[Cluster-devel] [PATCH] gfs2: Fix refcount leak in gfs2_glock_poke

[Bob Peterson <rpeterso@redhat.com>]

----- Original Message -----
> In gfs2_glock_poke, make sure gfs2_holder_uninit is called on the local
> glock holder.  Without that, we're leaking a glock and a pid reference.
> 
> Fixes: 9e8990dea926 ("gfs2: Smarter iopen glock waiting")
> Cc: stable at vger.kernel.org # v5.8+
> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
> ---
>  fs/gfs2/glock.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
> index 57134d326cfa..f13b136654ca 100644
> --- a/fs/gfs2/glock.c
> +++ b/fs/gfs2/glock.c
> @@ -790,9 +790,11 @@ static void gfs2_glock_poke(struct gfs2_glock *gl)
>  	struct gfs2_holder gh;
>  	int error;
>  
> -	error = gfs2_glock_nq_init(gl, LM_ST_SHARED, flags, &gh);
> +	gfs2_holder_init(gl, LM_ST_SHARED, flags, &gh);
> +	error = gfs2_glock_nq(&gh);
>  	if (!error)
>  		gfs2_glock_dq(&gh);
> +	gfs2_holder_uninit(&gh);
>  }
>  
>  static bool gfs2_try_evict(struct gfs2_glock *gl)
> --
> 2.26.2

Hi,

Looks okay.
I'd rather use nq_init and dq_uninit (and change similar code throughout
to use it more consistently) but in the past you've expressed a
concern that the compiler generates less efficient code.

Maybe it makes more sense to rework gfs2_glock_nq_init and dq_uninit
so the compiler does a more reasonable job? Or I suppose we could
delay that work for later.

Regards,

Bob Peterson