* CDDL HEADER END
*/
/*
- * Copyright 2009 Sun Microsystems, Inc. All rights reserved.
- * Use is subject to license terms.
+ * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
*/
/* Portions Copyright 2007 Jeremy Teo */
+/* Portions Copyright 2010 Robert Milkowski */
+
#include <sys/types.h>
#include <sys/param.h>
#include <sys/resource.h>
#include <sys/vfs.h>
#include <sys/vfs_opreg.h>
-#include <sys/vnode.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 <sys/vm.h>
-#include <vm/seg_vn.h>
#include <vm/pvn.h>
-#include <vm/as.h>
-#include <vm/kpm.h>
-#include <vm/seg_kpm.h>
-#include <sys/mman.h>
#include <sys/pathname.h>
#include <sys/cmn_err.h>
#include <sys/errno.h>
#include <sys/zfs_ioctl.h>
#include <sys/fs/zfs.h>
#include <sys/dmu.h>
+#include <sys/dmu_objset.h>
#include <sys/spa.h>
#include <sys/txg.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/filio.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_impl.h>
+#include <sys/cred.h>
#include <sys/attr.h>
/*
* 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(zfsvfs).
- * A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes
+ * This is done avoiding races using ZFS_ENTER(zsb).
+ * A ZFS_EXIT(zsb) 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.
*
- * (2) VN_RELE() should always be the last thing except for zil_commit()
+ * (2) iput() should always be the last thing except for zil_commit()
* (if necessary) and ZFS_EXIT(). This is for 3 reasons:
* First, if it's the last reference, the vnode/znode
* can be freed, so the zp may point to freed memory. Second, the last
* pushing cached pages (which acquires range locks) and syncing out
* cached atime changes. Third, zfs_zinactive() may require a new tx,
* which could deadlock the system if you were already holding one.
- * If you must call VN_RELE() within a tx then use VN_RELE_ASYNC().
+ * If you must call iput() within a tx then use iput_ASYNC().
*
* (3) All range locks must be grabbed before calling dmu_tx_assign(),
* as they can span dmu_tx_assign() calls.
* 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 zfsvfs->z_assign is TXG_NOWAIT,
+ * If dmu_tx_assign() returns ERESTART and zsb->z_assign is TXG_NOWAIT,
* then drop all locks, call dmu_tx_wait(), and try again.
*
* (5) If the operation succeeded, generate the intent log entry for it
* (6) At the end of each vnode op, the DMU tx must always commit,
* regardless of whether there were any errors.
*
- * (7) After dropping all locks, invoke zil_commit(zilog, seq, foid)
+ * (7) After dropping all locks, invoke zil_commit(zilog, foid)
* to ensure that synchronous semantics are provided when necessary.
*
* In general, this is how things should be ordered in each vnode op:
*
- * ZFS_ENTER(zfsvfs); // exit if unmounted
+ * ZFS_ENTER(zsb); // exit if unmounted
* top:
- * zfs_dirent_lock(&dl, ...) // lock directory entry (may VN_HOLD())
+ * 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
* if (error) {
* rw_exit(...); // drop locks
* zfs_dirent_unlock(dl); // unlock directory entry
- * VN_RELE(...); // release held vnodes
+ * iput(...); // release held vnodes
* if (error == ERESTART) {
* dmu_tx_wait(tx);
* dmu_tx_abort(tx);
* goto top;
* }
* dmu_tx_abort(tx); // abort DMU tx
- * ZFS_EXIT(zfsvfs); // finished in zfs
+ * ZFS_EXIT(zsb); // finished in zfs
* return (error); // really out of space
* }
* error = do_real_work(); // do whatever this VOP does
* dmu_tx_commit(tx); // commit DMU tx -- error or not
* rw_exit(...); // drop locks
* zfs_dirent_unlock(dl); // unlock directory entry
- * VN_RELE(...); // release held vnodes
- * zil_commit(zilog, seq, foid); // synchronous when necessary
- * ZFS_EXIT(zfsvfs); // finished in zfs
+ * iput(...); // release held vnodes
+ * zil_commit(zilog, foid); // synchronous when necessary
+ * ZFS_EXIT(zsb); // finished in zfs
* return (error); // done, report error
*/
-/* ARGSUSED */
-static int
-zfs_open(vnode_t **vpp, int flag, cred_t *cr, caller_context_t *ct)
-{
- znode_t *zp = VTOZ(*vpp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
-
- ZFS_ENTER(zfsvfs);
- ZFS_VERIFY_ZP(zp);
-
- if ((flag & FWRITE) && (zp->z_phys->zp_flags & ZFS_APPENDONLY) &&
- ((flag & FAPPEND) == 0)) {
- ZFS_EXIT(zfsvfs);
- return (EPERM);
- }
-
- if (!zfs_has_ctldir(zp) && zp->z_zfsvfs->z_vscan &&
- ZTOV(zp)->v_type == VREG &&
- !(zp->z_phys->zp_flags & ZFS_AV_QUARANTINED) &&
- zp->z_phys->zp_size > 0) {
- if (fs_vscan(*vpp, cr, 0) != 0) {
- ZFS_EXIT(zfsvfs);
- return (EACCES);
- }
- }
-
- /* Keep a count of the synchronous opens in the znode */
- if (flag & (FSYNC | FDSYNC))
- atomic_inc_32(&zp->z_sync_cnt);
-
- ZFS_EXIT(zfsvfs);
- return (0);
-}
-
-/* ARGSUSED */
-static int
-zfs_close(vnode_t *vp, int flag, int count, offset_t offset, cred_t *cr,
- caller_context_t *ct)
-{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
-
- ZFS_ENTER(zfsvfs);
- ZFS_VERIFY_ZP(zp);
-
- /* Decrement the synchronous opens in the znode */
- if ((flag & (FSYNC | FDSYNC)) && (count == 1))
- atomic_dec_32(&zp->z_sync_cnt);
-
- /*
- * Clean up any locks held by this process on the vp.
- */
- cleanlocks(vp, ddi_get_pid(), 0);
- cleanshares(vp, ddi_get_pid());
-
- if (!zfs_has_ctldir(zp) && zp->z_zfsvfs->z_vscan &&
- ZTOV(zp)->v_type == VREG &&
- !(zp->z_phys->zp_flags & ZFS_AV_QUARANTINED) &&
- zp->z_phys->zp_size > 0)
- VERIFY(fs_vscan(vp, cr, 1) == 0);
-
- ZFS_EXIT(zfsvfs);
- return (0);
-}
-
-/*
- * Lseek support for finding holes (cmd == _FIO_SEEK_HOLE) and
- * data (cmd == _FIO_SEEK_DATA). "off" is an in/out parameter.
- */
-static int
-zfs_holey(vnode_t *vp, int cmd, offset_t *off)
-{
- znode_t *zp = VTOZ(vp);
- uint64_t noff = (uint64_t)*off; /* new offset */
- uint64_t file_sz;
- int error;
- boolean_t hole;
-
- file_sz = zp->z_phys->zp_size;
- if (noff >= file_sz) {
- return (ENXIO);
- }
-
- if (cmd == _FIO_SEEK_HOLE)
- hole = B_TRUE;
- else
- hole = B_FALSE;
-
- error = dmu_offset_next(zp->z_zfsvfs->z_os, zp->z_id, hole, &noff);
-
- /* end of file? */
- if ((error == ESRCH) || (noff > file_sz)) {
- /*
- * Handle the virtual hole at the end of file.
- */
- if (hole) {
- *off = file_sz;
- return (0);
- }
- return (ENXIO);
- }
-
- if (noff < *off)
- return (error);
- *off = noff;
- return (error);
-}
-
-/* ARGSUSED */
-static int
-zfs_ioctl(vnode_t *vp, int com, intptr_t data, int flag, cred_t *cred,
- int *rvalp, caller_context_t *ct)
-{
- offset_t off;
- int error;
- zfsvfs_t *zfsvfs;
- znode_t *zp;
-
- switch (com) {
- case _FIOFFS:
- return (zfs_sync(vp->v_vfsp, 0, cred));
-
- /*
- * The following two ioctls are used by bfu. Faking out,
- * necessary to avoid bfu errors.
- */
- case _FIOGDIO:
- case _FIOSDIO:
- return (0);
-
- case _FIO_SEEK_DATA:
- case _FIO_SEEK_HOLE:
- if (ddi_copyin((void *)data, &off, sizeof (off), flag))
- return (EFAULT);
-
- zp = VTOZ(vp);
- zfsvfs = zp->z_zfsvfs;
- ZFS_ENTER(zfsvfs);
- ZFS_VERIFY_ZP(zp);
-
- /* offset parameter is in/out */
- error = zfs_holey(vp, com, &off);
- ZFS_EXIT(zfsvfs);
- if (error)
- return (error);
- if (ddi_copyout(&off, (void *)data, sizeof (off), flag))
- return (EFAULT);
- return (0);
- }
- return (ENOTTY);
-}
-
-/*
- * Utility functions to map and unmap a single physical page. These
- * are used to manage the mappable copies of ZFS file data, and therefore
- * do not update ref/mod bits.
- */
-caddr_t
-zfs_map_page(page_t *pp, enum seg_rw rw)
-{
- if (kpm_enable)
- return (hat_kpm_mapin(pp, 0));
- ASSERT(rw == S_READ || rw == S_WRITE);
- return (ppmapin(pp, PROT_READ | ((rw == S_WRITE) ? PROT_WRITE : 0),
- (caddr_t)-1));
-}
-
-void
-zfs_unmap_page(page_t *pp, caddr_t addr)
-{
- if (kpm_enable) {
- hat_kpm_mapout(pp, 0, addr);
- } else {
- ppmapout(addr);
- }
-}
-
+#if defined(_KERNEL)
/*
* When a file is memory mapped, we must keep the IO data synchronized
* between the DMU cache and the memory mapped pages. What this means:
* the page and the dmu buffer.
*/
static void
-update_pages(vnode_t *vp, int64_t start, int len, objset_t *os, uint64_t oid)
+update_pages(struct inode *ip, int64_t start, int len,
+ objset_t *os, uint64_t oid)
{
+ struct address_space *mp = ip->i_mapping;
+ struct page *pp;
+ uint64_t nbytes;
int64_t off;
+ void *pb;
- off = start & PAGEOFFSET;
- for (start &= PAGEMASK; len > 0; start += PAGESIZE) {
- page_t *pp;
- uint64_t nbytes = MIN(PAGESIZE - off, len);
+ off = start & (PAGE_CACHE_SIZE-1);
+ for (start &= PAGE_CACHE_MASK; len > 0; start += PAGE_CACHE_SIZE) {
+ nbytes = MIN(PAGE_CACHE_SIZE - off, len);
- if (pp = page_lookup(vp, start, SE_SHARED)) {
- caddr_t va;
+ pp = find_lock_page(mp, start >> PAGE_CACHE_SHIFT);
+ if (pp) {
+ if (mapping_writably_mapped(mp))
+ flush_dcache_page(pp);
- va = zfs_map_page(pp, S_WRITE);
- (void) dmu_read(os, oid, start+off, nbytes, va+off,
+ pb = kmap(pp);
+ (void) dmu_read(os, oid, start+off, nbytes, pb+off,
DMU_READ_PREFETCH);
- zfs_unmap_page(pp, va);
- page_unlock(pp);
+ kunmap(pp);
+
+ if (mapping_writably_mapped(mp))
+ flush_dcache_page(pp);
+
+ mark_page_accessed(pp);
+ SetPageUptodate(pp);
+ ClearPageError(pp);
+ unlock_page(pp);
+ page_cache_release(pp);
}
+
len -= nbytes;
off = 0;
}
* the file is memory mapped.
*/
static int
-mappedread(vnode_t *vp, int nbytes, uio_t *uio)
+mappedread(struct inode *ip, int nbytes, uio_t *uio)
{
- znode_t *zp = VTOZ(vp);
- objset_t *os = zp->z_zfsvfs->z_os;
+ struct address_space *mp = ip->i_mapping;
+ struct page *pp;
+ znode_t *zp = ITOZ(ip);
+ objset_t *os = ITOZSB(ip)->z_os;
int64_t start, off;
+ uint64_t bytes;
int len = nbytes;
int error = 0;
+ void *pb;
start = uio->uio_loffset;
- off = start & PAGEOFFSET;
- for (start &= PAGEMASK; len > 0; start += PAGESIZE) {
- page_t *pp;
- uint64_t bytes = MIN(PAGESIZE - off, len);
-
- if (pp = page_lookup(vp, start, SE_SHARED)) {
- caddr_t va;
-
- va = zfs_map_page(pp, S_READ);
- error = uiomove(va + off, bytes, UIO_READ, uio);
- zfs_unmap_page(pp, va);
- page_unlock(pp);
+ off = start & (PAGE_CACHE_SIZE-1);
+ for (start &= PAGE_CACHE_MASK; len > 0; start += PAGE_CACHE_SIZE) {
+ bytes = MIN(PAGE_CACHE_SIZE - off, len);
+
+ pp = find_lock_page(mp, start >> PAGE_CACHE_SHIFT);
+ if (pp) {
+ ASSERT(PageUptodate(pp));
+
+ pb = kmap(pp);
+ error = uiomove(pb + off, bytes, UIO_READ, uio);
+ kunmap(pp);
+
+ if (mapping_writably_mapped(mp))
+ flush_dcache_page(pp);
+
+ mark_page_accessed(pp);
+ unlock_page(pp);
+ page_cache_release(pp);
} else {
error = dmu_read_uio(os, zp->z_id, uio, bytes);
}
+
len -= bytes;
off = 0;
if (error)
}
return (error);
}
+#endif /* _KERNEL */
offset_t zfs_read_chunk_size = 1024 * 1024; /* Tunable */
/*
* Read bytes from specified file into supplied buffer.
*
- * IN: vp - vnode of file to be read from.
+ * IN: ip - inode of file to be read from.
* uio - structure supplying read location, range info,
* and return buffer.
- * ioflag - SYNC flags; used to provide FRSYNC semantics.
+ * ioflag - FSYNC flags; used to provide FRSYNC semantics.
+ * O_DIRECT flag; used to bypass page cache.
* cr - credentials of caller.
- * ct - caller context
*
* OUT: uio - updated offset and range, buffer filled.
*
* error code if failure
*
* Side Effects:
- * vp - atime updated if byte count > 0
+ * inode - atime updated if byte count > 0
*/
/* ARGSUSED */
-static int
-zfs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
+int
+zfs_read(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
objset_t *os;
ssize_t n, nbytes;
- int error;
+ int error = 0;
rl_t *rl;
+#ifdef HAVE_UIO_ZEROCOPY
+ xuio_t *xuio = NULL;
+#endif /* HAVE_UIO_ZEROCOPY */
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
- os = zfsvfs->z_os;
+ os = zsb->z_os;
- if (zp->z_phys->zp_flags & ZFS_AV_QUARANTINED) {
- ZFS_EXIT(zfsvfs);
+ if (zp->z_pflags & ZFS_AV_QUARANTINED) {
+ ZFS_EXIT(zsb);
return (EACCES);
}
* Validate file offset
*/
if (uio->uio_loffset < (offset_t)0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EINVAL);
}
* Fasttrack empty reads
*/
if (uio->uio_resid == 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (0);
}
+#ifdef HAVE_MANDLOCKS
/*
* Check for mandatory locks
*/
- if (MANDMODE((mode_t)zp->z_phys->zp_mode)) {
- if (error = chklock(vp, FREAD,
+ if (MANDMODE(zp->z_mode)) {
+ if (error = chklock(ip, FREAD,
uio->uio_loffset, uio->uio_resid, uio->uio_fmode, ct)) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
}
+#endif /* HAVE_MANDLOCK */
/*
* If we're in FRSYNC mode, sync out this znode before reading it.
*/
- if (ioflag & FRSYNC)
- zil_commit(zfsvfs->z_log, zp->z_last_itx, zp->z_id);
+ if (ioflag & FRSYNC || zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zsb->z_log, zp->z_id);
/*
* Lock the range against changes.
* If we are reading past end-of-file we can skip
* to the end; but we might still need to set atime.
*/
- if (uio->uio_loffset >= zp->z_phys->zp_size) {
+ if (uio->uio_loffset >= zp->z_size) {
error = 0;
goto out;
}
- ASSERT(uio->uio_loffset < zp->z_phys->zp_size);
- n = MIN(uio->uio_resid, zp->z_phys->zp_size - uio->uio_loffset);
+ ASSERT(uio->uio_loffset < zp->z_size);
+ n = MIN(uio->uio_resid, zp->z_size - uio->uio_loffset);
+
+#ifdef HAVE_UIO_ZEROCOPY
+ if ((uio->uio_extflg == UIO_XUIO) &&
+ (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) {
+ int nblk;
+ int blksz = zp->z_blksz;
+ uint64_t offset = uio->uio_loffset;
+
+ xuio = (xuio_t *)uio;
+ if ((ISP2(blksz))) {
+ nblk = (P2ROUNDUP(offset + n, blksz) - P2ALIGN(offset,
+ blksz)) / blksz;
+ } else {
+ ASSERT(offset + n <= blksz);
+ nblk = 1;
+ }
+ (void) dmu_xuio_init(xuio, nblk);
+
+ if (vn_has_cached_data(ip)) {
+ /*
+ * For simplicity, we always allocate a full buffer
+ * even if we only expect to read a portion of a block.
+ */
+ while (--nblk >= 0) {
+ (void) dmu_xuio_add(xuio,
+ dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
+ blksz), 0, blksz);
+ }
+ }
+ }
+#endif /* HAVE_UIO_ZEROCOPY */
while (n > 0) {
nbytes = MIN(n, zfs_read_chunk_size -
P2PHASE(uio->uio_loffset, zfs_read_chunk_size));
- if (vn_has_cached_data(vp))
- error = mappedread(vp, nbytes, uio);
+ if (zp->z_is_mapped && !(ioflag & O_DIRECT))
+ error = mappedread(ip, nbytes, uio);
else
error = dmu_read_uio(os, zp->z_id, uio, nbytes);
+
if (error) {
/* convert checksum errors into IO errors */
if (error == ECKSUM)
n -= nbytes;
}
-
out:
zfs_range_unlock(rl);
- ZFS_ACCESSTIME_STAMP(zfsvfs, zp);
- ZFS_EXIT(zfsvfs);
+ ZFS_ACCESSTIME_STAMP(zsb, zp);
+ zfs_inode_update(zp);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_read);
/*
* Write the bytes to a file.
*
- * IN: vp - vnode of file to be written to.
+ * IN: ip - inode of file to be written to.
* uio - structure supplying write location, range info,
* and data buffer.
* ioflag - FAPPEND flag set if in append mode.
+ * O_DIRECT flag; used to bypass page cache.
* cr - credentials of caller.
- * ct - caller context (NFS/CIFS fem monitor only)
*
* OUT: uio - updated offset and range.
*
* error code if failure
*
* Timestamps:
- * vp - ctime|mtime updated if byte count > 0
+ * ip - ctime|mtime updated if byte count > 0
*/
+
/* ARGSUSED */
-static int
-zfs_write(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
+int
+zfs_write(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- rlim64_t limit = uio->uio_llimit;
+ 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;
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ zfs_sb_t *zsb = ZTOZSB(zp);
zilog_t *zilog;
offset_t woff;
ssize_t n, nbytes;
rl_t *rl;
- int max_blksz = zfsvfs->z_max_blksz;
- uint64_t pflags;
- int error;
+ int max_blksz = zsb->z_max_blksz;
+ int error = 0;
arc_buf_t *abuf;
+ iovec_t *aiov = NULL;
+ xuio_t *xuio = NULL;
+ int i_iov = 0;
+ 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);
/*
* Fasttrack empty write
if (limit == RLIM64_INFINITY || limit > MAXOFFSET_T)
limit = MAXOFFSET_T;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
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,
+ &zp->z_pflags, 8);
+
/*
* If immutable or not appending then return EPERM
*/
- pflags = zp->z_phys->zp_flags;
- if ((pflags & (ZFS_IMMUTABLE | ZFS_READONLY)) ||
- ((pflags & ZFS_APPENDONLY) && !(ioflag & FAPPEND) &&
- (uio->uio_loffset < zp->z_phys->zp_size))) {
- ZFS_EXIT(zfsvfs);
+ if ((zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY)) ||
+ ((zp->z_pflags & ZFS_APPENDONLY) && !(ioflag & FAPPEND) &&
+ (uio->uio_loffset < zp->z_size))) {
+ ZFS_EXIT(zsb);
return (EPERM);
}
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
+
+ /*
+ * Validate file offset
+ */
+ woff = ioflag & FAPPEND ? zp->z_size : uio->uio_loffset;
+ if (woff < 0) {
+ ZFS_EXIT(zsb);
+ return (EINVAL);
+ }
+
+#ifdef HAVE_MANDLOCKS
+ /*
+ * Check for mandatory locks before calling zfs_range_lock()
+ * in order to prevent a deadlock with locks set via fcntl().
