[Cluster-devel] [GFS2 PATCH v2] GFS2: dirty inode correctly in gfs2_iomap_end for jdata

[Andreas Gruenbacher <agruenba@redhat.com>]

On 3 July 2018 at 16:54, Bob Peterson <rpeterso@redhat.com> wrote:
> Hi,
>
> I posted an equivalent patch on 18 June which was largely ignored.
> This is revised for the recent iomap changes.
> ---
> In patch 0c9018097f, function gfs2_write_end was changed so that it
> marked inodes dirty if they had new buffers queued to the current
> transaction. Unfortunately, it was only done for ordered data files.
> The check was never done for jdata files, so the inode was never
> properly marked dirty. The affected code has since been moved to
> function gfs2_iomap_end. This patch adds the additional check so
> that jdata files are also marked dirty.

Like I just said in the previous posting of this fix, I'd be much
happier to have a test case.

> Signed-off-by: Bob Peterson <rpeterso@redhat.com>
> ---
>  fs/gfs2/bmap.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
> index 49a6ab919bd7..afbee7e4ae5b 100644
> --- a/fs/gfs2/bmap.c
> +++ b/fs/gfs2/bmap.c
> @@ -1116,7 +1116,7 @@ static int gfs2_iomap_end(struct inode *inode, loff_t pos, loff_t length,
>         if (iomap->type != IOMAP_INLINE) {
>                 gfs2_ordered_add_inode(ip);
>
> -               if (tr->tr_num_buf_new)
> +               if (tr->tr_num_buf_new || tr->tr_num_databuf_new)
>                         __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
>                 else
>                         gfs2_trans_add_meta(ip->i_gl, dibh);
>

Thanks,
Andreas