[PATCH 09/26] xfs: make XFS_DQUOT_CLUSTER_SIZE_FSB part of the ondisk format

From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: darrick.wong@oracle.com
Cc: Christoph Hellwig <hch@lst.de>, linux-xfs@vger.kernel.org
Subject: [PATCH 09/26] xfs: make XFS_DQUOT_CLUSTER_SIZE_FSB part of the ondisk format
Date: Tue, 14 Jul 2020 18:51:33 -0700	[thread overview]
Message-ID: <159477789339.3263162.3626376647868941894.stgit@magnolia> (raw)
In-Reply-To: <159477783164.3263162.2564345443708779029.stgit@magnolia>

From: Darrick J. Wong <darrick.wong@oracle.com>

Move the dquot cluster size #define to xfs_format.h.  It is an important
part of the ondisk format because the ondisk dquot record size is not an
even power of two, which means that the buffer size we use is
significant here because the kernel leaves slack space at the end of the
buffer to avoid having to deal with a dquot record crossing a block
boundary.

This is also an excuse to fix one of the longstanding discrepancies
between kernel and userspace libxfs headers.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/libxfs/xfs_format.h |   16 ++++++++++++++++
 fs/xfs/xfs_qm.h            |   11 -----------
 2 files changed, 16 insertions(+), 11 deletions(-)

diff --git a/fs/xfs/libxfs/xfs_format.h b/fs/xfs/libxfs/xfs_format.h
index 79fbabeb476c..76d34b77031a 100644
--- a/fs/xfs/libxfs/xfs_format.h
+++ b/fs/xfs/libxfs/xfs_format.h
@@ -1209,6 +1209,22 @@ typedef struct xfs_dqblk {
 
 #define XFS_DQUOT_CRC_OFF	offsetof(struct xfs_dqblk, dd_crc)
 
+/*
+ * This defines the unit of allocation of dquots.
+ *
+ * Currently, it is just one file system block, and a 4K blk contains 30
+ * (136 * 30 = 4080) dquots. It's probably not worth trying to make
+ * this more dynamic.
+ *
+ * However, if this number is changed, we have to make sure that we don't
+ * implicitly assume that we do allocations in chunks of a single filesystem
+ * block in the dquot/xqm code.
+ *
+ * This is part of the ondisk format because the structure size is not a power
+ * of two, which leaves slack at the end of the disk block.
+ */
+#define XFS_DQUOT_CLUSTER_SIZE_FSB	(xfs_filblks_t)1
+
 /*
  * Remote symlink format and access functions.
  */
diff --git a/fs/xfs/xfs_qm.h b/fs/xfs/xfs_qm.h
index 27789272da95..c5d0716b378e 100644
--- a/fs/xfs/xfs_qm.h
+++ b/fs/xfs/xfs_qm.h
@@ -30,17 +30,6 @@ extern struct kmem_zone	*xfs_qm_dqtrxzone;
 	!dqp->q_core.d_rtbcount && \
 	!dqp->q_core.d_icount)
 
-/*
- * This defines the unit of allocation of dquots.
- * Currently, it is just one file system block, and a 4K blk contains 30
- * (136 * 30 = 4080) dquots. It's probably not worth trying to make
- * this more dynamic.
- * XXXsup However, if this number is changed, we have to make sure that we don't
- * implicitly assume that we do allocations in chunks of a single filesystem
- * block in the dquot/xqm code.
- */
-#define XFS_DQUOT_CLUSTER_SIZE_FSB	(xfs_filblks_t)1
-
 /* Defaults for each quota type: time limits, warn limits, usage limits */
 struct xfs_def_quota {
 	time64_t	btimelimit;	/* limit for blks timer */