+ */
+ if (MANDMODE((mode_t)zp->z_mode) &&
+ (error = chklock(ip, FWRITE, woff, n, uio->uio_fmode, ct)) != 0) {
+ ZFS_EXIT(zsb);
+ return (error);
+ }
+#endif /* HAVE_MANDLOCKS */
+#ifdef HAVE_UIO_ZEROCOPY
/*
* Pre-fault the pages to ensure slow (eg NFS) pages
* don't hold up txg.
+ * Skip this if uio contains loaned arc_buf.
*/
- uio_prefaultpages(n, uio);
+ if ((uio->uio_extflg == UIO_XUIO) &&
+ (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY))
+ xuio = (xuio_t *)uio;
+ else
+ uio_prefaultpages(MIN(n, max_blksz), uio);
+#endif /* HAVE_UIO_ZEROCOPY */
/*
* If in append mode, set the io offset pointer to eof.
*/
if (ioflag & FAPPEND) {
/*
- * Range lock for a file append:
- * The value for the start of range will be determined by
- * zfs_range_lock() (to guarantee append semantics).
- * If this write will cause the block size to increase,
- * zfs_range_lock() will lock the entire file, so we must
- * later reduce the range after we grow the block size.
+ * 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, 0, n, RL_APPEND);
+ woff = rl->r_off;
if (rl->r_len == UINT64_MAX) {
- /* overlocked, zp_size can't change */
- woff = uio->uio_loffset = zp->z_phys->zp_size;
- } else {
- woff = uio->uio_loffset = rl->r_off;
+ /*
+ * We overlocked the file because this write will cause
+ * the file block size to increase.
+ * Note that zp_size cannot change with this lock held.
+ */
+ woff = zp->z_size;
}
+ uio->uio_loffset = woff;
} else {
- woff = uio->uio_loffset;
/*
- * Validate file offset
- */
- if (woff < 0) {
- ZFS_EXIT(zfsvfs);
- return (EINVAL);
- }
-
- /*
- * If we need to grow the block size then zfs_range_lock()
- * will lock a wider range than we request here.
- * Later after growing the block size we reduce the range.
+ * Note that if the file block size will change as a result of
+ * 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, woff, n, RL_WRITER);
}
if (woff >= limit) {
zfs_range_unlock(rl);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EFBIG);
}
if ((woff + n) > limit || woff > (limit - n))
n = limit - woff;
- /*
- * Check for mandatory locks
- */
- if (MANDMODE((mode_t)zp->z_phys->zp_mode) &&
- (error = chklock(vp, FWRITE, woff, n, uio->uio_fmode, ct)) != 0) {
- zfs_range_unlock(rl);
- ZFS_EXIT(zfsvfs);
- return (error);
- }
- end_size = MAX(zp->z_phys->zp_size, woff + n);
+ /* Will this write extend the file length? */
+ write_eof = (woff + n > zp->z_size);
+
+ end_size = MAX(zp->z_size, woff + n);
/*
* Write the file in reasonable size chunks. Each chunk is written
while (n > 0) {
abuf = NULL;
woff = uio->uio_loffset;
-
again:
- if (zfs_usergroup_overquota(zfsvfs,
- B_FALSE, zp->z_phys->zp_uid) ||
- zfs_usergroup_overquota(zfsvfs,
- B_TRUE, zp->z_phys->zp_gid)) {
+ if (zfs_owner_overquota(zsb, zp, B_FALSE) ||
+ zfs_owner_overquota(zsb, zp, B_TRUE)) {
if (abuf != NULL)
dmu_return_arcbuf(abuf);
error = EDQUOT;
break;
}
- /*
- * If dmu_assign_arcbuf() is expected to execute with minimum
- * overhead loan an arc buffer and copy user data to it before
- * we enter a txg. This avoids holding a txg forever while we
- * pagefault on a hanging NFS server mapping.
- */
- if (abuf == NULL && n >= max_blksz &&
- woff >= zp->z_phys->zp_size &&
+ if (xuio && abuf == NULL) {
+ ASSERT(i_iov < iovcnt);
+ aiov = &iovp[i_iov];
+ abuf = dmu_xuio_arcbuf(xuio, i_iov);
+ dmu_xuio_clear(xuio, i_iov);
+ ASSERT((aiov->iov_base == abuf->b_data) ||
+ ((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 &&
P2PHASE(woff, max_blksz) == 0 &&
zp->z_blksz == max_blksz) {
+ /*
+ * This write covers a full block. "Borrow" a buffer
+ * from the dmu so that we can fill it before we enter
+ * a transaction. This avoids the possibility of
+ * holding up the transaction if the data copy hangs
+ * up on a pagefault (e.g., from an NFS server mapping).
+ */
size_t cbytes;
- abuf = dmu_request_arcbuf(zp->z_dbuf, max_blksz);
+ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
+ max_blksz);
ASSERT(abuf != NULL);
ASSERT(arc_buf_size(abuf) == max_blksz);
- if (error = uiocopy(abuf->b_data, max_blksz,
- UIO_WRITE, uio, &cbytes)) {
+ if ((error = uiocopy(abuf->b_data, max_blksz,
+ UIO_WRITE, uio, &cbytes))) {
dmu_return_arcbuf(abuf);
break;
}
/*
* Start a transaction.
*/
- tx = dmu_tx_create(zfsvfs->z_os);
- dmu_tx_hold_bonus(tx, zp->z_id);
+ tx = dmu_tx_create(zsb->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);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
if (error == ERESTART) {
if (abuf == NULL) {
tx_bytes = uio->uio_resid;
- error = dmu_write_uio(zfsvfs->z_os, zp->z_id, uio,
- nbytes, tx);
+ error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl),
+ uio, nbytes, tx);
tx_bytes -= uio->uio_resid;
} else {
tx_bytes = nbytes;
- ASSERT(tx_bytes == max_blksz);
- dmu_assign_arcbuf(zp->z_dbuf, woff, abuf, tx);
+ ASSERT(xuio == NULL || tx_bytes == aiov->iov_len);
+ /*
+ * If this is not a full block write, but we are
+ * extending the file past EOF and this data starts
+ * block-aligned, use assign_arcbuf(). Otherwise,
+ * write via dmu_write().
+ */
+ if (tx_bytes < max_blksz && (!write_eof ||
+ aiov->iov_base != abuf->b_data)) {
+ ASSERT(xuio);
+ dmu_write(zsb->z_os, zp->z_id, woff,
+ 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);
+ }
ASSERT(tx_bytes <= uio->uio_resid);
uioskip(uio, tx_bytes);
}
- if (tx_bytes && vn_has_cached_data(vp)) {
- update_pages(vp, woff,
- tx_bytes, zfsvfs->z_os, zp->z_id);
- }
+
+ if (tx_bytes && zp->z_is_mapped && !(ioflag & O_DIRECT))
+ update_pages(ip, woff, tx_bytes, zsb->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 *)&zp->z_size, sizeof (uint64_t), tx);
dmu_tx_commit(tx);
ASSERT(error != 0);
break;
* user 0 is not an ephemeral uid.
*/
mutex_enter(&zp->z_acl_lock);
- if ((zp->z_phys->zp_mode & (S_IXUSR | (S_IXUSR >> 3) |
+ if ((zp->z_mode & (S_IXUSR | (S_IXUSR >> 3) |
(S_IXUSR >> 6))) != 0 &&
- (zp->z_phys->zp_mode & (S_ISUID | S_ISGID)) != 0 &&
+ (zp->z_mode & (S_ISUID | S_ISGID)) != 0 &&
secpolicy_vnode_setid_retain(cr,
- (zp->z_phys->zp_mode & S_ISUID) != 0 &&
- zp->z_phys->zp_uid == 0) != 0) {
- zp->z_phys->zp_mode &= ~(S_ISUID | S_ISGID);
+ (zp->z_mode & S_ISUID) != 0 && zp->z_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),
+ (void *)&newmode, sizeof (uint64_t), tx);
}
mutex_exit(&zp->z_acl_lock);
- /*
- * Update time stamp. NOTE: This marks the bonus buffer as
- * dirty, so we don't have to do it again for zp_size.
- */
- zfs_time_stamper(zp, CONTENT_MODIFIED, tx);
+ zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime,
+ B_TRUE);
/*
* Update the file size (zp_size) if it has changed;
* account for possible concurrent updates.
*/
- while ((end_size = zp->z_phys->zp_size) < uio->uio_loffset)
- (void) atomic_cas_64(&zp->z_phys->zp_size, end_size,
+ while ((end_size = zp->z_size) < uio->uio_loffset) {
+ (void) atomic_cas_64(&zp->z_size, end_size,
uio->uio_loffset);
+ ASSERT(error == 0);
+ }
+ /*
+ * If we are replaying and eof is non zero then force
+ * 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;
+
+ error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
+
zfs_log_write(zilog, tx, TX_WRITE, zp, woff, tx_bytes, ioflag);
dmu_tx_commit(tx);
break;
ASSERT(tx_bytes == nbytes);
n -= nbytes;
+
+ if (!xuio && n > 0)
+ uio_prefaultpages(MIN(n, max_blksz), uio);
}
zfs_range_unlock(rl);
* 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 (zfsvfs->z_replay || uio->uio_resid == start_resid) {
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_replay || uio->uio_resid == start_resid) {
+ ZFS_EXIT(zsb);
return (error);
}
- if (ioflag & (FSYNC | FDSYNC))
- zil_commit(zilog, zp->z_last_itx, zp->z_id);
+ if (ioflag & (FSYNC | FDSYNC) ||
+ zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, zp->z_id);
- ZFS_EXIT(zfsvfs);
+ zfs_inode_update(zp);
+ ZFS_EXIT(zsb);
return (0);
}
+EXPORT_SYMBOL(zfs_write);
+
+static void
+iput_async(struct inode *ip, taskq_t *taskq)
+{
+ ASSERT(atomic_read(&ip->i_count) > 0);
+ if (atomic_read(&ip->i_count) == 1)
+ taskq_dispatch(taskq, (task_func_t *)iput, ip, TQ_SLEEP);
+ else
+ iput(ip);
+}
void
-zfs_get_done(dmu_buf_t *db, void *vzgd)
+zfs_get_done(zgd_t *zgd, int error)
{
- zgd_t *zgd = (zgd_t *)vzgd;
- rl_t *rl = zgd->zgd_rl;
- vnode_t *vp = ZTOV(rl->r_zp);
- objset_t *os = rl->r_zp->z_zfsvfs->z_os;
+ znode_t *zp = zgd->zgd_private;
+ objset_t *os = ZTOZSB(zp)->z_os;
+
+ if (zgd->zgd_db)
+ dmu_buf_rele(zgd->zgd_db, zgd);
+
+ zfs_range_unlock(zgd->zgd_rl);
- dmu_buf_rele(db, vzgd);
- zfs_range_unlock(rl);
/*
* Release the vnode asynchronously as we currently have the
* txg stopped from syncing.
*/
- VN_RELE_ASYNC(vp, dsl_pool_vnrele_taskq(dmu_objset_pool(os)));
- zil_add_block(zgd->zgd_zilog, zgd->zgd_bp);
+ iput_async(ZTOI(zp), dsl_pool_iput_taskq(dmu_objset_pool(os)));
+
+ if (error == 0 && zgd->zgd_bp)
+ zil_add_block(zgd->zgd_zilog, zgd->zgd_bp);
+
kmem_free(zgd, sizeof (zgd_t));
}
+#ifdef DEBUG
+static int zil_fault_io = 0;
+#endif
+
/*
* 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)
{
- zfsvfs_t *zfsvfs = arg;
- objset_t *os = zfsvfs->z_os;
+ zfs_sb_t *zsb = arg;
+ objset_t *os = zsb->z_os;
znode_t *zp;
- uint64_t off = lr->lr_offset;
+ 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;
- rl_t *rl;
zgd_t *zgd;
- int dlen = lr->lr_length; /* length of user data */
int error = 0;
- ASSERT(zio);
- ASSERT(dlen != 0);
+ ASSERT(zio != NULL);
+ ASSERT(size != 0);
/*
* Nothing to do if the file has been removed
*/
- if (zfs_zget(zfsvfs, lr->lr_foid, &zp) != 0)
+ if (zfs_zget(zsb, object, &zp) != 0)
return (ENOENT);
if (zp->z_unlinked) {
/*
* Release the vnode asynchronously as we currently have the
* txg stopped from syncing.
*/
- VN_RELE_ASYNC(ZTOV(zp),
- dsl_pool_vnrele_taskq(dmu_objset_pool(os)));
+ iput_async(ZTOI(zp), dsl_pool_iput_taskq(dmu_objset_pool(os)));
return (ENOENT);
}
+ zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP);
+ zgd->zgd_zilog = zsb->z_log;
+ zgd->zgd_private = zp;
+
/*
* Write records come in two flavors: immediate and indirect.
* For small writes it's cheaper to store the data with the
* we don't have to write the data twice.
*/
if (buf != NULL) { /* immediate write */
- rl = zfs_range_lock(zp, off, dlen, RL_READER);
+ zgd->zgd_rl = zfs_range_lock(zp, offset, size, RL_READER);
/* test for truncation needs to be done while range locked */
- if (off >= zp->z_phys->zp_size) {
+ if (offset >= zp->z_size) {
error = ENOENT;
- goto out;
+ } else {
+ error = dmu_read(os, object, offset, size, buf,
+ DMU_READ_NO_PREFETCH);
}
- VERIFY(0 == dmu_read(os, lr->lr_foid, off, dlen, buf,
- DMU_READ_NO_PREFETCH));
+ ASSERT(error == 0 || error == ENOENT);
} else { /* indirect write */
- uint64_t boff; /* block starting offset */
-
/*
* Have to lock the whole block to ensure when it's
* written out and it's checksum is being calculated
* blocksize after we get the lock in case it's changed!
*/
for (;;) {
- if (ISP2(zp->z_blksz)) {
- boff = P2ALIGN_TYPED(off, zp->z_blksz,
- uint64_t);
- } else {
- boff = 0;
- }
- dlen = zp->z_blksz;
- rl = zfs_range_lock(zp, boff, dlen, RL_READER);
- if (zp->z_blksz == dlen)
+ uint64_t blkoff;
+ size = zp->z_blksz;
+ blkoff = ISP2(size) ? P2PHASE(offset, size) : offset;
+ offset -= blkoff;
+ zgd->zgd_rl = zfs_range_lock(zp, offset, size,
+ RL_READER);
+ if (zp->z_blksz == size)
break;
- zfs_range_unlock(rl);
+ offset += blkoff;
+ zfs_range_unlock(zgd->zgd_rl);
}
/* test for truncation needs to be done while range locked */
- if (off >= zp->z_phys->zp_size) {
+ if (lr->lr_offset >= zp->z_size)
error = ENOENT;
- goto out;
+#ifdef DEBUG
+ if (zil_fault_io) {
+ error = EIO;
+ zil_fault_io = 0;
}
- zgd = (zgd_t *)kmem_alloc(sizeof (zgd_t), KM_SLEEP);
- zgd->zgd_rl = rl;
- zgd->zgd_zilog = zfsvfs->z_log;
- zgd->zgd_bp = &lr->lr_blkptr;
- VERIFY(0 == dmu_buf_hold(os, lr->lr_foid, boff, zgd, &db));
- ASSERT(boff == db->db_offset);
- lr->lr_blkoff = off - boff;
- error = dmu_sync(zio, db, &lr->lr_blkptr,
- lr->lr_common.lrc_txg, zfs_get_done, zgd);
- ASSERT((error && error != EINPROGRESS) ||
- lr->lr_length <= zp->z_blksz);
+#endif
if (error == 0)
- zil_add_block(zfsvfs->z_log, &lr->lr_blkptr);
- /*
- * If we get EINPROGRESS, then we need to wait for a
- * write IO initiated by dmu_sync() to complete before
- * we can release this dbuf. We will finish everything
- * up in the zfs_get_done() callback.
- */
- if (error == EINPROGRESS)
- return (0);
- dmu_buf_rele(db, zgd);
- kmem_free(zgd, sizeof (zgd_t));
+ error = dmu_buf_hold(os, object, offset, zgd, &db,
+ DMU_READ_NO_PREFETCH);
+
+ if (error == 0) {
+ zgd->zgd_db = db;
+ zgd->zgd_bp = bp;
+
+ ASSERT(db->db_offset == offset);
+ ASSERT(db->db_size == size);
+
+ error = dmu_sync(zio, lr->lr_common.lrc_txg,
+ zfs_get_done, zgd);
+ ASSERT(error || lr->lr_length <= zp->z_blksz);
+
+ /*
+ * On success, we need to wait for the write I/O
+ * initiated by dmu_sync() to complete before we can
+ * release this dbuf. We will finish everything up
+ * in the zfs_get_done() callback.
+ */
+ if (error == 0)
+ return (0);
+
+ if (error == EALREADY) {
+ lr->lr_common.lrc_txtype = TX_WRITE2;
+ error = 0;
+ }
+ }
}
-out:
- zfs_range_unlock(rl);
- /*
- * Release the vnode asynchronously as we currently have the
- * txg stopped from syncing.
- */
- VN_RELE_ASYNC(ZTOV(zp), dsl_pool_vnrele_taskq(dmu_objset_pool(os)));
+
+ zfs_get_done(zgd, error);
+
return (error);
}
/*ARGSUSED*/
-static int
-zfs_access(vnode_t *vp, int mode, int flag, cred_t *cr,
- caller_context_t *ct)
+int
+zfs_access(struct inode *ip, int mode, int flag, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
int error;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
if (flag & V_ACE_MASK)
else
error = zfs_zaccess_rwx(zp, mode, flag, cr);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_access);
/*
* Lookup an entry in a directory, or an extended attribute directory.
- * If it exists, return a held vnode reference for it.
+ * If it exists, return a held inode reference for it.
*
- * IN: dvp - vnode of directory to search.
+ * IN: dip - inode of directory to search.
* nm - name of entry to lookup.
- * pnp - full pathname to lookup [UNUSED].
* flags - LOOKUP_XATTR set if looking for an attribute.
- * rdir - root directory vnode [UNUSED].
* cr - credentials of caller.
- * ct - caller context
* direntflags - directory lookup flags
* realpnp - returned pathname.
*
- * OUT: vpp - vnode of located entry, NULL if not found.
+ * OUT: ipp - inode of located entry, NULL if not found.
*
* RETURN: 0 if success
* error code if failure
* NA
*/
/* ARGSUSED */
-static int
-zfs_lookup(vnode_t *dvp, char *nm, vnode_t **vpp, struct pathname *pnp,
- int flags, vnode_t *rdir, cred_t *cr, caller_context_t *ct,
- int *direntflags, pathname_t *realpnp)
+int
+zfs_lookup(struct inode *dip, char *nm, struct inode **ipp, int flags,
+ cred_t *cr, int *direntflags, pathname_t *realpnp)
{
- znode_t *zdp = VTOZ(dvp);
- zfsvfs_t *zfsvfs = zdp->z_zfsvfs;
- int error;
+ znode_t *zdp = ITOZ(dip);
+ zfs_sb_t *zsb = ITOZSB(dip);
+ int error = 0;
- ZFS_ENTER(zfsvfs);
+ /* fast path */
+ if (!(flags & (LOOKUP_XATTR | FIGNORECASE))) {
+
+ if (!S_ISDIR(dip->i_mode)) {
+ return (ENOTDIR);
+ } else if (zdp->z_sa_hdl == NULL) {
+ return (EIO);
+ }
+
+ if (nm[0] == 0 || (nm[0] == '.' && nm[1] == '\0')) {
+ error = zfs_fastaccesschk_execute(zdp, cr);
+ if (!error) {
+ *ipp = dip;
+ igrab(*ipp);
+ 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 (ENOENT);
+ } else {
+ *vpp = tvp;
+ return (specvp_check(vpp, cr));
+ }
+ }
+#endif /* HAVE_DNLC */
+ }
+ }
+
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zdp);
- *vpp = NULL;
+ *ipp = NULL;
if (flags & LOOKUP_XATTR) {
/*
* If the xattr property is off, refuse the lookup request.
*/
- if (!(zfsvfs->z_vfs->vfs_flag & VFS_XATTR)) {
- ZFS_EXIT(zfsvfs);
+ if (!(zsb->z_flags & ZSB_XATTR_USER)) {
+ ZFS_EXIT(zsb);
return (EINVAL);
}
* We don't allow recursive attributes..
* Maybe someday we will.
*/
- if (zdp->z_phys->zp_flags & ZFS_XATTR) {
- ZFS_EXIT(zfsvfs);
+ if (zdp->z_pflags & ZFS_XATTR) {
+ ZFS_EXIT(zsb);
return (EINVAL);
}
- if (error = zfs_get_xattrdir(VTOZ(dvp), vpp, cr, flags)) {
- ZFS_EXIT(zfsvfs);
+ if ((error = zfs_get_xattrdir(zdp, ipp, cr, flags))) {
+ ZFS_EXIT(zsb);
return (error);
}
* Do we have permission to get into attribute directory?
