xfs_vfsops.c

linux 内核源代码

	}
	if (!error && rtdev)
		error = xfs_setsize_buftarg(mp->m_rtdev_targp,
					    mp->m_sb.sb_blocksize,
					    mp->m_sb.sb_sectsize);
	if (error)
		goto error2;

	if (mp->m_flags & XFS_MOUNT_BARRIER)
		xfs_mountfs_check_barriers(mp);

	if ((error = xfs_filestream_mount(mp)))
		goto error2;

	error = XFS_IOINIT(mp, args, flags);
	if (error)
		goto error2;

	XFS_SEND_MOUNT(mp, DM_RIGHT_NULL, args->mtpt, args->fsname);

	return 0;

error2:
	if (mp->m_sb_bp)
		xfs_freesb(mp);
error1:
	xfs_binval(mp->m_ddev_targp);
	if (logdev && logdev != ddev)
		xfs_binval(mp->m_logdev_targp);
	if (rtdev)
		xfs_binval(mp->m_rtdev_targp);
error0:
	xfs_unmountfs_close(mp, credp);
	xfs_qmops_put(mp);
	xfs_dmops_put(mp);
	return error;
}

int
xfs_unmount(
	xfs_mount_t	*mp,
	int		flags,
	cred_t		*credp)
{
	xfs_inode_t	*rip;
	bhv_vnode_t	*rvp;
	int		unmount_event_wanted = 0;
	int		unmount_event_flags = 0;
	int		xfs_unmountfs_needed = 0;
	int		error;

	rip = mp->m_rootip;
	rvp = XFS_ITOV(rip);

#ifdef HAVE_DMAPI
	if (mp->m_flags & XFS_MOUNT_DMAPI) {
		error = XFS_SEND_PREUNMOUNT(mp,
				rvp, DM_RIGHT_NULL, rvp, DM_RIGHT_NULL,
				NULL, NULL, 0, 0,
				(mp->m_dmevmask & (1<<DM_EVENT_PREUNMOUNT))?
					0:DM_FLAGS_UNWANTED);
			if (error)
				return XFS_ERROR(error);
		unmount_event_wanted = 1;
		unmount_event_flags = (mp->m_dmevmask & (1<<DM_EVENT_UNMOUNT))?
					0 : DM_FLAGS_UNWANTED;
	}
#endif
	/*
	 * First blow any referenced inode from this file system
	 * out of the reference cache, and delete the timer.
	 */
	xfs_refcache_purge_mp(mp);

	/*
	 * Blow away any referenced inode in the filestreams cache.
	 * This can and will cause log traffic as inodes go inactive
	 * here.
	 */
	xfs_filestream_unmount(mp);

	XFS_bflush(mp->m_ddev_targp);
	error = xfs_unmount_flush(mp, 0);
	if (error)
		goto out;

	ASSERT(vn_count(rvp) == 1);

	/*
	 * Drop the reference count
	 */
	VN_RELE(rvp);

	/*
	 * If we're forcing a shutdown, typically because of a media error,
	 * we want to make sure we invalidate dirty pages that belong to
	 * referenced vnodes as well.
	 */
	if (XFS_FORCED_SHUTDOWN(mp)) {
		error = xfs_sync(mp, SYNC_WAIT | SYNC_CLOSE);
		ASSERT(error != EFSCORRUPTED);
	}
	xfs_unmountfs_needed = 1;

out:
	/*	Send DMAPI event, if required.
	 *	Then do xfs_unmountfs() if needed.
	 *	Then return error (or zero).
	 */
	if (unmount_event_wanted) {
		/* Note: mp structure must still exist for
		 * XFS_SEND_UNMOUNT() call.
		 */
		XFS_SEND_UNMOUNT(mp, error == 0 ? rvp : NULL,
			DM_RIGHT_NULL, 0, error, unmount_event_flags);
	}
	if (xfs_unmountfs_needed) {
		/*
		 * Call common unmount function to flush to disk
		 * and free the super block buffer & mount structures.
		 */
		xfs_unmountfs(mp, credp);
		xfs_qmops_put(mp);
		xfs_dmops_put(mp);
		kmem_free(mp, sizeof(xfs_mount_t));
	}

	return XFS_ERROR(error);
}

STATIC int
xfs_quiesce_fs(
	xfs_mount_t		*mp)
{
	int			count = 0, pincount;

	xfs_refcache_purge_mp(mp);
	xfs_flush_buftarg(mp->m_ddev_targp, 0);
	xfs_finish_reclaim_all(mp, 0);

