fa9e4066f08beec538e775443c5be79dd423fcabahrens * CDDL HEADER START
fa9e4066f08beec538e775443c5be79dd423fcabahrens * The contents of this file are subject to the terms of the
736b9155b17c80a2b71c813ccfe4524852cc7f6fmarks * Common Development and Distribution License (the "License").
736b9155b17c80a2b71c813ccfe4524852cc7f6fmarks * You may not use this file except in compliance with the License.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
fa9e4066f08beec538e775443c5be79dd423fcabahrens * See the License for the specific language governing permissions
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and limitations under the License.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * When distributing Covered Code, include this CDDL HEADER in each
fa9e4066f08beec538e775443c5be79dd423fcabahrens * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If applicable, add the following below this CDDL HEADER, with the
fa9e4066f08beec538e775443c5be79dd423fcabahrens * fields enclosed by brackets "[]" replaced with your own identifying
fa9e4066f08beec538e775443c5be79dd423fcabahrens * information: Portions Copyright [yyyy] [name of copyright owner]
fa9e4066f08beec538e775443c5be79dd423fcabahrens * CDDL HEADER END
d39ee142a97a7c58f60f7b52c62409f2ff64b234Mark Shellenbaum * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
9a686fbc186e8e2a64e9a5094d44c7d6fa0ea167Paul Dagnelie * Copyright (c) 2012, 2015 by Delphix. All rights reserved.
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
c3d26abc9ee97b4f60233556aadeb57e0bd30bb9Matthew Ahrens * Copyright (c) 2014 Integros [integros.com]
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek * Copyright 2015 Joyent, Inc.
75c7619736838143dd41b3c5c9d7adec4683488fpeteh/* Portions Copyright 2007 Jeremy Teo */
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante/* Portions Copyright 2010 Robert Milkowski */
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Programming rules.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Each vnode op performs some logical unit of work. To do this, the ZPL must
fa9e4066f08beec538e775443c5be79dd423fcabahrens * properly lock its in-core state, create a DMU transaction, do the work,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * record this work in the intent log (ZIL), commit the DMU transaction,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * and wait for the intent log to commit if it is a synchronous operation.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Moreover, the vnode ops must work in both normal and log replay context.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * The ordering of events is important to avoid deadlocks and references
fa9e4066f08beec538e775443c5be79dd423fcabahrens * to freed memory. The example below illustrates the following Big Rules:
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * (1) A check must be made in each zfs thread for a mounted file system.
3cb34c601f3ef3016f638574f5982e80c3735c71ahrens * This is done avoiding races using ZFS_ENTER(zfsvfs).
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * must be checked with ZFS_VERIFY_ZP(zp). Both of these macros
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * can return EIO from the calling function.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * (2) VN_RELE() should always be the last thing except for zil_commit()
b19a79ec1a527828a60c4d325ccd8dcbeb2b2e8bperrin * (if necessary) and ZFS_EXIT(). This is for 3 reasons:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * First, if it's the last reference, the vnode/znode
fa9e4066f08beec538e775443c5be79dd423fcabahrens * can be freed, so the zp may point to freed memory. Second, the last
fa9e4066f08beec538e775443c5be79dd423fcabahrens * reference will call zfs_zinactive(), which may induce a lot of work --
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * pushing cached pages (which acquires range locks) and syncing out
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cached atime changes. Third, zfs_zinactive() may require a new tx,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * which could deadlock the system if you were already holding one.
9d3574bff0b382b89dd380c1e23d742ebc156772Neil Perrin * If you must call VN_RELE() within a tx then use VN_RELE_ASYNC().
7885c754e607231ad1f14c765820d601690b6258perrin * (3) All range locks must be grabbed before calling dmu_tx_assign(),
7885c754e607231ad1f14c765820d601690b6258perrin * as they can span dmu_tx_assign() calls.
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * (4) If ZPL locks are held, pass TXG_NOWAIT as the second argument to
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * dmu_tx_assign(). This is critical because we don't want to block
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * while holding locks.
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * If no ZPL locks are held (aside from ZFS_ENTER()), use TXG_WAIT. This
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * reduces lock contention and CPU usage when we must wait (note that if
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * throughput is constrained by the storage, nearly every transaction
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * must wait).
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * Note, in particular, that if a lock is sometimes acquired before
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * the tx assigns, and sometimes after (e.g. z_lock), then failing
e722410c49fe67cbf0f639cbcc288bd6cbcf7dd1Matthew Ahrens * to use a non-blocking assign can deadlock the system. The scenario:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Thread A has grabbed a lock before calling dmu_tx_assign().
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Thread B is in an already-assigned tx, and blocks for this lock.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open()
fa9e4066f08beec538e775443c5be79dd423fcabahrens * forever, because the previous txg can't quiesce until B's tx commits.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT,
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens * then drop all locks, call dmu_tx_wait(), and try again. On subsequent
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens * calls to dmu_tx_assign(), pass TXG_WAITED rather than TXG_NOWAIT,
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens * to indicate that this operation has already called dmu_tx_wait().
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens * This will ensure that we don't retry forever, waiting a short bit
7885c754e607231ad1f14c765820d601690b6258perrin * (5) If the operation succeeded, generate the intent log entry for it
fa9e4066f08beec538e775443c5be79dd423fcabahrens * before dropping locks. This ensures that the ordering of events
fa9e4066f08beec538e775443c5be79dd423fcabahrens * in the intent log matches the order in which they actually occurred.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * During ZIL replay the zfs_log_* functions will update the sequence
1209a471b5681c43d839d4b890f708f500da7346Neil Perrin * number to indicate the zil transaction has replayed.
7885c754e607231ad1f14c765820d601690b6258perrin * (6) At the end of each vnode op, the DMU tx must always commit,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * regardless of whether there were any errors.
5002558f6bfef3915c7f3b4ecb7c19c7f044bf5bNeil Perrin * (7) After dropping all locks, invoke zil_commit(zilog, foid)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * to ensure that synchronous semantics are provided when necessary.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * In general, this is how things should be ordered in each vnode op:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * ZFS_ENTER(zfsvfs); // exit if unmounted
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_dirent_lock(&dl, ...) // lock directory entry (may VN_HOLD())
fa9e4066f08beec538e775443c5be79dd423fcabahrens * rw_enter(...); // grab any other locks you need
fa9e4066f08beec538e775443c5be79dd423fcabahrens * tx = dmu_tx_create(...); // get DMU tx
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dmu_tx_hold_*(); // hold each object you might modify
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens * error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * if (error) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * rw_exit(...); // drop locks
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_dirent_unlock(dl); // unlock directory entry
fa9e4066f08beec538e775443c5be79dd423fcabahrens * VN_RELE(...); // release held vnodes
1209a471b5681c43d839d4b890f708f500da7346Neil Perrin * if (error == ERESTART) {
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens * waited = B_TRUE;
8a2f1b9190d1dc288470a1fd2776d79ce82cb129ahrens * dmu_tx_wait(tx);
8a2f1b9190d1dc288470a1fd2776d79ce82cb129ahrens * dmu_tx_abort(tx);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * goto top;
8a2f1b9190d1dc288470a1fd2776d79ce82cb129ahrens * dmu_tx_abort(tx); // abort DMU tx
fa9e4066f08beec538e775443c5be79dd423fcabahrens * ZFS_EXIT(zfsvfs); // finished in zfs
fa9e4066f08beec538e775443c5be79dd423fcabahrens * return (error); // really out of space
fa9e4066f08beec538e775443c5be79dd423fcabahrens * error = do_real_work(); // do whatever this VOP does
fa9e4066f08beec538e775443c5be79dd423fcabahrens * if (error == 0)
b19a79ec1a527828a60c4d325ccd8dcbeb2b2e8bperrin * zfs_log_*(...); // on success, make ZIL entry
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dmu_tx_commit(tx); // commit DMU tx -- error or not
fa9e4066f08beec538e775443c5be79dd423fcabahrens * rw_exit(...); // drop locks
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_dirent_unlock(dl); // unlock directory entry
fa9e4066f08beec538e775443c5be79dd423fcabahrens * VN_RELE(...); // release held vnodes
5002558f6bfef3915c7f3b4ecb7c19c7f044bf5bNeil Perrin * zil_commit(zilog, foid); // synchronous when necessary
fa9e4066f08beec538e775443c5be79dd423fcabahrens * ZFS_EXIT(zfsvfs); // finished in zfs
fa9e4066f08beec538e775443c5be79dd423fcabahrens * return (error); // done, report error
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_open(vnode_t **vpp, int flag, cred_t *cr, caller_context_t *ct)
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((flag & FWRITE) && (zp->z_pflags & ZFS_APPENDONLY) &&
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum !(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) {
67bd71c6cc629bab3aa0d595c624a667f1574254perrin /* Keep a count of the synchronous opens in the znode */
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_close(vnode_t *vp, int flag, int count, offset_t offset, cred_t *cr,
ee8143cb5c45d521d6600e82791abc3978b9e9a7Chris Kirby * Clean up any locks held by this process on the vp.
67bd71c6cc629bab3aa0d595c624a667f1574254perrin /* Decrement the synchronous opens in the znode */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum !(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0)
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Lseek support for finding holes (cmd == _FIO_SEEK_HOLE) and
fa9e4066f08beec538e775443c5be79dd423fcabahrens * data (cmd == _FIO_SEEK_DATA). "off" is an in/out parameter.
fa9e4066f08beec538e775443c5be79dd423fcabahrens error = dmu_offset_next(zp->z_zfsvfs->z_os, zp->z_id, hole, &noff);
0fbc0cd0e52a11f6c4397a1714f94412cbf98b60Matthew Ahrens * We could find a hole that begins after the logical end-of-file,
0fbc0cd0e52a11f6c4397a1714f94412cbf98b60Matthew Ahrens * because dmu_offset_next() only works on whole blocks. If the
0fbc0cd0e52a11f6c4397a1714f94412cbf98b60Matthew Ahrens * EOF falls mid-block, then indicate that the "virtual hole"
0fbc0cd0e52a11f6c4397a1714f94412cbf98b60Matthew Ahrens * at the end of the file begins at the logical EOF, rather than
0fbc0cd0e52a11f6c4397a1714f94412cbf98b60Matthew Ahrens * at the end of the last block.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_ioctl(vnode_t *vp, int com, intptr_t data, int flag, cred_t *cred,
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * The following two ioctls are used by bfu. Faking out,
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * necessary to avoid bfu errors.
