xfs_vnodeops.c

linux 内核源代码

	*tpp = tp;
	return 0;
}

int
xfs_release(
	xfs_inode_t	*ip)
{
	bhv_vnode_t	*vp = XFS_ITOV(ip);
	xfs_mount_t	*mp = ip->i_mount;
	int		error;

	if (!VN_ISREG(vp) || (ip->i_d.di_mode == 0))
		return 0;

	/* If this is a read-only mount, don't do this (would generate I/O) */
	if (mp->m_flags & XFS_MOUNT_RDONLY)
		return 0;

	if (!XFS_FORCED_SHUTDOWN(mp)) {
		int truncated;

		/*
		 * If we are using filestreams, and we have an unlinked
		 * file that we are processing the last close on, then nothing
		 * will be able to reopen and write to this file. Purge this
		 * inode from the filestreams cache so that it doesn't delay
		 * teardown of the inode.
		 */
		if ((ip->i_d.di_nlink == 0) && xfs_inode_is_filestream(ip))
			xfs_filestream_deassociate(ip);

		/*
		 * If we previously truncated this file and removed old data
		 * in the process, we want to initiate "early" writeout on
		 * the last close.  This is an attempt to combat the notorious
		 * NULL files problem which is particularly noticable from a
		 * truncate down, buffered (re-)write (delalloc), followed by
		 * a crash.  What we are effectively doing here is
		 * significantly reducing the time window where we'd otherwise
		 * be exposed to that problem.
		 */
		truncated = xfs_iflags_test_and_clear(ip, XFS_ITRUNCATED);
		if (truncated && VN_DIRTY(vp) && ip->i_delayed_blks > 0)
			xfs_flush_pages(ip, 0, -1, XFS_B_ASYNC, FI_NONE);
	}

#ifdef HAVE_REFCACHE
	/* If we are in the NFS reference cache then don't do this now */
	if (ip->i_refcache)
		return 0;
#endif

	if (ip->i_d.di_nlink != 0) {
		if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
		     ((ip->i_size > 0) || (VN_CACHED(vp) > 0 ||
		       ip->i_delayed_blks > 0)) &&
		     (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
		    (!(ip->i_d.di_flags &
				(XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)))) {
			error = xfs_free_eofblocks(mp, ip, XFS_FREE_EOF_LOCK);
			if (error)
				return error;
			/* Update linux inode block count after free above */
			vn_to_inode(vp)->i_blocks = XFS_FSB_TO_BB(mp,
				ip->i_d.di_nblocks + ip->i_delayed_blks);
		}
	}

	return 0;
}

/*
 * xfs_inactive
 *
 * This is called when the vnode reference count for the vnode
 * goes to zero.  If the file has been unlinked, then it must
 * now be truncated.  Also, we clear all of the read-ahead state
 * kept for the inode here since the file is now closed.
 */
int
xfs_inactive(
	xfs_inode_t	*ip)
{
	bhv_vnode_t	*vp = XFS_ITOV(ip);
	xfs_bmap_free_t	free_list;
	xfs_fsblock_t	first_block;
	int		committed;
	xfs_trans_t	*tp;
	xfs_mount_t	*mp;
	int		error;
	int		truncate;

	vn_trace_entry(ip, __FUNCTION__, (inst_t *)__return_address);

	/*
	 * If the inode is already free, then there can be nothing
	 * to clean up here.
	 */
	if (ip->i_d.di_mode == 0 || VN_BAD(vp)) {
		ASSERT(ip->i_df.if_real_bytes == 0);
		ASSERT(ip->i_df.if_broot_bytes == 0);
		return VN_INACTIVE_CACHE;
	}

	/*
	 * Only do a truncate if it's a regular file with
	 * some actual space in it.  It's OK to look at the
	 * inode's fields without the lock because we're the
	 * only one with a reference to the inode.
	 */
	truncate = ((ip->i_d.di_nlink == 0) &&
	    ((ip->i_d.di_size != 0) || (ip->i_size != 0) ||
	     (ip->i_d.di_nextents > 0) || (ip->i_delayed_blks > 0)) &&
	    ((ip->i_d.di_mode & S_IFMT) == S_IFREG));

	mp = ip->i_mount;

	if (ip->i_d.di_nlink == 0 && DM_EVENT_ENABLED(ip, DM_EVENT_DESTROY)) {
		(void) XFS_SEND_DESTROY(mp, vp, DM_RIGHT_NULL);
	}

	error = 0;