	/* This loop must run at least twice.
	 * The first instance of the loop will flush
	 * most meta data but that will generate more
	 * meta data (typically directory updates).
	 * Which then must be flushed and logged before
	 * we can write the unmount record.
	 */
	do {
		xfs_syncsub(mp, SYNC_INODE_QUIESCE, NULL);
		pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
		if (!pincount) {
			delay(50);
			count++;
		}
	} while (count < 2);

	return 0;
}

/*
 * Second stage of a quiesce. The data is already synced, now we have to take
 * care of the metadata. New transactions are already blocked, so we need to
 * wait for any remaining transactions to drain out before proceding.
 */
STATIC void
xfs_attr_quiesce(
	xfs_mount_t	*mp)
{
	/* wait for all modifications to complete */
	while (atomic_read(&mp->m_active_trans) > 0)
		delay(100);

	/* flush inodes and push all remaining buffers out to disk */
	xfs_quiesce_fs(mp);

	ASSERT_ALWAYS(atomic_read(&mp->m_active_trans) == 0);

	/* Push the superblock and write an unmount record */
	xfs_log_sbcount(mp, 1);
	xfs_log_unmount_write(mp);
	xfs_unmountfs_writesb(mp);
}

int
xfs_mntupdate(
	struct xfs_mount		*mp,
	int				*flags,
	struct xfs_mount_args		*args)
{
	if (!(*flags & MS_RDONLY)) {			/* rw/ro -> rw */
		if (mp->m_flags & XFS_MOUNT_RDONLY)
			mp->m_flags &= ~XFS_MOUNT_RDONLY;
		if (args->flags & XFSMNT_BARRIER) {
			mp->m_flags |= XFS_MOUNT_BARRIER;
			xfs_mountfs_check_barriers(mp);
		} else {
			mp->m_flags &= ~XFS_MOUNT_BARRIER;
		}
	} else if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {	/* rw -> ro */
		xfs_filestream_flush(mp);
		xfs_sync(mp, SYNC_DATA_QUIESCE);
		xfs_attr_quiesce(mp);
		mp->m_flags |= XFS_MOUNT_RDONLY;
	}
	return 0;
}

/*
 * xfs_unmount_flush implements a set of flush operation on special
 * inodes, which are needed as a separate set of operations so that
 * they can be called as part of relocation process.
 */
int
xfs_unmount_flush(
	xfs_mount_t	*mp,		/* Mount structure we are getting
					   rid of. */
	int             relocation)	/* Called from vfs relocation. */
{
	xfs_inode_t	*rip = mp->m_rootip;
	xfs_inode_t	*rbmip;
	xfs_inode_t	*rsumip = NULL;
	bhv_vnode_t	*rvp = XFS_ITOV(rip);
	int		error;

	xfs_ilock(rip, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
	xfs_iflock(rip);

	/*
	 * Flush out the real time inodes.
	 */
	if ((rbmip = mp->m_rbmip) != NULL) {
		xfs_ilock(rbmip, XFS_ILOCK_EXCL);
		xfs_iflock(rbmip);
		error = xfs_iflush(rbmip, XFS_IFLUSH_SYNC);
		xfs_iunlock(rbmip, XFS_ILOCK_EXCL);

		if (error == EFSCORRUPTED)
			goto fscorrupt_out;

		ASSERT(vn_count(XFS_ITOV(rbmip)) == 1);

		rsumip = mp->m_rsumip;
		xfs_ilock(rsumip, XFS_ILOCK_EXCL);
		xfs_iflock(rsumip);
		error = xfs_iflush(rsumip, XFS_IFLUSH_SYNC);
		xfs_iunlock(rsumip, XFS_ILOCK_EXCL);

		if (error == EFSCORRUPTED)
			goto fscorrupt_out;

		ASSERT(vn_count(XFS_ITOV(rsumip)) == 1);
	}

	/*
	 * Synchronously flush root inode to disk
	 */
	error = xfs_iflush(rip, XFS_IFLUSH_SYNC);
	if (error == EFSCORRUPTED)
		goto fscorrupt_out2;

	if (vn_count(rvp) != 1 && !relocation) {
		xfs_iunlock(rip, XFS_ILOCK_EXCL);
		return XFS_ERROR(EBUSY);
	}