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (ddi_copyin((void *)data, &off, sizeof (off), flag))
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* offset parameter is in/out */
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (ddi_copyout(&off, (void *)data, sizeof (off), flag))
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman * _FIO_COUNT_FILLED adds a new ioctl command which
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman * exposes the number of filled blocks in a
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman * ZFS object.
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman * Wait for all dirty blocks for this object
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman * to get synced out to disk, and the DMU info
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman error = dmu_object_wait_synced(zfsvfs->z_os, zp->z_id);
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman * Retrieve fill count from DMU object.
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman error = dmu_object_info(zfsvfs->z_os, zp->z_id, &doi);
2bcf0248e992f292c7b814458bcdce2f004925d6Max Grossman if (ddi_copyout(&ndata, (void *)data, sizeof (ndata), flag))
0fab61babcfe4e703c15518d8ef8d07e1a545650Jonathan W Adams * Utility functions to map and unmap a single physical page. These
0fab61babcfe4e703c15518d8ef8d07e1a545650Jonathan W Adams * are used to manage the mappable copies of ZFS file data, and therefore
0fab61babcfe4e703c15518d8ef8d07e1a545650Jonathan W Adams * do not update ref/mod bits.
0fab61babcfe4e703c15518d8ef8d07e1a545650Jonathan W Adams return (ppmapin(pp, PROT_READ | ((rw == S_WRITE) ? PROT_WRITE : 0),
fa9e4066f08beec538e775443c5be79dd423fcabahrens * When a file is memory mapped, we must keep the IO data synchronized
fa9e4066f08beec538e775443c5be79dd423fcabahrens * between the DMU cache and the memory mapped pages. What this means:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * On Write: If we find a memory mapped page, we write to *both*
fa9e4066f08beec538e775443c5be79dd423fcabahrens * the page and the dmu buffer.
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybeeupdate_pages(vnode_t *vp, int64_t start, int len, objset_t *os, uint64_t oid)
fa9e4066f08beec538e775443c5be79dd423fcabahrens for (start &= PAGEMASK; len > 0; start += PAGESIZE) {
7bfdf011e081684f853a3242d0296695110d9d84Neil Perrin (void) dmu_read(os, oid, start+off, nbytes, va+off,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * When a file is memory mapped, we must keep the IO data synchronized
fa9e4066f08beec538e775443c5be79dd423fcabahrens * between the DMU cache and the memory mapped pages. What this means:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * On Read: We "read" preferentially from memory mapped pages,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * else we default from the dmu buffer.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * NOTE: We will always "break up" the IO into PAGESIZE uiomoves when
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * the file is memory mapped.
fa9e4066f08beec538e775443c5be79dd423fcabahrens for (start &= PAGEMASK; len > 0; start += PAGESIZE) {
f8554bb92be6be88a684bea58293b41a6168a538Matthew Ahrens error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billmoffset_t zfs_read_chunk_size = 1024 * 1024; /* Tunable */
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Read bytes from specified file into supplied buffer.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file to be read from.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * uio - structure supplying read location, range info,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and return buffer.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * ioflag - SYNC flags; used to provide FRSYNC semantics.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: uio - updated offset and range, buffer filled.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Side Effects:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - atime updated if byte count > 0
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States xuio_t *xuio = NULL;
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Validate file offset
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Fasttrack empty reads
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * Check for mandatory locks
fa9e4066f08beec538e775443c5be79dd423fcabahrens uio->uio_loffset, uio->uio_resid, uio->uio_fmode, ct)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If we're in FRSYNC mode, sync out this znode before reading it.
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (ioflag & FRSYNC || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * Lock the range against changes.
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin rl = zfs_range_lock(zp, uio->uio_loffset, uio->uio_resid, RL_READER);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If we are reading past end-of-file we can skip
fa9e4066f08beec538e775443c5be79dd423fcabahrens * to the end; but we might still need to set atime.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum n = MIN(uio->uio_resid, zp->z_size - uio->uio_loffset);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if ((uio->uio_extflg == UIO_XUIO) &&
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int nblk;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int blksz = zp->z_blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States uint64_t offset = uio->uio_loffset;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States xuio = (xuio_t *)uio;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if ((ISP2(blksz))) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States nblk = (P2ROUNDUP(offset + n, blksz) - P2ALIGN(offset,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States blksz)) / blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States } else {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(offset + n <= blksz);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States nblk = 1;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (vn_has_cached_data(vp)) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * For simplicity, we always allocate a full buffer
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * even if we only expect to read a portion of a block.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States while (--nblk >= 0) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm while (n > 0) {
f8554bb92be6be88a684bea58293b41a6168a538Matthew Ahrens error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
b87f3af36bb994656da117319f5129ddfd05ed21perrin /* convert checksum errors into IO errors */
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Write the bytes to a file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file to be written to.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * uio - structure supplying write location, range info,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and data buffer.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * ioflag - FAPPEND, FSYNC, and/or FDSYNC. FAPPEND is
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * set if in append mode.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context (NFS/CIFS fem monitor only)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: uio - updated offset and range.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime|mtime updated if byte count > 0
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_write(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States xuio_t *xuio = NULL;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int i_iov = 0;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int iovcnt = uio->uio_iovcnt;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States iovec_t *iovp = uio->uio_iov;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int write_eof;
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Fasttrack empty write
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (n == 0)
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * In a case vp->v_vfsp != zp->z_zfsvfs->z_vfs (e.g. snapshots) our
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * callers might not be able to detect properly that we are read-only,
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * so check it explicitly here.
c09193bfec93d8f7a15edeb9b043ede9dc5d82f0marks * If immutable or not appending then return EPERM
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY)) ||
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum ((zp->z_pflags & ZFS_APPENDONLY) && !(ioflag & FAPPEND) &&
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * Validate file offset
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum woff = ioflag & FAPPEND ? zp->z_size : uio->uio_loffset;
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * Check for mandatory locks before calling zfs_range_lock()
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * in order to prevent a deadlock with locks set via fcntl().
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell (error = chklock(vp, FWRITE, woff, n, uio->uio_fmode, ct)) != 0) {
c5c6ffa0498b9c8555798756141b4a3061a138c1maybee * Pre-fault the pages to ensure slow (eg NFS) pages
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * don't hold up txg.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * Skip this if uio contains loaned arc_buf.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if ((uio->uio_extflg == UIO_XUIO) &&
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY))
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States xuio = (xuio_t *)uio;
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If in append mode, set the io offset pointer to eof.
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * Obtain an appending range lock to guarantee file append
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * semantics. We reset the write offset once we have the lock.
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * We overlocked the file because this write will cause
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * the file block size to increase.
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * Note that zp_size cannot change with this lock held.
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * Note that if the file block size will change as a result of
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * this write, then this range lock will lock the entire file
41865f27c2601ae8f94e790349e5f67ae3c359acWilliam Gorrell * so that we can re-write the block safely.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States /* Will this write extend the file length? */
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * Write the file in reasonable size chunks. Each chunk is written
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * in a separate transaction; this keeps the intent log records small
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * and allows us to do more fine-grained space accounting.
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm while (n > 0) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) ||
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (xuio && abuf == NULL) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(i_iov < iovcnt);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States aiov = &iovp[i_iov];
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States abuf = dmu_xuio_arcbuf(xuio, i_iov);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States dmu_xuio_clear(xuio, i_iov);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States DTRACE_PROBE3(zfs_cp_write, int, i_iov,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States iovec_t *, aiov, arc_buf_t *, abuf);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT((aiov->iov_base == abuf->b_data) ||
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ((char *)aiov->iov_base - (char *)abuf->b_data +
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States aiov->iov_len == arc_buf_size(abuf)));
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States i_iov++;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States } else if (abuf == NULL && n >= max_blksz &&
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * This write covers a full block. "Borrow" a buffer
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * from the dmu so that we can fill it before we enter
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * a transaction. This avoids the possibility of
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * holding up the transaction if the data copy hangs
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * up on a pagefault (e.g., from an NFS server mapping).
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
2fdbea25c2ba89186b8a6b7c6840ebc9f4dff245Aleksandr Guzovskiy ASSERT(arc_buf_size(abuf) == max_blksz);
2fdbea25c2ba89186b8a6b7c6840ebc9f4dff245Aleksandr Guzovskiy if (error = uiocopy(abuf->b_data, max_blksz,
2fdbea25c2ba89186b8a6b7c6840ebc9f4dff245Aleksandr Guzovskiy * Start a transaction.
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm dmu_tx_hold_write(tx, zp->z_id, woff, MIN(n, max_blksz));
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * If zfs_range_lock() over-locked we grow the blocksize
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * and then reduce the lock range. This will only happen
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * on the first iteration since zfs_range_reduce() will
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * shrink down r_len to the appropriate size.
b515258426fed6c7311fd3f1dea697cfbd4085c6Matthew Ahrens * File's blocksize is already larger than the
b515258426fed6c7311fd3f1dea697cfbd4085c6Matthew Ahrens * "recordsize" property. Only let it grow to
b515258426fed6c7311fd3f1dea697cfbd4085c6Matthew Ahrens * the next power of 2.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * XXX - should we really limit each write to z_max_blksz?
