Re: [PATCH 2/6] xfs: actually set aside enough space to handle a bmbt split

[Brian Foster <bfoster@redhat.com>]

On Thu, Mar 17, 2022 at 02:21:06PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <djwong@kernel.org>
> 
> 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@kernel.org>
> ---
>  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/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.
> +	 */

"... to handle btree splits near -ENOSPC ..." ?

Otherwise LGTM:

Reviewed-by: Brian Foster <bfoster@redhat.com>

> +	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
>