	/* If this is a read-only mount, don't do this (would generate I/O) */
	if (mp->m_flags & XFS_MOUNT_RDONLY)
		goto out;

	if (ip->i_d.di_nlink != 0) {
		if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
                     ((ip->i_size > 0) || (VN_CACHED(vp) > 0 ||
                       ip->i_delayed_blks > 0)) &&
		      (ip->i_df.if_flags & XFS_IFEXTENTS) &&
		     (!(ip->i_d.di_flags &
				(XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)) ||
		      (ip->i_delayed_blks != 0)))) {
			error = xfs_free_eofblocks(mp, ip, XFS_FREE_EOF_LOCK);
			if (error)
				return VN_INACTIVE_CACHE;
			/* Update linux inode block count after free above */
			vn_to_inode(vp)->i_blocks = XFS_FSB_TO_BB(mp,
				ip->i_d.di_nblocks + ip->i_delayed_blks);
		}
		goto out;
	}

	ASSERT(ip->i_d.di_nlink == 0);

	if ((error = XFS_QM_DQATTACH(mp, ip, 0)))
		return VN_INACTIVE_CACHE;

	tp = xfs_trans_alloc(mp, XFS_TRANS_INACTIVE);
	if (truncate) {
		/*
		 * Do the xfs_itruncate_start() call before
		 * reserving any log space because itruncate_start
		 * will call into the buffer cache and we can't
		 * do that within a transaction.
		 */
		xfs_ilock(ip, XFS_IOLOCK_EXCL);

		error = xfs_itruncate_start(ip, XFS_ITRUNC_DEFINITE, 0);
		if (error) {
			xfs_trans_cancel(tp, 0);
			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
			return VN_INACTIVE_CACHE;
		}

		error = xfs_trans_reserve(tp, 0,
					  XFS_ITRUNCATE_LOG_RES(mp),
					  0, XFS_TRANS_PERM_LOG_RES,
					  XFS_ITRUNCATE_LOG_COUNT);
		if (error) {
			/* Don't call itruncate_cleanup */
			ASSERT(XFS_FORCED_SHUTDOWN(mp));
			xfs_trans_cancel(tp, 0);
			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
			return VN_INACTIVE_CACHE;
		}

		xfs_ilock(ip, XFS_ILOCK_EXCL);
		xfs_trans_ijoin(tp, ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
		xfs_trans_ihold(tp, ip);

		/*
		 * normally, we have to run xfs_itruncate_finish sync.
		 * But if filesystem is wsync and we're in the inactive
		 * path, then we know that nlink == 0, and that the
		 * xaction that made nlink == 0 is permanently committed
		 * since xfs_remove runs as a synchronous transaction.
		 */
		error = xfs_itruncate_finish(&tp, ip, 0, XFS_DATA_FORK,
				(!(mp->m_flags & XFS_MOUNT_WSYNC) ? 1 : 0));

		if (error) {
			xfs_trans_cancel(tp,
				XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
			xfs_iunlock(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
			return VN_INACTIVE_CACHE;
		}
	} else if ((ip->i_d.di_mode & S_IFMT) == S_IFLNK) {

		/*
		 * If we get an error while cleaning up a
		 * symlink we bail out.
		 */
		error = (ip->i_d.di_size > XFS_IFORK_DSIZE(ip)) ?
			xfs_inactive_symlink_rmt(ip, &tp) :
			xfs_inactive_symlink_local(ip, &tp);

		if (error) {
			ASSERT(tp == NULL);
			return VN_INACTIVE_CACHE;
		}

		xfs_trans_ijoin(tp, ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
		xfs_trans_ihold(tp, ip);
	} else {
		error = xfs_trans_reserve(tp, 0,
					  XFS_IFREE_LOG_RES(mp),
					  0, XFS_TRANS_PERM_LOG_RES,
					  XFS_INACTIVE_LOG_COUNT);
		if (error) {
			ASSERT(XFS_FORCED_SHUTDOWN(mp));
			xfs_trans_cancel(tp, 0);
			return VN_INACTIVE_CACHE;
		}

		xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
		xfs_trans_ijoin(tp, ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
		xfs_trans_ihold(tp, ip);
	}