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Perhaps we should use SPA_MAXBLOCKSIZE chunks?
fa9e4066f08beec538e775443c5be79dd423fcabahrens nbytes = MIN(n, max_blksz - P2PHASE(woff, max_blksz));
94d1a2100edbb6781ea1c047a6334bb3f15640f5Tim Haley error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl),
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(xuio == NULL || tx_bytes == aiov->iov_len);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * If this is not a full block write, but we are
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * extending the file past EOF and this data starts
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * block-aligned, use assign_arcbuf(). Otherwise,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * write via dmu_write().
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (tx_bytes < max_blksz && (!write_eof ||
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States aiov->iov_base != abuf->b_data)) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(xuio);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States dmu_write(zfsvfs->z_os, zp->z_id, woff,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States aiov->iov_len, aiov->iov_base, tx);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States dmu_return_arcbuf(abuf);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States xuio_stat_wbuf_copied();
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States } else {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(xuio || tx_bytes == max_blksz);
2fdbea25c2ba89186b8a6b7c6840ebc9f4dff245Aleksandr Guzovskiy if (tx_bytes && vn_has_cached_data(vp)) {
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * If we made no progress, we're done. If we made even
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * partial progress, update the znode and ZIL accordingly.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void *)&zp->z_size, sizeof (uint64_t), tx);
169cdae232f15e542d6af0a9ce30c3f84222bc0fmarks * Clear Set-UID/Set-GID bits on successful write if not
169cdae232f15e542d6af0a9ce30c3f84222bc0fmarks * privileged and at least one of the excute bits is set.
169cdae232f15e542d6af0a9ce30c3f84222bc0fmarks * It would be nice to to this after all writes have
169cdae232f15e542d6af0a9ce30c3f84222bc0fmarks * been done, but that would still expose the ISUID/ISGID
169cdae232f15e542d6af0a9ce30c3f84222bc0fmarks * to another app after the partial write is committed.
f1696b231bb1209e1cea95b7ff29b8c60493daa0Mark Shellenbaum * Note: we don't call zfs_fuid_map_id() here because
f1696b231bb1209e1cea95b7ff29b8c60493daa0Mark Shellenbaum * user 0 is not an ephemeral uid.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((zp->z_mode & (S_IXUSR | (S_IXUSR >> 3) |
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (zp->z_mode & S_ISUID) != 0 && zp->z_uid == 0) != 0) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_update(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime,
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * Update the file size (zp_size) if it has changed;
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm * account for possible concurrent updates.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum while ((end_size = zp->z_size) < uio->uio_loffset) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) atomic_cas_64(&zp->z_size, end_size,
c0e50c98ecbf7fb0783a0091a1077922afe09138Neil Perrin * If we are replaying and eof is non zero then force
c0e50c98ecbf7fb0783a0091a1077922afe09138Neil Perrin * the file size to the specified eof. Note, there's no
c0e50c98ecbf7fb0783a0091a1077922afe09138Neil Perrin * concurrency during replay.
c0e50c98ecbf7fb0783a0091a1077922afe09138Neil Perrin if (zfsvfs->z_replay && zfsvfs->z_replay_eof != 0)
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
feb08c6bb47ceffa5bb7a70f2c0549113f40fcc5billm zfs_log_write(zilog, tx, TX_WRITE, zp, woff, tx_bytes, ioflag);
ff8669478b050de13ef76d09f743b43bd3c04fe5Sanjeev Bagewadi if (!xuio && n > 0)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If we're in replay mode, or we made no progress, return error.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Otherwise, it's at least a partial write, so it's successful.
1209a471b5681c43d839d4b890f708f500da7346Neil Perrin if (zfsvfs->z_replay || uio->uio_resid == start_resid) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
9d3574bff0b382b89dd380c1e23d742ebc156772Neil Perrin * Release the vnode asynchronously as we currently have the
9d3574bff0b382b89dd380c1e23d742ebc156772Neil Perrin * txg stopped from syncing.
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick VN_RELE_ASYNC(ZTOV(zp), dsl_pool_vnrele_taskq(dmu_objset_pool(os)));
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Get data to generate a TX_WRITE intent log record.
c5c6ffa0498b9c8555798756141b4a3061a138c1maybeezfs_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio)
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * Nothing to do if the file has been removed
9d3574bff0b382b89dd380c1e23d742ebc156772Neil Perrin * Release the vnode asynchronously as we currently have the
9d3574bff0b382b89dd380c1e23d742ebc156772Neil Perrin * txg stopped from syncing.
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Write records come in two flavors: immediate and indirect.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * For small writes it's cheaper to store the data with the
fa9e4066f08beec538e775443c5be79dd423fcabahrens * log record (immediate); for large writes it's cheaper to
fa9e4066f08beec538e775443c5be79dd423fcabahrens * sync the data and get a pointer to it (indirect) so that
fa9e4066f08beec538e775443c5be79dd423fcabahrens * we don't have to write the data twice.
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick zgd->zgd_rl = zfs_range_lock(zp, offset, size, RL_READER);
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin /* test for truncation needs to be done while range locked */
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick error = dmu_read(os, object, offset, size, buf,
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin } else { /* indirect write */
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * Have to lock the whole block to ensure when it's
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * written out and it's checksum is being calculated
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * that no one can change the data. We need to re-check
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin * blocksize after we get the lock in case it's changed!
dfe73b3d6f9191b75fe71a92c8854f83c6d16a63Jeff Bonwick blkoff = ISP2(size) ? P2PHASE(offset, size) : offset;
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin /* test for truncation needs to be done while range locked */
47cb52daa729f19e298c85a84e8df069365c5232Jeff Bonwick error = dmu_buf_hold(os, object, offset, zgd, &db,
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick * On success, we need to wait for the write I/O
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick * initiated by dmu_sync() to complete before we can
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick * release this dbuf. We will finish everything up
b24ab6762772a3f6a89393947930c7fa61306783Jeff Bonwick * in the zfs_get_done() callback.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/*ARGSUSED*/
d47621a49c68c359358f6630aa45cc320762f51fTim Haley * If vnode is for a device return a specfs vnode instead.
d47621a49c68c359358f6630aa45cc320762f51fTim Haley svp = specvp(*vpp, (*vpp)->v_rdev, (*vpp)->v_type, cr);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Lookup an entry in a directory, or an extended attribute directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If it exists, return a held vnode reference for it.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: dvp - vnode of directory to search.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * nm - name of entry to lookup.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * pnp - full pathname to lookup [UNUSED].
fa9e4066f08beec538e775443c5be79dd423fcabahrens * flags - LOOKUP_XATTR set if looking for an attribute.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * rdir - root directory vnode [UNUSED].
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * direntflags - directory lookup flags
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * realpnp - returned pathname.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: vpp - vnode of located entry, NULL if not found.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_lookup(vnode_t *dvp, char *nm, vnode_t **vpp, struct pathname *pnp,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw int flags, vnode_t *rdir, cred_t *cr, caller_context_t *ct,
d47621a49c68c359358f6630aa45cc320762f51fTim Haley /* fast path */
d47621a49c68c359358f6630aa45cc320762f51fTim Haley if (nm[0] == 0 || (nm[0] == '.' && nm[1] == '\0')) {
d47621a49c68c359358f6630aa45cc320762f51fTim Haley DTRACE_PROBE2(zfs__fastpath__lookup__miss, vnode_t *, dvp, char *, nm);
7b55fa8ea6046becb3b72f8886a503979c322084ck * If the xattr property is off, refuse the lookup request.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * We don't allow recursive attributes..
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Maybe someday we will.
3f063a9d046be1602e680274441fdc168bba9f61ck if (error = zfs_get_xattrdir(VTOZ(dvp), vpp, cr, flags)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Do we have permission to get into attribute directory?
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Check accessibility of directory.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr)) {
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw error = zfs_dirlook(zdp, nm, vpp, flags, direntflags, realpnp);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Attempt to create a new entry in a directory. If the entry
fa9e4066f08beec538e775443c5be79dd423fcabahrens * already exists, truncate the file if permissible, else return
fa9e4066f08beec538e775443c5be79dd423fcabahrens * an error. Return the vp of the created or trunc'd file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: dvp - vnode of directory to put new file entry in.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * name - name of new file entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vap - attributes of new file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * excl - flag indicating exclusive or non-exclusive mode.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * mode - mode to open file with.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * flag - large file flag [UNUSED].
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens * vsecp - ACL to be set
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: vpp - vnode of created or trunc'd entry.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dvp - ctime|mtime updated if new entry created
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime|mtime always, atime if new
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_create(vnode_t *dvp, char *name, vattr_t *vap, vcexcl_t excl,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw int mode, vnode_t **vpp, cred_t *cr, int flag, caller_context_t *ct,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If we have an ephemeral id, ACL, or XVATTR then
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * make sure file system is at proper version
fa9e4066f08beec538e775443c5be79dd423fcabahrens if ((vap->va_mode & VSVTX) && secpolicy_vnode_stky_modify(cr))
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Null component name refers to the directory itself.