*/
- if (error = zfs_zaccess(VTOZ(*vpp), ACE_EXECUTE, 0,
- B_FALSE, cr)) {
- VN_RELE(*vpp);
- *vpp = NULL;
+ if ((error = zfs_zaccess(ITOZ(*ipp), ACE_EXECUTE, 0,
+ B_FALSE, cr))) {
+ iput(*ipp);
+ *ipp = NULL;
}
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
- if (dvp->v_type != VDIR) {
- ZFS_EXIT(zfsvfs);
+ if (!S_ISDIR(dip->i_mode)) {
+ ZFS_EXIT(zsb);
return (ENOTDIR);
}
* Check accessibility of directory.
*/
- if (error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr)) {
- ZFS_EXIT(zfsvfs);
+ if ((error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr))) {
+ ZFS_EXIT(zsb);
return (error);
}
- if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm),
+ if (zsb->z_utf8 && u8_validate(nm, strlen(nm),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EILSEQ);
}
- error = zfs_dirlook(zdp, nm, vpp, flags, direntflags, realpnp);
- if (error == 0) {
- /*
- * Convert device special files
- */
- if (IS_DEVVP(*vpp)) {
- vnode_t *svp;
-
- svp = specvp(*vpp, (*vpp)->v_rdev, (*vpp)->v_type, cr);
- VN_RELE(*vpp);
- if (svp == NULL)
- error = ENOSYS;
- else
- *vpp = svp;
- }
- }
+ error = zfs_dirlook(zdp, nm, ipp, flags, direntflags, realpnp);
+ if ((error == 0) && (*ipp))
+ zfs_inode_update(ITOZ(*ipp));
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_lookup);
/*
* Attempt to create a new entry in a directory. If the entry
* already exists, truncate the file if permissible, else return
- * an error. Return the vp of the created or trunc'd file.
+ * an error. Return the ip of the created or trunc'd file.
*
- * IN: dvp - vnode of directory to put new file entry in.
+ * IN: dip - inode of directory to put new file entry in.
* name - name of new file entry.
* vap - attributes of new file.
* excl - flag indicating exclusive or non-exclusive mode.
* mode - mode to open file with.
* cr - credentials of caller.
* flag - large file flag [UNUSED].
- * ct - caller context
- * vsecp - ACL to be set
+ * vsecp - ACL to be set
*
- * OUT: vpp - vnode of created or trunc'd entry.
+ * OUT: ipp - inode of created or trunc'd entry.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * dvp - ctime|mtime updated if new entry created
- * vp - ctime|mtime always, atime if new
+ * dip - ctime|mtime updated if new entry created
+ * ip - ctime|mtime always, atime if new
*/
/* ARGSUSED */
-static int
-zfs_create(vnode_t *dvp, char *name, vattr_t *vap, vcexcl_t excl,
- int mode, vnode_t **vpp, cred_t *cr, int flag, caller_context_t *ct,
- vsecattr_t *vsecp)
+int
+zfs_create(struct inode *dip, char *name, vattr_t *vap, int excl,
+ int mode, struct inode **ipp, cred_t *cr, int flag, vsecattr_t *vsecp)
{
- znode_t *zp, *dzp = VTOZ(dvp);
- zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
+ znode_t *zp, *dzp = ITOZ(dip);
+ zfs_sb_t *zsb = ITOZSB(dip);
zilog_t *zilog;
objset_t *os;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
int error;
- ksid_t *ksid;
uid_t uid;
- gid_t gid = crgetgid(cr);
- zfs_acl_ids_t acl_ids;
+ gid_t gid;
+ zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
+ boolean_t have_acl = B_FALSE;
/*
* If we have an ephemeral id, ACL, or XVATTR then
* make sure file system is at proper version
*/
- ksid = crgetsid(cr, KSID_OWNER);
- if (ksid)
- uid = ksid_getid(ksid);
- else
- uid = crgetuid(cr);
+ gid = crgetgid(cr);
+ uid = crgetuid(cr);
- if (zfsvfs->z_use_fuids == B_FALSE &&
- (vsecp || (vap->va_mask & AT_XVATTR) ||
- IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
+ if (zsb->z_use_fuids == B_FALSE &&
+ (vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (EINVAL);
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(dzp);
- os = zfsvfs->z_os;
- zilog = zfsvfs->z_log;
+ os = zsb->z_os;
+ zilog = zsb->z_log;
- if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
+ if (zsb->z_utf8 && u8_validate(name, strlen(name),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EILSEQ);
}
- if (vap->va_mask & AT_XVATTR) {
+ if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
- crgetuid(cr), cr, vap->va_type)) != 0) {
- ZFS_EXIT(zfsvfs);
+ crgetuid(cr), cr, vap->va_mode)) != 0) {
+ ZFS_EXIT(zsb);
return (error);
}
}
-top:
- *vpp = NULL;
-
- if ((vap->va_mode & VSVTX) && secpolicy_vnode_stky_modify(cr))
- vap->va_mode &= ~VSVTX;
+top:
+ *ipp = NULL;
if (*name == '\0') {
/*
* Null component name refers to the directory itself.
*/
- VN_HOLD(dvp);
+ igrab(dip);
zp = dzp;
dl = NULL;
error = 0;
} else {
- /* possible VN_HOLD(zp) */
+ /* possible igrab(zp) */
int zflg = 0;
if (flag & FIGNORECASE)
error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
NULL, NULL);
if (error) {
+ if (have_acl)
+ zfs_acl_ids_free(&acl_ids);
if (strcmp(name, "..") == 0)
error = EISDIR;
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
}
+
if (zp == NULL) {
uint64_t txtype;
* 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 ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
+ if (have_acl)
+ zfs_acl_ids_free(&acl_ids);
goto out;
}
* We only support the creation of regular files in
* extended attribute directories.
*/
- if ((dzp->z_phys->zp_flags & ZFS_XATTR) &&
- (vap->va_type != VREG)) {
+
+ if ((dzp->z_pflags & ZFS_XATTR) && !S_ISREG(vap->va_mode)) {
+ if (have_acl)
+ zfs_acl_ids_free(&acl_ids);
error = EINVAL;
goto out;
}
- if ((error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp,
- &acl_ids)) != 0)
+ if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
+ cr, vsecp, &acl_ids)) != 0)
goto out;
- if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
+ have_acl = B_TRUE;
+
+ if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
+ zfs_acl_ids_free(&acl_ids);
error = EDQUOT;
goto out;
}
tx = dmu_tx_create(os);
- dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
- fuid_dirtied = zfsvfs->z_fuid_dirty;
+
+ dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
+ ZFS_SA_BASE_ATTR_SIZE);
+
+ fuid_dirtied = zsb->z_fuid_dirty;
if (fuid_dirtied)
- zfs_fuid_txhold(zfsvfs, tx);
- dmu_tx_hold_bonus(tx, dzp->z_id);
+ zfs_fuid_txhold(zsb, tx);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
- if (acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
+ 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) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
- 0, SPA_MAXBLOCKSIZE);
+ 0, acl_ids.z_aclp->z_acl_bytes);
}
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
- zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
+ zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
- zfs_mknode(dzp, vap, tx, cr, 0, &zp, 0, &acl_ids);
+ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
if (fuid_dirtied)
- zfs_fuid_sync(zfsvfs, tx);
+ zfs_fuid_sync(zsb, tx);
(void) zfs_link_create(dl, zp, tx, ZNEW);
-
txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap);
if (flag & FIGNORECASE)
txtype |= TX_CI;
} else {
int aflags = (flag & FAPPEND) ? V_APPEND : 0;
+ if (have_acl)
+ zfs_acl_ids_free(&acl_ids);
+ have_acl = B_FALSE;
+
/*
* A directory entry already exists for this name.
*/
/*
* Can't truncate an existing file if in exclusive mode.
*/
- if (excl == EXCL) {
+ if (excl) {
error = EEXIST;
goto out;
}
/*
* Can't open a directory for writing.
*/
- if ((ZTOV(zp)->v_type == VDIR) && (mode & S_IWRITE)) {
+ if (S_ISDIR(ZTOI(zp)->i_mode)) {
error = EISDIR;
goto out;
}
/*
* Truncate regular files if requested.
*/
- if ((ZTOV(zp)->v_type == VREG) &&
- (vap->va_mask & AT_SIZE) && (vap->va_size == 0)) {
+ 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;
error = zfs_freesp(zp, 0, 0, mode, TRUE);
- if (error == 0) {
- vnevent_create(ZTOV(zp), ct);
- }
}
}
out:
if (error) {
if (zp)
- VN_RELE(ZTOV(zp));
+ iput(ZTOI(zp));
} else {
- *vpp = ZTOV(zp);
- /*
- * If vnode is for a device return a specfs vnode instead.
- */
- if (IS_DEVVP(*vpp)) {
- struct vnode *svp;
-
- svp = specvp(*vpp, (*vpp)->v_rdev, (*vpp)->v_type, cr);
- VN_RELE(*vpp);
- if (svp == NULL) {
- error = ENOSYS;
- }
- *vpp = svp;
- }
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ *ipp = ZTOI(zp);
}
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_create);
/*
* Remove an entry from a directory.
*
- * IN: dvp - vnode of directory to remove entry from.
+ * IN: dip - inode of directory to remove entry from.
* name - name of entry to remove.
* cr - credentials of caller.
- * ct - caller context
- * flags - case flags
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * dvp - ctime|mtime
- * vp - ctime (if nlink > 0)
+ * dip - ctime|mtime
+ * ip - ctime (if nlink > 0)
*/
+
+uint64_t null_xattr = 0;
+
/*ARGSUSED*/
-static int
-zfs_remove(vnode_t *dvp, char *name, cred_t *cr, caller_context_t *ct,
- int flags)
+int
+zfs_remove(struct inode *dip, char *name, cred_t *cr)
{
- znode_t *zp, *dzp = VTOZ(dvp);
- znode_t *xzp = NULL;
- vnode_t *vp;
- zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
+ znode_t *zp, *dzp = ITOZ(dip);
+ znode_t *xzp;
+ struct inode *ip;
+ zfs_sb_t *zsb = ITOZSB(dip);
zilog_t *zilog;
- uint64_t acl_obj, xattr_obj;
+ uint64_t xattr_obj;
+ uint64_t xattr_obj_unlinked = 0;
+ uint64_t obj = 0;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
- boolean_t may_delete_now, delete_now = FALSE;
- boolean_t unlinked, toobig = FALSE;
+ boolean_t unlinked;
uint64_t txtype;
pathname_t *realnmp = NULL;
+#ifdef HAVE_PN_UTILS
pathname_t realnm;
+#endif /* HAVE_PN_UTILS */
int error;
int zflg = ZEXISTS;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(dzp);
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
+#ifdef HAVE_PN_UTILS
if (flags & FIGNORECASE) {
zflg |= ZCILOOK;
pn_alloc(&realnm);
realnmp = &realnm;
}
+#endif /* HAVE_PN_UTILS */
top:
+ xattr_obj = 0;
+ xzp = NULL;
/*
* Attempt to lock directory; fail if entry doesn't exist.
*/
- if (error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
- NULL, realnmp)) {
+ if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
+ NULL, realnmp))) {
+#ifdef HAVE_PN_UTILS
if (realnmp)
pn_free(realnmp);
- ZFS_EXIT(zfsvfs);
+#endif /* HAVE_PN_UTILS */
+ ZFS_EXIT(zsb);
return (error);
}
- vp = ZTOV(zp);
+ ip = ZTOI(zp);
- if (error = zfs_zaccess_delete(dzp, zp, cr)) {
+ if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
goto out;
}
/*
* Need to use rmdir for removing directories.
*/
- if (vp->v_type == VDIR) {
+ if (S_ISDIR(ip->i_mode)) {
error = EPERM;
goto out;
}
- vnevent_remove(vp, dvp, name, ct);
-
+#ifdef HAVE_DNLC
if (realnmp)
dnlc_remove(dvp, realnmp->pn_buf);
else
dnlc_remove(dvp, name);
-
- mutex_enter(&vp->v_lock);
- may_delete_now = vp->v_count == 1 && !vn_has_cached_data(vp);
- mutex_exit(&vp->v_lock);
+#endif /* HAVE_DNLC */
/*
- * We may delete the znode now, or we may put it in the unlinked set;
- * it depends on whether we're the last link, and on whether there are
- * other holds on the vnode. So we dmu_tx_hold() the right things to
- * allow for either case.
+ * We never delete the znode and always place it in the unlinked
+ * set. The dentry cache will always hold the last reference and
+ * is responsible for safely freeing the znode.
*/
- tx = dmu_tx_create(zfsvfs->z_os);
+ obj = zp->z_id;
+ tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
- dmu_tx_hold_bonus(tx, zp->z_id);
- if (may_delete_now) {
- toobig =
- zp->z_phys->zp_size > zp->z_blksz * DMU_MAX_DELETEBLKCNT;
- /* if the file is too big, only hold_free a token amount */
- dmu_tx_hold_free(tx, zp->z_id, 0,
- (toobig ? DMU_MAX_ACCESS : DMU_OBJECT_END));
- }
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
+ zfs_sa_upgrade_txholds(tx, zp);
+ zfs_sa_upgrade_txholds(tx, dzp);
/* are there any extended attributes? */
- if ((xattr_obj = zp->z_phys->zp_xattr) != 0) {
- /* XXX - do we need this if we are deleting? */
- dmu_tx_hold_bonus(tx, xattr_obj);
+ error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
+ &xattr_obj, sizeof (xattr_obj));
+ if (error == 0 && xattr_obj) {
+ error = zfs_zget(zsb, xattr_obj, &xzp);
+ ASSERT3U(error, ==, 0);
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
+ dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
}
- /* are there any additional acls */
- if ((acl_obj = zp->z_phys->zp_acl.z_acl_extern_obj) != 0 &&
- may_delete_now)
- dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
-
/* charge as an update -- would be nice not to charge at all */
- dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
+ dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, FALSE, NULL);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
- VN_RELE(vp);
+ iput(ip);
+ if (xzp)
+ iput(ZTOI(xzp));
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
+#ifdef HAVE_PN_UTILS
if (realnmp)
pn_free(realnmp);
+#endif /* HAVE_PN_UTILS */
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
}
if (unlinked) {
- mutex_enter(&vp->v_lock);
- delete_now = may_delete_now && !toobig &&
- vp->v_count == 1 && !vn_has_cached_data(vp) &&
- zp->z_phys->zp_xattr == xattr_obj &&
- zp->z_phys->zp_acl.z_acl_extern_obj == acl_obj;
- mutex_exit(&vp->v_lock);
- }
-
- if (delete_now) {
- if (zp->z_phys->zp_xattr) {
- error = zfs_zget(zfsvfs, zp->z_phys->zp_xattr, &xzp);
- ASSERT3U(error, ==, 0);
- ASSERT3U(xzp->z_phys->zp_links, ==, 2);
- dmu_buf_will_dirty(xzp->z_dbuf, tx);
- mutex_enter(&xzp->z_lock);
- xzp->z_unlinked = 1;
- xzp->z_phys->zp_links = 0;
- mutex_exit(&xzp->z_lock);
- zfs_unlinked_add(xzp, tx);
- zp->z_phys->zp_xattr = 0; /* probably unnecessary */
- }
+ /*
+ * Hold z_lock so that we can make sure that the ACL obj
+ * hasn't changed. Could have been deleted due to
+ * zfs_sa_upgrade().
+ */
mutex_enter(&zp->z_lock);
- mutex_enter(&vp->v_lock);
- vp->v_count--;
- ASSERT3U(vp->v_count, ==, 0);
- mutex_exit(&vp->v_lock);
+ (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
+ &xattr_obj_unlinked, sizeof (xattr_obj_unlinked));
mutex_exit(&zp->z_lock);
- zfs_znode_delete(zp, tx);
- } else if (unlinked) {
zfs_unlinked_add(zp, tx);
}
txtype = TX_REMOVE;
+#ifdef HAVE_PN_UTILS
if (flags & FIGNORECASE)
txtype |= TX_CI;
- zfs_log_remove(zilog, tx, txtype, dzp, name);
+#endif /* HAVE_PN_UTILS */
+ zfs_log_remove(zilog, tx, txtype, dzp, name, obj);
dmu_tx_commit(tx);
out:
+#ifdef HAVE_PN_UTILS
if (realnmp)
pn_free(realnmp);
+#endif /* HAVE_PN_UTILS */
zfs_dirent_unlock(dl);
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ if (xzp)
+ zfs_inode_update(xzp);
- if (!delete_now) {
- VN_RELE(vp);
- } else if (xzp) {
- /* this rele is delayed to prevent nesting transactions */
- VN_RELE(ZTOV(xzp));
- }
+ iput(ip);
+ if (xzp)
+ iput(ZTOI(xzp));
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_remove);
/*
- * Create a new directory and insert it into dvp using the name
+ * Create a new directory and insert it into dip using the name
* provided. Return a pointer to the inserted directory.
*
- * IN: dvp - vnode of directory to add subdir to.
+ * IN: dip - inode of directory to add subdir to.
* dirname - name of new directory.
* vap - attributes of new directory.
* cr - credentials of caller.
- * ct - caller context
* vsecp - ACL to be set
*
- * OUT: vpp - vnode of created directory.
+ * OUT: ipp - inode of created directory.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * dvp - ctime|mtime updated
- * vp - ctime|mtime|atime updated
+ * dip - ctime|mtime updated
+ * ipp - ctime|mtime|atime updated
*/
/*ARGSUSED*/
-static int
-zfs_mkdir(vnode_t *dvp, char *dirname, vattr_t *vap, vnode_t **vpp, cred_t *cr,
- caller_context_t *ct, int flags, vsecattr_t *vsecp)
+int
+zfs_mkdir(struct inode *dip, char *dirname, vattr_t *vap, struct inode **ipp,
+ cred_t *cr, int flags, vsecattr_t *vsecp)
{
- znode_t *zp, *dzp = VTOZ(dvp);
- zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
+ znode_t *zp, *dzp = ITOZ(dip);
+ zfs_sb_t *zsb = ITOZSB(dip);
zilog_t *zilog;
zfs_dirlock_t *dl;
uint64_t txtype;
dmu_tx_t *tx;
int error;
int zf = ZNEW;
- ksid_t *ksid;
uid_t uid;
gid_t gid = crgetgid(cr);
- zfs_acl_ids_t acl_ids;
+ zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
- ASSERT(vap->va_type == VDIR);
+ ASSERT(S_ISDIR(vap->va_mode));
/*
* If we have an ephemeral id, ACL, or XVATTR then
* make sure file system is at proper version
*/
- ksid = crgetsid(cr, KSID_OWNER);
- if (ksid)
- uid = ksid_getid(ksid);
- else
- uid = crgetuid(cr);
- if (zfsvfs->z_use_fuids == B_FALSE &&
- (vsecp || (vap->va_mask & AT_XVATTR) ||
- IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
+ uid = crgetuid(cr);
+ if (zsb->z_use_fuids == B_FALSE &&
+ (vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (EINVAL);
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(dzp);
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
- if (dzp->z_phys->zp_flags & ZFS_XATTR) {
- ZFS_EXIT(zfsvfs);
+ if (dzp->z_pflags & ZFS_XATTR) {
+ ZFS_EXIT(zsb);
return (EINVAL);
}
- if (zfsvfs->z_utf8 && u8_validate(dirname,
+ if (zsb->z_utf8 && u8_validate(dirname,
strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EILSEQ);
}
if (flags & FIGNORECASE)
zf |= ZCILOOK;
- if (vap->va_mask & AT_XVATTR)
+ if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
- crgetuid(cr), cr, vap->va_type)) != 0) {
- ZFS_EXIT(zfsvfs);
+ crgetuid(cr), cr, vap->va_mode)) != 0) {
+ ZFS_EXIT(zsb);
return (error);
}
+ }
+ if ((error = zfs_acl_ids_create(dzp, 0, vap, cr,
+ vsecp, &acl_ids)) != 0) {
+ ZFS_EXIT(zsb);
+ return (error);
+ }
/*
* First make sure the new directory doesn't exist.
+ *
+ * Existence is checked first to make sure we don't return
+ * EACCES instead of EEXIST which can cause some applications
+ * to fail.
*/
top:
- *vpp = NULL;
+ *ipp = NULL;
- if (error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf,
- NULL, NULL)) {
- ZFS_EXIT(zfsvfs);
+ if ((error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf,
+ NULL, NULL))) {
+ zfs_acl_ids_free(&acl_ids);
+ ZFS_EXIT(zsb);
return (error);
}
- if (error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr)) {
+ if ((error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr))) {
+ zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
- if ((error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp,
- &acl_ids)) != 0) {
- zfs_dirent_unlock(dl);
- ZFS_EXIT(zfsvfs);
- return (error);
- }
- if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
+ if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
+ zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EDQUOT);
}
/*
* Add a new entry to the directory.
