Review: Be smarter about handling ENOSPC during writeback

During delayed allocation extent conversion or unwritten extent conversion, we need to reserve some blocks for transactions reservations. We need to reserve these blocks in case a btree split occurs and we need to allocate some blocks.

Unfortunately, we've only ever reserved the number of data blocks we are allocating, so in both the unwritten and delalloc case we can get ENOSPC to the transaction reservation. This is bad because in both cases we cannot report the failure to the writing application.

The fix is two-fold:

1 - leverage the reserved block infrastructure XFS already has to reserve a small pool of blocks by default to allow specially marked transactions to dip into when we are at ENOSPC.

Default setting is min(5%, 1024 blocks).

2 - convert critical transaction reservations to be allowed to dip into this pool. Spots changed are delalloc conversion, unwritten extent conversion and growing a filesystem at ENOSPC.

Comments?

Cheers,

Dave.
--
Dave Chinner
Principal Engineer
SGI Australian Software Group

---
 fs/xfs/xfs_fsops.c |   10 +++++++---
 fs/xfs/xfs_iomap.c |   22 ++++++++--------------
 fs/xfs/xfs_mount.c |   37 +++++++++++++++++++++++++++++++++++--
 3 files changed, 50 insertions(+), 19 deletions(-)

Index: 2.6.x-xfs-new/fs/xfs/xfs_fsops.c
===================================================================

fs/xfs/xfs_fsops.c +7 -3

@@ -179,6 +179,7 @@ xfs_growfs_data_private(
 		up_write(&mp->m_peraglock);
 	}
 	tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFS);
+	tp->t_flags |= XFS_TRANS_RESERVE;
 	if ((error = xfs_trans_reserve(tp, XFS_GROWFS_SPACE_RES(mp),
 			XFS_GROWDATA_LOG_RES(mp), 0, 0, 0))) {
 		xfs_trans_cancel(tp, 0);

@@ -500,8 +501,9 @@ xfs_reserve_blocks(
 	unsigned long		s;
 
 	/* If inval is null, report current values and return */
-
 	if (inval == (__uint64_t *)NULL) {
+		if (!outval)
+			return EINVAL;
 		outval->resblks = mp->m_resblks;
 		outval->resblks_avail = mp->m_resblks_avail;
 		return 0;

@@ -564,8 +566,10 @@ retry:
 		}
 	}
 out:
-	outval->resblks = mp->m_resblks;
-	outval->resblks_avail = mp->m_resblks_avail;
+	if (outval) {
+		outval->resblks = mp->m_resblks;
+		outval->resblks_avail = mp->m_resblks_avail;
+	}
 	XFS_SB_UNLOCK(mp, s);
 
 	if (fdblks_delta) {

Index: 2.6.x-xfs-new/fs/xfs/xfs_mount.c
===================================================================

fs/xfs/xfs_mount.c +35 -2

@@ -718,7 +718,7 @@ xfs_mountfs(
 	bhv_vnode_t	*rvp = NULL;
 	int		readio_log, writeio_log;
 	xfs_daddr_t	d;
-	__uint64_t	ret64;
+	__uint64_t	resblks;
 	__int64_t	update_flags;
 	uint		quotamount, quotaflags;
 	int		agno;

@@ -835,6 +835,7 @@ xfs_mountfs(
 	 */
 	if ((mfsi_flags & XFS_MFSI_SECOND) == 0 &&
 	    (mp->m_flags & XFS_MOUNT_NOUUID) == 0) {
+		__uint64_t	ret64;
 		if (xfs_uuid_mount(mp)) {
 			error = XFS_ERROR(EINVAL);
 			goto error1;

@@ -1127,13 +1128,27 @@ xfs_mountfs(
 		goto error4;
 	}
 
-
 	/*
 	 * Complete the quota initialisation, post-log-replay component.
 	 */
 	if ((error = XFS_QM_MOUNT(mp, quotamount, quotaflags, mfsi_flags)))
 		goto error4;
 
