ZTOI(zp)->i_ctime = zpl_inode_timespec_trunc(vap->va_ctime,
+ ZTOI(zp)->i_sb->s_time_gran);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
+ ctime, sizeof (ctime));
}
+
+ if (projid != ZFS_INVALID_PROJID) {
+ zp->z_projid = projid;
+ SA_ADD_BULK_ATTR(bulk, count,
+ SA_ZPL_PROJID(zfsvfs), NULL, &zp->z_projid,
+ sizeof (zp->z_projid));
+ }
+
+ if (attrzp && mask) {
+ SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
+ SA_ZPL_CTIME(zfsvfs), NULL, &ctime,
+ sizeof (ctime));
+ }
+
/*
* Do this after setting timestamps to prevent timestamp
* update from toggling bit
if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_QUARANTINED)) {
XVA_SET_REQ(xvap, XAT_AV_QUARANTINED);
}
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_PROJINHERIT)) {
+ XVA_SET_REQ(xvap, XAT_PROJINHERIT);
+ }
if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
ASSERT(S_ISREG(ip->i_mode));
}
if (fuid_dirtied)
- zfs_fuid_sync(zsb, tx);
+ zfs_fuid_sync(zfsvfs, tx);
if (mask != 0)
zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp);
mutex_exit(&attrzp->z_lock);
}
out:
- if (err == 0 && attrzp) {
+ if (err == 0 && xattr_count > 0) {
err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk,
xattr_count, tx);
ASSERT(err2 == 0);
}
- if (attrzp)
- iput(ZTOI(attrzp));
if (aclp)
zfs_acl_free(aclp);
if (err) {
dmu_tx_abort(tx);
+ if (attrzp)
+ iput(ZTOI(attrzp));
if (err == ERESTART)
goto top;
} else {
- err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
+ if (count > 0)
+ err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
dmu_tx_commit(tx);
+ if (attrzp) {
+ if (err2 == 0 && handle_eadir)
+ err2 = zfs_setattr_dir(attrzp);
+ iput(ZTOI(attrzp));
+ }
zfs_inode_update(zp);
}
out2:
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
out3:
- kmem_free(xattr_bulk, sizeof (sa_bulk_attr_t) * 7);
- kmem_free(bulk, sizeof (sa_bulk_attr_t) * 7);
+ kmem_free(xattr_bulk, sizeof (sa_bulk_attr_t) * bulks);
+ kmem_free(bulk, sizeof (sa_bulk_attr_t) * bulks);
kmem_free(tmpxvattr, sizeof (xvattr_t));
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (err);
}
-EXPORT_SYMBOL(zfs_setattr);
typedef struct zfs_zlock {
krwlock_t *zl_rwlock; /* lock we acquired */
while ((zl = *zlpp) != NULL) {
if (zl->zl_znode != NULL)
- iput(ZTOI(zl->zl_znode));
+ zfs_iput_async(ZTOI(zl->zl_znode));
rw_exit(zl->zl_rwlock);
*zlpp = zl->zl_next;
kmem_free(zl, sizeof (*zl));
{
znode_t *tdzp, *szp, *tzp;
znode_t *sdzp = ITOZ(sdip);
- zfs_sb_t *zsb = ITOZSB(sdip);
+ zfsvfs_t *zfsvfs = ITOZSB(sdip);
zilog_t *zilog;
zfs_dirlock_t *sdl, *tdl;
dmu_tx_t *tx;
int zflg = 0;
boolean_t waited = B_FALSE;
- ZFS_ENTER(zsb);
+ if (snm == NULL || tnm == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(sdzp);
- zilog = zsb->z_log;
+ zilog = zfsvfs->z_log;
tdzp = ITOZ(tdip);
ZFS_VERIFY_ZP(tdzp);
* super blocks.
*/
if (tdip->i_sb != sdip->i_sb || zfsctl_is_node(tdip)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EXDEV));
}
- if (zsb->z_utf8 && u8_validate(tnm,
+ if (zfsvfs->z_utf8 && u8_validate(tnm,
strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
* See the comment in zfs_link() for why this is considered bad.
*/
if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
* First compare the two name arguments without
* considering any case folding.