*/
- tx = dmu_tx_create(zfsvfs->z_os);
+ tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
- fuid_dirtied = zfsvfs->z_fuid_dirty;
+ fuid_dirtied = zsb->z_fuid_dirty;
if (fuid_dirtied)
- zfs_fuid_txhold(zfsvfs, tx);
- if (acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE)
- dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
- 0, SPA_MAXBLOCKSIZE);
+ zfs_fuid_txhold(zsb, tx);
+ if (!zsb->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, TXG_NOWAIT);
if (error) {
- zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
+ zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
/*
* Create new node.
*/
- zfs_mknode(dzp, vap, tx, cr, 0, &zp, 0, &acl_ids);
+ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
if (fuid_dirtied)
- zfs_fuid_sync(zfsvfs, tx);
+ zfs_fuid_sync(zsb, tx);
+
/*
* Now put new name in parent dir.
*/
(void) zfs_link_create(dl, zp, tx, ZNEW);
- *vpp = ZTOV(zp);
+ *ipp = ZTOI(zp);
txtype = zfs_log_create_txtype(Z_DIR, vsecp, vap);
if (flags & FIGNORECASE)
acl_ids.z_fuidp, vap);
zfs_acl_ids_free(&acl_ids);
+
dmu_tx_commit(tx);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ ZFS_EXIT(zsb);
return (0);
}
+EXPORT_SYMBOL(zfs_mkdir);
/*
* Remove a directory subdir entry. If the current working
* directory is the same as the subdir to be removed, the
* remove will fail.
*
- * IN: dvp - vnode of directory to remove from.
+ * IN: dip - inode of directory to remove from.
* name - name of directory to be removed.
- * cwd - vnode of current working directory.
+ * cwd - inode of current working directory.
* cr - credentials of caller.
- * ct - caller context
* flags - case flags
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * dvp - ctime|mtime updated
+ * dip - ctime|mtime updated
*/
/*ARGSUSED*/
-static int
-zfs_rmdir(vnode_t *dvp, char *name, vnode_t *cwd, cred_t *cr,
- caller_context_t *ct, int flags)
+int
+zfs_rmdir(struct inode *dip, char *name, struct inode *cwd, cred_t *cr,
+ int flags)
{
- znode_t *dzp = VTOZ(dvp);
+ znode_t *dzp = ITOZ(dip);
znode_t *zp;
- vnode_t *vp;
- zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
+ struct inode *ip;
+ zfs_sb_t *zsb = ITOZSB(dip);
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
int error;
int zflg = ZEXISTS;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(dzp);
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
/*
* Attempt to lock directory; fail if entry doesn't exist.
*/
- if (error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
- NULL, NULL)) {
- ZFS_EXIT(zfsvfs);
+ if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
+ NULL, NULL))) {
+ ZFS_EXIT(zsb);
return (error);
}
- vp = ZTOV(zp);
+ ip = ZTOI(zp);
- if (error = zfs_zaccess_delete(dzp, zp, cr)) {
+ if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
goto out;
}
- if (vp->v_type != VDIR) {
+ if (!S_ISDIR(ip->i_mode)) {
error = ENOTDIR;
goto out;
}
- if (vp == cwd) {
+ if (ip == cwd) {
error = EINVAL;
goto out;
}
- vnevent_rmdir(vp, dvp, name, ct);
-
/*
* Grab a lock on the directory to make sure that noone is
* trying to add (or lookup) entries while we are removing it.
*/
rw_enter(&zp->z_parent_lock, RW_WRITER);
- tx = dmu_tx_create(zfsvfs->z_os);
+ tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
- dmu_tx_hold_bonus(tx, zp->z_id);
- dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
+ dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, FALSE, NULL);
+ zfs_sa_upgrade_txholds(tx, zp);
+ zfs_sa_upgrade_txholds(tx, dzp);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
rw_exit(&zp->z_parent_lock);
rw_exit(&zp->z_name_lock);
zfs_dirent_unlock(dl);
- VN_RELE(vp);
+ iput(ip);
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
uint64_t txtype = TX_RMDIR;
if (flags & FIGNORECASE)
txtype |= TX_CI;
- zfs_log_remove(zilog, tx, txtype, dzp, name);
+ zfs_log_remove(zilog, tx, txtype, dzp, name, ZFS_NO_OBJECT);
}
dmu_tx_commit(tx);
out:
zfs_dirent_unlock(dl);
- VN_RELE(vp);
+ iput(ip);
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ zfs_inode_update(dzp);
+ zfs_inode_update(zp);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_rmdir);
/*
* Read as many directory entries as will fit into the provided
- * buffer from the given directory cursor position (specified in
- * the uio structure.
+ * dirent buffer from the given directory cursor position.
*
- * IN: vp - vnode of directory to read.
- * uio - structure supplying read location, range info,
- * and return buffer.
- * cr - credentials of caller.
- * ct - caller context
- * flags - case flags
+ * IN: ip - inode of directory to read.
+ * dirent - buffer for directory entries.
*
- * OUT: uio - updated offset and range, buffer filled.
- * eofp - set to true if end-of-file detected.
+ * OUT: dirent - filler buffer of directory entries.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * vp - atime updated
+ * ip - atime updated
*
* Note that the low 4 bits of the cookie returned by zap is always zero.
* This allows us to use the low range for "special" directory entries:
* we use the offset 2 for the '.zfs' directory.
*/
/* ARGSUSED */
-static int
-zfs_readdir(vnode_t *vp, uio_t *uio, cred_t *cr, int *eofp,
- caller_context_t *ct, int flags)
+int
+zfs_readdir(struct inode *ip, void *dirent, filldir_t filldir,
+ loff_t *pos, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- iovec_t *iovp;
- edirent_t *eodp;
- dirent64_t *odp;
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
objset_t *os;
- caddr_t outbuf;
- size_t bufsize;
zap_cursor_t zc;
zap_attribute_t zap;
- uint_t bytes_wanted;
- uint64_t offset; /* must be unsigned; checks for < 1 */
- int local_eof;
int outcount;
int error;
uint8_t prefetch;
- boolean_t check_sysattrs;
+ int done = 0;
+ uint64_t parent;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
- /*
- * If we are not given an eof variable,
- * use a local one.
- */
- if (eofp == NULL)
- eofp = &local_eof;
-
- /*
- * Check for valid iov_len.
- */
- if (uio->uio_iov->iov_len <= 0) {
- ZFS_EXIT(zfsvfs);
- return (EINVAL);
- }
+ if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zsb),
+ &parent, sizeof (parent))) != 0)
+ goto out;
/*
* Quit if directory has been removed (posix)
*/
- if ((*eofp = zp->z_unlinked) != 0) {
- ZFS_EXIT(zfsvfs);
- return (0);
- }
-
error = 0;
- os = zfsvfs->z_os;
- offset = uio->uio_loffset;
+ if (zp->z_unlinked)
+ goto out;
+
+ os = zsb->z_os;
prefetch = zp->z_zn_prefetch;
/*
* Initialize the iterator cursor.
*/
- if (offset <= 3) {
+ if (*pos <= 3) {
/*
* Start iteration from the beginning of the directory.
*/
/*
* The offset is a serialized cursor.
*/
- zap_cursor_init_serialized(&zc, os, zp->z_id, offset);
+ zap_cursor_init_serialized(&zc, os, zp->z_id, *pos);
}
/*
- * Get space to change directory entries into fs independent format.
- */
- iovp = uio->uio_iov;
- bytes_wanted = iovp->iov_len;
- if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1) {
- bufsize = bytes_wanted;
- outbuf = kmem_alloc(bufsize, KM_SLEEP);
- odp = (struct dirent64 *)outbuf;
- } else {
- bufsize = bytes_wanted;
- odp = (struct dirent64 *)iovp->iov_base;
- }
- eodp = (struct edirent *)odp;
-
- /*
- * If this VFS supports the system attribute view interface; and
- * we're looking at an extended attribute directory; and we care
- * about normalization conflicts on this vfs; then we must check
- * for normalization conflicts with the sysattr name space.
- */
- check_sysattrs = vfs_has_feature(vp->v_vfsp, VFSFT_SYSATTR_VIEWS) &&
- (vp->v_flag & V_XATTRDIR) && zfsvfs->z_norm &&
- (flags & V_RDDIR_ENTFLAGS);
-
- /*
* Transform to file-system independent format
*/
outcount = 0;
- while (outcount < bytes_wanted) {
- ino64_t objnum;
- ushort_t reclen;
- off64_t *next;
+ while (!done) {
+ uint64_t objnum;
/*
* Special case `.', `..', and `.zfs'.
*/
- if (offset == 0) {
+ if (*pos == 0) {
(void) strcpy(zap.za_name, ".");
zap.za_normalization_conflict = 0;
objnum = zp->z_id;
- } else if (offset == 1) {
+ } else if (*pos == 1) {
(void) strcpy(zap.za_name, "..");
zap.za_normalization_conflict = 0;
- objnum = zp->z_phys->zp_parent;
- } else if (offset == 2 && zfs_show_ctldir(zp)) {
+ objnum = parent;
+ } else if (*pos == 2 && zfs_show_ctldir(zp)) {
(void) strcpy(zap.za_name, ZFS_CTLDIR_NAME);
zap.za_normalization_conflict = 0;
- objnum = ZFSCTL_INO_ROOT;
- } else {
- /*
- * Grab next entry.
- */
- if (error = zap_cursor_retrieve(&zc, &zap)) {
- if ((*eofp = (error == ENOENT)) != 0)
- break;
- else
- goto update;
- }
-
- if (zap.za_integer_length != 8 ||
- zap.za_num_integers != 1) {
- cmn_err(CE_WARN, "zap_readdir: bad directory "
- "entry, obj = %lld, offset = %lld\n",
- (u_longlong_t)zp->z_id,
- (u_longlong_t)offset);
- error = ENXIO;
- goto update;
- }
-
- objnum = ZFS_DIRENT_OBJ(zap.za_first_integer);
- /*
- * MacOS X can extract the object type here such as:
- * uint8_t type = ZFS_DIRENT_TYPE(zap.za_first_integer);
- */
-
- if (check_sysattrs && !zap.za_normalization_conflict) {
- zap.za_normalization_conflict =
- xattr_sysattr_casechk(zap.za_name);
- }
- }
-
- if (flags & V_RDDIR_ACCFILTER) {
- /*
- * If we have no access at all, don't include
- * this entry in the returned information
- */
- znode_t *ezp;
- if (zfs_zget(zp->z_zfsvfs, objnum, &ezp) != 0)
- goto skip_entry;
- if (!zfs_has_access(ezp, cr)) {
- VN_RELE(ZTOV(ezp));
- goto skip_entry;
- }
- VN_RELE(ZTOV(ezp));
- }
-
- if (flags & V_RDDIR_ENTFLAGS)
- reclen = EDIRENT_RECLEN(strlen(zap.za_name));
- else
- reclen = DIRENT64_RECLEN(strlen(zap.za_name));
-
- /*
- * Will this entry fit in the buffer?
- */
- if (outcount + reclen > bufsize) {
+ objnum = ZFSCTL_INO_ROOT;
+ } else {
/*
- * Did we manage to fit anything in the buffer?
+ * Grab next entry.
*/
- if (!outcount) {
- error = EINVAL;
+ if ((error = zap_cursor_retrieve(&zc, &zap))) {
+ if (error == ENOENT)
+ break;
+ else
+ goto update;
+ }
+
+ if (zap.za_integer_length != 8 ||
+ zap.za_num_integers != 1) {
+ cmn_err(CE_WARN, "zap_readdir: bad directory "
+ "entry, obj = %lld, offset = %lld\n",
+ (u_longlong_t)zp->z_id,
+ (u_longlong_t)*pos);
+ error = ENXIO;
goto update;
}
- break;
+
+ objnum = ZFS_DIRENT_OBJ(zap.za_first_integer);
}
- if (flags & V_RDDIR_ENTFLAGS) {
- /*
- * Add extended flag entry:
- */
- eodp->ed_ino = objnum;
- eodp->ed_reclen = reclen;
- /* NOTE: ed_off is the offset for the *next* entry */
- next = &(eodp->ed_off);
- eodp->ed_eflags = zap.za_normalization_conflict ?
- ED_CASE_CONFLICT : 0;
- (void) strncpy(eodp->ed_name, zap.za_name,
- EDIRENT_NAMELEN(reclen));
- eodp = (edirent_t *)((intptr_t)eodp + reclen);
- } else {
- /*
- * Add normal entry:
- */
- odp->d_ino = objnum;
- odp->d_reclen = reclen;
- /* NOTE: d_off is the offset for the *next* entry */
- next = &(odp->d_off);
- (void) strncpy(odp->d_name, zap.za_name,
- DIRENT64_NAMELEN(reclen));
- odp = (dirent64_t *)((intptr_t)odp + reclen);
+ done = filldir(dirent, zap.za_name, strlen(zap.za_name),
+ zap_cursor_serialize(&zc), objnum, 0);
+ if (done) {
+ break;
}
- outcount += reclen;
-
- ASSERT(outcount <= bufsize);
/* Prefetch znode */
- if (prefetch)
+ if (prefetch) {
dmu_prefetch(os, objnum, 0, 0);
+ }
- skip_entry:
- /*
- * Move to the next entry, fill in the previous offset.
- */
- if (offset > 2 || (offset == 2 && !zfs_show_ctldir(zp))) {
+ if (*pos >= 2) {
zap_cursor_advance(&zc);
- offset = zap_cursor_serialize(&zc);
+ *pos = zap_cursor_serialize(&zc);
} else {
- offset += 1;
+ (*pos)++;
}
- *next = offset;
}
zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */
- if (uio->uio_segflg == UIO_SYSSPACE && uio->uio_iovcnt == 1) {
- iovp->iov_base += outcount;
- iovp->iov_len -= outcount;
- uio->uio_resid -= outcount;
- } else if (error = uiomove(outbuf, (long)outcount, UIO_READ, uio)) {
- /*
- * Reset the pointer.
- */
- offset = uio->uio_loffset;
- }
-
update:
zap_cursor_fini(&zc);
- if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1)
- kmem_free(outbuf, bufsize);
-
if (error == ENOENT)
error = 0;
- ZFS_ACCESSTIME_STAMP(zfsvfs, zp);
+ ZFS_ACCESSTIME_STAMP(zsb, zp);
+ zfs_inode_update(zp);
+
+out:
+ ZFS_EXIT(zsb);
- uio->uio_loffset = offset;
- ZFS_EXIT(zfsvfs);
return (error);
}
+EXPORT_SYMBOL(zfs_readdir);
ulong_t zfs_fsync_sync_cnt = 4;
-static int
-zfs_fsync(vnode_t *vp, int syncflag, cred_t *cr, caller_context_t *ct)
+int
+zfs_fsync(struct inode *ip, int syncflag, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
-
- /*
- * Regardless of whether this is required for standards conformance,
- * this is the logical behavior when fsync() is called on a file with
- * dirty pages. We use B_ASYNC since the ZIL transactions are already
- * going to be pushed out as part of the zil_commit().
- */
- if (vn_has_cached_data(vp) && !(syncflag & FNODSYNC) &&
- (vp->v_type == VREG) && !(IS_SWAPVP(vp)))
- (void) VOP_PUTPAGE(vp, (offset_t)0, (size_t)0, B_ASYNC, cr, ct);
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
(void) tsd_set(zfs_fsyncer_key, (void *)zfs_fsync_sync_cnt);
- ZFS_ENTER(zfsvfs);
- ZFS_VERIFY_ZP(zp);
- zil_commit(zfsvfs->z_log, zp->z_last_itx, zp->z_id);
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_os->os_sync != ZFS_SYNC_DISABLED) {
+ ZFS_ENTER(zsb);
+ ZFS_VERIFY_ZP(zp);
+ zil_commit(zsb->z_log, zp->z_id);
+ ZFS_EXIT(zsb);
+ }
return (0);
}
+EXPORT_SYMBOL(zfs_fsync);
/*
* Get the requested file attributes and place them in the provided
* vattr structure.
*
- * IN: vp - vnode of file.
+ * IN: ip - inode of file.
* vap - va_mask identifies requested attributes.
- * If AT_XVATTR set, then optional attrs are requested
+ * If ATTR_XVATTR set, then optional attrs are requested
* flags - ATTR_NOACLCHECK (CIFS server context)
* cr - credentials of caller.
- * ct - caller context
*
* OUT: vap - attribute values.
*
* RETURN: 0 (always succeeds)
*/
/* ARGSUSED */
-static int
-zfs_getattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr,
- caller_context_t *ct)
+int
+zfs_getattr(struct inode *ip, vattr_t *vap, int flags, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
- znode_phys_t *pzp;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
int error = 0;
uint64_t links;
+ uint64_t mtime[2], ctime[2];
xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */
xoptattr_t *xoap = NULL;
boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
+ sa_bulk_attr_t bulk[2];
+ int count = 0;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
- pzp = zp->z_phys;
+
+ zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid);
+
+ 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);
+
+ if ((error = sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) != 0) {
+ ZFS_EXIT(zsb);
+ return (error);
+ }
/*
* If ACL is trivial don't bother looking for ACE_READ_ATTRIBUTES.
* Also, if we are the owner don't bother, since owner should
* always be allowed to read basic attributes of file.
*/
- if (!(pzp->zp_flags & ZFS_ACL_TRIVIAL) &&
- (pzp->zp_uid != crgetuid(cr))) {
- if (error = zfs_zaccess(zp, ACE_READ_ATTRIBUTES, 0,
- skipaclchk, cr)) {
- ZFS_EXIT(zfsvfs);
+ if (!(zp->z_pflags & ZFS_ACL_TRIVIAL) &&
+ (vap->va_uid != crgetuid(cr))) {
+ if ((error = zfs_zaccess(zp, ACE_READ_ATTRIBUTES, 0,
+ skipaclchk, cr))) {
+ ZFS_EXIT(zsb);
return (error);
}
}
*/
mutex_enter(&zp->z_lock);
- vap->va_type = vp->v_type;
- vap->va_mode = pzp->zp_mode & MODEMASK;
- zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid);
- vap->va_fsid = zp->z_zfsvfs->z_vfs->vfs_dev;
+ vap->va_type = vn_mode_to_vtype(zp->z_mode);
+ vap->va_mode = zp->z_mode;
+ vap->va_fsid = 0;
vap->va_nodeid = zp->z_id;
- if ((vp->v_flag & VROOT) && zfs_show_ctldir(zp))
- links = pzp->zp_links + 1;
+ if ((zp->z_id == zsb->z_root) && zfs_show_ctldir(zp))
+ links = zp->z_links + 1;
else
- links = pzp->zp_links;
- vap->va_nlink = MIN(links, UINT32_MAX); /* nlink_t limit! */
- vap->va_size = pzp->zp_size;
- vap->va_rdev = vp->v_rdev;
- vap->va_seq = zp->z_seq;
+ links = zp->z_links;
+ vap->va_nlink = MIN(links, ZFS_LINK_MAX);
+ vap->va_size = i_size_read(ip);
+ vap->va_rdev = ip->i_rdev;
+ vap->va_seq = ip->i_generation;
/*
* 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 && zfsvfs->z_use_fuids) {
+ if ((xoap = xva_getxoptattr(xvap)) != NULL && zsb->z_use_fuids) {
if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) {
xoap->xoa_archive =
- ((pzp->zp_flags & ZFS_ARCHIVE) != 0);
+ ((zp->z_pflags & ZFS_ARCHIVE) != 0);
XVA_SET_RTN(xvap, XAT_ARCHIVE);
}
if (XVA_ISSET_REQ(xvap, XAT_READONLY)) {
xoap->xoa_readonly =
- ((pzp->zp_flags & ZFS_READONLY) != 0);
+ ((zp->z_pflags & ZFS_READONLY) != 0);
XVA_SET_RTN(xvap, XAT_READONLY);
}
if (XVA_ISSET_REQ(xvap, XAT_SYSTEM)) {
xoap->xoa_system =
- ((pzp->zp_flags & ZFS_SYSTEM) != 0);
+ ((zp->z_pflags & ZFS_SYSTEM) != 0);
XVA_SET_RTN(xvap, XAT_SYSTEM);
}
if (XVA_ISSET_REQ(xvap, XAT_HIDDEN)) {
xoap->xoa_hidden =
- ((pzp->zp_flags & ZFS_HIDDEN) != 0);
+ ((zp->z_pflags & ZFS_HIDDEN) != 0);
XVA_SET_RTN(xvap, XAT_HIDDEN);
}
if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
xoap->xoa_nounlink =
- ((pzp->zp_flags & ZFS_NOUNLINK) != 0);
+ ((zp->z_pflags & ZFS_NOUNLINK) != 0);
XVA_SET_RTN(xvap, XAT_NOUNLINK);
}
if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) {
xoap->xoa_immutable =
- ((pzp->zp_flags & ZFS_IMMUTABLE) != 0);
+ ((zp->z_pflags & ZFS_IMMUTABLE) != 0);
XVA_SET_RTN(xvap, XAT_IMMUTABLE);
}
if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) {
xoap->xoa_appendonly =
- ((pzp->zp_flags & ZFS_APPENDONLY) != 0);
+ ((zp->z_pflags & ZFS_APPENDONLY) != 0);
XVA_SET_RTN(xvap, XAT_APPENDONLY);
}
if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) {
xoap->xoa_nodump =
- ((pzp->zp_flags & ZFS_NODUMP) != 0);
+ ((zp->z_pflags & ZFS_NODUMP) != 0);
XVA_SET_RTN(xvap, XAT_NODUMP);
}
if (XVA_ISSET_REQ(xvap, XAT_OPAQUE)) {
xoap->xoa_opaque =
- ((pzp->zp_flags & ZFS_OPAQUE) != 0);
+ ((zp->z_pflags & ZFS_OPAQUE) != 0);
XVA_SET_RTN(xvap, XAT_OPAQUE);
}
if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
xoap->xoa_av_quarantined =
- ((pzp->zp_flags & ZFS_AV_QUARANTINED) != 0);
+ ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0);
XVA_SET_RTN(xvap, XAT_AV_QUARANTINED);
}
if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
xoap->xoa_av_modified =
- ((pzp->zp_flags & ZFS_AV_MODIFIED) != 0);
+ ((zp->z_pflags & ZFS_AV_MODIFIED) != 0);
XVA_SET_RTN(xvap, XAT_AV_MODIFIED);
}
if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) &&
- vp->v_type == VREG &&
- (pzp->zp_flags & ZFS_BONUS_SCANSTAMP)) {
- size_t len;
- dmu_object_info_t doi;
-
- /*
- * Only VREG files have anti-virus scanstamps, so we
- * won't conflict with symlinks in the bonus buffer.