+	/*
+	 * Now we are mounted, reserve a small amount of unused space for
+	 * privileged transactions. This is needed so that transaction
+	 * space required for critical operations can dip into this pool
+	 * when at ENOSPC. This is needed for operations like create with
+	 * attr, unwritten extent conversion at ENOSPC, etc. Data allocations
+	 * are not allowed to use this reserved space.
+	 *
+	 * We default to 5% or 1024 fsbs of space reserved, whichever is smaller.
+	 * This may drive us straight to ENOSPC on mount, but that implies
+	 * we were already there on the last unmount.
+	 */
+	resblks = min_t(__uint64_t, mp->m_sb.sb_dblocks / 20, 1024);
+	xfs_reserve_blocks(mp, &resblks, NULL);
+
 	return 0;
 
  error4:

@@ -1172,6 +1187,7 @@ xfs_unmountfs(xfs_mount_t *mp, struct cr
 #if defined(DEBUG) || defined(INDUCE_IO_ERROR)
 	int64_t		fsid;
 #endif
+	__uint64_t	resblks;
 
 	/*
 	 * We can potentially deadlock here if we have an inode cluster

@@ -1200,6 +1216,23 @@ xfs_unmountfs(xfs_mount_t *mp, struct cr
 		xfs_binval(mp->m_rtdev_targp);
 	}
 
+	/*
+	 * Unreserve any blocks we have so that when we unmount we don't account
+	 * the reserved free space as used. This is really only necessary for
+	 * lazy superblock counting because it trusts the incore superblock
+	 * counters to be aboslutely correct on clean unmount.
+	 *
+	 * We don't bother correcting this elsewhere for lazy superblock
+	 * counting because on mount of an unclean filesystem we reconstruct the
+	 * correct counter value and this is irrelevant.
+	 *
+	 * For non-lazy counter filesystems, this doesn't matter at all because
+	 * we only every apply deltas to the superblock and hence the incore
+	 * value does not matter....
+	 */
+	resblks = 0;
+	xfs_reserve_blocks(mp, &resblks, NULL);
+
 	xfs_log_sbcount(mp, 1);
 	xfs_unmountfs_writesb(mp);
 	xfs_unmountfs_wait(mp); 		/* wait for async bufs */

Index: 2.6.x-xfs-new/fs/xfs/xfs_iomap.c
===================================================================

fs/xfs/xfs_iomap.c +8 -14

@@ -489,13 +489,13 @@ xfs_iomap_write_direct(
 	if (unlikely(rt)) {
 		resrtextents = qblocks = resaligned;
 		resrtextents /= mp->m_sb.sb_rextsize;
-  		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
-  		quota_flag = XFS_QMOPT_RES_RTBLKS;
-  	} else {
-  		resrtextents = 0;
+		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+		quota_flag = XFS_QMOPT_RES_RTBLKS;
+	} else {
+		resrtextents = 0;
 		resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
-  		quota_flag = XFS_QMOPT_RES_REGBLKS;
-  	}
+		quota_flag = XFS_QMOPT_RES_REGBLKS;
+	}
 
 	/*
 	 * Allocate and setup the transaction

@@ -788,18 +788,12 @@ xfs_iomap_write_allocate(
 		nimaps = 0;
 		while (nimaps == 0) {
 			tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE);
+			tp->t_flags |= XFS_TRANS_RESERVE;
 			nres = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK);
 			error = xfs_trans_reserve(tp, nres,
 					XFS_WRITE_LOG_RES(mp),
 					0, XFS_TRANS_PERM_LOG_RES,
 					XFS_WRITE_LOG_COUNT);
-			if (error == ENOSPC) {
-				error = xfs_trans_reserve(tp, 0,
-						XFS_WRITE_LOG_RES(mp),
-						0,
-						XFS_TRANS_PERM_LOG_RES,
-						XFS_WRITE_LOG_COUNT);
-			}
 			if (error) {
 				xfs_trans_cancel(tp, 0);
 				return XFS_ERROR(error);

@@ -917,8 +911,8 @@ xfs_iomap_write_unwritten(
 		 * from unwritten to real. Do allocations in a loop until
 		 * we have covered the range passed in.
 		 */
-
 		tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE);
+		tp->t_flags |= XFS_TRANS_RESERVE;
 		error = xfs_trans_reserve(tp, resblks,
 				XFS_WRITE_LOG_RES(mp), 0,
 				XFS_TRANS_PERM_LOG_RES,

During delayed allocation extent conversion or unwritten extent conversion, we need to reserve some blocks for transactions reservations. We need to reserve these blocks in case a btree split occurs and we need to allocate some blocks.

