Re: [PATCH 1/2] rxrpc: Fix rxrpc_peer leak in rxrpc_look_up_bundle()

[Marc Dionne]

On Wed, Nov 24, 2021 at 10:37 AM David Howells <dhowells@xxxxxxxxxx> wrote:
>
> From: Eiichi Tsukata <eiichi.tsukata@xxxxxxxxxxx>
>
> Need to call rxrpc_put_peer() for bundle candidate before kfree() as it
> holds a ref to rxrpc_peer.
>
> [DH: v2: Changed to abstract out the bundle freeing code into a function]
>
> Fixes: 245500d853e9 ("rxrpc: Rewrite the client connection manager")
> Signed-off-by: Eiichi Tsukata <eiichi.tsukata@xxxxxxxxxxx>
> Signed-off-by: David Howells <dhowells@xxxxxxxxxx>
> cc: Marc Dionne <marc.dionne@xxxxxxxxxxxx>
> cc: linux-afs@xxxxxxxxxxxxxxxxxxx
> Link: https://lore.kernel.org/r/20211121041608.133740-1-eiichi.tsukata@xxxxxxxxxxx/ # v1
> ---
>
>  net/rxrpc/conn_client.c |   14 +++++++++-----
>  1 file changed, 9 insertions(+), 5 deletions(-)
>
> diff --git a/net/rxrpc/conn_client.c b/net/rxrpc/conn_client.c
> index dbea0bfee48e..8120138dac01 100644
> --- a/net/rxrpc/conn_client.c
> +++ b/net/rxrpc/conn_client.c
> @@ -135,16 +135,20 @@ struct rxrpc_bundle *rxrpc_get_bundle(struct rxrpc_bundle *bundle)
>         return bundle;
>  }
>
> +static void rxrpc_free_bundle(struct rxrpc_bundle *bundle)
> +{
> +       rxrpc_put_peer(bundle->params.peer);
> +       kfree(bundle);
> +}
> +
>  void rxrpc_put_bundle(struct rxrpc_bundle *bundle)
>  {
>         unsigned int d = bundle->debug_id;
>         unsigned int u = atomic_dec_return(&bundle->usage);
>
>         _debug("PUT B=%x %u", d, u);
> -       if (u == 0) {
> -               rxrpc_put_peer(bundle->params.peer);
> -               kfree(bundle);
> -       }
> +       if (u == 0)
> +               rxrpc_free_bundle(bundle);
>  }
>
>  /*
> @@ -328,7 +332,7 @@ static struct rxrpc_bundle *rxrpc_look_up_bundle(struct rxrpc_conn_parameters *c
>         return candidate;
>
>  found_bundle_free:
> -       kfree(candidate);
> +       rxrpc_free_bundle(candidate);
>  found_bundle:
>         rxrpc_get_bundle(bundle);
>         spin_unlock(&local->client_bundles_lock);

Reviewed-by: Marc Dionne <marc.dionne@xxxxxxxxxxxx>

Marc