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* possible VN_HOLD(zp) */
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Create a new file object and update the directory
fa9e4066f08beec538e775443c5be79dd423fcabahrens * to reference it.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * We only support the creation of regular files in
fa9e4066f08beec538e775443c5be79dd423fcabahrens * extended attribute directories.
c8c24165eb6f3690a1b5b8877619e069efe5199bMark Shellenbaum if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
148434217c040ea38dc844384f6ba68d9b325906Matthew Ahrens if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * A directory entry already exists for this name.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Can't truncate an existing file if in exclusive mode.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Can't open a directory for writing.
fa9e4066f08beec538e775443c5be79dd423fcabahrens if ((ZTOV(zp)->v_type == VDIR) && (mode & S_IWRITE)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Verify requested access to file.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (mode && (error = zfs_zaccess_rwx(zp, mode, aflags, cr))) {
5730cc9a43c5f11a472d7536ed81facfd10f1e2emaybee * Truncate regular files if requested.
cdb0ab79ea1af7b8fc339a04d4bf7426dc77ec4emaybee /* we can't hold any locks when calling zfs_freesp() */
df2381bfa5cf7dd654bcf30b2f5af53f34f3043apraks if (error == 0) {
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Remove an entry from a directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: dvp - vnode of directory to remove entry from.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * name - name of entry to remove.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * flags - case flags
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dvp - ctime|mtime
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime (if nlink > 0)
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_remove(vnode_t *dvp, char *name, cred_t *cr, caller_context_t *ct,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Attempt to lock directory; fail if entry doesn't exist.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Need to use rmdir for removing directories.
fa9e4066f08beec538e775443c5be79dd423fcabahrens may_delete_now = vp->v_count == 1 && !vn_has_cached_data(vp);
893a6d32980d24be1349478f44169009d4801c25ahrens * We may delete the znode now, or we may put it in the unlinked set;
fa9e4066f08beec538e775443c5be79dd423fcabahrens * it depends on whether we're the last link, and on whether there are
fa9e4066f08beec538e775443c5be79dd423fcabahrens * other holds on the vnode. So we dmu_tx_hold() the right things to
fa9e4066f08beec538e775443c5be79dd423fcabahrens * allow for either case.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zp->z_size > zp->z_blksz * DMU_MAX_DELETEBLKCNT;
cdb0ab79ea1af7b8fc339a04d4bf7426dc77ec4emaybee /* if the file is too big, only hold_free a token amount */
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* are there any extended attributes? */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
1412a1a223b7a94990edf5114c108b0a29c3f881Mark Shellenbaum if ((acl_obj = zfs_external_acl(zp)) != 0 && may_delete_now)
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* charge as an update -- would be nice not to charge at all */
893a6d32980d24be1349478f44169009d4801c25ahrens dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
6575bca01367958c7237253d88e5fa9ef0b1650aSimon Klinkert * Mark this transaction as typically resulting in a net free of space
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Remove the directory entry.
1412a1a223b7a94990edf5114c108b0a29c3f881Mark Shellenbaum * Hold z_lock so that we can make sure that the ACL obj
1412a1a223b7a94990edf5114c108b0a29c3f881Mark Shellenbaum * hasn't changed. Could have been deleted due to
1412a1a223b7a94990edf5114c108b0a29c3f881Mark Shellenbaum * zfs_sa_upgrade().
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum &xattr_obj_unlinked, sizeof (xattr_obj_unlinked));
1412a1a223b7a94990edf5114c108b0a29c3f881Mark Shellenbaum xattr_obj == xattr_obj_unlinked && zfs_external_acl(zp) ==
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
893a6d32980d24be1349478f44169009d4801c25ahrens } else if (unlinked) {
51bd2f974defeffad92baed420db2d05d55e1481Neil Perrin zfs_log_remove(zilog, tx, txtype, dzp, name, obj);
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Create a new directory and insert it into dvp using the name
fa9e4066f08beec538e775443c5be79dd423fcabahrens * provided. Return a pointer to the inserted directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: dvp - vnode of directory to add subdir to.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dirname - name of new directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vap - attributes of new directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * flags - case flags
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * vsecp - ACL to be set
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: vpp - vnode of created directory.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dvp - ctime|mtime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime|mtime|atime updated
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_mkdir(vnode_t *dvp, char *dirname, vattr_t *vap, vnode_t **vpp, cred_t *cr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If we have an ephemeral id, ACL, or XVATTR then
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * make sure file system is at proper version
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
c8c24165eb6f3690a1b5b8877619e069efe5199bMark Shellenbaum if ((error = zfs_acl_ids_create(dzp, 0, vap, cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * First make sure the new directory doesn't exist.
c8c24165eb6f3690a1b5b8877619e069efe5199bMark Shellenbaum * Existence is checked first to make sure we don't return
c8c24165eb6f3690a1b5b8877619e069efe5199bMark Shellenbaum * EACCES instead of EEXIST which can cause some applications
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr)) {
148434217c040ea38dc844384f6ba68d9b325906Matthew Ahrens if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Add a new entry to the directory.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Create new node.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Now put new name in parent dir.
89459e17032b6bb1d59eebd2b7c0d06859d4657cMark Shellenbaum zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, vsecp,
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Remove a directory subdir entry. If the current working
fa9e4066f08beec538e775443c5be79dd423fcabahrens * directory is the same as the subdir to be removed, the
fa9e4066f08beec538e775443c5be79dd423fcabahrens * remove will fail.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: dvp - vnode of directory to remove from.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * name - name of directory to be removed.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cwd - vnode of current working directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * flags - case flags
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dvp - ctime|mtime updated
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_rmdir(vnode_t *dvp, char *name, vnode_t *cwd, cred_t *cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Attempt to lock directory; fail if entry doesn't exist.
af2c4821c0a23e873f2a63bca4145080aa2183e3maybee * Grab a lock on the directory to make sure that noone is
af2c4821c0a23e873f2a63bca4145080aa2183e3maybee * trying to add (or lookup) entries while we are removing it.
af2c4821c0a23e873f2a63bca4145080aa2183e3maybee * Grab a lock on the parent pointer to make sure we play well
fa9e4066f08beec538e775443c5be79dd423fcabahrens * with the treewalk and directory rename code.
893a6d32980d24be1349478f44169009d4801c25ahrens dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (error == 0) {
5002558f6bfef3915c7f3b4ecb7c19c7f044bf5bNeil Perrin zfs_log_remove(zilog, tx, txtype, dzp, name, ZFS_NO_OBJECT);
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Read as many directory entries as will fit into the provided
fa9e4066f08beec538e775443c5be79dd423fcabahrens * buffer from the given directory cursor position (specified in
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * the uio structure).
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of directory to read.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * uio - structure supplying read location, range info,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and return buffer.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * flags - case flags
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: uio - updated offset and range, buffer filled.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * eofp - set to true if end-of-file detected.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - atime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Note that the low 4 bits of the cookie returned by zap is always zero.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * This allows us to use the low range for "special" directory entries:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * We use 0 for '.', and 1 for '..'. If this is the root of the filesystem,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * we use the offset 2 for the '.zfs' directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_readdir(vnode_t *vp, uio_t *uio, cred_t *cr, int *eofp,
fa9e4066f08beec538e775443c5be79dd423fcabahrens uint64_t offset; /* must be unsigned; checks for < 1 */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If we are not given an eof variable,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * use a local one.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Check for valid iov_len.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Quit if directory has been removed (posix)
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Initialize the iterator cursor.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Start iteration from the beginning of the directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * The offset is a serialized cursor.
7f6e3e7d4ebf3d6d45073854bef004ca26f8e918perrin zap_cursor_init_serialized(&zc, os, zp->z_id, offset);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Get space to change directory entries into fs independent format.
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1) {
9660e5cb810c5efc22db0f1459461c1f1eec15b9Janice Chang * If this VFS supports the system attribute view interface; and
9660e5cb810c5efc22db0f1459461c1f1eec15b9Janice Chang * we're looking at an extended attribute directory; and we care
9660e5cb810c5efc22db0f1459461c1f1eec15b9Janice Chang * about normalization conflicts on this vfs; then we must check
9660e5cb810c5efc22db0f1459461c1f1eec15b9Janice Chang * for normalization conflicts with the sysattr name space.
9660e5cb810c5efc22db0f1459461c1f1eec15b9Janice Chang check_sysattrs = vfs_has_feature(vp->v_vfsp, VFSFT_SYSATTR_VIEWS) &&
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Transform to file-system independent format
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Special case `.', `..', and `.zfs'.
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (offset == 0) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Grab next entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens "entry, obj = %lld, offset = %lld\n",
b1b8ab34de515a5e83206da22c3d7e563241b021lling * MacOS X can extract the object type here such as:
b1b8ab34de515a5e83206da22c3d7e563241b021lling * uint8_t type = ZFS_DIRENT_TYPE(zap.za_first_integer);
e802abbda8c322f24d47835734f4a793ef15ddc8Tim Haley * If we have no access at all, don't include
e802abbda8c322f24d47835734f4a793ef15ddc8Tim Haley * this entry in the returned information
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Will this entry fit in the buffer?
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Did we manage to fit anything in the buffer?
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Add extended flag entry:
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw /* NOTE: ed_off is the offset for the *next* entry */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Add normal entry:
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw /* NOTE: d_off is the offset for the *next* entry */
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* Prefetch znode */
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Move to the next entry, fill in the previous offset.
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (offset > 2 || (offset == 2 && !zfs_show_ctldir(zp))) {
7f6e3e7d4ebf3d6d45073854bef004ca26f8e918perrin zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (uio->uio_segflg == UIO_SYSSPACE && uio->uio_iovcnt == 1) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens } else if (error = uiomove(outbuf, (long)outcount, UIO_READ, uio)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Reset the pointer.
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1)
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_fsync(vnode_t *vp, int syncflag, cred_t *cr, caller_context_t *ct)
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * Regardless of whether this is required for standards conformance,
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * this is the logical behavior when fsync() is called on a file with
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * dirty pages. We use B_ASYNC since the ZIL transactions are already
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * going to be pushed out as part of the zil_commit().
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock if (vn_has_cached_data(vp) && !(syncflag & FNODSYNC) &&
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw (void) VOP_PUTPAGE(vp, (offset_t)0, (size_t)0, B_ASYNC, cr, ct);
ec533521f091387e2911769818ae4be26219fca2fr (void) tsd_set(zfs_fsyncer_key, (void *)zfs_fsync_sync_cnt);
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Get the requested file attributes and place them in the provided
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vattr structure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vap - va_mask identifies requested attributes.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If AT_XVATTR set, then optional attrs are requested
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * flags - ATTR_NOACLCHECK (CIFS server context)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: vap - attribute values.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 (always succeeds).