*/
- int nofold = (zsb->z_norm & ~U8_TEXTPREP_TOUPPER);
+ int nofold = (zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER);
cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error);
- ASSERT(error == 0 || !zsb->z_utf8);
+ ASSERT(error == 0 || !zfsvfs->z_utf8);
if (cmp == 0) {
/*
* POSIX: "If the old argument and the new argument
* the rename() function shall return successfully
* and perform no other action."
*/
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
/*
* is an exact match, we will allow this to proceed as
* a name-change request.
*/
- if ((zsb->z_case == ZFS_CASE_INSENSITIVE ||
- (zsb->z_case == ZFS_CASE_MIXED &&
+ if ((zfsvfs->z_case == ZFS_CASE_INSENSITIVE ||
+ (zfsvfs->z_case == ZFS_CASE_MIXED &&
flags & FIGNORECASE)) &&
- u8_strcmp(snm, tnm, 0, zsb->z_norm, U8_UNICODE_LATEST,
+ u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST,
&error) == 0) {
/*
* case preserving rename request, require exact
if (strcmp(snm, "..") == 0)
serr = EINVAL;
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (serr);
}
if (terr) {
if (strcmp(tnm, "..") == 0)
terr = EINVAL;
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (terr);
}
+ /*
+ * If we are using project inheritance, means if the directory has
+ * ZFS_PROJINHERIT set, then its descendant directories will inherit
+ * not only the project ID, but also the ZFS_PROJINHERIT flag. Under
+ * such case, we only allow renames into our tree when the project
+ * IDs are the same.
+ */
+ if (tdzp->z_pflags & ZFS_PROJINHERIT &&
+ tdzp->z_projid != szp->z_projid) {
+ error = SET_ERROR(EXDEV);
+ goto out;
+ }
+
/*
* Must have write access at the source to remove the old entry
* and write access at the target to create the new entry.
}
}
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, sdzp->z_id, FALSE, snm);
}
zfs_sa_upgrade_txholds(tx, szp);
- dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, FALSE, NULL);
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
if (zl != NULL)
zfs_rename_unlock(&zl);
if (sdzp == tdzp)
rw_exit(&sdzp->z_name_lock);
- iput(ZTOI(szp));
- if (tzp)
- iput(ZTOI(tzp));
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
+ iput(ZTOI(szp));
+ if (tzp)
+ iput(ZTOI(tzp));
goto top;
}
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ iput(ZTOI(szp));
+ if (tzp)
+ iput(ZTOI(tzp));
+ ZFS_EXIT(zfsvfs);
return (error);
}
error = zfs_link_create(tdl, szp, tx, ZRENAMING);
if (error == 0) {
szp->z_pflags |= ZFS_AV_MODIFIED;
+ if (tdzp->z_pflags & ZFS_PROJINHERIT)
+ szp->z_pflags |= ZFS_PROJINHERIT;
- error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zsb),
+ error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs),
(void *)&szp->z_pflags, sizeof (uint64_t), tx);
ASSERT0(error);
VERIFY3U(zfs_link_destroy(tdl, szp, tx,
ZRENAMING, NULL), ==, 0);
}
+ } else {
+ /*
+ * If we had removed the existing target, subsequent
+ * call to zfs_link_create() to add back the same entry
+ * but, the new dnode (szp) should not fail.
+ */
+ ASSERT(tzp == NULL);
}
}
iput(ZTOI(tzp));
}
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_rename);
/*
* Insert the indicated symbolic reference entry into the directory.
znode_t *zp, *dzp = ITOZ(dip);
zfs_dirlock_t *dl;
dmu_tx_t *tx;
- zfs_sb_t *zsb = ITOZSB(dip);
+ zfsvfs_t *zfsvfs = ITOZSB(dip);
zilog_t *zilog;
uint64_t len = strlen(link);
int error;
ASSERT(S_ISLNK(vap->va_mode));
- ZFS_ENTER(zsb);
+ if (name == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
- zilog = zsb->z_log;
+ zilog = zfsvfs->z_log;
- if (zsb->z_utf8 && u8_validate(name, strlen(name),
+ if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
if (len > MAXPATHLEN) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENAMETOOLONG));
}
if ((error = zfs_acl_ids_create(dzp, 0,
vap, cr, NULL, &acl_ids)) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
top:
error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL);
if (error) {
zfs_acl_ids_free(&acl_ids);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
- if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
+ if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, ZFS_DEFAULT_PROJID)) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EDQUOT));
}
- tx = dmu_tx_create(zsb->z_os);
- fuid_dirtied = zsb->z_fuid_dirty;
+ tx = dmu_tx_create(zfsvfs->z_os);
+ fuid_dirtied = zfsvfs->z_fuid_dirty;
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len));
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE + len);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
- if (!zsb->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
+ if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
acl_ids.z_aclp->z_acl_bytes);
}
if (fuid_dirtied)
- zfs_fuid_txhold(zsb, tx);
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ zfs_fuid_txhold(zfsvfs, tx);
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
if (fuid_dirtied)
- zfs_fuid_sync(zsb, tx);
+ zfs_fuid_sync(zfsvfs, tx);
mutex_enter(&zp->z_lock);
if (zp->z_is_sa)
- error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zsb),
+ error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs),
link, len, tx);
else
zfs_sa_symlink(zp, link, len, tx);
mutex_exit(&zp->z_lock);
zp->z_size = len;
- (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zsb),
+ (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
&zp->z_size, sizeof (zp->z_size), tx);
/*
* Insert the new object into the directory.