Unfortunately, we've only ever reserved the number of data blocks we are allocating, so in both the unwritten and delalloc case we can get ENOSPC to the transaction reservation. This is bad because in both cases we cannot report the failure to the writing application.

The fix is two-fold:

1 - leverage the reserved block infrastructure XFS already has to reserve a small pool of blocks by default to allow specially marked transactions to dip into when we are at ENOSPC.

Default setting is min(5%, 1024 blocks).

2 - convert critical transaction reservations to be allowed to dip into this pool. Spots changed are delalloc conversion, unwritten extent conversion and growing a filesystem at ENOSPC.

Comments?

Cheers,

Dave.
--
Dave Chinner
Principal Engineer
SGI Australian Software Group

---

 fs/xfs/xfs_fsops.c |   10 +++++++---

 fs/xfs/xfs_mount.c |   37 +++++++++++++++++++++++++++++++++++--

 fs/xfs/xfs_iomap.c |   22 ++++++++--------------

 3 files changed, 50 insertions(+), 19 deletions(-)

Index: 2.6.x-xfs-new/fs/xfs/xfs_fsops.c
===================================================================

fs/xfs/xfs_fsops.c +7 -3

@@ -179,6 +179,7 @@ xfs_growfs_data_private(
 		up_write(&mp->m_peraglock);
 	}
 	tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFS);
+	tp->t_flags |= XFS_TRANS_RESERVE;
 	if ((error = xfs_trans_reserve(tp, XFS_GROWFS_SPACE_RES(mp),
 			XFS_GROWDATA_LOG_RES(mp), 0, 0, 0))) {
 		xfs_trans_cancel(tp, 0);

@@ -500,8 +501,9 @@ xfs_reserve_blocks(
 	unsigned long		s;
 
 	/* If inval is null, report current values and return */
-
 	if (inval == (__uint64_t *)NULL) {
+		if (!outval)
+			return EINVAL;
 		outval->resblks = mp->m_resblks;
 		outval->resblks_avail = mp->m_resblks_avail;
 		return 0;

@@ -564,8 +566,10 @@ retry:
 		}
 	}
 out:
-	outval->resblks = mp->m_resblks;
-	outval->resblks_avail = mp->m_resblks_avail;
+	if (outval) {
+		outval->resblks = mp->m_resblks;
+		outval->resblks_avail = mp->m_resblks_avail;
+	}
 	XFS_SB_UNLOCK(mp, s);
 
 	if (fdblks_delta) {

Index: 2.6.x-xfs-new/fs/xfs/xfs_mount.c
===================================================================

fs/xfs/xfs_mount.c +35 -2

@@ -718,7 +718,7 @@ xfs_mountfs(
 	bhv_vnode_t	*rvp = NULL;
 	int		readio_log, writeio_log;
 	xfs_daddr_t	d;
-	__uint64_t	ret64;
+	__uint64_t	resblks;
 	__int64_t	update_flags;
 	uint		quotamount, quotaflags;
 	int		agno;

@@ -835,6 +835,7 @@ xfs_mountfs(
 	 */
 	if ((mfsi_flags & XFS_MFSI_SECOND) == 0 &&
 	    (mp->m_flags & XFS_MOUNT_NOUUID) == 0) {
+		__uint64_t	ret64;
 		if (xfs_uuid_mount(mp)) {
 			error = XFS_ERROR(EINVAL);
 			goto error1;

@@ -1127,13 +1128,27 @@ xfs_mountfs(
 		goto error4;
 	}
 
-
 	/*
 	 * Complete the quota initialisation, post-log-replay component.
 	 */
 	if ((error = XFS_QM_MOUNT(mp, quotamount, quotaflags, mfsi_flags)))
 		goto error4;
 
