xfs_vfsops.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 1,927 行 · 第 1/4 页

				xfs_alloc_buftarg(logdev) : mp->m_ddev_targp;
	if (!mp->m_logdev_targp)
		goto error0;

	/*
	 * Setup flags based on mount(2) options and then the superblock
	 */
	error = xfs_start_flags(vfsp, args, mp);
	if (error)
		goto error1;
	error = xfs_readsb(mp);
	if (error)
		goto error1;
	error = xfs_finish_flags(vfsp, args, mp);
	if (error)
		goto error2;

	/*
	 * Setup xfs_mount buffer target pointers based on superblock
	 */
	error = xfs_setsize_buftarg(mp->m_ddev_targp, mp->m_sb.sb_blocksize,
				    mp->m_sb.sb_sectsize);
	if (!error && logdev && logdev != ddev) {
		unsigned int	log_sector_size = BBSIZE;

		if (XFS_SB_VERSION_HASSECTOR(&mp->m_sb))
			log_sector_size = mp->m_sb.sb_logsectsize;
		error = xfs_setsize_buftarg(mp->m_logdev_targp,
					    mp->m_sb.sb_blocksize,
					    log_sector_size);
	}
	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;

	error = XFS_IOINIT(vfsp, args, flags);
	if (!error)
		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);
	return error;
}

STATIC int
xfs_unmount(
	bhv_desc_t	*bdp,
	int		flags,
	cred_t		*credp)
{
	struct vfs	*vfsp = bhvtovfs(bdp);
	xfs_mount_t	*mp = XFS_BHVTOM(bdp);
	xfs_inode_t	*rip;
	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);

	if (vfsp->vfs_flag & VFS_DMI) {
		error = XFS_SEND_PREUNMOUNT(mp, vfsp,
				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;
	}

	/*
	 * First blow any referenced inode from this file system
	 * out of the reference cache, and delete the timer.
	 */
	xfs_refcache_purge_mp(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->m_bhv,
			 (SYNC_WAIT | SYNC_CLOSE), credp);
		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, vfsp, 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);
	}

	return XFS_ERROR(error);
}

#define REMOUNT_READONLY_FLAGS	(SYNC_REMOUNT|SYNC_ATTR|SYNC_WAIT)

STATIC int
xfs_mntupdate(
	bhv_desc_t			*bdp,
	int				*flags,
	struct xfs_mount_args		*args)
{
	struct vfs	*vfsp = bhvtovfs(bdp);
	xfs_mount_t	*mp = XFS_BHVTOM(bdp);
	int		pincount, error;
	int		count = 0;

	if (args->flags & XFSMNT_NOATIME)
		mp->m_flags |= XFS_MOUNT_NOATIME;
	else
		mp->m_flags &= ~XFS_MOUNT_NOATIME;

	if (!(vfsp->vfs_flag & VFS_RDONLY)) {
		VFS_SYNC(vfsp, SYNC_FSDATA|SYNC_BDFLUSH|SYNC_ATTR, NULL, error);
	}

	if (*flags & MS_RDONLY) {
		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 {
			VFS_SYNC(vfsp, REMOUNT_READONLY_FLAGS, NULL, error);
			pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
			if (!pincount) {
				delay(50);
				count++;
			}
		} while (count < 2);

		/* Ok now write out an unmount record */
		xfs_log_unmount_write(mp);
		xfs_unmountfs_writesb(mp);
		vfsp->vfs_flag |= VFS_RDONLY;
	} else {
		vfsp->vfs_flag &= ~VFS_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;
	vnode_t		*rvp = XFS_ITOV(rip);
	int		error;

	xfs_ilock(rip, XFS_ILOCK_EXCL);
	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
 */
STATIC int
xfs_root(
	bhv_desc_t	*bdp,
	vnode_t		**vpp)
{
	vnode_t		*vp;

	vp = XFS_ITOV((XFS_BHVTOM(bdp))->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.
 */
STATIC int
xfs_statvfs(
	bhv_desc_t	*bdp,
	xfs_statfs_t	*statp,
	vnode_t		*vp)
{
	__uint64_t	fakeinos;
	xfs_extlen_t	lsize;
	xfs_mount_t	*mp;
	xfs_sb_t	*sbp;
	unsigned long	s;
	u64 id;

	mp = XFS_BHVTOM(bdp);
	sbp = &(mp->m_sb);

	statp->f_type = XFS_SB_MAGIC;

	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;
	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);

	id = huge_encode_dev(mp->m_dev);
	statp->f_fsid.val[0] = (u32)id;
	statp->f_fsid.val[1] = (u32)(id >> 32);
	statp->f_namelen = MAXNAMELEN - 1;

	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 everthing.
 *      SYNC_FSDATA  - This indicates that the caller would like to make
 *		       sure the superblock is safe on disk.  We can ensure
 *		       this by simply makeing sure the log gets flushed
 *		       if SYNC_BDFLUSH is set, and by actually writing it
 *		       out otherwise.
 *
 */
/*ARGSUSED*/
STATIC int
xfs_sync(
	bhv_desc_t	*bdp,
	int		flags,
	cred_t		*credp)
{
	xfs_mount_t	*mp;

	mp = XFS_BHVTOM(bdp);
	return (xfs_syncsub(mp, flags, 0, 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.  In the interests of not
 * changing interfaces within the 6.5 family, additional internallly-
 * required functions are specified within a separate xflags parameter,
 * only available by calling this routine.
 *
 */
STATIC int
xfs_sync_inodes(
	xfs_mount_t	*mp,
	int		flags,
	int             xflags,
	int             *bypassed)
{
	xfs_inode_t	*ip = NULL;
	xfs_inode_t	*ip_next;
	xfs_buf_t	*bp;
	vnode_t		*vp = NULL;
	vmap_t		vmap;
	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; \
	}