*/
- (void) zfs_link_create(dl, zp, tx, ZNEW);
-
- if (flags & FIGNORECASE)
- txtype |= TX_CI;
- zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);
+ error = zfs_link_create(dl, zp, tx, ZNEW);
+ if (error != 0) {
+ zfs_znode_delete(zp, tx);
+ remove_inode_hash(ZTOI(zp));
+ } else {
+ if (flags & FIGNORECASE)
+ txtype |= TX_CI;
+ zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);
- zfs_inode_update(dzp);
- zfs_inode_update(zp);
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ }
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- *ipp = ZTOI(zp);
+ if (error == 0) {
+ *ipp = ZTOI(zp);
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
- zil_commit(zilog, 0);
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+ } else {
+ iput(ZTOI(zp));
+ }
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_symlink);
/*
* Return, in the buffer contained in the provided uio structure,
zfs_readlink(struct inode *ip, uio_t *uio, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int error;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
mutex_enter(&zp->z_lock);
if (zp->z_is_sa)
error = sa_lookup_uio(zp->z_sa_hdl,
- SA_ZPL_SYMLINK(zsb), uio);
+ SA_ZPL_SYMLINK(zfsvfs), uio);
else
error = zfs_sa_readlink(zp, uio);
mutex_exit(&zp->z_lock);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_readlink);
/*
* Insert a new entry into directory tdip referencing sip.
{
znode_t *dzp = ITOZ(tdip);
znode_t *tzp, *szp;
- zfs_sb_t *zsb = ITOZSB(tdip);
+ zfsvfs_t *zfsvfs = ITOZSB(tdip);
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
uint64_t parent;
uid_t owner;
boolean_t waited = B_FALSE;
-
+ boolean_t is_tmpfile = 0;
+ uint64_t txg;
+#ifdef HAVE_TMPFILE
+ is_tmpfile = (sip->i_nlink == 0 && (sip->i_state & I_LINKABLE));
+#endif
ASSERT(S_ISDIR(tdip->i_mode));
- ZFS_ENTER(zsb);
+ if (name == NULL)
+ return (SET_ERROR(EINVAL));
+
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
- zilog = zsb->z_log;
+ zilog = zfsvfs->z_log;
/*
* POSIX dictates that we return EPERM here.
* Better choices include ENOTSUP or EISDIR.
*/
if (S_ISDIR(sip->i_mode)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
szp = ITOZ(sip);
ZFS_VERIFY_ZP(szp);
+ /*
+ * If we are using project inheritance, means if the directory has
+ * ZFS_PROJINHERIT set, then its descendant directories will inherit
+ * not only the project ID, but also the ZFS_PROJINHERIT flag. Under
+ * such case, we only allow hard link creation in our tree when the
+ * project IDs are the same.
+ */
+ if (dzp->z_pflags & ZFS_PROJINHERIT && dzp->z_projid != szp->z_projid) {
+ ZFS_EXIT(zfsvfs);
+ return (SET_ERROR(EXDEV));
+ }
+
/*
* We check i_sb because snapshots and the ctldir must have different
* super blocks.
