Re: [PATCH 3/3] xfs_repair: clear the needsrepair flag

[Brian Foster <bfoster@redhat.com>]

On Fri, Jan 08, 2021 at 10:28:16PM -0800, Darrick J. Wong wrote:
> From: Darrick J. Wong <djwong@kernel.org>
> 
> Clear the needsrepair flag, since it's used to prevent mounting of an
> inconsistent filesystem.
> 
> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
> ---
>  repair/agheader.c |   11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> 
> diff --git a/repair/agheader.c b/repair/agheader.c
> index 8bb99489..f6174dbf 100644
> --- a/repair/agheader.c
> +++ b/repair/agheader.c
> @@ -452,6 +452,17 @@ secondary_sb_whack(
>  			rval |= XR_AG_SB_SEC;
>  	}
>  
> +	if (xfs_sb_version_needsrepair(sb)) {
> +		if (!no_modify)
> +			sb->sb_features_incompat &=
> +					~XFS_SB_FEAT_INCOMPAT_NEEDSREPAIR;
> +		if (!do_bzero) {
> +			rval |= XR_AG_SB;
> +			do_warn(_("needsrepair flag set in sb %d\n"), i);
> +		} else
> +			rval |= XR_AG_SB_SEC;
> +	}
> +

Looks reasonable modulo the questions on the previous patch. When I give
this a test, one thing that stands out is that the needsrepair state
itself sort of presents as corruption. I.e.,

# ./db/xfs_db -x -c "version needsrepair" <dev>
Upgrading V5 filesystem
Upgraded V5 filesystem.  Please run xfs_repair.
versionnum [0xb4a5+0x18a] =
V5,NLINK,DIRV2,ALIGN,LOGV2,EXTFLG,MOREBITS,ATTR2,LAZYSBCOUNT,PROJID32BIT,CRC,FTYPE,FINOBT,SPARSE_INODES,REFLINK,NEEDSREPAIR
# ./repair/xfs_repair <dev>
Phase 1 - find and verify superblock...
Phase 2 - using internal log
        - zero log...
        - scan filesystem freespace and inode maps...
needsrepair flag set in sb 1
reset bad sb for ag 1
needsrepair flag set in sb 2
reset bad sb for ag 2
needsrepair flag set in sb 0
reset bad sb for ag 0
needsrepair flag set in sb 3
reset bad sb for ag 3
        - found root inode chunk
Phase 3 - for each AG...
...

So nothing was ever done to this fs besides set and clear the bit. Not a
huge deal, but I wonder if we should print something more user friendly
to indicate that repair found and cleared the needsrepair state, or at
least just avoid the "reset bad sb ..." message for the needsrepair
case.

Brian

>  	return(rval);
>  }
>  
>