[PATCH v3] xfs: check da node block pad field during scrub

[PATCH v3] xfs: check da node block pad field during scrub

[Yuto Ohnuki]

The da node block header (xfs_da3_node_hdr) contains a __pad32 field
that should always be zero. Add a check for this during directory and
attribute btree scrubbing.

Since old kernels may have written non-zero padding without issues, flag
this as an optimization opportunity (preen) rather than corruption.

Signed-off-by: Yuto Ohnuki <ytohnuki@amazon.com>
---
Changes in v3:
- Drop unnecessary be32_to_cpu() on comparison to zero.
- Link to v2: https://lore.kernel.org/all/20260408174033.4165-2-ytohnuki@amazon.com/
Changes in v2:
- Use xchk_da_set_preen instead of xchk_da_set_corrupt since
  old kernels may have written non-zero padding without issues.
- Link to v1: https://lore.kernel.org/all/20260404143043.65268-2-ytohnuki@amazon.com/
---
 fs/xfs/scrub/dabtree.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/scrub/dabtree.c b/fs/xfs/scrub/dabtree.c
index 1a71d36898b1..c2d6ad59d03e 100644
--- a/fs/xfs/scrub/dabtree.c
+++ b/fs/xfs/scrub/dabtree.c
@@ -454,7 +454,12 @@ xchk_da_btree_block(
 			}
 		}
 
-		/* XXX: Check hdr3.pad32 once we know how to fix it. */
+		if (xfs_has_crc(ip->i_mount)) {
+			struct xfs_da3_node_hdr *nodehdr3 = blk->bp->b_addr;
+
+			if (nodehdr3->__pad32)
+				xchk_da_set_preen(ds, level);
+		}
 		break;
 	default:
 		xchk_da_set_corrupt(ds, level);
-- 
2.50.1




Amazon Web Services EMEA SARL, 38 avenue John F. Kennedy, L-1855 Luxembourg, R.C.S. Luxembourg B186284

Amazon Web Services EMEA SARL, Irish Branch, One Burlington Plaza, Burlington Road, Dublin 4, Ireland, branch registration number 908705

Re: [PATCH v3] xfs: check da node block pad field during scrub

[Darrick J. Wong]

On Fri, Apr 10, 2026 at 06:06:15PM +0100, Yuto Ohnuki wrote:
> The da node block header (xfs_da3_node_hdr) contains a __pad32 field
> that should always be zero. Add a check for this during directory and
> attribute btree scrubbing.
> 
> Since old kernels may have written non-zero padding without issues, flag
> this as an optimization opportunity (preen) rather than corruption.
> 
> Signed-off-by: Yuto Ohnuki <ytohnuki@amazon.com>

Should the dabtree write verifier also set __pad32 to zero, in case the
kernel encounters a dabtree block writen out before the memset was added
to xfs_da3_node_create in commit a45086e27dfa21 ("xfs: validate metadata
LSNs against log on v5 superblocks")?

This change on its own is correct though, so
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>

--D

> ---
> Changes in v3:
> - Drop unnecessary be32_to_cpu() on comparison to zero.
> - Link to v2: https://lore.kernel.org/all/20260408174033.4165-2-ytohnuki@amazon.com/
> Changes in v2:
> - Use xchk_da_set_preen instead of xchk_da_set_corrupt since
>   old kernels may have written non-zero padding without issues.
> - Link to v1: https://lore.kernel.org/all/20260404143043.65268-2-ytohnuki@amazon.com/
> ---
>  fs/xfs/scrub/dabtree.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/xfs/scrub/dabtree.c b/fs/xfs/scrub/dabtree.c
> index 1a71d36898b1..c2d6ad59d03e 100644
> --- a/fs/xfs/scrub/dabtree.c
> +++ b/fs/xfs/scrub/dabtree.c
> @@ -454,7 +454,12 @@ xchk_da_btree_block(
>  			}
>  		}
>  
> -		/* XXX: Check hdr3.pad32 once we know how to fix it. */
> +		if (xfs_has_crc(ip->i_mount)) {
> +			struct xfs_da3_node_hdr *nodehdr3 = blk->bp->b_addr;
> +
> +			if (nodehdr3->__pad32)
> +				xchk_da_set_preen(ds, level);
> +		}
>  		break;
>  	default:
>  		xchk_da_set_corrupt(ds, level);
> -- 
> 2.50.1
> 
> 
> 
> 
> Amazon Web Services EMEA SARL, 38 avenue John F. Kennedy, L-1855 Luxembourg, R.C.S. Luxembourg B186284
> 
> Amazon Web Services EMEA SARL, Irish Branch, One Burlington Plaza, Burlington Road, Dublin 4, Ireland, branch registration number 908705
> 
> 
> 
>