*/
if (sip->i_sb != tdip->i_sb || zfsctl_is_node(sip)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EXDEV));
}
/* Prevent links to .zfs/shares files */
- if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zsb),
+ if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (uint64_t))) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
- if (parent == zsb->z_shares_dir) {
- ZFS_EXIT(zsb);
+ if (parent == zfsvfs->z_shares_dir) {
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
- if (zsb->z_utf8 && u8_validate(name,
+ if (zfsvfs->z_utf8 && u8_validate(name,
strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
* imposed in attribute space.
*/
if ((szp->z_pflags & ZFS_XATTR) != (dzp->z_pflags & ZFS_XATTR)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
- owner = zfs_fuid_map_id(zsb, szp->z_uid, cr, ZFS_OWNER);
+ owner = zfs_fuid_map_id(zfsvfs, KUID_TO_SUID(sip->i_uid),
+ cr, ZFS_OWNER);
if (owner != crgetuid(cr) && secpolicy_basic_link(cr) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
*/
error = zfs_dirent_lock(&dl, dzp, name, &tzp, zf, NULL, NULL);
if (error) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
+ if (is_tmpfile)
+ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
+
zfs_sa_upgrade_txholds(tx, szp);
zfs_sa_upgrade_txholds(tx, dzp);
- error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
+ error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
goto top;
}
dmu_tx_abort(tx);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-
+ /* unmark z_unlinked so zfs_link_create will not reject */
+ if (is_tmpfile)
+ szp->z_unlinked = 0;
error = zfs_link_create(dl, szp, tx, 0);
if (error == 0) {
uint64_t txtype = TX_LINK;
- if (flags & FIGNORECASE)
- txtype |= TX_CI;
- zfs_log_link(zilog, tx, txtype, dzp, szp, name);
+ /*
+ * tmpfile is created to be in z_unlinkedobj, so remove it.
+ * Also, we don't log in ZIL, be cause all previous file
+ * operation on the tmpfile are ignored by ZIL. Instead we
+ * always wait for txg to sync to make sure all previous
+ * operation are sync safe.
+ */
+ if (is_tmpfile) {
+ VERIFY(zap_remove_int(zfsvfs->z_os,
+ zfsvfs->z_unlinkedobj, szp->z_id, tx) == 0);
+ } else {
+ if (flags & FIGNORECASE)
+ txtype |= TX_CI;
+ zfs_log_link(zilog, tx, txtype, dzp, szp, name);
+ }
+ } else if (is_tmpfile) {
+ /* restore z_unlinked since when linking failed */
+ szp->z_unlinked = 1;
}
-
+ txg = dmu_tx_get_txg(tx);
dmu_tx_commit(tx);
zfs_dirent_unlock(dl);
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (!is_tmpfile && zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
+ if (is_tmpfile)
+ txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), txg);
+
zfs_inode_update(dzp);
zfs_inode_update(szp);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_link);
static void
zfs_putpage_commit_cb(void *arg)
zfs_putpage(struct inode *ip, struct page *pp, struct writeback_control *wbc)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
loff_t offset;
loff_t pgoff;
unsigned int pglen;
- rl_t *rl;
dmu_tx_t *tx;
caddr_t va;
int err = 0;
int cnt = 0;
struct address_space *mapping;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
ASSERT(PageLocked(pp));
/* Page is beyond end of file */
if (pgoff >= offset) {
unlock_page(pp);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
* is to register a page_mkwrite() handler to count the page
* against its quota when it is about to be dirtied.