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_getattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
f1696b231bb1209e1cea95b7ff29b8c60493daa0Mark Shellenbaum zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((error = sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) != 0) {
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If ACL is trivial don't bother looking for ACE_READ_ATTRIBUTES.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Also, if we are the owner don't bother, since owner should
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * always be allowed to read basic attributes of file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Return all attributes. It's cheaper to provide the answer
fa9e4066f08beec538e775443c5be79dd423fcabahrens * than to determine whether we were asked the question.
ecd6cf800b63704be73fb264c3f5b6e0dafc068dmarks vap->va_nlink = MIN(links, UINT32_MAX); /* nlink_t limit! */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Add in any requested optional attributes and the create time.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Also set the corresponding bits in the returned attribute bitmap.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if ((xoap = xva_getxoptattr(xvap)) != NULL && zfsvfs->z_use_fuids) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum ZFS_TIME_DECODE(&xoap->xoa_createtime, times);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum xoap->xoa_reparse = ((zp->z_pflags & ZFS_REPARSE) != 0);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum ZFS_TIME_DECODE(&vap->va_atime, zp->z_atime);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum sa_object_size(zp->z_sa_hdl, &vap->va_blksize, &vap->va_nblocks);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Block size hasn't been set; suggest maximal I/O transfers.
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Set the file attributes to the values contained in the
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vattr structure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file to be modified.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vap - new attribute values.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If AT_XVATTR set, then optional attrs are being set
fa9e4066f08beec538e775443c5be79dd423fcabahrens * flags - ATTR_UTIME set if non-default time values provided.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * - ATTR_NOACLCHECK (CIFS context only).
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime updated, mtime updated if size changed.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_setattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Make sure that if we have ephemeral uid/gid or xvattr specified
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * that file system is at proper version level
02dcba3bb864e9d2365296acd67cb9b81d6ee30dtimh if (mask & AT_SIZE && vp->v_type != VREG && vp->v_type != VFIFO) {
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If this is an xvattr_t, then get a pointer to the structure of
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * optional attributes. If this is NULL, then we have a vattr_t.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Immutable files can only alter immutable bit and atime
02dcba3bb864e9d2365296acd67cb9b81d6ee30dtimh ((mask & AT_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((mask & AT_SIZE) && (zp->z_pflags & ZFS_READONLY)) {
931293416c5362b3d50b217b4390d1c88f7c7500marks * Verify timestamps doesn't overflow 32 bits.
931293416c5362b3d50b217b4390d1c88f7c7500marks * ZFS can handle large timestamps, but 32bit syscalls can't
931293416c5362b3d50b217b4390d1c88f7c7500marks * handle times greater than 2039. This check should be removed
931293416c5362b3d50b217b4390d1c88f7c7500marks * once large timestamps are fully supported.
931293416c5362b3d50b217b4390d1c88f7c7500marks if (((mask & AT_ATIME) && TIMESPEC_OVERFLOW(&vap->va_atime)) ||
931293416c5362b3d50b217b4390d1c88f7c7500marks ((mask & AT_MTIME) && TIMESPEC_OVERFLOW(&vap->va_mtime))) {
d47621a49c68c359358f6630aa45cc320762f51fTim Haley /* Can this be moved to before the top label? */
fa9e4066f08beec538e775443c5be79dd423fcabahrens * First validate permissions
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw err = zfs_zaccess(zp, ACE_WRITE_DATA, 0, skipaclchk, cr);
5730cc9a43c5f11a472d7536ed81facfd10f1e2emaybee * XXX - Note, we are not providing any open
5730cc9a43c5f11a472d7536ed81facfd10f1e2emaybee * mode flags here (like FNDELAY), so we may
5730cc9a43c5f11a472d7536ed81facfd10f1e2emaybee * block if there are locks present... this
5730cc9a43c5f11a472d7536ed81facfd10f1e2emaybee * should be addressed in openat().
cdb0ab79ea1af7b8fc339a04d4bf7426dc77ec4emaybee /* XXX - would it be OK to generate a log record here? */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw ((mask & AT_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) ||
a933bc4144669256cbe2aaf2261aed3432f8e7femarks * NOTE: even if a new mode is being set,
a933bc4144669256cbe2aaf2261aed3432f8e7femarks * we may clear S_ISUID/S_ISGID bits.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Take ownership or chgrp to group we are a member of
fa9e4066f08beec538e775443c5be79dd423fcabahrens take_owner = (mask & AT_UID) && (vap->va_uid == crgetuid(cr));
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If both AT_UID and AT_GID are set then take_owner and
fa9e4066f08beec538e775443c5be79dd423fcabahrens * take_group must both be set in order to allow taking
fa9e4066f08beec538e775443c5be79dd423fcabahrens * ownership.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Otherwise, send the check through secpolicy_vnode_setattr()
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (((idmask == (AT_UID|AT_GID)) && take_owner && take_group) ||
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Remove setuid/setgid for non-privileged users
f1696b231bb1209e1cea95b7ff29b8c60493daa0Mark Shellenbaum zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid);
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum * Update xvattr mask to include only those attributes
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum * that are actually changing.
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum * the bits will be restored prior to actually setting
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum * the attributes so the caller thinks they were set.
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum XVA_SET_REQ(&tmpxvattr, XAT_AV_QUARANTINED);
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) {
13f9f30ebea4836394531786ea98e70b31d85829marks * If trim_mask is set then take ownership
f92daba9919b6e68875ccdc9a5532cadf37959f1marks * has been granted or write_acl is present and user
f92daba9919b6e68875ccdc9a5532cadf37959f1marks * has the ability to modify mode. In that case remove
f92daba9919b6e68875ccdc9a5532cadf37959f1marks * UID|GID and or MODE from mask so that
13f9f30ebea4836394531786ea98e70b31d85829marks * secpolicy_vnode_setattr() doesn't revoke it.
fa9e4066f08beec538e775443c5be79dd423fcabahrens err = secpolicy_vnode_setattr(cr, vp, vap, &oldva, flags,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * secpolicy_vnode_setattr, or take ownership may have
fa9e4066f08beec538e775443c5be79dd423fcabahrens * changed va_mask
0b2a8171b47dd85552e914b19eea8ce3c3c731ffMark Shellenbaum err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum err = zfs_zget(zp->z_zfsvfs, xattr_obj, &attrzp);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_fuid_overquota(zfsvfs, B_FALSE, new_uid)) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum new_gid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_gid,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_fuid_overquota(zfsvfs, B_TRUE, new_gid)) {
169cdae232f15e542d6af0a9ce30c3f84222bc0fmarks new_mode = (pmode & S_IFMT) | (vap->va_mode & ~S_IFMT);
71dbfc287ce623fa58f67b024d14ef6924fa1b2aPaul B. Henson if (zp->z_zfsvfs->z_acl_mode == ZFS_ACL_RESTRICTED &&
a3c49ce110f325a563c245bedc4d533adddb7211Albert Lee if (err = zfs_acl_chmod_setattr(zp, &aclp, new_mode))
1412a1a223b7a94990edf5114c108b0a29c3f881Mark Shellenbaum if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum * Are we upgrading ACL from old V0 format
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum * to V1 format?
2bd6c4de2748ab3787a6e1ab244159d738d831e0Mark Shellenbaum if (zfsvfs->z_version >= ZPL_VERSION_FUID &&
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum } else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Set each attribute requested.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * We group settings according to the locks they need to acquire.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Note: you cannot set ctime directly, although it will be
fa9e4066f08beec538e775443c5be79dd423fcabahrens * updated as a side-effect of calling this function.
db9986c78dc7730f3091d7780ef0ba0937183a24Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
db9986c78dc7730f3091d7780ef0ba0937183a24Mark Shellenbaum SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags,
27dd1e87cd3d939264769dd4af7e6a529cde001fMark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
27dd1e87cd3d939264769dd4af7e6a529cde001fMark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs),
27dd1e87cd3d939264769dd4af7e6a529cde001fMark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum ZFS_TIME_ENCODE(&vap->va_atime, zp->z_atime);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
cdb0ab79ea1af7b8fc339a04d4bf7426dc77ec4emaybee /* XXX - shouldn't this be done *before* the ATIME/MTIME checks? */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (mask & AT_SIZE && !(mask & AT_MTIME)) {
db9986c78dc7730f3091d7780ef0ba0937183a24Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum } else if (mask != 0) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime, ctime,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_tstamp_update_setup(attrzp, STATE_CHANGED,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * Do this after setting timestamps to prevent timestamp
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * update from toggling bit
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum * restore trimmed off masks
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum * so that return masks can be set for caller.
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(&tmpxvattr, XAT_APPENDONLY)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(&tmpxvattr, XAT_NOUNLINK)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(&tmpxvattr, XAT_IMMUTABLE)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(&tmpxvattr, XAT_NODUMP)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_MODIFIED)) {
ae4caef845148f81ab1dd46f11a76c2b59e2353dMark Shellenbaum if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_QUARANTINED)) {
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Drop locks and release vnodes that were held by zfs_rename_lock().
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Search back through the directory tree, using the ".." entries.
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Lock each directory in the chain to prevent concurrent renames.
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Fail any attempt to move a directory into one of its own descendants.
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * XXX - z_parent_lock can overlap with map or grow locks
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_rename_lock(znode_t *szp, znode_t *tdzp, znode_t *sdzp, zfs_zlock_t **zlpp)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * First pass write-locks szp and compares to zp->z_id.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Later passes read-lock zp and compare to zp->z_parent.
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Another thread is renaming in this path.
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Note that if we are a WRITER, we don't have any
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * parent_locks held yet.
