On Wed, Dec 21, 2022 at 10:22:21AM -0500, Hironori Shiina wrote:
> The root inode number should be set to `breq->startino` for getting stat
> information of the root when XFS_BULK_IREQ_SPECIAL_ROOT is used.
> Otherwise, the inode search is started from 1
> (XFS_BULK_IREQ_SPECIAL_ROOT) and the inode with the lowest number in a
> filesystem is returned.
>
> Fixes: bf3cb3944792 ("xfs: allow single bulkstat of special inodes")
> Signed-off-by: Hironori Shiina <shiina.hironori@xxxxxxxxxxx>
D'oh. Nobody noticed this because .... the root dir is usually the
first inode in the first inode chunk, and XFS_BULK_IREQ_SPECIAL_ROOT==1,
so that usually lines us up to find the root dir.
Except in that weird case where we format with a big stripe alignment,
mount with zeroed stripe alignment, and then stuff gets allocated before
the ichunk containing the root inode, right?
> ---
> fs/xfs/xfs_ioctl.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> index 13f1b2add390..020111f0f2a2 100644
> --- a/fs/xfs/xfs_ioctl.c
> +++ b/fs/xfs/xfs_ioctl.c
> @@ -780,7 +780,7 @@ xfs_bulk_ireq_setup(
>
> switch (hdr->ino) {
I might change @hdr to be a const pointer to prevent further accidents
here when I commit this.
Reviewed-by: Darrick J. Wong <djwong@xxxxxxxxxx>
--D
> case XFS_BULK_IREQ_SPECIAL_ROOT:
> - hdr->ino = mp->m_sb.sb_rootino;
> + breq->startino = mp->m_sb.sb_rootino;
> break;
> default:
> return -EINVAL;
> --
> 2.38.1
>