*/
- if (zfs_owner_overquota(zsb, zp, B_FALSE) ||
- zfs_owner_overquota(zsb, zp, B_TRUE)) {
+ if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT,
+ KUID_TO_SUID(ip->i_uid)) ||
+ zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT,
+ KGID_TO_SGID(ip->i_gid)) ||
+ (zp->z_projid != ZFS_DEFAULT_PROJID &&
+ zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT,
+ zp->z_projid))) {
err = EDQUOT;
}
#endif
redirty_page_for_writepage(wbc, pp);
unlock_page(pp);
- rl = zfs_range_lock(&zp->z_range_lock, pgoff, pglen, RL_WRITER);
+ locked_range_t *lr = rangelock_enter(&zp->z_rangelock,
+ pgoff, pglen, RL_WRITER);
lock_page(pp);
/* Page mapping changed or it was no longer dirty, we're done */
if (unlikely((mapping != pp->mapping) || !PageDirty(pp))) {
unlock_page(pp);
- zfs_range_unlock(rl);
- ZFS_EXIT(zsb);
+ rangelock_exit(lr);
+ ZFS_EXIT(zfsvfs);
return (0);
}
/* Another process started write block if required */
if (PageWriteback(pp)) {
unlock_page(pp);
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
if (wbc->sync_mode != WB_SYNC_NONE)
wait_on_page_writeback(pp);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
/* Clear the dirty flag the required locks are held */
if (!clear_page_dirty_for_io(pp)) {
unlock_page(pp);
- zfs_range_unlock(rl);
- ZFS_EXIT(zsb);
+ rangelock_exit(lr);
+ ZFS_EXIT(zfsvfs);
return (0);
}
set_page_writeback(pp);
unlock_page(pp);
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_write(tx, zp->z_id, pgoff, pglen);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
__set_page_dirty_nobuffers(pp);
ClearPageError(pp);
end_page_writeback(pp);
- zfs_range_unlock(rl);
- ZFS_EXIT(zsb);
+ rangelock_exit(lr);
+ ZFS_EXIT(zfsvfs);
return (err);
}
va = kmap(pp);
ASSERT3U(pglen, <=, PAGE_SIZE);
- dmu_write(zsb->z_os, zp->z_id, pgoff, pglen, va, tx);
+ dmu_write(zfsvfs->z_os, zp->z_id, pgoff, pglen, va, tx);
kunmap(pp);
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zsb), NULL, &mtime, 16);
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zsb), NULL, &ctime, 16);
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_FLAGS(zfsvfs), NULL,
+ &zp->z_pflags, 8);
/* Preserve the mtime and ctime provided by the inode */
ZFS_TIME_ENCODE(&ip->i_mtime, mtime);
err = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx);
- zfs_log_write(zsb->z_log, tx, TX_WRITE, zp, pgoff, pglen, 0,
+ zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, pgoff, pglen, 0,
zfs_putpage_commit_cb, pp);
dmu_tx_commit(tx);
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
if (wbc->sync_mode != WB_SYNC_NONE) {
/*
* writepages() normally handles the entire commit for
* performance reasons.
*/
- if (zsb->z_log != NULL)
- zil_commit(zsb->z_log, zp->z_id);
+ zil_commit(zfsvfs->z_log, zp->z_id);
}
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (err);
}
zfs_dirty_inode(struct inode *ip, int flags)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
dmu_tx_t *tx;
uint64_t mode, atime[2], mtime[2], ctime[2];
sa_bulk_attr_t bulk[4];
int error = 0;
int cnt = 0;
- if (zfs_is_readonly(zsb) || dmu_objset_is_snapshot(zsb->z_os))
+ if (zfs_is_readonly(zfsvfs) || dmu_objset_is_snapshot(zfsvfs->z_os))
return (0);
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
#ifdef I_DIRTY_TIME
}
#endif
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
mutex_enter(&zp->z_lock);
zp->z_atime_dirty = 0;
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MODE(zsb), NULL, &mode, 8);
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_ATIME(zsb), NULL, &atime, 16);
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zsb), NULL, &mtime, 16);
- SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zsb), NULL, &ctime, 16);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
+ SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
/* Preserve the mode, mtime and ctime provided by the inode */
ZFS_TIME_ENCODE(&ip->i_atime, atime);
dmu_tx_commit(tx);
out:
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_dirty_inode);
/*ARGSUSED*/
void
zfs_inactive(struct inode *ip)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
uint64_t atime[2];
int error;
int need_unlock = 0;
/* Only read lock if we haven't already write locked, e.g. rollback */
- if (!RW_WRITE_HELD(&zsb->z_teardown_inactive_lock)) {
+ if (!RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)) {
need_unlock = 1;
- rw_enter(&zsb->z_teardown_inactive_lock, RW_READER);
+ rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_READER);
}
if (zp->z_sa_hdl == NULL) {
if (need_unlock)
- rw_exit(&zsb->z_teardown_inactive_lock);
+ rw_exit(&zfsvfs->z_teardown_inactive_lock);
return;
}
if (zp->z_atime_dirty && zp->z_unlinked == 0) {
- dmu_tx_t *tx = dmu_tx_create(zsb->z_os);
+ dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
} else {
ZFS_TIME_ENCODE(&ip->i_atime, atime);
mutex_enter(&zp->z_lock);
- (void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zsb),
+ (void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs),
(void *)&atime, sizeof (atime), tx);
zp->z_atime_dirty = 0;
mutex_exit(&zp->z_lock);
zfs_zinactive(zp);
if (need_unlock)
- rw_exit(&zsb->z_teardown_inactive_lock);
+ rw_exit(&zfsvfs->z_teardown_inactive_lock);
}
-EXPORT_SYMBOL(zfs_inactive);
/*
* Bounds-check the seek operation.