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Drop our locks and restart
ff008e00c5e0385b59fdceb4802bb4c6f77fda5amaybee * Wait for other thread to drop its locks
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (oidp == szp->z_id) /* We're a descendant of szp */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (oidp == rootid) /* We've hit the top */
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum int error = zfs_zget(zp->z_zfsvfs, oidp, &zp);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zp->z_zfsvfs),
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Move an entry from the provided source directory to the target
fa9e4066f08beec538e775443c5be79dd423fcabahrens * directory. Change the entry name as indicated.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: sdvp - Source directory containing the "old entry".
fa9e4066f08beec538e775443c5be79dd423fcabahrens * snm - Old entry name.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * tdvp - Target directory to contain the "new entry".
fa9e4066f08beec538e775443c5be79dd423fcabahrens * tnm - New entry name.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * flags - case flags
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * sdvp,tdvp - ctime|mtime updated
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_rename(vnode_t *sdvp, char *snm, vnode_t *tdvp, char *tnm, cred_t *cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Make sure we have the real vp for the target directory.
18e649782fcf9da52d755b2074b4b9382426c818Marcel Telka * We check z_zfsvfs rather than v_vfsp here, because snapshots and the
18e649782fcf9da52d755b2074b4b9382426c818Marcel Telka * ctldir appear to have the same v_vfsp.
18e649782fcf9da52d755b2074b4b9382426c818Marcel Telka if (tdzp->z_zfsvfs != zfsvfs || zfsctl_is_node(tdvp)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * This is to prevent the creation of links into attribute space
fa9e4066f08beec538e775443c5be79dd423fcabahrens * by renaming a linked file into/outof an attribute directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * See the comment in zfs_link() for why this is considered bad.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Lock source and target directory entries. To prevent deadlock,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * a lock ordering must be defined. We lock the directory with
fa9e4066f08beec538e775443c5be79dd423fcabahrens * the smallest object id first, or if it's a tie, the one with
fa9e4066f08beec538e775443c5be79dd423fcabahrens * the lexically first name.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * First compare the two name arguments without
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * considering any case folding.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error);
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (cmp == 0) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * POSIX: "If the old argument and the new argument
fa9e4066f08beec538e775443c5be79dd423fcabahrens * both refer to links to the same existing file,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * the rename() function shall return successfully
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and perform no other action."
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If the file system is case-folding, then we may
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * have some more checking to do. A case-folding file
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * system is either supporting mixed case sensitivity
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * access or is completely case-insensitive. Note
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * that the file system is always case preserving.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * In mixed sensitivity mode case sensitive behavior
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * is the default. FIGNORECASE must be used to
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * explicitly request case insensitive behavior.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * If the source and target names provided differ only
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * by case (e.g., a request to rename 'tim' to 'Tim'),
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * we will treat this as a special case in the
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * case-insensitive mode: as long as the source name
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * is an exact match, we will allow this to proceed as
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * a name-change request.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * case preserving rename request, require exact
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * name matches
afefc7e4027e02291d2f534f969cfd38b9dd1db3Sanjeev Bagewadi * If the source and destination directories are the same, we should
afefc7e4027e02291d2f534f969cfd38b9dd1db3Sanjeev Bagewadi * grab the z_name_lock of that directory only once.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Source entry invalid or not there.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Must have write access at the source to remove the old entry
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and write access at the target to create the new entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Note that if target and source are the same, this can be
fa9e4066f08beec538e775443c5be79dd423fcabahrens * done in a single check.
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr))
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Check to make sure rename is valid.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Can't do a move like this: /usr/a/b to /usr/a/b/c/d
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Does target exist?
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Source and target must be the same type.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * POSIX dictates that when the source and target
fa9e4066f08beec538e775443c5be79dd423fcabahrens * entries refer to the same file object, rename
fa9e4066f08beec538e775443c5be79dd423fcabahrens * must do nothing and exit without error.
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek vnevent_pre_rename_src(ZTOV(szp), sdvp, snm, ct);
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek vnevent_pre_rename_dest(ZTOV(tzp), tdvp, tnm, ct);
df2381bfa5cf7dd654bcf30b2f5af53f34f3043apraks * notify the target directory if it is not the same
df2381bfa5cf7dd654bcf30b2f5af53f34f3043apraks * as source directory.
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek vnevent_pre_rename_dest_dir(tdvp, ZTOV(szp), tnm, ct);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE);
893a6d32980d24be1349478f44169009d4801c25ahrens dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek error = rm_err = zfs_link_destroy(tdl, tzp, tx, zflg, NULL);
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (error == 0) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (error == 0) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void *)&szp->z_pflags, sizeof (uint64_t), tx);
fa9e4066f08beec538e775443c5be79dd423fcabahrens error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL);
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * Update path information for the target vnode
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * At this point, we have successfully created
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * the target name, but have failed to remove
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * the source name. Since the create was done
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * with the ZRENAMING flag, there are
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * complications; for one, the link count is
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * wrong. The easiest way to deal with this
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * is to remove the newly created target, and
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * return the original error. This must
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * succeed; fortunately, it is very unlikely to
6ed5e6ab20c9a0a54cdfbb139e0ccae3f8527e9bSam Falkner * fail, since we just created it.
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek vnevent_rename_dest(ZTOV(tzp), tdvp, tnm, ct);
54207fd2e1e7ed01d0416da8cf296dbef920fbfcJerry Jelinek /* notify the target dir if it is not the same as source dir */
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Insert the indicated symbolic reference entry into the directory.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: dvp - Directory to contain new symbolic link.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * link - Name for new symlink entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vap - Attributes of new entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * flags - case flags
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * dvp - ctime|mtime updated
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_symlink(vnode_t *dvp, char *name, vattr_t *vap, char *link, cred_t *cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Attempt to lock directory; fail if entry already exists.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL);
c8c24165eb6f3690a1b5b8877619e069efe5199bMark Shellenbaum if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) {
148434217c040ea38dc844384f6ba68d9b325906Matthew Ahrens if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len));
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Create a new object for the symlink.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum * for version 4 ZPL datsets the symlink will be an SA attribute
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Insert the new object into the directory.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Return, in the buffer contained in the provided uio structure,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * the symbolic path referred to by vp.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of symbolic link.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * uio - structure to contain the link path.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * OUT: uio - structure containing the link path.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - atime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_readlink(vnode_t *vp, uio_t *uio, cred_t *cr, caller_context_t *ct)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Insert a new entry into directory tdvp referencing svp.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: tdvp - Directory to contain new entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * svp - vnode of new entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * name - name of new entry.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * tdvp - ctime|mtime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens * svp - ctime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_link(vnode_t *tdvp, vnode_t *svp, char *name, cred_t *cr,
d39ee142a97a7c58f60f7b52c62409f2ff64b234Mark Shellenbaum * POSIX dictates that we return EPERM here.
d39ee142a97a7c58f60f7b52c62409f2ff64b234Mark Shellenbaum * Better choices include ENOTSUP or EISDIR.
18e649782fcf9da52d755b2074b4b9382426c818Marcel Telka * We check z_zfsvfs rather than v_vfsp here, because snapshots and the
18e649782fcf9da52d755b2074b4b9382426c818Marcel Telka * ctldir appear to have the same v_vfsp.
18e649782fcf9da52d755b2074b4b9382426c818Marcel Telka if (szp->z_zfsvfs != zfsvfs || zfsctl_is_node(svp)) {
d39ee142a97a7c58f60f7b52c62409f2ff64b234Mark Shellenbaum /* Prevent links to .zfs/shares files */
d39ee142a97a7c58f60f7b52c62409f2ff64b234Mark Shellenbaum if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
fa9e4066f08beec538e775443c5be79dd423fcabahrens * We do not support links between attributes and non-attributes
fa9e4066f08beec538e775443c5be79dd423fcabahrens * because of the potential security risk of creating links
fa9e4066f08beec538e775443c5be79dd423fcabahrens * into "normal" file space in order to circumvent restrictions
fa9e4066f08beec538e775443c5be79dd423fcabahrens * imposed in attribute space.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((szp->z_pflags & ZFS_XATTR) != (dzp->z_pflags & ZFS_XATTR)) {
f1696b231bb1209e1cea95b7ff29b8c60493daa0Mark Shellenbaum owner = zfs_fuid_map_id(zfsvfs, szp->z_uid, cr, ZFS_OWNER);
f1696b231bb1209e1cea95b7ff29b8c60493daa0Mark Shellenbaum if (owner != crgetuid(cr) && secpolicy_basic_link(cr) != 0) {
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Attempt to lock directory; fail if entry already exists.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw error = zfs_dirent_lock(&dl, dzp, name, &tzp, zf, NULL, NULL);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
69962b5647e4a8b9b14998733b765925381b727eMatthew Ahrens error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT);
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw if (error == 0) {
df2381bfa5cf7dd654bcf30b2f5af53f34f3043apraks if (error == 0) {
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_null_putapage() is used when the file system has been force
fa9e4066f08beec538e775443c5be79dd423fcabahrens * unmounted. It just drops the pages.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_null_putapage(vnode_t *vp, page_t *pp, u_offset_t *offp,
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * Push a page out to disk, klustering if possible.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * IN: vp - file to push page to.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * pp - page to push.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * flags - additional flags.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * cr - credentials of caller.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * OUT: offp - start of range pushed.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * lenp - len of range pushed.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * NOTE: callers must have locked the page to be pushed. On
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * exit, the page (and all other pages in the kluster) must be
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * unlocked.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_putapage(vnode_t *vp, page_t *pp, u_offset_t *offp,
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * If our blocksize is bigger than the page size, try to kluster
1209a471b5681c43d839d4b890f708f500da7346Neil Perrin * multiple pages so that we write a full block (thus avoiding
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee * a read-modify-write).