+	/*
+	 * Now we are mounted, reserve a small amount of unused space for
+	 * privileged transactions. This is needed so that transaction
+	 * space required for critical operations can dip into this pool
+	 * when at ENOSPC. This is needed for operations like create with
+	 * attr, unwritten extent conversion at ENOSPC, etc. Data allocations
+	 * are not allowed to use this reserved space.
+	 *
+	 * We default to 5% or 1024 fsbs of space reserved, whichever is smaller.
+	 * This may drive us straight to ENOSPC on mount, but that implies
+	 * we were already there on the last unmount.
+	 */
+	resblks = min_t(__uint64_t, mp->m_sb.sb_dblocks / 20, 1024);
+	xfs_reserve_blocks(mp, &resblks, NULL);
+
 	return 0;
 
  error4:

@@ -1172,6 +1187,7 @@ xfs_unmountfs(xfs_mount_t *mp, struct cr
 #if defined(DEBUG) || defined(INDUCE_IO_ERROR)
 	int64_t		fsid;
 #endif
+	__uint64_t	resblks;
 
 	/*
 	 * We can potentially deadlock here if we have an inode cluster

@@ -1200,6 +1216,23 @@ xfs_unmountfs(xfs_mount_t *mp, struct cr
 		xfs_binval(mp->m_rtdev_targp);
 	}
 
+	/*
+	 * Unreserve any blocks we have so that when we unmount we don't account
+	 * the reserved free space as used. This is really only necessary for
+	 * lazy superblock counting because it trusts the incore superblock
+	 * counters to be aboslutely correct on clean unmount.
+	 *
+	 * We don't bother correcting this elsewhere for lazy superblock
+	 * counting because on mount of an unclean filesystem we reconstruct the
+	 * correct counter value and this is irrelevant.
+	 *
+	 * For non-lazy counter filesystems, this doesn't matter at all because
+	 * we only every apply deltas to the superblock and hence the incore
+	 * value does not matter....
+	 */
+	resblks = 0;
+	xfs_reserve_blocks(mp, &resblks, NULL);
+
 	xfs_log_sbcount(mp, 1);
 	xfs_unmountfs_writesb(mp);
 	xfs_unmountfs_wait(mp); 		/* wait for async bufs */

Index: 2.6.x-xfs-new/fs/xfs/xfs_iomap.c
===================================================================

fs/xfs/xfs_iomap.c +8 -14

@@ -489,13 +489,13 @@ xfs_iomap_write_direct(
 	if (unlikely(rt)) {
 		resrtextents = qblocks = resaligned;
 		resrtextents /= mp->m_sb.sb_rextsize;
-  		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
-  		quota_flag = XFS_QMOPT_RES_RTBLKS;
-  	} else {
-  		resrtextents = 0;
+		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+		quota_flag = XFS_QMOPT_RES_RTBLKS;
+	} else {
+		resrtextents = 0;
 		resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
-  		quota_flag = XFS_QMOPT_RES_REGBLKS;
-  	}
+		quota_flag = XFS_QMOPT_RES_REGBLKS;
+	}
 
 	/*
 	 * Allocate and setup the transaction

@@ -788,18 +788,12 @@ xfs_iomap_write_allocate(
 		nimaps = 0;
 		while (nimaps == 0) {
 			tp = xfs_trans_alloc(mp, XFS_TRANS_START_WRITE);
+			tp->t_flags |= XFS_TRANS_RESERVE;
 			nres = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK);
 			error = xfs_trans_reserve(tp, nres,
 					XFS_WRITE_LOG_RES(mp),
 					0, XFS_TRANS_PERM_LOG_RES,
 					XFS_WRITE_LOG_COUNT);
-			if (error == ENOSPC) {
-				error = xfs_trans_reserve(tp, 0,
-						XFS_WRITE_LOG_RES(mp),
-						0,
-						XFS_TRANS_PERM_LOG_RES,
-						XFS_WRITE_LOG_COUNT);
-			}
 			if (error) {
 				xfs_trans_cancel(tp, 0);
 				return XFS_ERROR(error);