- */
- dmu_object_info_from_db(zp->z_dbuf, &doi);
- len = sizeof (xoap->xoa_av_scanstamp) +
- sizeof (znode_phys_t);
- if (len <= doi.doi_bonus_size) {
- /*
- * pzp points to the start of the
- * znode_phys_t. pzp + 1 points to the
- * first byte after the znode_phys_t.
- */
- (void) memcpy(xoap->xoa_av_scanstamp,
- pzp + 1,
- sizeof (xoap->xoa_av_scanstamp));
- XVA_SET_RTN(xvap, XAT_AV_SCANSTAMP);
- }
+ S_ISREG(ip->i_mode)) {
+ zfs_sa_get_scanstamp(zp, xvap);
}
if (XVA_ISSET_REQ(xvap, XAT_CREATETIME)) {
- ZFS_TIME_DECODE(&xoap->xoa_createtime, pzp->zp_crtime);
+ uint64_t times[2];
+
+ (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(zsb),
+ times, sizeof (times));
+ ZFS_TIME_DECODE(&xoap->xoa_createtime, times);
XVA_SET_RTN(xvap, XAT_CREATETIME);
}
+
+ if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
+ xoap->xoa_reparse = ((zp->z_pflags & ZFS_REPARSE) != 0);
+ XVA_SET_RTN(xvap, XAT_REPARSE);
+ }
+ if (XVA_ISSET_REQ(xvap, XAT_GEN)) {
+ xoap->xoa_generation = zp->z_gen;
+ XVA_SET_RTN(xvap, XAT_GEN);
+ }
+
+ if (XVA_ISSET_REQ(xvap, XAT_OFFLINE)) {
+ xoap->xoa_offline =
+ ((zp->z_pflags & ZFS_OFFLINE) != 0);
+ XVA_SET_RTN(xvap, XAT_OFFLINE);
+ }
+
+ if (XVA_ISSET_REQ(xvap, XAT_SPARSE)) {
+ xoap->xoa_sparse =
+ ((zp->z_pflags & ZFS_SPARSE) != 0);
+ XVA_SET_RTN(xvap, XAT_SPARSE);
+ }
}
- ZFS_TIME_DECODE(&vap->va_atime, pzp->zp_atime);
- ZFS_TIME_DECODE(&vap->va_mtime, pzp->zp_mtime);
- ZFS_TIME_DECODE(&vap->va_ctime, pzp->zp_ctime);
+ ZFS_TIME_DECODE(&vap->va_atime, zp->z_atime);
+ ZFS_TIME_DECODE(&vap->va_mtime, mtime);
+ ZFS_TIME_DECODE(&vap->va_ctime, ctime);
mutex_exit(&zp->z_lock);
- dmu_object_size_from_db(zp->z_dbuf, &vap->va_blksize, &vap->va_nblocks);
+ sa_object_size(zp->z_sa_hdl, &vap->va_blksize, &vap->va_nblocks);
if (zp->z_blksz == 0) {
/*
* Block size hasn't been set; suggest maximal I/O transfers.
*/
- vap->va_blksize = zfsvfs->z_max_blksz;
+ vap->va_blksize = zsb->z_max_blksz;
}
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (0);
}
+EXPORT_SYMBOL(zfs_getattr);
/*
* Set the file attributes to the values contained in the
* vattr structure.
*
- * IN: vp - vnode of file to be modified.
+ * IN: ip - inode of file to be modified.
* vap - new attribute values.
- * If AT_XVATTR set, then optional attrs are being set
+ * If ATTR_XVATTR set, then optional attrs are being set
* flags - ATTR_UTIME set if non-default time values provided.
* - ATTR_NOACLCHECK (CIFS context only).
* cr - credentials of caller.
- * ct - caller context
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * vp - ctime updated, mtime updated if size changed.
+ * ip - ctime updated, mtime updated if size changed.
*/
/* ARGSUSED */
-static int
-zfs_setattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr,
- caller_context_t *ct)
+int
+zfs_setattr(struct inode *ip, vattr_t *vap, int flags, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- znode_phys_t *pzp;
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
zilog_t *zilog;
dmu_tx_t *tx;
vattr_t oldva;
- xvattr_t tmpxvattr;
+ xvattr_t *tmpxvattr;
uint_t mask = vap->va_mask;
uint_t saved_mask;
int trim_mask = 0;
uint64_t new_mode;
uint64_t new_uid, new_gid;
+ uint64_t xattr_obj;
+ uint64_t mtime[2], ctime[2];
znode_t *attrzp;
int need_policy = FALSE;
- int err;
+ int err, err2;
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 = NULL;
+ zfs_acl_t *aclp;
boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
- boolean_t fuid_dirtied = B_FALSE;
+ boolean_t fuid_dirtied = B_FALSE;
+ sa_bulk_attr_t bulk[7], xattr_bulk[7];
+ int count = 0, xattr_count = 0;
if (mask == 0)
return (0);
- if (mask & AT_NOSET)
- return (EINVAL);
-
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
- pzp = zp->z_phys;
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
/*
* Make sure that if we have ephemeral uid/gid or xvattr specified
* that file system is at proper version level
*/
- if (zfsvfs->z_use_fuids == B_FALSE &&
- (((mask & AT_UID) && IS_EPHEMERAL(vap->va_uid)) ||
- ((mask & AT_GID) && IS_EPHEMERAL(vap->va_gid)) ||
- (mask & AT_XVATTR))) {
- ZFS_EXIT(zfsvfs);
+ if (zsb->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);
return (EINVAL);
}
- if (mask & AT_SIZE && vp->v_type == VDIR) {
- ZFS_EXIT(zfsvfs);
+ if (mask & ATTR_SIZE && S_ISDIR(ip->i_mode)) {
+ ZFS_EXIT(zsb);
return (EISDIR);
}
- if (mask & AT_SIZE && vp->v_type != VREG && vp->v_type != VFIFO) {
- ZFS_EXIT(zfsvfs);
+ if (mask & ATTR_SIZE && !S_ISREG(ip->i_mode) && !S_ISFIFO(ip->i_mode)) {
+ ZFS_EXIT(zsb);
return (EINVAL);
}
*/
xoap = xva_getxoptattr(xvap);
- xva_init(&tmpxvattr);
+ tmpxvattr = kmem_alloc(sizeof(xvattr_t), KM_SLEEP);
+ xva_init(tmpxvattr);
/*
* Immutable files can only alter immutable bit and atime
*/
- if ((pzp->zp_flags & ZFS_IMMUTABLE) &&
- ((mask & (AT_SIZE|AT_UID|AT_GID|AT_MTIME|AT_MODE)) ||
- ((mask & AT_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) {
- ZFS_EXIT(zfsvfs);
- return (EPERM);
+ 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;
+ goto out3;
}
- if ((mask & AT_SIZE) && (pzp->zp_flags & ZFS_READONLY)) {
- ZFS_EXIT(zfsvfs);
- return (EPERM);
+ if ((mask & ATTR_SIZE) && (zp->z_pflags & ZFS_READONLY)) {
+ err = EPERM;
+ goto out3;
}
/*
* handle times greater than 2039. This check should be removed
* once large timestamps are fully supported.
*/
- if (mask & (AT_ATIME | AT_MTIME)) {
- if (((mask & AT_ATIME) && TIMESPEC_OVERFLOW(&vap->va_atime)) ||
- ((mask & AT_MTIME) && TIMESPEC_OVERFLOW(&vap->va_mtime))) {
- ZFS_EXIT(zfsvfs);
- return (EOVERFLOW);
+ if (mask & (ATTR_ATIME | ATTR_MTIME)) {
+ if (((mask & ATTR_ATIME) && TIMESPEC_OVERFLOW(&vap->va_atime)) ||
+ ((mask & ATTR_MTIME) && TIMESPEC_OVERFLOW(&vap->va_mtime))) {
+ err = EOVERFLOW;
+ goto out3;
}
}
top:
attrzp = NULL;
+ aclp = NULL;
- if (zfsvfs->z_vfs->vfs_flag & VFS_RDONLY) {
- ZFS_EXIT(zfsvfs);
- return (EROFS);
+ /* Can this be moved to before the top label? */
+ if (zsb->z_vfs->mnt_flags & MNT_READONLY) {
+ err = EROFS;
+ goto out3;
}
/*
* First validate permissions
*/
- if (mask & AT_SIZE) {
+ if (mask & ATTR_SIZE) {
err = zfs_zaccess(zp, ACE_WRITE_DATA, 0, skipaclchk, cr);
- if (err) {
- ZFS_EXIT(zfsvfs);
- return (err);
- }
+ if (err)
+ goto out3;
+
/*
* XXX - Note, we are not providing any open
* mode flags here (like FNDELAY), so we may
*/
/* XXX - would it be OK to generate a log record here? */
err = zfs_freesp(zp, vap->va_size, 0, 0, FALSE);
- if (err) {
- ZFS_EXIT(zfsvfs);
- return (err);
- }
+ if (err)
+ goto out3;
+
+ /* Careful negative Linux return code here */
+ err = -vmtruncate(ip, vap->va_size);
+ if (err)
+ goto out3;
}
- if (mask & (AT_ATIME|AT_MTIME) ||
- ((mask & AT_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) ||
+ if (mask & (ATTR_ATIME|ATTR_MTIME) ||
+ ((mask & ATTR_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) ||
XVA_ISSET_REQ(xvap, XAT_READONLY) ||
XVA_ISSET_REQ(xvap, XAT_ARCHIVE) ||
+ XVA_ISSET_REQ(xvap, XAT_OFFLINE) ||
+ XVA_ISSET_REQ(xvap, XAT_SPARSE) ||
XVA_ISSET_REQ(xvap, XAT_CREATETIME) ||
- XVA_ISSET_REQ(xvap, XAT_SYSTEM))))
+ XVA_ISSET_REQ(xvap, XAT_SYSTEM)))) {
need_policy = zfs_zaccess(zp, ACE_WRITE_ATTRIBUTES, 0,
skipaclchk, cr);
+ }
- if (mask & (AT_UID|AT_GID)) {
- int idmask = (mask & (AT_UID|AT_GID));
+ if (mask & (ATTR_UID|ATTR_GID)) {
+ int idmask = (mask & (ATTR_UID|ATTR_GID));
int take_owner;
int take_group;
* we may clear S_ISUID/S_ISGID bits.
*/
- if (!(mask & AT_MODE))
- vap->va_mode = pzp->zp_mode;
+ if (!(mask & ATTR_MODE))
+ vap->va_mode = zp->z_mode;
/*
* Take ownership or chgrp to group we are a member of
*/
- take_owner = (mask & AT_UID) && (vap->va_uid == crgetuid(cr));
- take_group = (mask & AT_GID) &&
- zfs_groupmember(zfsvfs, vap->va_gid, cr);
+ take_owner = (mask & ATTR_UID) && (vap->va_uid == crgetuid(cr));
+ take_group = (mask & ATTR_GID) &&
+ zfs_groupmember(zsb, vap->va_gid, cr);
/*
- * If both AT_UID and AT_GID are set then take_owner and
+ * If both ATTR_UID and ATTR_GID are set then take_owner and
* take_group must both be set in order to allow taking
* ownership.
*
*
*/
- if (((idmask == (AT_UID|AT_GID)) && take_owner && take_group) ||
- ((idmask == AT_UID) && take_owner) ||
- ((idmask == AT_GID) && take_group)) {
+ if (((idmask == (ATTR_UID|ATTR_GID)) &&
+ take_owner && take_group) ||
+ ((idmask == ATTR_UID) && take_owner) ||
+ ((idmask == ATTR_GID) && take_group)) {
if (zfs_zaccess(zp, ACE_WRITE_OWNER, 0,
skipaclchk, cr) == 0) {
/*
* Remove setuid/setgid for non-privileged users
*/
- secpolicy_setid_clear(vap, cr);
- trim_mask = (mask & (AT_UID|AT_GID));
+ (void) secpolicy_setid_clear(vap, cr);
+ trim_mask = (mask & (ATTR_UID|ATTR_GID));
} else {
need_policy = TRUE;
}
}
mutex_enter(&zp->z_lock);
- oldva.va_mode = pzp->zp_mode;
+ oldva.va_mode = zp->z_mode;
zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid);
- if (mask & AT_XVATTR) {
+ if (mask & ATTR_XVATTR) {
/*
* Update xvattr mask to include only those attributes
* that are actually changing.
*/
if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) {
if (xoap->xoa_appendonly !=
- ((pzp->zp_flags & ZFS_APPENDONLY) != 0)) {
+ ((zp->z_pflags & ZFS_APPENDONLY) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_APPENDONLY);
- XVA_SET_REQ(&tmpxvattr, XAT_APPENDONLY);
+ XVA_SET_REQ(tmpxvattr, XAT_APPENDONLY);
}
}
if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
if (xoap->xoa_nounlink !=
- ((pzp->zp_flags & ZFS_NOUNLINK) != 0)) {
+ ((zp->z_pflags & ZFS_NOUNLINK) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_NOUNLINK);
- XVA_SET_REQ(&tmpxvattr, XAT_NOUNLINK);
+ XVA_SET_REQ(tmpxvattr, XAT_NOUNLINK);
}
}
if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) {
if (xoap->xoa_immutable !=
- ((pzp->zp_flags & ZFS_IMMUTABLE) != 0)) {
+ ((zp->z_pflags & ZFS_IMMUTABLE) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_IMMUTABLE);
- XVA_SET_REQ(&tmpxvattr, XAT_IMMUTABLE);
+ XVA_SET_REQ(tmpxvattr, XAT_IMMUTABLE);
}
}
if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) {
if (xoap->xoa_nodump !=
- ((pzp->zp_flags & ZFS_NODUMP) != 0)) {
+ ((zp->z_pflags & ZFS_NODUMP) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_NODUMP);
- XVA_SET_REQ(&tmpxvattr, XAT_NODUMP);
+ XVA_SET_REQ(tmpxvattr, XAT_NODUMP);
}
}
if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
if (xoap->xoa_av_modified !=
- ((pzp->zp_flags & ZFS_AV_MODIFIED) != 0)) {
+ ((zp->z_pflags & ZFS_AV_MODIFIED) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_AV_MODIFIED);
- XVA_SET_REQ(&tmpxvattr, XAT_AV_MODIFIED);
+ XVA_SET_REQ(tmpxvattr, XAT_AV_MODIFIED);
}
}
if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
- if ((vp->v_type != VREG &&
+ if ((!S_ISREG(ip->i_mode) &&
xoap->xoa_av_quarantined) ||
xoap->xoa_av_quarantined !=
- ((pzp->zp_flags & ZFS_AV_QUARANTINED) != 0)) {
+ ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_AV_QUARANTINED);
- XVA_SET_REQ(&tmpxvattr, XAT_AV_QUARANTINED);
+ XVA_SET_REQ(tmpxvattr, XAT_AV_QUARANTINED);
}
}
+ if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
+ mutex_exit(&zp->z_lock);
+ err = EPERM;
+ goto out3;
+ }
+
if (need_policy == FALSE &&
(XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) ||
XVA_ISSET_REQ(xvap, XAT_OPAQUE))) {
mutex_exit(&zp->z_lock);
- if (mask & AT_MODE) {
+ if (mask & ATTR_MODE) {
if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) {
- err = secpolicy_setid_setsticky_clear(vp, vap,
+ err = secpolicy_setid_setsticky_clear(ip, vap,
&oldva, cr);
- if (err) {
- ZFS_EXIT(zfsvfs);
- return (err);
- }
- trim_mask |= AT_MODE;
+ if (err)
+ goto out3;
+
+ trim_mask |= ATTR_MODE;
} else {
need_policy = TRUE;
}
saved_mask = vap->va_mask;
vap->va_mask &= ~trim_mask;
}
- err = secpolicy_vnode_setattr(cr, vp, vap, &oldva, flags,
+ err = secpolicy_vnode_setattr(cr, ip, vap, &oldva, flags,
(int (*)(void *, int, cred_t *))zfs_zaccess_unix, zp);
- if (err) {
- ZFS_EXIT(zfsvfs);
- return (err);
- }
+ if (err)
+ goto out3;
if (trim_mask)
vap->va_mask |= saved_mask;
*/
mask = vap->va_mask;
- tx = dmu_tx_create(zfsvfs->z_os);
- dmu_tx_hold_bonus(tx, zp->z_id);
+ if ((mask & (ATTR_UID | ATTR_GID))) {
+ err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
+ &xattr_obj, sizeof (xattr_obj));
+
+ if (err == 0 && xattr_obj) {
+ err = zfs_zget(ZTOZSB(zp), xattr_obj, &attrzp);
+ if (err)
+ goto out2;
+ }
+ if (mask & ATTR_UID) {
+ new_uid = zfs_fuid_create(zsb,
+ (uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
+ if (new_uid != zp->z_uid &&
+ zfs_fuid_overquota(zsb, B_FALSE, new_uid)) {
+ if (attrzp)
+ iput(ZTOI(attrzp));
+ err = EDQUOT;
+ goto out2;
+ }
+ }
- if (mask & AT_MODE) {
- uint64_t pmode = pzp->zp_mode;
+ 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)) {
+ if (attrzp)
+ iput(ZTOI(attrzp));
+ err = EDQUOT;
+ goto out2;
+ }
+ }
+ }
+ tx = dmu_tx_create(zsb->z_os);
+ if (mask & ATTR_MODE) {
+ uint64_t pmode = zp->z_mode;
+ uint64_t acl_obj;
new_mode = (pmode & S_IFMT) | (vap->va_mode & ~S_IFMT);
- if (err = zfs_acl_chmod_setattr(zp, &aclp, new_mode))
- goto out;
- if (pzp->zp_acl.z_acl_extern_obj) {
- /* Are we upgrading ACL from old V0 format to new V1 */
- if (zfsvfs->z_version <= ZPL_VERSION_FUID &&
- pzp->zp_acl.z_acl_version ==
+ zfs_acl_chmod_setattr(zp, &aclp, new_mode);
+
+ mutex_enter(&zp->z_lock);
+ if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) {
+ /*
+ * Are we upgrading ACL from old V0 format
+ * to V1 format?