	/*
	 * Release dquot that rootinode, rbmino and rsumino might be holding,
	 * flush and purge the quota inodes.
	 */
	error = XFS_QM_UNMOUNT(mp);
	if (error == EFSCORRUPTED)
		goto fscorrupt_out2;

	if (rbmip) {
		VN_RELE(XFS_ITOV(rbmip));
		VN_RELE(XFS_ITOV(rsumip));
	}

	xfs_iunlock(rip, XFS_ILOCK_EXCL);
	return 0;

fscorrupt_out:
	xfs_ifunlock(rip);

fscorrupt_out2:
	xfs_iunlock(rip, XFS_ILOCK_EXCL);

	return XFS_ERROR(EFSCORRUPTED);
}

/*
 * xfs_root extracts the root vnode from a vfs.
 *
 * vfsp -- the vfs struct for the desired file system
 * vpp  -- address of the caller's vnode pointer which should be
 *         set to the desired fs root vnode
 */
int
xfs_root(
	xfs_mount_t	*mp,
	bhv_vnode_t	**vpp)
{
	bhv_vnode_t	*vp;

	vp = XFS_ITOV(mp->m_rootip);
	VN_HOLD(vp);
	*vpp = vp;
	return 0;
}

/*
 * xfs_statvfs
 *
 * Fill in the statvfs structure for the given file system.  We use
 * the superblock lock in the mount structure to ensure a consistent
 * snapshot of the counters returned.
 */
int
xfs_statvfs(
	xfs_mount_t	*mp,
	bhv_statvfs_t	*statp,
	bhv_vnode_t	*vp)
{
	__uint64_t	fakeinos;
	xfs_extlen_t	lsize;
	xfs_sb_t	*sbp;
	unsigned long	s;

	sbp = &(mp->m_sb);

	statp->f_type = XFS_SB_MAGIC;

	xfs_icsb_sync_counters_flags(mp, XFS_ICSB_LAZY_COUNT);
	s = XFS_SB_LOCK(mp);
	statp->f_bsize = sbp->sb_blocksize;
	lsize = sbp->sb_logstart ? sbp->sb_logblocks : 0;
	statp->f_blocks = sbp->sb_dblocks - lsize;
	statp->f_bfree = statp->f_bavail =
				sbp->sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
	fakeinos = statp->f_bfree << sbp->sb_inopblog;
#if XFS_BIG_INUMS
	fakeinos += mp->m_inoadd;
#endif
	statp->f_files =
	    MIN(sbp->sb_icount + fakeinos, (__uint64_t)XFS_MAXINUMBER);
	if (mp->m_maxicount)
#if XFS_BIG_INUMS
		if (!mp->m_inoadd)
#endif
			statp->f_files = min_t(typeof(statp->f_files),
						statp->f_files,
						mp->m_maxicount);
	statp->f_ffree = statp->f_files - (sbp->sb_icount - sbp->sb_ifree);
	XFS_SB_UNLOCK(mp, s);

	xfs_statvfs_fsid(statp, mp);
	statp->f_namelen = MAXNAMELEN - 1;

	if (vp)
		XFS_QM_DQSTATVFS(xfs_vtoi(vp), statp);
	return 0;
}


/*
 * xfs_sync flushes any pending I/O to file system vfsp.
 *
 * This routine is called by vfs_sync() to make sure that things make it
 * out to disk eventually, on sync() system calls to flush out everything,
 * and when the file system is unmounted.  For the vfs_sync() case, all
 * we really need to do is sync out the log to make all of our meta-data
 * updates permanent (except for timestamps).  For calls from pflushd(),
 * dirty pages are kept moving by calling pdflush() on the inodes
 * containing them.  We also flush the inodes that we can lock without
 * sleeping and the superblock if we can lock it without sleeping from
 * vfs_sync() so that items at the tail of the log are always moving out.
 *
 * Flags:
 *      SYNC_BDFLUSH - We're being called from vfs_sync() so we don't want
 *		       to sleep if we can help it.  All we really need
 *		       to do is ensure that the log is synced at least
 *		       periodically.  We also push the inodes and
 *		       superblock if we can lock them without sleeping
 *			and they are not pinned.
 *      SYNC_ATTR    - We need to flush the inodes.  If SYNC_BDFLUSH is not
 *		       set, then we really want to lock each inode and flush
 *		       it.
 *      SYNC_WAIT    - All the flushes that take place in this call should
 *		       be synchronous.
 *      SYNC_DELWRI  - This tells us to push dirty pages associated with
 *		       inodes.  SYNC_WAIT and SYNC_BDFLUSH are used to
 *		       determine if they should be flushed sync, async, or
 *		       delwri.
 *      SYNC_CLOSE   - This flag is passed when the system is being
 *		       unmounted.  We should sync and invalidate everything.
 *      SYNC_FSDATA  - This indicates that the caller would like to make
 *		       sure the superblock is safe on disk.  We can ensure
 *		       this by simply making sure the log gets flushed
 *		       if SYNC_BDFLUSH is set, and by actually writing it
 *		       out otherwise.
 *	SYNC_IOWAIT  - The caller wants us to wait for all data I/O to complete
 *		       before we return (including direct I/O). Forms the drain
 *		       side of the write barrier needed to safely quiesce the
 *		       filesystem.
 *
 */
int
xfs_sync(
	xfs_mount_t	*mp,
	int		flags)
{
	int		error;