@@ -917,8 +911,8 @@ xfs_iomap_write_unwritten(
 		 * from unwritten to real. Do allocations in a loop until
 		 * we have covered the range passed in.
 		 */
-
 		tp = xfs_trans_alloc(mp, XFS_TRANS_START_WRITE);
+		tp->t_flags |= XFS_TRANS_RESERVE;
 		error = xfs_trans_reserve(tp, resblks,
 				XFS_WRITE_LOG_RES(mp), 0,
 				XFS_TRANS_PERM_LOG_RES,

Hi Dave,

Putting the xfs_reserve_blocks discussion to the side.... (discussed separately)

Quoted 1 line

fs/xfs/xfs_fsops.c | 10 +++++++---

* allow xfs_reserve_blocks() to handle a null outval so that we can call xfs_reserve_blocks other than thru ioctl,
where we don't care about outval
* xfs_growfs_data_private() or's in XFS_TRANS_RESERVE like we do for root EAs
-> allow growfs transaction to dip in to reserve space

Quoted 1 line

fs/xfs/xfs_mount.c | 37 +++++++++++++++++++++++++++++++++++--

* xfs_mountfs(): cleanup - restrict a variable (ret64) to the block its used in
* xfs_mountfs(): do our xfs_reserve_blocks() for what we think we'll need
- pass NULL for 2nd param to it as we don't care (why we changed xfs_fsops.c)
- defaults to min(1024 FSBs, 5% dblocks)
-> not sure how one would choose this but it sounds big enough

* xfs_unmountfs(): xfs_reserve_blocks of zero and so restoring the sb free counter

Q: so I guess, for DMF systems which presumably turn this stuff on using the ioctl;

we should tell them to stop doing this - they could stuff us up by overriding it
maybe and they don't need to.

Quoted 1 line

fs/xfs/xfs_iomap.c | 22 ++++++++--------------

* some whitespace cleanup
xfs_iomap_write_allocate():
* delalloc extent conversion - mark transaction for reserved blocks space
* don't handle ENOSPC here, as we shouldn't get it now I presume

xfs_iomap_write_unwritten
* unwritten extent conversion - mark trans for reserved blocks

Seems simple enough :)

Will we get questions from people about reduced space from df? :)

Quoted 1 line · Timothy Shimmin

Will we get questions from people about reduced space from df? :)

If we do, I think you just volunteered to write the FAQ entry ;)

Quoted 4 lines · David Chinner

Quoted 1 line · Timothy Shimmin

Will we get questions from people about reduced space from df? :)

Quoted 1 line

If we do, I think you just volunteered to write the FAQ entry ;)

It would be more correct of XFS to start doing the right thing by reporting different values for b_free and b_avail in statfs(2) - this code in xfs_mount.c::xfs_statvfs() ...

    statp->f_bfree = statp->f_bavail =
                            sbp->sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);

During delayed allocation extent conversion or unwritten extent conversion, we need to reserve some blocks for transactions reservations. We need to reserve these blocks in case a btree split occurs and we need to allocate some blocks.

Unfortunately, we've only ever reserved the number of data blocks we are allocating, so in both the unwritten and delalloc case we can get ENOSPC to the transaction reservation. This is bad because in both cases we cannot report the failure to the writing application.

The fix is two-fold:

1 - leverage the reserved block infrastructure XFS already has to reserve a small pool of blocks by default to allow specially marked transactions to dip into when we are at ENOSPC.

Default setting is min(5%, 1024 blocks).

2 - convert critical transaction reservations to be allowed to dip into this pool. Spots changed are delalloc conversion, unwritten extent conversion and growing a filesystem at ENOSPC.

Comments?