+ */
+ if (zsb->z_version >= ZPL_VERSION_FUID &&
+ zfs_znode_acl_version(zp) ==
ZFS_ACL_VERSION_INITIAL) {
- dmu_tx_hold_free(tx,
- pzp->zp_acl.z_acl_extern_obj, 0,
+ dmu_tx_hold_free(tx, acl_obj, 0,
DMU_OBJECT_END);
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, aclp->z_acl_bytes);
} else {
- dmu_tx_hold_write(tx,
- pzp->zp_acl.z_acl_extern_obj, 0,
+ dmu_tx_hold_write(tx, acl_obj, 0,
aclp->z_acl_bytes);
}
- } else if (aclp->z_acl_bytes > ZFS_ACE_SPACE) {
+ } else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, aclp->z_acl_bytes);
}
+ 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))
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
+ else
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
}
- if (mask & (AT_UID | AT_GID)) {
- if (pzp->zp_xattr) {
- err = zfs_zget(zp->z_zfsvfs, pzp->zp_xattr, &attrzp);
- if (err)
- goto out;
- dmu_tx_hold_bonus(tx, attrzp->z_id);
- }
- if (mask & AT_UID) {
- new_uid = zfs_fuid_create(zfsvfs,
- (uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
- if (new_uid != pzp->zp_uid &&
- zfs_usergroup_overquota(zfsvfs, B_FALSE, new_uid)) {
- err = EDQUOT;
- goto out;
- }
- }
-
- if (mask & AT_GID) {
- new_gid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_gid,
- cr, ZFS_GROUP, &fuidp);
- if (new_gid != pzp->zp_gid &&
- zfs_usergroup_overquota(zfsvfs, B_TRUE, new_gid)) {
- err = EDQUOT;
- goto out;
- }
- }
- fuid_dirtied = zfsvfs->z_fuid_dirty;
- if (fuid_dirtied) {
- if (zfsvfs->z_fuid_obj == 0) {
- dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
- dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
- FUID_SIZE_ESTIMATE(zfsvfs));
- dmu_tx_hold_zap(tx, MASTER_NODE_OBJ,
- FALSE, NULL);
- } else {
- dmu_tx_hold_bonus(tx, zfsvfs->z_fuid_obj);
- dmu_tx_hold_write(tx, zfsvfs->z_fuid_obj, 0,
- FUID_SIZE_ESTIMATE(zfsvfs));
- }
- }
+ if (attrzp) {
+ dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE);
}
+ fuid_dirtied = zsb->z_fuid_dirty;
+ if (fuid_dirtied)
+ zfs_fuid_txhold(zsb, tx);
+
+ zfs_sa_upgrade_txholds(tx, zp);
+
err = dmu_tx_assign(tx, TXG_NOWAIT);
if (err) {
if (err == ERESTART)
goto out;
}
- dmu_buf_will_dirty(zp->z_dbuf, tx);
-
+ count = 0;
/*
* Set each attribute requested.
* We group settings according to the locks they need to acquire.
* updated as a side-effect of calling this function.
*/
- mutex_enter(&zp->z_lock);
- if (mask & AT_MODE) {
+ if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
mutex_enter(&zp->z_acl_lock);
- zp->z_phys->zp_mode = new_mode;
- err = zfs_aclset_common(zp, aclp, cr, tx);
- ASSERT3U(err, ==, 0);
- mutex_exit(&zp->z_acl_lock);
- }
+ mutex_enter(&zp->z_lock);
- if (attrzp)
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL,
+ &zp->z_pflags, sizeof (zp->z_pflags));
+
+ if (attrzp) {
+ if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
+ 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,
+ sizeof (attrzp->z_pflags));
+ }
+
+ if (mask & (ATTR_UID|ATTR_GID)) {
+
+ if (mask & ATTR_UID) {
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), 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,
+ sizeof (new_uid));
+ attrzp->z_uid = new_uid;
+ }
+ }
- if (mask & AT_UID) {
- pzp->zp_uid = new_uid;
- if (attrzp)
- attrzp->z_phys->zp_uid = new_uid;
+ if (mask & ATTR_GID) {
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb),
+ 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,
+ sizeof (new_gid));
+ attrzp->z_gid = new_gid;
+ }
+ }
+ if (!(mask & ATTR_MODE)) {
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb),
+ NULL, &new_mode, sizeof (new_mode));
+ new_mode = zp->z_mode;
+ }
+ err = zfs_acl_chown_setattr(zp);
+ ASSERT(err == 0);
+ if (attrzp) {
+ err = zfs_acl_chown_setattr(attrzp);
+ ASSERT(err == 0);
+ }
}
- if (mask & AT_GID) {
- pzp->zp_gid = new_gid;
- if (attrzp)
- attrzp->z_phys->zp_gid = new_gid;
+ if (mask & ATTR_MODE) {
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL,
+ &new_mode, sizeof (new_mode));
+ zp->z_mode = new_mode;
+ ASSERT3P(aclp, !=, NULL);
+ err = zfs_aclset_common(zp, aclp, cr, tx);
+ ASSERT3U(err, ==, 0);
+ if (zp->z_acl_cached)
+ zfs_acl_free(zp->z_acl_cached);
+ zp->z_acl_cached = aclp;
+ aclp = NULL;
}
- if (attrzp)
- mutex_exit(&attrzp->z_lock);
- if (mask & AT_ATIME)
- ZFS_TIME_ENCODE(&vap->va_atime, pzp->zp_atime);
+ if (mask & ATTR_ATIME) {
+ ZFS_TIME_ENCODE(&vap->va_atime, zp->z_atime);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL,
+ &zp->z_atime, sizeof (zp->z_atime));
+ }
- if (mask & AT_MTIME)
- ZFS_TIME_ENCODE(&vap->va_mtime, pzp->zp_mtime);
+ if (mask & ATTR_MTIME) {
+ ZFS_TIME_ENCODE(&vap->va_mtime, mtime);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL,
+ mtime, sizeof (mtime));
+ }
/* XXX - shouldn't this be done *before* the ATIME/MTIME checks? */
- if (mask & AT_SIZE)
- zfs_time_stamper_locked(zp, CONTENT_MODIFIED, tx);
- else if (mask != 0)
- zfs_time_stamper_locked(zp, STATE_CHANGED, tx);
+ 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,
+ B_TRUE);
+ } 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,
+ B_TRUE);
+ 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, B_TRUE);
+ }
+ }
/*
* Do this after setting timestamps to prevent timestamp
* update from toggling bit
*/
- if (xoap && (mask & AT_XVATTR)) {
+ if (xoap && (mask & ATTR_XVATTR)) {
/*
* restore trimmed off masks
* so that return masks can be set for caller.
*/
- if (XVA_ISSET_REQ(&tmpxvattr, XAT_APPENDONLY)) {
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_APPENDONLY)) {
XVA_SET_REQ(xvap, XAT_APPENDONLY);
}
- if (XVA_ISSET_REQ(&tmpxvattr, XAT_NOUNLINK)) {
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_NOUNLINK)) {
XVA_SET_REQ(xvap, XAT_NOUNLINK);
}
- if (XVA_ISSET_REQ(&tmpxvattr, XAT_IMMUTABLE)) {
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_IMMUTABLE)) {
XVA_SET_REQ(xvap, XAT_IMMUTABLE);
}
- if (XVA_ISSET_REQ(&tmpxvattr, XAT_NODUMP)) {
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_NODUMP)) {
XVA_SET_REQ(xvap, XAT_NODUMP);
}
- if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_MODIFIED)) {
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_MODIFIED)) {
XVA_SET_REQ(xvap, XAT_AV_MODIFIED);
}
- if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_QUARANTINED)) {
+ if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_QUARANTINED)) {
XVA_SET_REQ(xvap, XAT_AV_QUARANTINED);
}
- if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) {
- size_t len;
- dmu_object_info_t doi;
+ if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
+ ASSERT(S_ISREG(ip->i_mode));
- ASSERT(vp->v_type == VREG);
-
- /* Grow the bonus buffer if necessary. */
- dmu_object_info_from_db(zp->z_dbuf, &doi);
- len = sizeof (xoap->xoa_av_scanstamp) +
- sizeof (znode_phys_t);
- if (len > doi.doi_bonus_size)
- VERIFY(dmu_set_bonus(zp->z_dbuf, len, tx) == 0);
- }
- zfs_xvattr_set(zp, xvap);
+ zfs_xvattr_set(zp, xvap, tx);
}
if (fuid_dirtied)
- zfs_fuid_sync(zfsvfs, tx);
+ zfs_fuid_sync(zsb, tx);
if (mask != 0)
zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp);
mutex_exit(&zp->z_lock);
+ if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
+ mutex_exit(&zp->z_acl_lock);
+ if (attrzp) {
+ if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
+ mutex_exit(&attrzp->z_acl_lock);
+ mutex_exit(&attrzp->z_lock);
+ }
out:
- if (attrzp)
- VN_RELE(ZTOV(attrzp));
+ if (err == 0 && attrzp) {
+ err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk,
+ xattr_count, tx);
+ ASSERT(err2 == 0);
+ }
- if (aclp) {
+ if (attrzp)
+ iput(ZTOI(attrzp));
+ if (aclp)
zfs_acl_free(aclp);
- aclp = NULL;
- }
if (fuidp) {
zfs_fuid_info_free(fuidp);
fuidp = NULL;
}
- if (err)
+ if (err) {
dmu_tx_abort(tx);
- else
+ if (err == ERESTART)
+ goto top;
+ } else {
+ err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
dmu_tx_commit(tx);
+ zfs_inode_update(zp);
+ }
- if (err == ERESTART)
- goto top;
+out2:
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
- ZFS_EXIT(zfsvfs);
+out3:
+ kmem_free(tmpxvattr, sizeof(xvattr_t));
+ ZFS_EXIT(zsb);
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)
- VN_RELE(ZTOV(zl->zl_znode));
+ iput(ZTOI(zl->zl_znode));
rw_exit(zl->zl_rwlock);
*zlpp = zl->zl_next;
kmem_free(zl, sizeof (*zl));
{
zfs_zlock_t *zl;
znode_t *zp = tdzp;
- uint64_t rootid = zp->z_zfsvfs->z_root;
- uint64_t *oidp = &zp->z_id;
+ uint64_t rootid = ZTOZSB(zp)->z_root;
+ uint64_t oidp = zp->z_id;
krwlock_t *rwlp = &szp->z_parent_lock;
krw_t rw = RW_WRITER;
zfs_rename_unlock(&zl);
*zlpp = NULL;
zp = tdzp;
- oidp = &zp->z_id;
+ oidp = zp->z_id;
rwlp = &szp->z_parent_lock;
rw = RW_WRITER;
continue;
zl->zl_next = *zlpp;
*zlpp = zl;
- if (*oidp == szp->z_id) /* We're a descendant of szp */
+ if (oidp == szp->z_id) /* We're a descendant of szp */
return (EINVAL);
- if (*oidp == rootid) /* We've hit the top */
+ if (oidp == rootid) /* We've hit the top */
return (0);
if (rw == RW_READER) { /* i.e. not the first pass */
- int error = zfs_zget(zp->z_zfsvfs, *oidp, &zp);
+ int error = zfs_zget(ZTOZSB(zp), oidp, &zp);
if (error)
return (error);
zl->zl_znode = zp;
}
- oidp = &zp->z_phys->zp_parent;
+ (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(ZTOZSB(zp)),
+ &oidp, sizeof (oidp));
rwlp = &zp->z_parent_lock;
rw = RW_READER;
* Move an entry from the provided source directory to the target
* directory. Change the entry name as indicated.
*
- * IN: sdvp - Source directory containing the "old entry".
+ * IN: sdip - Source directory containing the "old entry".
* snm - Old entry name.
- * tdvp - Target directory to contain the "new entry".
+ * tdip - Target directory to contain the "new entry".
* tnm - New entry name.
* cr - credentials of caller.
- * ct - caller context
* flags - case flags
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * sdvp,tdvp - ctime|mtime updated
+ * sdip,tdip - ctime|mtime updated
*/
/*ARGSUSED*/
-static int
-zfs_rename(vnode_t *sdvp, char *snm, vnode_t *tdvp, char *tnm, cred_t *cr,
- caller_context_t *ct, int flags)
+int
+zfs_rename(struct inode *sdip, char *snm, struct inode *tdip, char *tnm,
+ cred_t *cr, int flags)
{
znode_t *tdzp, *szp, *tzp;
- znode_t *sdzp = VTOZ(sdvp);
- zfsvfs_t *zfsvfs = sdzp->z_zfsvfs;
+ znode_t *sdzp = ITOZ(sdip);
+ zfs_sb_t *zsb = ITOZSB(sdip);
zilog_t *zilog;
- vnode_t *realvp;
zfs_dirlock_t *sdl, *tdl;
dmu_tx_t *tx;
zfs_zlock_t *zl;
int error = 0;
int zflg = 0;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(sdzp);
- zilog = zfsvfs->z_log;
-
- /*
- * Make sure we have the real vp for the target directory.
- */
- if (VOP_REALVP(tdvp, &realvp, ct) == 0)
- tdvp = realvp;
+ zilog = zsb->z_log;
- if (tdvp->v_vfsp != sdvp->v_vfsp) {
- ZFS_EXIT(zfsvfs);
+ if (tdip->i_sb != sdip->i_sb) {
+ ZFS_EXIT(zsb);
return (EXDEV);
}
- tdzp = VTOZ(tdvp);
+ tdzp = ITOZ(tdip);
ZFS_VERIFY_ZP(tdzp);
- if (zfsvfs->z_utf8 && u8_validate(tnm,
+ if (zsb->z_utf8 && u8_validate(tnm,
strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EILSEQ);
}
* by renaming a linked file into/outof an attribute directory.
* See the comment in zfs_link() for why this is considered bad.
*/
- if ((tdzp->z_phys->zp_flags & ZFS_XATTR) !=
- (sdzp->z_phys->zp_flags & ZFS_XATTR)) {
- ZFS_EXIT(zfsvfs);
+ if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) {
+ ZFS_EXIT(zsb);
return (EINVAL);
}
* First compare the two name arguments without
* considering any case folding.
*/
- int nofold = (zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER);
+ int nofold = (zsb->z_norm & ~U8_TEXTPREP_TOUPPER);
cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error);
- ASSERT(error == 0 || !zfsvfs->z_utf8);
+ ASSERT(error == 0 || !zsb->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(zfsvfs);
+ ZFS_EXIT(zsb);
return (0);
}
/*
* is an exact match, we will allow this to proceed as
* a name-change request.
*/
- if ((zfsvfs->z_case == ZFS_CASE_INSENSITIVE ||
- (zfsvfs->z_case == ZFS_CASE_MIXED &&
+ if ((zsb->z_case == ZFS_CASE_INSENSITIVE ||
+ (zsb->z_case == ZFS_CASE_MIXED &&
flags & FIGNORECASE)) &&
- u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST,
+ u8_strcmp(snm, tnm, 0, zsb->z_norm, U8_UNICODE_LATEST,
&error) == 0) {
/*
* case preserving rename request, require exact
}
}
+ /*
+ * If the source and destination directories are the same, we should
+ * grab the z_name_lock of that directory only once.
+ */
+ if (sdzp == tdzp) {
+ zflg |= ZHAVELOCK;
+ rw_enter(&sdzp->z_name_lock, RW_READER);
+ }
+
if (cmp < 0) {
serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp,
ZEXISTS | zflg, NULL, NULL);
if (!terr) {
zfs_dirent_unlock(tdl);
if (tzp)
- VN_RELE(ZTOV(tzp));
+ iput(ZTOI(tzp));
}
+
+ if (sdzp == tdzp)
+ rw_exit(&sdzp->z_name_lock);
+
if (strcmp(snm, "..") == 0)
serr = EINVAL;
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (serr);
}
if (terr) {
zfs_dirent_unlock(sdl);
- VN_RELE(ZTOV(szp));
+ iput(ZTOI(szp));
+
+ if (sdzp == tdzp)
+ rw_exit(&sdzp->z_name_lock);
+
if (strcmp(tnm, "..") == 0)
terr = EINVAL;
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (terr);
}
* done in a single check.
*/
- if (error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr))
+ if ((error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr)))
goto out;
- if (ZTOV(szp)->v_type == VDIR) {
+ if (S_ISDIR(ZTOI(szp)->i_mode)) {
/*
* Check to make sure rename is valid.
* Can't do a move like this: /usr/a/b to /usr/a/b/c/d
*/
- if (error = zfs_rename_lock(szp, tdzp, sdzp, &zl))
+ if ((error = zfs_rename_lock(szp, tdzp, sdzp, &zl)))
goto out;
}
/*
* Source and target must be the same type.
*/
- if (ZTOV(szp)->v_type == VDIR) {
- if (ZTOV(tzp)->v_type != VDIR) {
+ if (S_ISDIR(ZTOI(szp)->i_mode)) {
+ if (!S_ISDIR(ZTOI(tzp)->i_mode)) {
error = ENOTDIR;
goto out;
}
} else {
- if (ZTOV(tzp)->v_type == VDIR) {
+ if (S_ISDIR(ZTOI(tzp)->i_mode)) {
error = EISDIR;
goto out;
}
}
}
- vnevent_rename_src(ZTOV(szp), sdvp, snm, ct);
- if (tzp)
- vnevent_rename_dest(ZTOV(tzp), tdvp, tnm, ct);
-
- /*
- * notify the target directory if it is not the same
- * as source directory.
- */
- if (tdvp != sdvp) {
- vnevent_rename_dest_dir(tdvp, ct);
- }
-
- tx = dmu_tx_create(zfsvfs->z_os);
- dmu_tx_hold_bonus(tx, szp->z_id); /* nlink changes */
- dmu_tx_hold_bonus(tx, sdzp->z_id); /* nlink changes */
+ tx = dmu_tx_create(zsb->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);
dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, tnm);
- if (sdzp != tdzp)
- dmu_tx_hold_bonus(tx, tdzp->z_id); /* nlink changes */
- if (tzp)
- dmu_tx_hold_bonus(tx, tzp->z_id); /* parent changes */
- dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
+ if (sdzp != tdzp) {
+ dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE);
+ zfs_sa_upgrade_txholds(tx, tdzp);
+ }
+ if (tzp) {
+ dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE);
+ zfs_sa_upgrade_txholds(tx, tzp);
+ }
+
+ zfs_sa_upgrade_txholds(tx, szp);
+ dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, FALSE, NULL);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
if (zl != NULL)
zfs_rename_unlock(&zl);
zfs_dirent_unlock(sdl);
zfs_dirent_unlock(tdl);
- VN_RELE(ZTOV(szp));
+
+ if (sdzp == tdzp)
+ rw_exit(&sdzp->z_name_lock);
+
+ iput(ZTOI(szp));
if (tzp)
- VN_RELE(ZTOV(tzp));
+ iput(ZTOI(tzp));
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
if (error == 0) {
error = zfs_link_create(tdl, szp, tx, ZRENAMING);
if (error == 0) {
- szp->z_phys->zp_flags |= ZFS_AV_MODIFIED;
-
- error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL);
- ASSERT(error == 0);
+ szp->z_pflags |= ZFS_AV_MODIFIED;
- zfs_log_rename(zilog, tx,
- TX_RENAME | (flags & FIGNORECASE ? TX_CI : 0),
- sdzp, sdl->dl_name, tdzp, tdl->dl_name, szp);
+ error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zsb),
+ (void *)&szp->z_pflags, sizeof (uint64_t), tx);
+ ASSERT3U(error, ==, 0);
- /* Update path information for the target vnode */
- vn_renamepath(tdvp, ZTOV(szp), tnm, strlen(tnm));
+ error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL);
+ if (error == 0) {
+ zfs_log_rename(zilog, tx, TX_RENAME |
+ (flags & FIGNORECASE ? TX_CI : 0), sdzp,
+ sdl->dl_name, tdzp, tdl->dl_name, szp);
+ } else {
+ /*
+ * At this point, we have successfully created
+ * the target name, but have failed to remove
+ * the source name. Since the create was done
+ * with the ZRENAMING flag, there are
+ * complications; for one, the link count is
+ * wrong. The easiest way to deal with this
+ * is to remove the newly created target, and
+ * return the original error. This must
+ * succeed; fortunately, it is very unlikely to
+ * fail, since we just created it.
+ */
+ VERIFY3U(zfs_link_destroy(tdl, szp, tx,
+ ZRENAMING, NULL), ==, 0);
+ }
}
}
zfs_dirent_unlock(sdl);
zfs_dirent_unlock(tdl);
- VN_RELE(ZTOV(szp));
- if (tzp)
- VN_RELE(ZTOV(tzp));
+ zfs_inode_update(sdzp);
+ if (sdzp == tdzp)
+ rw_exit(&sdzp->z_name_lock);
- ZFS_EXIT(zfsvfs);
+ if (sdzp != tdzp)
+ zfs_inode_update(tdzp);
+
+ zfs_inode_update(szp);
+ iput(ZTOI(szp));
+ if (tzp) {
+ zfs_inode_update(tzp);
+ iput(ZTOI(tzp));
+ }
+
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_rename);
/*
* Insert the indicated symbolic reference entry into the directory.
*
- * IN: dvp - Directory to contain new symbolic link.
+ * IN: dip - Directory to contain new symbolic link.
* link - Name for new symlink entry.
* vap - Attributes of new entry.
* target - Target path of new symlink.
+ *
* cr - credentials of caller.