return (0);
return ((*noffp < 0 || *noffp > MAXOFFSET_T) ? EINVAL : 0);
}
-EXPORT_SYMBOL(zfs_seek);
/*
* Fill pages with data from the disk.
zfs_fillpage(struct inode *ip, struct page *pl[], int nr_pages)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
objset_t *os;
struct page *cur_pp;
u_offset_t io_off, total;
unsigned page_idx;
int err;
- os = zsb->z_os;
+ os = zfsvfs->z_os;
io_len = nr_pages << PAGE_SHIFT;
i_size = i_size_read(ip);
io_off = page_offset(pl[0]);
* Iterate over list of pages and read each page individually.
*/
page_idx = 0;
- cur_pp = pl[0];
for (total = io_off + io_len; io_off < total; io_off += PAGESIZE) {
caddr_t va;
+ cur_pp = pl[page_idx++];
va = kmap(cur_pp);
err = dmu_read(os, zp->z_id, io_off, PAGESIZE, va,
DMU_READ_PREFETCH);
err = SET_ERROR(EIO);
return (err);
}
- cur_pp = pl[++page_idx];
}
return (0);
zfs_getpage(struct inode *ip, struct page *pl[], int nr_pages)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int err;
if (pl == NULL)
return (0);
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
err = zfs_fillpage(ip, pl, nr_pages);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (err);
}
-EXPORT_SYMBOL(zfs_getpage);
/*
* Check ZFS specific permissions to memory map a section of a file.
unsigned long vm_flags)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if ((vm_flags & VM_WRITE) && (zp->z_pflags &
(ZFS_IMMUTABLE | ZFS_READONLY | ZFS_APPENDONLY))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
if ((vm_flags & (VM_READ | VM_EXEC)) &&
(zp->z_pflags & ZFS_AV_QUARANTINED)) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EACCES));
}
if (off < 0 || len > MAXOFFSET_T - off) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENXIO));
}
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_map);
/*
* convoff - converts the given data (start, whence) to the
int error;
if ((lckdat->l_whence == 2) || (whence == 2)) {
- if ((error = zfs_getattr(ip, &vap, 0, CRED()) != 0))
+ if ((error = zfs_getattr(ip, &vap, 0, CRED())))
return (error);
}
offset_t offset, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
uint64_t off, len;
int error;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if (cmd != F_FREESP) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
* Callers might not be able to detect properly that we are read-only,
* so check it explicitly here.
*/
- if (zfs_is_readonly(zsb)) {
- ZFS_EXIT(zsb);
+ if (zfs_is_readonly(zfsvfs)) {
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EROFS));
}
if ((error = convoff(ip, bfp, 0, offset))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
if (bfp->l_len < 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
* operates directly on inodes, so we need to check access rights.