	/*
	 * If there are attributes associated with the file
	 * then blow them away now.  The code calls a routine
	 * that recursively deconstructs the attribute fork.
	 * We need to just commit the current transaction
	 * because we can't use it for xfs_attr_inactive().
	 */
	if (ip->i_d.di_anextents > 0) {
		error = xfs_inactive_attrs(ip, &tp);
		/*
		 * If we got an error, the transaction is already
		 * cancelled, and the inode is unlocked. Just get out.
		 */
		 if (error)
			 return VN_INACTIVE_CACHE;
	} else if (ip->i_afp) {
		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
	}

	/*
	 * Free the inode.
	 */
	XFS_BMAP_INIT(&free_list, &first_block);
	error = xfs_ifree(tp, ip, &free_list);
	if (error) {
		/*
		 * If we fail to free the inode, shut down.  The cancel
		 * might do that, we need to make sure.  Otherwise the
		 * inode might be lost for a long time or forever.
		 */
		if (!XFS_FORCED_SHUTDOWN(mp)) {
			cmn_err(CE_NOTE,
		"xfs_inactive:	xfs_ifree() returned an error = %d on %s",
				error, mp->m_fsname);
			xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
		}
		xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
	} else {
		/*
		 * Credit the quota account(s). The inode is gone.
		 */
		XFS_TRANS_MOD_DQUOT_BYINO(mp, tp, ip, XFS_TRANS_DQ_ICOUNT, -1);

		/*
		 * Just ignore errors at this point.  There is
		 * nothing we can do except to try to keep going.
		 */
		(void) xfs_bmap_finish(&tp,  &free_list, &committed);
		(void) xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
	}
	/*
	 * Release the dquots held by inode, if any.
	 */
	XFS_QM_DQDETACH(mp, ip);

	xfs_iunlock(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);

 out:
	return VN_INACTIVE_CACHE;
}


int
xfs_lookup(
	xfs_inode_t		*dp,
	bhv_vname_t		*dentry,
	bhv_vnode_t		**vpp)
{
	xfs_inode_t		*ip;
	xfs_ino_t		e_inum;
	int			error;
	uint			lock_mode;

	vn_trace_entry(dp, __FUNCTION__, (inst_t *)__return_address);

	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
		return XFS_ERROR(EIO);

	lock_mode = xfs_ilock_map_shared(dp);
	error = xfs_dir_lookup_int(dp, lock_mode, dentry, &e_inum, &ip);
	if (!error) {
		*vpp = XFS_ITOV(ip);
		ITRACE(ip);
	}
	xfs_iunlock_map_shared(dp, lock_mode);
	return error;
}

int
xfs_create(
	xfs_inode_t		*dp,
	bhv_vname_t		*dentry,
	mode_t			mode,
	xfs_dev_t		rdev,
	bhv_vnode_t		**vpp,
	cred_t			*credp)
{
	char			*name = VNAME(dentry);
	xfs_mount_t	        *mp = dp->i_mount;
	bhv_vnode_t		*dir_vp = XFS_ITOV(dp);
	xfs_inode_t		*ip;
	bhv_vnode_t	        *vp = NULL;
	xfs_trans_t		*tp;
	int                     error;
	xfs_bmap_free_t		free_list;
	xfs_fsblock_t		first_block;
	boolean_t		unlock_dp_on_error = B_FALSE;
	int			dm_event_sent = 0;
	uint			cancel_flags;
	int			committed;
	xfs_prid_t		prid;
	struct xfs_dquot	*udqp, *gdqp;
	uint			resblks;
	int			namelen;

	ASSERT(!*vpp);
	vn_trace_entry(dp, __FUNCTION__, (inst_t *)__return_address);

	namelen = VNAMELEN(dentry);

	if (DM_EVENT_ENABLED(dp, DM_EVENT_CREATE)) {
		error = XFS_SEND_NAMESP(mp, DM_EVENT_CREATE,
				dir_vp, DM_RIGHT_NULL, NULL,
				DM_RIGHT_NULL, name, NULL,
				mode, 0, 0);

		if (error)
			return error;
		dm_event_sent = 1;
	}

	if (XFS_FORCED_SHUTDOWN(mp))
		return XFS_ERROR(EIO);