- * ct - caller context
* flags - case flags
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * dvp - ctime|mtime updated
+ * dip - ctime|mtime updated
*/
/*ARGSUSED*/
-static int
-zfs_symlink(vnode_t *dvp, char *name, vattr_t *vap, char *link, cred_t *cr,
- caller_context_t *ct, int flags)
+int
+zfs_symlink(struct inode *dip, char *name, vattr_t *vap, char *link,
+ struct inode **ipp, cred_t *cr, int flags)
{
- znode_t *zp, *dzp = VTOZ(dvp);
+ znode_t *zp, *dzp = ITOZ(dip);
zfs_dirlock_t *dl;
dmu_tx_t *tx;
- zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
+ zfs_sb_t *zsb = ITOZSB(dip);
zilog_t *zilog;
- int len = strlen(link);
+ uint64_t len = strlen(link);
int error;
int zflg = ZNEW;
zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
+ uint64_t txtype = TX_SYMLINK;
- ASSERT(vap->va_type == VLNK);
+ ASSERT(S_ISLNK(vap->va_mode));
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(dzp);
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
- if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
+ if (zsb->z_utf8 && u8_validate(name, strlen(name),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EILSEQ);
}
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
-top:
- if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) {
- ZFS_EXIT(zfsvfs);
- return (error);
- }
if (len > MAXPATHLEN) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (ENAMETOOLONG);
}
+ if ((error = zfs_acl_ids_create(dzp, 0,
+ vap, cr, NULL, &acl_ids)) != 0) {
+ ZFS_EXIT(zsb);
+ return (error);
+ }
+top:
+ *ipp = NULL;
+
/*
* Attempt to lock directory; fail if entry already exists.
*/
error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL);
if (error) {
- ZFS_EXIT(zfsvfs);
+ zfs_acl_ids_free(&acl_ids);
+ ZFS_EXIT(zsb);
return (error);
}
- VERIFY(0 == zfs_acl_ids_create(dzp, 0, vap, cr, NULL, &acl_ids));
- if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
+ if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
+ return (error);
+ }
+
+ if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
+ zfs_acl_ids_free(&acl_ids);
+ zfs_dirent_unlock(dl);
+ ZFS_EXIT(zsb);
return (EDQUOT);
}
- tx = dmu_tx_create(zfsvfs->z_os);
- fuid_dirtied = zfsvfs->z_fuid_dirty;
+ tx = dmu_tx_create(zsb->z_os);
+ fuid_dirtied = zsb->z_fuid_dirty;
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len));
- dmu_tx_hold_bonus(tx, dzp->z_id);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
- if (acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE)
- dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, SPA_MAXBLOCKSIZE);
+ 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) {
+ dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
+ acl_ids.z_aclp->z_acl_bytes);
+ }
if (fuid_dirtied)
- zfs_fuid_txhold(zfsvfs, tx);
+ zfs_fuid_txhold(zsb, tx);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
- zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
+ zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
- dmu_buf_will_dirty(dzp->z_dbuf, tx);
-
/*
* Create a new object for the symlink.
- * Put the link content into bonus buffer if it will fit;
- * otherwise, store it just like any other file data.
+ * for version 4 ZPL datsets the symlink will be an SA attribute
*/
- if (sizeof (znode_phys_t) + len <= dmu_bonus_max()) {
- zfs_mknode(dzp, vap, tx, cr, 0, &zp, len, &acl_ids);
- if (len != 0)
- bcopy(link, zp->z_phys + 1, len);
- } else {
- dmu_buf_t *dbp;
+ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
- zfs_mknode(dzp, vap, tx, cr, 0, &zp, 0, &acl_ids);
-
- if (fuid_dirtied)
- zfs_fuid_sync(zfsvfs, tx);
- /*
- * Nothing can access the znode yet so no locking needed
- * for growing the znode's blocksize.
- */
- zfs_grow_blocksize(zp, len, tx);
-
- VERIFY(0 == dmu_buf_hold(zfsvfs->z_os,
- zp->z_id, 0, FTAG, &dbp));
- dmu_buf_will_dirty(dbp, tx);
+ if (fuid_dirtied)
+ zfs_fuid_sync(zsb, tx);
- ASSERT3U(len, <=, dbp->db_size);
- bcopy(link, dbp->db_data, len);
- dmu_buf_rele(dbp, FTAG);
- }
- zp->z_phys->zp_size = len;
+ mutex_enter(&zp->z_lock);
+ if (zp->z_is_sa)
+ error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zsb),
+ 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),
+ &zp->z_size, sizeof (zp->z_size), tx);
/*
* Insert the new object into the directory.
*/
(void) zfs_link_create(dl, zp, tx, ZNEW);
- if (error == 0) {
- uint64_t txtype = TX_SYMLINK;
- if (flags & FIGNORECASE)
- txtype |= TX_CI;
- zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);
- }
+
+ 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_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
- VN_RELE(ZTOV(zp));
+ *ipp = ZTOI(zp);
- ZFS_EXIT(zfsvfs);
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_symlink);
/*
* Return, in the buffer contained in the provided uio structure,
- * the symbolic path referred to by vp.
+ * the symbolic path referred to by ip.
*
- * IN: vp - vnode of symbolic link.
- * uoip - structure to contain the link path.
+ * IN: ip - inode of symbolic link
+ * uio - structure to contain the link path.
* cr - credentials of caller.
- * ct - caller context
- *
- * OUT: uio - structure to contain the link path.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * vp - atime updated
+ * ip - atime updated
*/
/* ARGSUSED */
-static int
-zfs_readlink(vnode_t *vp, uio_t *uio, cred_t *cr, caller_context_t *ct)
+int
+zfs_readlink(struct inode *ip, uio_t *uio, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
- size_t bufsz;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
int error;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
- bufsz = (size_t)zp->z_phys->zp_size;
- if (bufsz + sizeof (znode_phys_t) <= zp->z_dbuf->db_size) {
- error = uiomove(zp->z_phys + 1,
- MIN((size_t)bufsz, uio->uio_resid), UIO_READ, uio);
- } else {
- dmu_buf_t *dbp;
- error = dmu_buf_hold(zfsvfs->z_os, zp->z_id, 0, FTAG, &dbp);
- if (error) {
- ZFS_EXIT(zfsvfs);
- return (error);
- }
- error = uiomove(dbp->db_data,
- MIN((size_t)bufsz, uio->uio_resid), UIO_READ, uio);
- dmu_buf_rele(dbp, FTAG);
- }
+ mutex_enter(&zp->z_lock);
+ if (zp->z_is_sa)
+ error = sa_lookup_uio(zp->z_sa_hdl,
+ SA_ZPL_SYMLINK(zsb), uio);
+ else
+ error = zfs_sa_readlink(zp, uio);
+ mutex_exit(&zp->z_lock);
- ZFS_ACCESSTIME_STAMP(zfsvfs, zp);
- ZFS_EXIT(zfsvfs);
+ ZFS_ACCESSTIME_STAMP(zsb, zp);
+ zfs_inode_update(zp);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_readlink);
/*
- * Insert a new entry into directory tdvp referencing svp.
+ * Insert a new entry into directory tdip referencing sip.
*
- * IN: tdvp - Directory to contain new entry.
- * svp - vnode of new entry.
+ * IN: tdip - Directory to contain new entry.
+ * sip - inode of new entry.
* name - name of new entry.
* cr - credentials of caller.
- * ct - caller context
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * tdvp - ctime|mtime updated
- * svp - ctime updated
+ * tdip - ctime|mtime updated
+ * sip - ctime updated
*/
/* ARGSUSED */
-static int
-zfs_link(vnode_t *tdvp, vnode_t *svp, char *name, cred_t *cr,
- caller_context_t *ct, int flags)
+int
+zfs_link(struct inode *tdip, struct inode *sip, char *name, cred_t *cr)
{
- znode_t *dzp = VTOZ(tdvp);
+ znode_t *dzp = ITOZ(tdip);
znode_t *tzp, *szp;
- zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
+ zfs_sb_t *zsb = ITOZSB(tdip);
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
- vnode_t *realvp;
int error;
int zf = ZNEW;
+ uint64_t parent;
uid_t owner;
- ASSERT(tdvp->v_type == VDIR);
+ ASSERT(S_ISDIR(tdip->i_mode));
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(dzp);
- zilog = zfsvfs->z_log;
+ zilog = zsb->z_log;
- if (VOP_REALVP(svp, &realvp, ct) == 0)
- svp = realvp;
+ /*
+ * POSIX dictates that we return EPERM here.
+ * Better choices include ENOTSUP or EISDIR.
+ */
+ if (S_ISDIR(sip->i_mode)) {
+ ZFS_EXIT(zsb);
+ return (EPERM);
+ }
- if (svp->v_vfsp != tdvp->v_vfsp) {
- ZFS_EXIT(zfsvfs);
+ if (sip->i_sb != tdip->i_sb) {
+ ZFS_EXIT(zsb);
return (EXDEV);
}
- szp = VTOZ(svp);
+
+ szp = ITOZ(sip);
ZFS_VERIFY_ZP(szp);
- if (zfsvfs->z_utf8 && u8_validate(name,
+ /* Prevent links to .zfs/shares files */
+
+ if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zsb),
+ &parent, sizeof (uint64_t))) != 0) {
+ ZFS_EXIT(zsb);
+ return (error);
+ }
+ if (parent == zsb->z_shares_dir) {
+ ZFS_EXIT(zsb);
+ return (EPERM);
+ }
+
+ if (zsb->z_utf8 && u8_validate(name,
strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EILSEQ);
}
+#ifdef HAVE_PN_UTILS
if (flags & FIGNORECASE)
zf |= ZCILOOK;
+#endif /* HAVE_PN_UTILS */
-top:
/*
* We do not support links between attributes and non-attributes
* because of the potential security risk of creating links
* into "normal" file space in order to circumvent restrictions
* imposed in attribute space.
*/
- if ((szp->z_phys->zp_flags & ZFS_XATTR) !=
- (dzp->z_phys->zp_flags & ZFS_XATTR)) {
- ZFS_EXIT(zfsvfs);
+ if ((szp->z_pflags & ZFS_XATTR) != (dzp->z_pflags & ZFS_XATTR)) {
+ ZFS_EXIT(zsb);
return (EINVAL);
}
- /*
- * POSIX dictates that we return EPERM here.
- * Better choices include ENOTSUP or EISDIR.
- */
- if (svp->v_type == VDIR) {
- ZFS_EXIT(zfsvfs);
- return (EPERM);
- }
-
- owner = zfs_fuid_map_id(zfsvfs, szp->z_phys->zp_uid, cr, ZFS_OWNER);
- if (owner != crgetuid(cr) &&
- secpolicy_basic_link(cr) != 0) {
- ZFS_EXIT(zfsvfs);
+ owner = zfs_fuid_map_id(zsb, szp->z_uid, cr, ZFS_OWNER);
+ if (owner != crgetuid(cr) && secpolicy_basic_link(cr) != 0) {
+ ZFS_EXIT(zsb);
return (EPERM);
}
- if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) {
- ZFS_EXIT(zfsvfs);
+ if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
+ ZFS_EXIT(zsb);
return (error);
}
+top:
/*
* Attempt to lock directory; fail if entry already exists.
*/
error = zfs_dirent_lock(&dl, dzp, name, &tzp, zf, NULL, NULL);
if (error) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
- tx = dmu_tx_create(zfsvfs->z_os);
- dmu_tx_hold_bonus(tx, szp->z_id);
+ tx = dmu_tx_create(zsb->z_os);
+ dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
+ zfs_sa_upgrade_txholds(tx, szp);
+ zfs_sa_upgrade_txholds(tx, dzp);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
goto top;
}
dmu_tx_abort(tx);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
if (error == 0) {
uint64_t txtype = TX_LINK;
+#ifdef HAVE_PN_UTILS
if (flags & FIGNORECASE)
txtype |= TX_CI;
+#endif /* HAVE_PN_UTILS */
zfs_log_link(zilog, tx, txtype, dzp, szp, name);
}
zfs_dirent_unlock(dl);
- if (error == 0) {
- vnevent_link(svp, ct);
- }
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
- ZFS_EXIT(zfsvfs);
+ zfs_inode_update(dzp);
+ zfs_inode_update(szp);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_link);
+#ifdef HAVE_MMAP
/*
* zfs_null_putapage() is used when the file system has been force
* unmounted. It just drops the pages.
dmu_tx_t *tx;
u_offset_t off, koff;
size_t len, klen;
- uint64_t filesz;
int err;
- filesz = zp->z_phys->zp_size;
off = pp->p_offset;
len = PAGESIZE;
/*
* multiple pages so that we write a full block (thus avoiding
* a read-modify-write).
*/
- if (off < filesz && zp->z_blksz > PAGESIZE) {
+ if (off < zp->z_size && zp->z_blksz > PAGESIZE) {
klen = P2ROUNDUP((ulong_t)zp->z_blksz, PAGESIZE);
koff = ISP2(klen) ? P2ALIGN(off, (u_offset_t)klen) : 0;
- ASSERT(koff <= filesz);
- if (koff + klen > filesz)
- klen = P2ROUNDUP(filesz - koff, (uint64_t)PAGESIZE);
+ ASSERT(koff <= zp->z_size);
+ if (koff + klen > zp->z_size)
+ klen = P2ROUNDUP(zp->z_size - koff, (uint64_t)PAGESIZE);
pp = pvn_write_kluster(vp, pp, &off, &len, koff, klen, flags);
}
ASSERT3U(btop(len), ==, btopr(len));
/*
* Can't push pages past end-of-file.
*/
- if (off >= filesz) {
+ if (off >= zp->z_size) {
/* ignore all pages */
err = 0;
goto out;
- } else if (off + len > filesz) {
- int npages = btopr(filesz - off);
+ } else if (off + len > zp->z_size) {
+ int npages = btopr(zp->z_size - off);
page_t *trunc;
page_list_break(&pp, &trunc, npages);
/* ignore pages past end of file */
if (trunc)
pvn_write_done(trunc, flags);
- len = filesz - off;
+ len = zp->z_size - off;
}
- if (zfs_usergroup_overquota(zfsvfs, B_FALSE, zp->z_phys->zp_uid) ||
- zfs_usergroup_overquota(zfsvfs, B_TRUE, zp->z_phys->zp_gid)) {
+ if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) ||
+ zfs_owner_overquota(zfsvfs, zp, B_TRUE)) {
err = EDQUOT;
goto out;
}
top:
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_write(tx, zp->z_id, off, len);
- dmu_tx_hold_bonus(tx, zp->z_id);
+
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
+ zfs_sa_upgrade_txholds(tx, zp);
err = dmu_tx_assign(tx, TXG_NOWAIT);
if (err != 0) {
if (err == ERESTART) {
}
if (err == 0) {
- zfs_time_stamper(zp, CONTENT_MODIFIED, tx);
+ uint64_t mtime[2], ctime[2];
+ sa_bulk_attr_t bulk[3];
+ int count = 0;
+
+ 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_FLAGS(zfsvfs), NULL,
+ &zp->z_pflags, 8);
+ zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime,
+ B_TRUE);
zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, off, len, 0);
- dmu_tx_commit(tx);
}
+ dmu_tx_commit(tx);
out:
pvn_write_done(pp, (err ? B_ERROR : 0) | flags);
*/
/*ARGSUSED*/
static int
-zfs_putpage(vnode_t *vp, offset_t off, size_t len, int flags, cred_t *cr,
- caller_context_t *ct)
+zfs_putpage(vnode_t *vp, offset_t off, size_t len, int flags, cred_t *cr)
{
znode_t *zp = VTOZ(vp);
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
}
rl = zfs_range_lock(zp, io_off, io_len, RL_WRITER);
- if (off > zp->z_phys->zp_size) {
+ if (off > zp->z_size) {
/* past end of file */
zfs_range_unlock(rl);
ZFS_EXIT(zfsvfs);
return (0);
}
- len = MIN(io_len, P2ROUNDUP(zp->z_phys->zp_size, PAGESIZE) - io_off);
+ len = MIN(io_len, P2ROUNDUP(zp->z_size, PAGESIZE) - io_off);
for (off = io_off; io_off < off + len; io_off += io_len) {
if ((flags & B_INVAL) || ((flags & B_ASYNC) == 0)) {
}
out:
zfs_range_unlock(rl);
- if ((flags & B_ASYNC) == 0)
- zil_commit(zfsvfs->z_log, UINT64_MAX, zp->z_id);
+ if ((flags & B_ASYNC) == 0 || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zfsvfs->z_log, zp->z_id);
ZFS_EXIT(zfsvfs);
return (error);
}
+#endif /* HAVE_MMAP */
/*ARGSUSED*/
void
-zfs_inactive(vnode_t *vp, cred_t *cr, caller_context_t *ct)
+zfs_inactive(struct inode *ip)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
int error;
- rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_READER);
- if (zp->z_dbuf == NULL) {
- /*
- * The fs has been unmounted, or we did a
- * suspend/resume and this file no longer exists.
- */
- if (vn_has_cached_data(vp)) {
- (void) pvn_vplist_dirty(vp, 0, zfs_null_putapage,
- B_INVAL, cr);
- }
+#ifdef HAVE_SNAPSHOT
+ /* Early return for snapshot inode? */
+#endif /* HAVE_SNAPSHOT */
- mutex_enter(&zp->z_lock);
- vp->v_count = 0; /* count arrives as 1 */
- mutex_exit(&zp->z_lock);
- rw_exit(&zfsvfs->z_teardown_inactive_lock);
- zfs_znode_free(zp);
+ rw_enter(&zsb->z_teardown_inactive_lock, RW_READER);
+ if (zp->z_sa_hdl == NULL) {
+ rw_exit(&zsb->z_teardown_inactive_lock);
return;
}
- /*
- * Attempt to push any data in the page cache. If this fails
- * we will get kicked out later in zfs_zinactive().
- */
- if (vn_has_cached_data(vp)) {
- (void) pvn_vplist_dirty(vp, 0, zfs_putapage, B_INVAL|B_ASYNC,
- cr);
- }
-
if (zp->z_atime_dirty && zp->z_unlinked == 0) {
- dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);
+ dmu_tx_t *tx = dmu_tx_create(zsb->z_os);
- dmu_tx_hold_bonus(tx, zp->z_id);
+ dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
+ zfs_sa_upgrade_txholds(tx, zp);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
} else {
- dmu_buf_will_dirty(zp->z_dbuf, tx);
mutex_enter(&zp->z_lock);
+ (void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zsb),
+ (void *)&zp->z_atime, sizeof (zp->z_atime), tx);
zp->z_atime_dirty = 0;
mutex_exit(&zp->z_lock);
dmu_tx_commit(tx);
}
zfs_zinactive(zp);
- rw_exit(&zfsvfs->z_teardown_inactive_lock);
+ rw_exit(&zsb->z_teardown_inactive_lock);
}
+EXPORT_SYMBOL(zfs_inactive);
/*
* Bounds-check the seek operation.
*
- * IN: vp - vnode seeking within
+ * IN: ip - inode seeking within
* ooff - old file offset
* noffp - pointer to new file offset
* ct - caller context
* EINVAL if new offset invalid
*/
/* ARGSUSED */
-static int
-zfs_seek(vnode_t *vp, offset_t ooff, offset_t *noffp,
- caller_context_t *ct)
+int
+zfs_seek(struct inode *ip, offset_t ooff, offset_t *noffp)
{
- if (vp->v_type == VDIR)
+ if (S_ISDIR(ip->i_mode))
return (0);
return ((*noffp < 0 || *noffp > MAXOFFSET_T) ? EINVAL : 0);
}
+EXPORT_SYMBOL(zfs_seek);
+#ifdef HAVE_MMAP
/*
* Pre-filter the generic locking function to trap attempts to place
* a mandatory lock on a memory mapped file.
*/
static int
zfs_frlock(vnode_t *vp, int cmd, flock64_t *bfp, int flag, offset_t offset,
- flk_callback_t *flk_cbp, cred_t *cr, caller_context_t *ct)
+ flk_callback_t *flk_cbp, cred_t *cr)
{
znode_t *zp = VTOZ(vp);
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
- int error;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
* return an error, but we don't worry about races between this
* function and zfs_map().
