From: Darrick J. Wong <djwong@xxxxxxxxxx>
The comment for xfs_alloc_set_aside indicates that we want to set aside
enough space to handle a bmap btree split. The code, unfortunately,
hardcodes this to 4.
This is incorrect, since file bmap btrees can be taller than that:
xfs_db> btheight bmapbt -n 4294967296 -b 512
bmapbt: worst case per 512-byte block: 13 records (leaf) / 13 keyptrs (node)
level 0: 4294967296 records, 330382100 blocks
level 1: 330382100 records, 25414008 blocks
level 2: 25414008 records, 1954924 blocks
level 3: 1954924 records, 150379 blocks
level 4: 150379 records, 11568 blocks
level 5: 11568 records, 890 blocks
level 6: 890 records, 69 blocks
level 7: 69 records, 6 blocks
level 8: 6 records, 1 block
9 levels, 357913945 blocks total
Fix this by using the actual bmap btree maxlevel value for the
set-aside. We subtract one because the root is always in the inode and
hence never splits.
Signed-off-by: Darrick J. Wong <djwong@xxxxxxxxxx>
---
fs/xfs/libxfs/xfs_alloc.c | 7 +++++--
fs/xfs/libxfs/xfs_sb.c | 2 --
fs/xfs/xfs_mount.c | 7 +++++++
3 files changed, 12 insertions(+), 4 deletions(-)
diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
index b0678e96ce61..747b3e45303f 100644
--- a/fs/xfs/libxfs/xfs_alloc.c
+++ b/fs/xfs/libxfs/xfs_alloc.c
@@ -107,13 +107,16 @@ xfs_prealloc_blocks(
* aside a few blocks which will not be reserved in delayed allocation.
*
* For each AG, we need to reserve enough blocks to replenish a totally empty
- * AGFL and 4 more to handle a potential split of the file's bmap btree.
+ * AGFL and enough to handle a potential split of a file's bmap btree.
*/
unsigned int
xfs_alloc_set_aside(
struct xfs_mount *mp)
{
- return mp->m_sb.sb_agcount * (XFS_ALLOCBT_AGFL_RESERVE + 4);
+ unsigned int bmbt_splits;
+
+ bmbt_splits = max(mp->m_bm_maxlevels[0], mp->m_bm_maxlevels[1]) - 1;
+ return mp->m_sb.sb_agcount * (XFS_ALLOCBT_AGFL_RESERVE + bmbt_splits);
}
/*
diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c
index f4e84aa1d50a..b823beb944e4 100644
--- a/fs/xfs/libxfs/xfs_sb.c
+++ b/fs/xfs/libxfs/xfs_sb.c
@@ -887,8 +887,6 @@ xfs_sb_mount_common(
mp->m_refc_mnr[1] = mp->m_refc_mxr[1] / 2;
mp->m_bsize = XFS_FSB_TO_BB(mp, 1);
- mp->m_alloc_set_aside = xfs_alloc_set_aside(mp);
- mp->m_ag_max_usable = xfs_alloc_ag_max_usable(mp);
}
/*
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index bed73e8002a5..9336176dc706 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -652,6 +652,13 @@ xfs_mountfs(
xfs_agbtree_compute_maxlevels(mp);
+ /*
+ * Compute the amount of space to set aside to handle btree splits now
+ * that we have calculated the btree maxlevels.
+ */
+ mp->m_alloc_set_aside = xfs_alloc_set_aside(mp);
+ mp->m_ag_max_usable = xfs_alloc_ag_max_usable(mp);
+
/*
* Check if sb_agblocks is aligned at stripe boundary. If sb_agblocks
* is NOT aligned turn off m_dalign since allocator alignment is within