Cheers,

Dave.
--
Dave Chinner
Principal Engineer
SGI Australian Software Group

---
 fs/xfs/xfs_fsops.c |   10 +++++++---
 fs/xfs/xfs_iomap.c |   22 ++++++++--------------
 fs/xfs/xfs_mount.c |   37 +++++++++++++++++++++++++++++++++++--
 3 files changed, 50 insertions(+), 19 deletions(-)

Index: 2.6.x-xfs-new/fs/xfs/xfs_fsops.c
===================================================================

fs/xfs/xfs_fsops.c +7 -3

@@ -179,6 +179,7 @@ xfs_growfs_data_private(
 		up_write(&mp->m_peraglock);
 	}
 	tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFS);
+	tp->t_flags |= XFS_TRANS_RESERVE;
 	if ((error = xfs_trans_reserve(tp, XFS_GROWFS_SPACE_RES(mp),
 			XFS_GROWDATA_LOG_RES(mp), 0, 0, 0))) {
 		xfs_trans_cancel(tp, 0);

@@ -500,8 +501,9 @@ xfs_reserve_blocks(
 	unsigned long		s;
 
 	/* If inval is null, report current values and return */
-
 	if (inval == (__uint64_t *)NULL) {
+		if (!outval)
+			return EINVAL;
 		outval->resblks = mp->m_resblks;
 		outval->resblks_avail = mp->m_resblks_avail;
 		return 0;

@@ -564,8 +566,10 @@ retry:
 		}
 	}
 out:
-	outval->resblks = mp->m_resblks;
-	outval->resblks_avail = mp->m_resblks_avail;
+	if (outval) {
+		outval->resblks = mp->m_resblks;
+		outval->resblks_avail = mp->m_resblks_avail;
+	}
 	XFS_SB_UNLOCK(mp, s);
 
 	if (fdblks_delta) {

Index: 2.6.x-xfs-new/fs/xfs/xfs_mount.c
===================================================================

fs/xfs/xfs_mount.c +35 -2

@@ -718,7 +718,7 @@ xfs_mountfs(
 	bhv_vnode_t	*rvp = NULL;
 	int		readio_log, writeio_log;
 	xfs_daddr_t	d;
-	__uint64_t	ret64;
+	__uint64_t	resblks;
 	__int64_t	update_flags;
 	uint		quotamount, quotaflags;
 	int		agno;

@@ -835,6 +835,7 @@ xfs_mountfs(
 	 */
 	if ((mfsi_flags & XFS_MFSI_SECOND) == 0 &&
 	    (mp->m_flags & XFS_MOUNT_NOUUID) == 0) {
+		__uint64_t	ret64;
 		if (xfs_uuid_mount(mp)) {
 			error = XFS_ERROR(EINVAL);
 			goto error1;

@@ -1127,13 +1128,27 @@ xfs_mountfs(
 		goto error4;
 	}
 
-
 	/*
 	 * Complete the quota initialisation, post-log-replay component.
 	 */
 	if ((error = XFS_QM_MOUNT(mp, quotamount, quotaflags, mfsi_flags)))
 		goto error4;
 
+	/*
+	 * Now we are mounted, reserve a small amount of unused space for
+	 * privileged transactions. This is needed so that transaction
+	 * space required for critical operations can dip into this pool
+	 * when at ENOSPC. This is needed for operations like create with
+	 * attr, unwritten extent conversion at ENOSPC, etc. Data allocations
+	 * are not allowed to use this reserved space.
+	 *
+	 * We default to 5% or 1024 fsbs of space reserved, whichever is smaller.
+	 * This may drive us straight to ENOSPC on mount, but that implies
+	 * we were already there on the last unmount.
+	 */
+	resblks = min_t(__uint64_t, mp->m_sb.sb_dblocks / 20, 1024);
+	xfs_reserve_blocks(mp, &resblks, NULL);
+
 	return 0;
 
  error4:

@@ -1172,6 +1187,7 @@ xfs_unmountfs(xfs_mount_t *mp, struct cr
 #if defined(DEBUG) || defined(INDUCE_IO_ERROR)
 	int64_t		fsid;
 #endif
+	__uint64_t	resblks;
 
 	/*
 	 * We can potentially deadlock here if we have an inode cluster

@@ -1200,6 +1216,23 @@ xfs_unmountfs(xfs_mount_t *mp, struct cr
 		xfs_binval(mp->m_rtdev_targp);
 	}
 