*/
- if (zp->z_mapcnt > 0 && MANDMODE((mode_t)zp->z_phys->zp_mode)) {
+ if (zp->z_mapcnt > 0 && MANDMODE(zp->z_mode)) {
ZFS_EXIT(zfsvfs);
return (EAGAIN);
}
- error = fs_frlock(vp, cmd, bfp, flag, offset, flk_cbp, cr, ct);
ZFS_EXIT(zfsvfs);
- return (error);
+ return (fs_frlock(vp, cmd, bfp, flag, offset, flk_cbp, cr, ct));
}
/*
static int
zfs_getpage(vnode_t *vp, offset_t off, size_t len, uint_t *protp,
page_t *pl[], size_t plsz, struct seg *seg, caddr_t addr,
- enum seg_rw rw, cred_t *cr, caller_context_t *ct)
+ enum seg_rw rw, cred_t *cr)
{
znode_t *zp = VTOZ(vp);
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
/*ARGSUSED*/
static int
zfs_map(vnode_t *vp, offset_t off, struct as *as, caddr_t *addrp,
- size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr,
- caller_context_t *ct)
+ size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr)
{
znode_t *zp = VTOZ(vp);
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
- if ((prot & PROT_WRITE) &&
- (zp->z_phys->zp_flags & (ZFS_IMMUTABLE | ZFS_READONLY |
- ZFS_APPENDONLY))) {
+ if ((prot & PROT_WRITE) && (zp->z_pflags &
+ (ZFS_IMMUTABLE | ZFS_READONLY | ZFS_APPENDONLY))) {
ZFS_EXIT(zfsvfs);
return (EPERM);
}
if ((prot & (PROT_READ | PROT_EXEC)) &&
- (zp->z_phys->zp_flags & ZFS_AV_QUARANTINED)) {
+ (zp->z_pflags & ZFS_AV_QUARANTINED)) {
ZFS_EXIT(zfsvfs);
return (EACCES);
}
/*
* If file is locked, disallow mapping.
*/
- if (MANDMODE((mode_t)zp->z_phys->zp_mode) && vn_has_flocks(vp)) {
+ if (MANDMODE(zp->z_mode) && vn_has_flocks(vp)) {
ZFS_EXIT(zfsvfs);
return (EAGAIN);
}
/* ARGSUSED */
static int
zfs_addmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
- size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr,
- caller_context_t *ct)
+ size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr)
{
uint64_t pages = btopr(len);
/* ARGSUSED */
static int
zfs_delmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
- size_t len, uint_t prot, uint_t maxprot, uint_t flags, cred_t *cr,
- caller_context_t *ct)
+ size_t len, uint_t prot, uint_t maxprot, uint_t flags, cred_t *cr)
{
uint64_t pages = btopr(len);
return (0);
}
+#endif /* HAVE_MMAP */
+
+/*
+ * convoff - converts the given data (start, whence) to the
+ * given whence.
+ */
+int
+convoff(struct inode *ip, flock64_t *lckdat, int whence, offset_t offset)
+{
+ vattr_t vap;
+ int error;
+
+ if ((lckdat->l_whence == 2) || (whence == 2)) {
+ if ((error = zfs_getattr(ip, &vap, 0, CRED()) != 0))
+ return (error);
+ }
+
+ switch (lckdat->l_whence) {
+ case 1:
+ lckdat->l_start += offset;
+ break;
+ case 2:
+ lckdat->l_start += vap.va_size;
+ /* FALLTHRU */
+ case 0:
+ break;
+ default:
+ return (EINVAL);
+ }
+
+ if (lckdat->l_start < 0)
+ return (EINVAL);
+
+ switch (whence) {
+ case 1:
+ lckdat->l_start -= offset;
+ break;
+ case 2:
+ lckdat->l_start -= vap.va_size;
+ /* FALLTHRU */
+ case 0:
+ break;
+ default:
+ return (EINVAL);
+ }
+
+ lckdat->l_whence = (short)whence;
+ return (0);
+}
/*
* Free or allocate space in a file. Currently, this function only
* misnamed, as its functionality includes the ability to allocate as
* well as free space.
*
- * IN: vp - vnode of file to free data in.
+ * IN: ip - inode of file to free data in.
* cmd - action to take (only F_FREESP supported).
* bfp - section of file to free/alloc.
* flag - current file open mode flags.
* offset - current file offset.
* cr - credentials of caller [UNUSED].
- * ct - caller context.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
- * vp - ctime|mtime updated
+ * ip - ctime|mtime updated
*/
/* ARGSUSED */
-static int
-zfs_space(vnode_t *vp, int cmd, flock64_t *bfp, int flag,
- offset_t offset, cred_t *cr, caller_context_t *ct)
+int
+zfs_space(struct inode *ip, int cmd, flock64_t *bfp, int flag,
+ offset_t offset, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
uint64_t off, len;
int error;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
if (cmd != F_FREESP) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EINVAL);
}
- if (error = convoff(vp, bfp, 0, offset)) {
- ZFS_EXIT(zfsvfs);
+ if ((error = convoff(ip, bfp, 0, offset))) {
+ ZFS_EXIT(zsb);
return (error);
}
if (bfp->l_len < 0) {
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (EINVAL);
}
error = zfs_freesp(zp, off, len, flag, TRUE);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_space);
/*ARGSUSED*/
-static int
-zfs_fid(vnode_t *vp, fid_t *fidp, caller_context_t *ct)
+int
+zfs_fid(struct inode *ip, fid_t *fidp)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
uint32_t gen;
+ uint64_t gen64;
uint64_t object = zp->z_id;
zfid_short_t *zfid;
- int size, i;
+ int size, i, error;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
- gen = (uint32_t)zp->z_gen;
- size = (zfsvfs->z_parent != zfsvfs) ? LONG_FID_LEN : SHORT_FID_LEN;
+ if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zsb),
+ &gen64, sizeof (uint64_t))) != 0) {
+ ZFS_EXIT(zsb);
+ 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(zfsvfs);
+ ZFS_EXIT(zsb);
return (ENOSPC);
}
zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i));
if (size == LONG_FID_LEN) {
- uint64_t objsetid = dmu_objset_id(zfsvfs->z_os);
+ uint64_t objsetid = dmu_objset_id(zsb->z_os);
zfid_long_t *zlfid;
zlfid = (zfid_long_t *)fidp;
zlfid->zf_setgen[i] = 0;
}
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (0);
}
-
-static int
-zfs_pathconf(vnode_t *vp, int cmd, ulong_t *valp, cred_t *cr,
- caller_context_t *ct)
-{
- znode_t *zp, *xzp;
- zfsvfs_t *zfsvfs;
- zfs_dirlock_t *dl;
- int error;
-
- switch (cmd) {
- case _PC_LINK_MAX:
- *valp = ULONG_MAX;
- return (0);
-
- case _PC_FILESIZEBITS:
- *valp = 64;
- return (0);
-
- case _PC_XATTR_EXISTS:
- zp = VTOZ(vp);
- zfsvfs = zp->z_zfsvfs;
- ZFS_ENTER(zfsvfs);
- ZFS_VERIFY_ZP(zp);
- *valp = 0;
- error = zfs_dirent_lock(&dl, zp, "", &xzp,
- ZXATTR | ZEXISTS | ZSHARED, NULL, NULL);
- if (error == 0) {
- zfs_dirent_unlock(dl);
- if (!zfs_dirempty(xzp))
- *valp = 1;
- VN_RELE(ZTOV(xzp));
- } else if (error == ENOENT) {
- /*
- * If there aren't extended attributes, it's the
- * same as having zero of them.
- */
- error = 0;
- }
- ZFS_EXIT(zfsvfs);
- return (error);
-
- case _PC_SATTR_ENABLED:
- case _PC_SATTR_EXISTS:
- *valp = vfs_has_feature(vp->v_vfsp, VFSFT_SYSATTR_VIEWS) &&
- (vp->v_type == VREG || vp->v_type == VDIR);
- return (0);
-
- case _PC_ACCESS_FILTERING:
- *valp = vfs_has_feature(vp->v_vfsp, VFSFT_ACCESS_FILTER) &&
- vp->v_type == VDIR;
- return (0);
-
- case _PC_ACL_ENABLED:
- *valp = _ACL_ACE_ENABLED;
- return (0);
-
- case _PC_MIN_HOLE_SIZE:
- *valp = (ulong_t)SPA_MINBLOCKSIZE;
- return (0);
-
- default:
- return (fs_pathconf(vp, cmd, valp, cr, ct));
- }
-}
+EXPORT_SYMBOL(zfs_fid);
/*ARGSUSED*/
-static int
-zfs_getsecattr(vnode_t *vp, vsecattr_t *vsecp, int flag, cred_t *cr,
- caller_context_t *ct)
+int
+zfs_getsecattr(struct inode *ip, vsecattr_t *vsecp, int flag, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
int error;
boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
error = zfs_getacl(zp, vsecp, skipaclchk, cr);
- ZFS_EXIT(zfsvfs);
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_getsecattr);
/*ARGSUSED*/
-static int
-zfs_setsecattr(vnode_t *vp, vsecattr_t *vsecp, int flag, cred_t *cr,
- caller_context_t *ct)
+int
+zfs_setsecattr(struct inode *ip, vsecattr_t *vsecp, int flag, cred_t *cr)
{
- znode_t *zp = VTOZ(vp);
- zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
int error;
boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
+ zilog_t *zilog = zsb->z_log;
- ZFS_ENTER(zfsvfs);
+ ZFS_ENTER(zsb);
ZFS_VERIFY_ZP(zp);
+
error = zfs_setacl(zp, vsecp, skipaclchk, cr);
- ZFS_EXIT(zfsvfs);
+
+ if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS)
+ zil_commit(zilog, 0);
+
+ ZFS_EXIT(zsb);
return (error);
}
+EXPORT_SYMBOL(zfs_setsecattr);
+#ifdef HAVE_UIO_ZEROCOPY
/*
- * Predeclare these here so that the compiler assumes that
- * this is an "old style" function declaration that does
- * not include arguments => we won't get type mismatch errors
- * in the initializations that follow.
+ * Tunable, both must be a power of 2.
+ *
+ * zcr_blksz_min: the smallest read we may consider to loan out an arcbuf
+ * zcr_blksz_max: if set to less than the file block size, allow loaning out of
+ * an arcbuf for a partial block read
*/
-static int zfs_inval();
-static int zfs_isdir();
+int zcr_blksz_min = (1 << 10); /* 1K */
+int zcr_blksz_max = (1 << 17); /* 128K */
+/*ARGSUSED*/
static int
-zfs_inval()
+zfs_reqzcbuf(struct inode *ip, enum uio_rw ioflag, xuio_t *xuio, cred_t *cr)
{
- return (EINVAL);
+ znode_t *zp = ITOZ(ip);
+ zfs_sb_t *zsb = ITOZSB(ip);
+ int max_blksz = zsb->z_max_blksz;
+ uio_t *uio = &xuio->xu_uio;
+ ssize_t size = uio->uio_resid;
+ offset_t offset = uio->uio_loffset;
+ int blksz;
+ int fullblk, i;
+ arc_buf_t *abuf;
+ ssize_t maxsize;
+ int preamble, postamble;
+
+ if (xuio->xu_type != UIOTYPE_ZEROCOPY)
+ return (EINVAL);
+
+ ZFS_ENTER(zsb);
+ ZFS_VERIFY_ZP(zp);
+ switch (ioflag) {
+ case UIO_WRITE:
+ /*
+ * Loan out an arc_buf for write if write size is bigger than
+ * max_blksz, and the file's block size is also max_blksz.
+ */
+ blksz = max_blksz;
+ if (size < blksz || zp->z_blksz != blksz) {
+ ZFS_EXIT(zsb);
+ return (EINVAL);
+ }
+ /*
+ * Caller requests buffers for write before knowing where the
+ * write offset might be (e.g. NFS TCP write).
+ */
+ if (offset == -1) {
+ preamble = 0;
+ } else {
+ preamble = P2PHASE(offset, blksz);
+ if (preamble) {
+ preamble = blksz - preamble;
+ size -= preamble;
+ }
+ }
+
+ postamble = P2PHASE(size, blksz);
+ size -= postamble;
+
+ fullblk = size / blksz;
+ (void) dmu_xuio_init(xuio,
+ (preamble != 0) + fullblk + (postamble != 0));
+
+ /*
+ * Have to fix iov base/len for partial buffers. They
+ * currently represent full arc_buf's.
+ */
+ if (preamble) {
+ /* data begins in the middle of the arc_buf */
+ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
+ blksz);
+ ASSERT(abuf);
+ (void) dmu_xuio_add(xuio, abuf,
+ blksz - preamble, preamble);
+ }
+
+ for (i = 0; i < fullblk; i++) {
+ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
+ blksz);
+ ASSERT(abuf);
+ (void) dmu_xuio_add(xuio, abuf, 0, blksz);
+ }
+
+ if (postamble) {
+ /* data ends in the middle of the arc_buf */
+ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
+ blksz);
+ ASSERT(abuf);
+ (void) dmu_xuio_add(xuio, abuf, 0, postamble);
+ }
+ break;
+ case UIO_READ:
+ /*
+ * Loan out an arc_buf for read if the read size is larger than
+ * the current file block size. Block alignment is not
+ * considered. Partial arc_buf will be loaned out for read.
+ */
+ blksz = zp->z_blksz;
+ if (blksz < zcr_blksz_min)
+ blksz = zcr_blksz_min;
+ if (blksz > zcr_blksz_max)
+ blksz = zcr_blksz_max;
+ /* avoid potential complexity of dealing with it */
+ if (blksz > max_blksz) {
+ ZFS_EXIT(zsb);
+ return (EINVAL);
+ }
+
+ maxsize = zp->z_size - uio->uio_loffset;
+ if (size > maxsize)
+ size = maxsize;
+
+ if (size < blksz) {
+ ZFS_EXIT(zsb);
+ return (EINVAL);
+ }
+ break;
+ default:
+ ZFS_EXIT(zsb);
+ return (EINVAL);
+ }
+
+ uio->uio_extflg = UIO_XUIO;
+ XUIO_XUZC_RW(xuio) = ioflag;
+ ZFS_EXIT(zsb);
+ return (0);
}
+/*ARGSUSED*/
static int
-zfs_isdir()
+zfs_retzcbuf(struct inode *ip, xuio_t *xuio, cred_t *cr)
{
- return (EISDIR);
-}
-/*
- * Directory vnode operations template
- */
-vnodeops_t *zfs_dvnodeops;
-const fs_operation_def_t zfs_dvnodeops_template[] = {
- VOPNAME_OPEN, { .vop_open = zfs_open },
- VOPNAME_CLOSE, { .vop_close = zfs_close },
- VOPNAME_READ, { .error = zfs_isdir },
- VOPNAME_WRITE, { .error = zfs_isdir },
- VOPNAME_IOCTL, { .vop_ioctl = zfs_ioctl },
- VOPNAME_GETATTR, { .vop_getattr = zfs_getattr },
- VOPNAME_SETATTR, { .vop_setattr = zfs_setattr },
- VOPNAME_ACCESS, { .vop_access = zfs_access },
- VOPNAME_LOOKUP, { .vop_lookup = zfs_lookup },
- VOPNAME_CREATE, { .vop_create = zfs_create },
- VOPNAME_REMOVE, { .vop_remove = zfs_remove },
- VOPNAME_LINK, { .vop_link = zfs_link },
- VOPNAME_RENAME, { .vop_rename = zfs_rename },
- VOPNAME_MKDIR, { .vop_mkdir = zfs_mkdir },
- VOPNAME_RMDIR, { .vop_rmdir = zfs_rmdir },
- VOPNAME_READDIR, { .vop_readdir = zfs_readdir },
- VOPNAME_SYMLINK, { .vop_symlink = zfs_symlink },
- VOPNAME_FSYNC, { .vop_fsync = zfs_fsync },
- VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
- VOPNAME_FID, { .vop_fid = zfs_fid },
- VOPNAME_SEEK, { .vop_seek = zfs_seek },
- VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
- VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
- VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
- VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
- NULL, NULL
-};
-
-/*
- * Regular file vnode operations template
- */
-vnodeops_t *zfs_fvnodeops;
-const fs_operation_def_t zfs_fvnodeops_template[] = {
- VOPNAME_OPEN, { .vop_open = zfs_open },
- VOPNAME_CLOSE, { .vop_close = zfs_close },
- VOPNAME_READ, { .vop_read = zfs_read },
- VOPNAME_WRITE, { .vop_write = zfs_write },
- VOPNAME_IOCTL, { .vop_ioctl = zfs_ioctl },
- VOPNAME_GETATTR, { .vop_getattr = zfs_getattr },
- VOPNAME_SETATTR, { .vop_setattr = zfs_setattr },
- VOPNAME_ACCESS, { .vop_access = zfs_access },
- VOPNAME_LOOKUP, { .vop_lookup = zfs_lookup },
- VOPNAME_RENAME, { .vop_rename = zfs_rename },
- VOPNAME_FSYNC, { .vop_fsync = zfs_fsync },
- VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
- VOPNAME_FID, { .vop_fid = zfs_fid },
- VOPNAME_SEEK, { .vop_seek = zfs_seek },
- VOPNAME_FRLOCK, { .vop_frlock = zfs_frlock },
- VOPNAME_SPACE, { .vop_space = zfs_space },
- VOPNAME_GETPAGE, { .vop_getpage = zfs_getpage },
- VOPNAME_PUTPAGE, { .vop_putpage = zfs_putpage },
- VOPNAME_MAP, { .vop_map = zfs_map },
- VOPNAME_ADDMAP, { .vop_addmap = zfs_addmap },
- VOPNAME_DELMAP, { .vop_delmap = zfs_delmap },
- VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
- VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
- VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
- VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
- NULL, NULL
-};
-
-/*
- * Symbolic link vnode operations template
- */
-vnodeops_t *zfs_symvnodeops;
-const fs_operation_def_t zfs_symvnodeops_template[] = {
- VOPNAME_GETATTR, { .vop_getattr = zfs_getattr },
- VOPNAME_SETATTR, { .vop_setattr = zfs_setattr },
- VOPNAME_ACCESS, { .vop_access = zfs_access },
- VOPNAME_RENAME, { .vop_rename = zfs_rename },
- VOPNAME_READLINK, { .vop_readlink = zfs_readlink },
- VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
- VOPNAME_FID, { .vop_fid = zfs_fid },
- VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
- VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
- NULL, NULL
-};
+ int i;
+ arc_buf_t *abuf;
+ int ioflag = XUIO_XUZC_RW(xuio);
-/*
- * special share hidden files vnode operations template
- */
-vnodeops_t *zfs_sharevnodeops;
-const fs_operation_def_t zfs_sharevnodeops_template[] = {
- VOPNAME_GETATTR, { .vop_getattr = zfs_getattr },
- VOPNAME_ACCESS, { .vop_access = zfs_access },
- VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
- VOPNAME_FID, { .vop_fid = zfs_fid },
- VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
- VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
- VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
- VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
- NULL, NULL
-};
+ ASSERT(xuio->xu_type == UIOTYPE_ZEROCOPY);
-/*
- * Extended attribute directory vnode operations template
- * This template is identical to the directory vnodes
- * operation template except for restricted operations:
- * VOP_MKDIR()
- * VOP_SYMLINK()
- * Note that there are other restrictions embedded in:
- * zfs_create() - restrict type to VREG
- * zfs_link() - no links into/out of attribute space
- * zfs_rename() - no moves into/out of attribute space
- */
-vnodeops_t *zfs_xdvnodeops;
-const fs_operation_def_t zfs_xdvnodeops_template[] = {
- VOPNAME_OPEN, { .vop_open = zfs_open },
- VOPNAME_CLOSE, { .vop_close = zfs_close },
- VOPNAME_IOCTL, { .vop_ioctl = zfs_ioctl },
- VOPNAME_GETATTR, { .vop_getattr = zfs_getattr },
- VOPNAME_SETATTR, { .vop_setattr = zfs_setattr },
- VOPNAME_ACCESS, { .vop_access = zfs_access },
- VOPNAME_LOOKUP, { .vop_lookup = zfs_lookup },
- VOPNAME_CREATE, { .vop_create = zfs_create },
- VOPNAME_REMOVE, { .vop_remove = zfs_remove },
- VOPNAME_LINK, { .vop_link = zfs_link },
- VOPNAME_RENAME, { .vop_rename = zfs_rename },
- VOPNAME_MKDIR, { .error = zfs_inval },
- VOPNAME_RMDIR, { .vop_rmdir = zfs_rmdir },
- VOPNAME_READDIR, { .vop_readdir = zfs_readdir },
- VOPNAME_SYMLINK, { .error = zfs_inval },
- VOPNAME_FSYNC, { .vop_fsync = zfs_fsync },
- VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
- VOPNAME_FID, { .vop_fid = zfs_fid },
- VOPNAME_SEEK, { .vop_seek = zfs_seek },
- VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
- VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
- VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
- VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
- NULL, NULL
-};
+ i = dmu_xuio_cnt(xuio);
+ while (i-- > 0) {
+ abuf = dmu_xuio_arcbuf(xuio, i);
+ /*
+ * if abuf == NULL, it must be a write buffer
+ * that has been returned in zfs_write().
+ */
+ if (abuf)
+ dmu_return_arcbuf(abuf);
+ ASSERT(abuf || ioflag == UIO_WRITE);
+ }
-/*
- * Error vnode operations template
- */
-vnodeops_t *zfs_evnodeops;
-const fs_operation_def_t zfs_evnodeops_template[] = {
- VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
- VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
- NULL, NULL
-};
+ dmu_xuio_fini(xuio);
+ return (0);
+}
+#endif /* HAVE_UIO_ZEROCOPY */