	/* Return through std_return after this point. */

	udqp = gdqp = NULL;
	if (dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
		prid = dp->i_d.di_projid;
	else
		prid = (xfs_prid_t)dfltprid;

	/*
	 * Make sure that we have allocated dquot(s) on disk.
	 */
	error = XFS_QM_DQVOPALLOC(mp, dp,
			current_fsuid(credp), current_fsgid(credp), prid,
			XFS_QMOPT_QUOTALL|XFS_QMOPT_INHERIT, &udqp, &gdqp);
	if (error)
		goto std_return;

	ip = NULL;

	tp = xfs_trans_alloc(mp, XFS_TRANS_CREATE);
	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
	resblks = XFS_CREATE_SPACE_RES(mp, namelen);
	/*
	 * Initially assume that the file does not exist and
	 * reserve the resources for that case.  If that is not
	 * the case we'll drop the one we have and get a more
	 * appropriate transaction later.
	 */
	error = xfs_trans_reserve(tp, resblks, XFS_CREATE_LOG_RES(mp), 0,
			XFS_TRANS_PERM_LOG_RES, XFS_CREATE_LOG_COUNT);
	if (error == ENOSPC) {
		resblks = 0;
		error = xfs_trans_reserve(tp, 0, XFS_CREATE_LOG_RES(mp), 0,
				XFS_TRANS_PERM_LOG_RES, XFS_CREATE_LOG_COUNT);
	}
	if (error) {
		cancel_flags = 0;
		goto error_return;
	}

	xfs_ilock(dp, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
	unlock_dp_on_error = B_TRUE;

	XFS_BMAP_INIT(&free_list, &first_block);

	ASSERT(ip == NULL);

	/*
	 * Reserve disk quota and the inode.
	 */
	error = XFS_TRANS_RESERVE_QUOTA(mp, tp, udqp, gdqp, resblks, 1, 0);
	if (error)
		goto error_return;

	if (resblks == 0 && (error = xfs_dir_canenter(tp, dp, name, namelen)))
		goto error_return;
	error = xfs_dir_ialloc(&tp, dp, mode, 1,
			rdev, credp, prid, resblks > 0,
			&ip, &committed);
	if (error) {
		if (error == ENOSPC)
			goto error_return;
		goto abort_return;
	}
	ITRACE(ip);

	/*
	 * At this point, we've gotten a newly allocated inode.
	 * It is locked (and joined to the transaction).
	 */

	ASSERT(ismrlocked (&ip->i_lock, MR_UPDATE));

	/*
	 * Now we join the directory inode to the transaction.  We do not do it
	 * earlier because xfs_dir_ialloc might commit the previous transaction
	 * (and release all the locks).  An error from here on will result in
	 * the transaction cancel unlocking dp so don't do it explicitly in the
	 * error path.
	 */
	VN_HOLD(dir_vp);
	xfs_trans_ijoin(tp, dp, XFS_ILOCK_EXCL);
	unlock_dp_on_error = B_FALSE;

	error = xfs_dir_createname(tp, dp, name, namelen, ip->i_ino,
					&first_block, &free_list, resblks ?
					resblks - XFS_IALLOC_SPACE_RES(mp) : 0);
	if (error) {
		ASSERT(error != ENOSPC);
		goto abort_return;
	}
	xfs_ichgtime(dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);

	/*
	 * If this is a synchronous mount, make sure that the
	 * create transaction goes to disk before returning to
	 * the user.
	 */
	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC)) {
		xfs_trans_set_sync(tp);
	}

	dp->i_gen++;

	/*
	 * Attach the dquot(s) to the inodes and modify them incore.
	 * These ids of the inode couldn't have changed since the new
	 * inode has been locked ever since it was created.
	 */
	XFS_QM_DQVOPCREATE(mp, tp, ip, udqp, gdqp);

	/*
	 * xfs_trans_commit normally decrements the vnode ref count
	 * when it unlocks the inode. Since we want to return the
	 * vnode to the caller, we bump the vnode ref count now.
	 */
	IHOLD(ip);
	vp = XFS_ITOV(ip);

	error = xfs_bmap_finish(&tp, &free_list, &committed);
	if (error) {
		xfs_bmap_cancel(&free_list);
		goto abort_rele;
	}

	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
	if (error) {
		IRELE(ip);
		tp = NULL;
		goto error_return;
	}