	/*
	 * Get the Quota Manager to flush the dquots.
	 *
	 * If XFS quota support is not enabled or this filesystem
	 * instance does not use quotas XFS_QM_DQSYNC will always
	 * return zero.
	 */
	error = XFS_QM_DQSYNC(mp, flags);
	if (error) {
		/*
		 * If we got an IO error, we will be shutting down.
		 * So, there's nothing more for us to do here.
		 */
		ASSERT(error != EIO || XFS_FORCED_SHUTDOWN(mp));
		if (XFS_FORCED_SHUTDOWN(mp))
			return XFS_ERROR(error);
	}

	if (flags & SYNC_IOWAIT)
		xfs_filestream_flush(mp);

	return xfs_syncsub(mp, flags, NULL);
}

/*
 * xfs sync routine for internal use
 *
 * This routine supports all of the flags defined for the generic vfs_sync
 * interface as explained above under xfs_sync.
 *
 */
int
xfs_sync_inodes(
	xfs_mount_t	*mp,
	int		flags,
	int             *bypassed)
{
	xfs_inode_t	*ip = NULL;
	xfs_inode_t	*ip_next;
	xfs_buf_t	*bp;
	bhv_vnode_t	*vp = NULL;
	int		error;
	int		last_error;
	uint64_t	fflag;
	uint		lock_flags;
	uint		base_lock_flags;
	boolean_t	mount_locked;
	boolean_t	vnode_refed;
	int		preempt;
	xfs_dinode_t	*dip;
	xfs_iptr_t	*ipointer;
#ifdef DEBUG
	boolean_t	ipointer_in = B_FALSE;

#define IPOINTER_SET	ipointer_in = B_TRUE
#define IPOINTER_CLR	ipointer_in = B_FALSE
#else
#define IPOINTER_SET
#define IPOINTER_CLR
#endif


/* Insert a marker record into the inode list after inode ip. The list
 * must be locked when this is called. After the call the list will no
 * longer be locked.
 */
#define IPOINTER_INSERT(ip, mp)	{ \
		ASSERT(ipointer_in == B_FALSE); \
		ipointer->ip_mnext = ip->i_mnext; \
		ipointer->ip_mprev = ip; \
		ip->i_mnext = (xfs_inode_t *)ipointer; \
		ipointer->ip_mnext->i_mprev = (xfs_inode_t *)ipointer; \
		preempt = 0; \
		XFS_MOUNT_IUNLOCK(mp); \
		mount_locked = B_FALSE; \
		IPOINTER_SET; \
	}

/* Remove the marker from the inode list. If the marker was the only item
 * in the list then there are no remaining inodes and we should zero out
 * the whole list. If we are the current head of the list then move the head
 * past us.
 */
#define IPOINTER_REMOVE(ip, mp)	{ \
		ASSERT(ipointer_in == B_TRUE); \
		if (ipointer->ip_mnext != (xfs_inode_t *)ipointer) { \
			ip = ipointer->ip_mnext; \
			ip->i_mprev = ipointer->ip_mprev; \
			ipointer->ip_mprev->i_mnext = ip; \
			if (mp->m_inodes == (xfs_inode_t *)ipointer) { \
				mp->m_inodes = ip; \
			} \
		} else { \
			ASSERT(mp->m_inodes == (xfs_inode_t *)ipointer); \
			mp->m_inodes = NULL; \
			ip = NULL; \
		} \
		IPOINTER_CLR; \
	}

#define XFS_PREEMPT_MASK	0x7f

	if (bypassed)
		*bypassed = 0;
	if (mp->m_flags & XFS_MOUNT_RDONLY)
		return 0;
	error = 0;
	last_error = 0;
	preempt = 0;

	/* Allocate a reference marker */
	ipointer = (xfs_iptr_t *)kmem_zalloc(sizeof(xfs_iptr_t), KM_SLEEP);

	fflag = XFS_B_ASYNC;		/* default is don't wait */
	if (flags & (SYNC_BDFLUSH | SYNC_DELWRI))
		fflag = XFS_B_DELWRI;
	if (flags & SYNC_WAIT)
		fflag = 0;		/* synchronous overrides all */