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (off < zp->z_size && zp->z_blksz > PAGESIZE) {
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee klen = P2ROUNDUP((ulong_t)zp->z_blksz, PAGESIZE);
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee koff = ISP2(klen) ? P2ALIGN(off, (u_offset_t)klen) : 0;
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum klen = P2ROUNDUP(zp->z_size - koff, (uint64_t)PAGESIZE);
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee pp = pvn_write_kluster(vp, pp, &off, &len, koff, klen, flags);
dd6ef5383c0b29543894f993c2ab3ab8ab6e6f20maybee * Can't push pages past end-of-file.
f4d2e9e65b9e8b1a74757024bec1678199b233b2maybee /* ignore all pages */
f4d2e9e65b9e8b1a74757024bec1678199b233b2maybee /* ignore pages past end of file */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) ||
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (err != 0) {
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee err = dmu_write_pages(zfsvfs->z_os, zp->z_id, off, len, pp, tx);
44eda4d76a9383a159e44aa60b63a17644ddd5b1maybee if (err == 0) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
db9986c78dc7730f3091d7780ef0ba0937183a24Mark Shellenbaum SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime,
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, off, len, 0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Copy the portion of the file indicated from pages into the file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * The pages are stored in a page list attached to the files vnode.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file to push page data to.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * off - position in file to put data.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * len - amount of data to write.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * flags - flags to control the operation.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime|mtime updated
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_putpage(vnode_t *vp, offset_t off, size_t len, int flags, cred_t *cr,
c4fc6b2110a79240282dafa6c538f2b42a6b8e2aGarrett D'Amore * There's nothing to do if no data is cached.
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * Align this request to the file block size in case we kluster.
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * XXX - this can result in pretty aggresive locking, which can
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * impact simultanious read/write access. One option might be
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * to break up long requests (len == 0) into block-by-block
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * operations to get narrower locking.
5a6f561952d9493613222dd61bab024a0d7d84a4Mark Maybee io_len = P2ROUNDUP_TYPED(len + (off - io_off), blksz, size_t);
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * Search the entire vp list for pages >= io_off.
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee rl = zfs_range_lock(zp, io_off, UINT64_MAX, RL_WRITER);
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee error = pvn_vplist_dirty(vp, io_off, zfs_putapage, flags, cr);
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee rl = zfs_range_lock(zp, io_off, io_len, RL_WRITER);
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* past end of file */
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum len = MIN(io_len, P2ROUNDUP(zp->z_size, PAGESIZE) - io_off);
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee for (off = io_off; io_off < off + len; io_off += io_len) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Found a dirty page to push
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin err = zfs_putapage(vp, pp, &io_off, &io_len, flags, cr);
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if ((flags & B_ASYNC) == 0 || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_inactive(vnode_t *vp, cred_t *cr, caller_context_t *ct)
874395d5f8cae2b9cd2d1fcbfcfe963a0c23966dmaybee * The fs has been unmounted, or we did a
874395d5f8cae2b9cd2d1fcbfcfe963a0c23966dmaybee * suspend/resume and this file no longer exists.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Attempt to push any data in the page cache. If this fails
fa9e4066f08beec538e775443c5be79dd423fcabahrens * we will get kicked out later in zfs_zinactive().
8afd4dd66b1dd487290f73b2f35040343a3cb821perrin (void) pvn_vplist_dirty(vp, 0, zfs_putapage, B_INVAL|B_ASYNC,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs),
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (void *)&zp->z_atime, sizeof (zp->z_atime), tx);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Bounds-check the seek operation.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode seeking within
fa9e4066f08beec538e775443c5be79dd423fcabahrens * ooff - old file offset
fa9e4066f08beec538e775443c5be79dd423fcabahrens * noffp - pointer to new file offset
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, EINVAL if new offset invalid.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens return ((*noffp < 0 || *noffp > MAXOFFSET_T) ? EINVAL : 0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Pre-filter the generic locking function to trap attempts to place
fa9e4066f08beec538e775443c5be79dd423fcabahrens * a mandatory lock on a memory mapped file.
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_frlock(vnode_t *vp, int cmd, flock64_t *bfp, int flag, offset_t offset,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw flk_callback_t *flk_cbp, cred_t *cr, caller_context_t *ct)
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * We are following the UFS semantics with respect to mapcnt
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * here: If we see that the file is mapped already, then we will
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * return an error, but we don't worry about races between this
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * function and zfs_map().
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (zp->z_mapcnt > 0 && MANDMODE(zp->z_mode)) {
04ce3d0b99f323222197d013257f04d5c65e2d61Mark Shellenbaum return (fs_frlock(vp, cmd, bfp, flag, offset, flk_cbp, cr, ct));
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If we can't find a page in the cache, we will create a new page
fa9e4066f08beec538e775443c5be79dd423fcabahrens * and fill it with file data. For efficiency, we may try to fill
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * multiple pages at once (klustering) to fill up the supplied page
ed8861874631957aa2528404968a3e366a5d8a49Mark Maybee * list. Note that the pages to be filled are held with an exclusive
ed8861874631957aa2528404968a3e366a5d8a49Mark Maybee * lock to prevent access by other threads while they are being filled.
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_fillpage(vnode_t *vp, u_offset_t off, struct seg *seg,
fa9e4066f08beec538e775443c5be79dd423fcabahrens caddr_t addr, page_t *pl[], size_t plsz, enum seg_rw rw)
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * We only have a single page, don't bother klustering
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * Try to find enough pages to fill the page list
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * The page already exists, nothing to do here.
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Fill the pages in the kluster.
fa9e4066f08beec538e775443c5be79dd423fcabahrens for (total = io_off + io_len; io_off < total; io_off += PAGESIZE) {
7bfdf011e081684f853a3242d0296695110d9d84Neil Perrin err = dmu_read(os, zp->z_id, io_off, PAGESIZE, va,
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* On error, toss the entire kluster */
b87f3af36bb994656da117319f5129ddfd05ed21perrin /* convert checksum errors into IO errors */
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * Fill in the page list array from the kluster starting
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * from the desired offset `off'.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * NOTE: the page list will always be null terminated.
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Return pointers to the pages for the file region [off, off + len]
fa9e4066f08beec538e775443c5be79dd423fcabahrens * in the pl array. If plsz is greater than len, this function may
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * also return page pointers from after the specified region
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * (i.e. the region [off, off + plsz]). These additional pages are
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * only returned if they are already in the cache, or were created as
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee * part of a klustered read.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file to get data from.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * off - position in file to get data from.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * len - amount of data to retrieve.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * plsz - length of provided page list.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * seg - segment to obtain pages for.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * addr - virtual address of fault.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * rw - mode of created pages.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller.
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * OUT: protp - protection mode of created pages.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * pl - list of pages created.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - atime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_getpage(vnode_t *vp, offset_t off, size_t len, uint_t *protp,
f7170741490edba9d1d9c697c177c887172bc741Will Andrews page_t *pl[], size_t plsz, struct seg *seg, caddr_t addr,
f7170741490edba9d1d9c697c177c887172bc741Will Andrews enum seg_rw rw, cred_t *cr, caller_context_t *ct)
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee /* we do our own caching, faultahead is unnecessary */
ed8861874631957aa2528404968a3e366a5d8a49Mark Maybee * Loop through the requested range [off, off + len) looking
fa9e4066f08beec538e775443c5be79dd423fcabahrens * for pages. If we don't find a page, we will need to create
fa9e4066f08beec538e775443c5be79dd423fcabahrens * a new page and fill it with data from the file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens while (len > 0) {
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee else if (err = zfs_fillpage(vp, off, seg, addr, pl, plsz, rw))
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Fill out the page array with any pages already in the cache.
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee while (plsz > 0 &&
ac05c741c43aa3e2f9b2f35878d03c299ff80d99Mark Maybee (*pl++ = page_lookup_nowait(vp, off, SE_SHARED))) {
fe2f476a78f63879c77fc29e41519eea3354941cperrin * Release any pages we have previously locked.
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * Request a memory map for a section of a file. This code interacts
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * with common code and the VM system as follows:
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * - common code calls mmap(), which ends up in smmap_common()
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * - this calls VOP_MAP(), which takes you into (say) zfs
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * - zfs_map() calls as_map(), passing segvn_create() as the callback
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * - segvn_create() creates the new segment and calls VOP_ADDMAP()
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * - zfs_addmap() updates z_mapcnt
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_map(vnode_t *vp, offset_t off, struct as *as, caddr_t *addrp,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr,
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum (ZFS_IMMUTABLE | ZFS_READONLY | ZFS_APPENDONLY))) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If file is locked, disallow mapping.
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if (MANDMODE(zp->z_mode) && vn_has_flocks(vp)) {
60946fe0e0cab23f683e9de92451aa45b7913135mec error = choose_addr(as, addrp, len, off, ADDR_VACALIGN, flags);
60946fe0e0cab23f683e9de92451aa45b7913135mec if (error != 0) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens error = as_map(as, *addrp, len, segvn_create, &vn_a);
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_addmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * The reason we push dirty pages as part of zfs_delmap() is so that we get a
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * more accurate mtime for the associated file. Since we don't have a way of
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * detecting when the data was actually modified, we have to resort to
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * heuristics. If an explicit msync() is done, then we mark the mtime when the
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * last page is pushed. The problem occurs when the msync() call is omitted,
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * which by far the most common case:
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens * <modify memory>
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens * <time lapse>
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens * putpage() via fsflush
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * If we wait until fsflush to come along, we can have a modification time that
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * is some arbitrary point in the future. In order to prevent this in the
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * common case, we flush pages whenever a (MAP_SHARED, PROT_WRITE) mapping is
b468a217b67dc26ce21da5d5a2ca09bb6249e4faeschrock * torn down.
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_delmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw size_t len, uint_t prot, uint_t maxprot, uint_t flags, cred_t *cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Free or allocate space in a file. Currently, this function only
fa9e4066f08beec538e775443c5be79dd423fcabahrens * supports the `F_FREESP' command. However, this command is somewhat
fa9e4066f08beec538e775443c5be79dd423fcabahrens * misnamed, as its functionality includes the ability to allocate as
fa9e4066f08beec538e775443c5be79dd423fcabahrens * well as free space.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * IN: vp - vnode of file to free data in.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cmd - action to take (only F_FREESP supported).
fa9e4066f08beec538e775443c5be79dd423fcabahrens * bfp - section of file to free/alloc.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * flag - current file open mode flags.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * offset - current file offset.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * cr - credentials of caller [UNUSED].