*/
if ((error = zfs_zaccess(zp, ACE_WRITE_DATA, 0, B_FALSE, cr))) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
error = zfs_freesp(zp, off, len, flag, TRUE);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_space);
/*ARGSUSED*/
int
zfs_fid(struct inode *ip, fid_t *fidp)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
uint32_t gen;
uint64_t gen64;
uint64_t object = zp->z_id;
zfid_short_t *zfid;
int size, i, error;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
- if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zsb),
+ if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs),
&gen64, sizeof (uint64_t))) != 0) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
gen = (uint32_t)gen64;
- size = (zsb->z_parent != zsb) ? LONG_FID_LEN : SHORT_FID_LEN;
- if (fidp->fid_len < size) {
- fidp->fid_len = size;
- ZFS_EXIT(zsb);
- return (SET_ERROR(ENOSPC));
- }
+ size = SHORT_FID_LEN;
zfid = (zfid_short_t *)fidp;
for (i = 0; i < sizeof (zfid->zf_gen); i++)
zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i));
- if (size == LONG_FID_LEN) {
- uint64_t objsetid = dmu_objset_id(zsb->z_os);
- zfid_long_t *zlfid;
-
- zlfid = (zfid_long_t *)fidp;
-
- for (i = 0; i < sizeof (zlfid->zf_setid); i++)
- zlfid->zf_setid[i] = (uint8_t)(objsetid >> (8 * i));
-
- /* XXX - this should be the generation number for the objset */
- for (i = 0; i < sizeof (zlfid->zf_setgen); i++)
- zlfid->zf_setgen[i] = 0;
- }
-
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
-EXPORT_SYMBOL(zfs_fid);
/*ARGSUSED*/
int
zfs_getsecattr(struct inode *ip, vsecattr_t *vsecp, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int error;
boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
error = zfs_getacl(zp, vsecp, skipaclchk, cr);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_getsecattr);
/*ARGSUSED*/
int
zfs_setsecattr(struct inode *ip, vsecattr_t *vsecp, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
int error;
boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
- zilog_t *zilog = zsb->z_log;
+ zilog_t *zilog = zfsvfs->z_log;
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
error = zfs_setacl(zp, vsecp, skipaclchk, cr);
- if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (error);
}
-EXPORT_SYMBOL(zfs_setsecattr);
#ifdef HAVE_UIO_ZEROCOPY
/*
zfs_reqzcbuf(struct inode *ip, enum uio_rw ioflag, xuio_t *xuio, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ITOZSB(ip);
- int max_blksz = zsb->z_max_blksz;
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
+ int max_blksz = zfsvfs->z_max_blksz;
uio_t *uio = &xuio->xu_uio;
ssize_t size = uio->uio_resid;
offset_t offset = uio->uio_loffset;
if (xuio->xu_type != UIOTYPE_ZEROCOPY)
return (SET_ERROR(EINVAL));
- ZFS_ENTER(zsb);
+ ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
switch (ioflag) {
case UIO_WRITE:
*/
blksz = max_blksz;
if (size < blksz || zp->z_blksz != blksz) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
/*
blksz = zcr_blksz_max;
/* avoid potential complexity of dealing with it */
if (blksz > max_blksz) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
size = maxsize;
if (size < blksz) {
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
break;
default:
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
uio->uio_extflg = UIO_XUIO;
XUIO_XUZC_RW(xuio) = ioflag;
- ZFS_EXIT(zsb);
+ ZFS_EXIT(zfsvfs);
return (0);
}
}
#endif /* HAVE_UIO_ZEROCOPY */
-#if defined(_KERNEL) && defined(HAVE_SPL)
+#if defined(_KERNEL)
+EXPORT_SYMBOL(zfs_open);
+EXPORT_SYMBOL(zfs_close);
+EXPORT_SYMBOL(zfs_read);
+EXPORT_SYMBOL(zfs_write);
+EXPORT_SYMBOL(zfs_access);
+EXPORT_SYMBOL(zfs_lookup);
+EXPORT_SYMBOL(zfs_create);
+EXPORT_SYMBOL(zfs_tmpfile);
+EXPORT_SYMBOL(zfs_remove);
+EXPORT_SYMBOL(zfs_mkdir);
+EXPORT_SYMBOL(zfs_rmdir);
+EXPORT_SYMBOL(zfs_readdir);
+EXPORT_SYMBOL(zfs_fsync);
+EXPORT_SYMBOL(zfs_getattr);
+EXPORT_SYMBOL(zfs_getattr_fast);
+EXPORT_SYMBOL(zfs_setattr);
+EXPORT_SYMBOL(zfs_rename);
+EXPORT_SYMBOL(zfs_symlink);
+EXPORT_SYMBOL(zfs_readlink);
+EXPORT_SYMBOL(zfs_link);
+EXPORT_SYMBOL(zfs_inactive);
+EXPORT_SYMBOL(zfs_space);
+EXPORT_SYMBOL(zfs_fid);
+EXPORT_SYMBOL(zfs_getsecattr);
+EXPORT_SYMBOL(zfs_setsecattr);
+EXPORT_SYMBOL(zfs_getpage);
+EXPORT_SYMBOL(zfs_putpage);
+EXPORT_SYMBOL(zfs_dirty_inode);
+EXPORT_SYMBOL(zfs_map);
+
+/* CSTYLED */
module_param(zfs_delete_blocks, ulong, 0644);
MODULE_PARM_DESC(zfs_delete_blocks, "Delete files larger than N blocks async");
module_param(zfs_read_chunk_size, long, 0644);
projects / mirror_zfs.git / blobdiff