+	/*
+	 * Unreserve any blocks we have so that when we unmount we don't account
+	 * the reserved free space as used. This is really only necessary for
+	 * lazy superblock counting because it trusts the incore superblock
+	 * counters to be aboslutely correct on clean unmount.
+	 *
+	 * We don't bother correcting this elsewhere for lazy superblock
+	 * counting because on mount of an unclean filesystem we reconstruct the
+	 * correct counter value and this is irrelevant.
+	 *
+	 * For non-lazy counter filesystems, this doesn't matter at all because
+	 * we only every apply deltas to the superblock and hence the incore
+	 * value does not matter....
+	 */
+	resblks = 0;
+	xfs_reserve_blocks(mp, &resblks, NULL);
+
 	xfs_log_sbcount(mp, 1);
 	xfs_unmountfs_writesb(mp);
 	xfs_unmountfs_wait(mp); 		/* wait for async bufs */

Index: 2.6.x-xfs-new/fs/xfs/xfs_iomap.c
===================================================================

fs/xfs/xfs_iomap.c +8 -14

@@ -489,13 +489,13 @@ xfs_iomap_write_direct(
 	if (unlikely(rt)) {
 		resrtextents = qblocks = resaligned;
 		resrtextents /= mp->m_sb.sb_rextsize;
-  		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
-  		quota_flag = XFS_QMOPT_RES_RTBLKS;
-  	} else {
-  		resrtextents = 0;
+		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+		quota_flag = XFS_QMOPT_RES_RTBLKS;
+	} else {
+		resrtextents = 0;
 		resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
-  		quota_flag = XFS_QMOPT_RES_REGBLKS;
-  	}
+		quota_flag = XFS_QMOPT_RES_REGBLKS;
+	}
 
 	/*
 	 * Allocate and setup the transaction

@@ -788,18 +788,12 @@ xfs_iomap_write_allocate(
 		nimaps = 0;
 		while (nimaps == 0) {
 			tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE);
+			tp->t_flags |= XFS_TRANS_RESERVE;
 			nres = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK);
 			error = xfs_trans_reserve(tp, nres,
 					XFS_WRITE_LOG_RES(mp),
 					0, XFS_TRANS_PERM_LOG_RES,
 					XFS_WRITE_LOG_COUNT);
-			if (error == ENOSPC) {
-				error = xfs_trans_reserve(tp, 0,
-						XFS_WRITE_LOG_RES(mp),
-						0,
-						XFS_TRANS_PERM_LOG_RES,
-						XFS_WRITE_LOG_COUNT);
-			}
 			if (error) {
 				xfs_trans_cancel(tp, 0);
 				return XFS_ERROR(error);

@@ -917,8 +911,8 @@ xfs_iomap_write_unwritten(
 		 * from unwritten to real. Do allocations in a loop until
 		 * we have covered the range passed in.
 		 */
-
 		tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE);
+		tp->t_flags |= XFS_TRANS_RESERVE;
 		error = xfs_trans_reserve(tp, resblks,
 				XFS_WRITE_LOG_RES(mp), 0,
 				XFS_TRANS_PERM_LOG_RES,

Hi Dave,

As an aside,
I don't understand the following bit of code in xfs_reserve_blocks():

       /*
        * If our previous reservation was larger than the current value,
        * then move any unused blocks back to the free pool.
        */
       fdblks_delta = 0;
       if (mp->m_resblks > request) {
               lcounter = mp->m_resblks_avail - request;
               if (lcounter  > 0) {            /* release unused blocks */
                       fdblks_delta = lcounter;
                       mp->m_resblks_avail -= lcounter;
               }

I can't see why it is not calculating a delta for:
delta = m_resblks - request

and using that to update the m_resblks_avail and fdblks_delta; like we do in the case below where the request is the larger one. Instead it is affectively doing:
m_resblks_avail = m_resblks_avail - m_resblks_avail + request
= request

It looks wrong to me.
What am I missing?
And why doesn't sb_fdblocks need to be updated in this case.