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw * ct - caller context.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * RETURN: 0 on success, error code on failure.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Timestamps:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * vp - ctime|mtime updated
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* ARGSUSED */
fa9e4066f08beec538e775443c5be79dd423fcabahrenszfs_space(vnode_t *vp, int cmd, flock64_t *bfp, int flag,
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * In a case vp->v_vfsp != zp->z_zfsvfs->z_vfs (e.g. snapshots) our
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * callers might not be able to detect properly that we are read-only,
2144b121c08e0eb676cc6ca4662ebbc9f9c22fe3Marcel Telka * so check it explicitly here.
104e2ed78d9ef0a0f89f320108b8ca29ca3850d5perrin len = bfp->l_len; /* 0 means from off to end of file */
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw/*ARGSUSED*/
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs),
fa9e4066f08beec538e775443c5be79dd423fcabahrens size = (zfsvfs->z_parent != zfsvfs) ? LONG_FID_LEN : SHORT_FID_LEN;
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* Must have a non-zero generation number to distinguish from .zfs */
fa9e4066f08beec538e775443c5be79dd423fcabahrens /* XXX - this should be the generation number for the objset */
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_pathconf(vnode_t *vp, int cmd, ulong_t *valp, cred_t *cr,
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens if (error == 0) {
fa9e4066f08beec538e775443c5be79dd423fcabahrens * If there aren't extended attributes, it's the
fa9e4066f08beec538e775443c5be79dd423fcabahrens * same as having zero of them.
9660e5cb810c5efc22db0f1459461c1f1eec15b9Janice Chang *valp = vfs_has_feature(vp->v_vfsp, VFSFT_SYSATTR_VIEWS) &&
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw return (0);
e802abbda8c322f24d47835734f4a793ef15ddc8Tim Haley *valp = vfs_has_feature(vp->v_vfsp, VFSFT_ACCESS_FILTER) &&
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens return (0);
3b862e9a9ce59d5dbf0177b9eb293109fde6bf36Roger A. Faulkner /* nanosecond timestamp resolution */
fa9e4066f08beec538e775443c5be79dd423fcabahrens/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_getsecattr(vnode_t *vp, vsecattr_t *vsecp, int flag, cred_t *cr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
fa9e4066f08beec538e775443c5be79dd423fcabahrens/*ARGSUSED*/
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amwzfs_setsecattr(vnode_t *vp, vsecattr_t *vsecp, int flag, cred_t *cr,
da6c28aaf62fa55f0fdb8004aa40f88f23bf53f0amw boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
55da60b91d96984f12de050ce428373ea25c7f35Mark J Musante if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * The smallest read we may consider to loan out an arcbuf.
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * This must be a power of 2.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United Statesint zcr_blksz_min = (1 << 10); /* 1K */
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * If set to less than the file block size, allow loaning out of an
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * arcbuf for a partial block read. This must be a power of 2.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United Statesint zcr_blksz_max = (1 << 17); /* 128K */
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States/*ARGSUSED*/
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United Statesstatic int
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United Stateszfs_reqzcbuf(vnode_t *vp, enum uio_rw ioflag, xuio_t *xuio, cred_t *cr,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States caller_context_t *ct)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States znode_t *zp = VTOZ(vp);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States zfsvfs_t *zfsvfs = zp->z_zfsvfs;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int max_blksz = zfsvfs->z_max_blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States uio_t *uio = &xuio->xu_uio;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ssize_t size = uio->uio_resid;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States offset_t offset = uio->uio_loffset;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int fullblk, i;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States arc_buf_t *abuf;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ssize_t maxsize;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int preamble, postamble;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (xuio->xu_type != UIOTYPE_ZEROCOPY)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_ENTER(zfsvfs);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_VERIFY_ZP(zp);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States switch (ioflag) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States case UIO_WRITE:
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * Loan out an arc_buf for write if write size is bigger than
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * max_blksz, and the file's block size is also max_blksz.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States blksz = max_blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (size < blksz || zp->z_blksz != blksz) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_EXIT(zfsvfs);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * Caller requests buffers for write before knowing where the
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * write offset might be (e.g. NFS TCP write).
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (offset == -1) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States preamble = 0;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States } else {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States preamble = P2PHASE(offset, blksz);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (preamble) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States preamble = blksz - preamble;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States size -= preamble;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States postamble = P2PHASE(size, blksz);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States size -= postamble;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States fullblk = size / blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States (preamble != 0) + fullblk + (postamble != 0));
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States DTRACE_PROBE3(zfs_reqzcbuf_align, int, preamble,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int, postamble, int,
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States (preamble != 0) + fullblk + (postamble != 0));
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * Have to fix iov base/len for partial buffers. They
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * currently represent full arc_buf's.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (preamble) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States /* data begins in the middle of the arc_buf */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(abuf);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States for (i = 0; i < fullblk; i++) {
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(abuf);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (postamble) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States /* data ends in the middle of the arc_buf */
0a586cea3ceec7e5e50e7e54c745082a7a333ac2Mark Shellenbaum abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl),
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(abuf);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States break;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States case UIO_READ:
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * Loan out an arc_buf for read if the read size is larger than
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * the current file block size. Block alignment is not
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * considered. Partial arc_buf will be loaned out for read.
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States blksz = zp->z_blksz;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (blksz < zcr_blksz_min)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States blksz = zcr_blksz_min;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (blksz > zcr_blksz_max)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States blksz = zcr_blksz_max;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States /* avoid potential complexity of dealing with it */
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (blksz > max_blksz) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_EXIT(zfsvfs);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (size > maxsize)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States size = maxsize;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (size < blksz || vn_has_cached_data(vp)) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_EXIT(zfsvfs);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States break;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States default:
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_EXIT(zfsvfs);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States uio->uio_extflg = UIO_XUIO;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States XUIO_XUZC_RW(xuio) = ioflag;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ZFS_EXIT(zfsvfs);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States return (0);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States/*ARGSUSED*/
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United Statesstatic int
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United Stateszfs_retzcbuf(vnode_t *vp, xuio_t *xuio, cred_t *cr, caller_context_t *ct)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int i;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States arc_buf_t *abuf;
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States int ioflag = XUIO_XUZC_RW(xuio);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(xuio->xu_type == UIOTYPE_ZEROCOPY);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States i = dmu_xuio_cnt(xuio);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States while (i-- > 0) {
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States abuf = dmu_xuio_arcbuf(xuio, i);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * if abuf == NULL, it must be a write buffer
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States * that has been returned in zfs_write().
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States if (abuf)
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States dmu_return_arcbuf(abuf);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States ASSERT(abuf || ioflag == UIO_WRITE);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States dmu_xuio_fini(xuio);
c242f9a02a2ef021449275ae0a1d2581ee77231dchunli zhang - Sun Microsystems - Irvine United States return (0);
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Predeclare these here so that the compiler assumes that
fa9e4066f08beec538e775443c5be79dd423fcabahrens * this is an "old style" function declaration that does
fa9e4066f08beec538e775443c5be79dd423fcabahrens * not include arguments => we won't get type mismatch errors
fa9e4066f08beec538e775443c5be79dd423fcabahrens * in the initializations that follow.
fa9e4066f08beec538e775443c5be79dd423fcabahrensstatic int zfs_inval();
fa9e4066f08beec538e775443c5be79dd423fcabahrensstatic int zfs_isdir();
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Directory vnode operations template
aa59c4cb15a6ac5d4e585dadf7a055b580abf579rsb VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
aa59c4cb15a6ac5d4e585dadf7a055b580abf579rsb VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Regular file vnode operations template
aa59c4cb15a6ac5d4e585dadf7a055b580abf579rsb VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
aa59c4cb15a6ac5d4e585dadf7a055b580abf579rsb VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens VOPNAME_REQZCBUF, { .vop_reqzcbuf = zfs_reqzcbuf },
4bb73804952172060c9efb163b89c17f56804fe8Matthew Ahrens VOPNAME_RETZCBUF, { .vop_retzcbuf = zfs_retzcbuf },
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Symbolic link vnode operations template
fa9e4066f08beec538e775443c5be79dd423fcabahrensconst fs_operation_def_t zfs_symvnodeops_template[] = {
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright * special share hidden files vnode operations template
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wrightconst fs_operation_def_t zfs_sharevnodeops_template[] = {
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright VOPNAME_GETATTR, { .vop_getattr = zfs_getattr },
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright VOPNAME_INACTIVE, { .vop_inactive = zfs_inactive },
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright VOPNAME_PATHCONF, { .vop_pathconf = zfs_pathconf },
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
743a77ed89085d3c232c4a2f65ab4e19576839e2Alan Wright VOPNAME_VNEVENT, { .vop_vnevent = fs_vnevent_support },
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Extended attribute directory vnode operations template
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * This template is identical to the directory vnodes
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * operation template except for restricted operations:
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * VOP_MKDIR()
f7170741490edba9d1d9c697c177c887172bc741Will Andrews * VOP_SYMLINK()
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Note that there are other restrictions embedded in:
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_create() - restrict type to VREG
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_link() - no links into/out of attribute space
fa9e4066f08beec538e775443c5be79dd423fcabahrens * zfs_rename() - no moves into/out of attribute space
fa9e4066f08beec538e775443c5be79dd423fcabahrensconst fs_operation_def_t zfs_xdvnodeops_template[] = {
aa59c4cb15a6ac5d4e585dadf7a055b580abf579rsb VOPNAME_GETSECATTR, { .vop_getsecattr = zfs_getsecattr },
aa59c4cb15a6ac5d4e585dadf7a055b580abf579rsb VOPNAME_SETSECATTR, { .vop_setsecattr = zfs_setsecattr },
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Error vnode operations template