Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <1e0877bc528f3e9218f0070889c7288a8aaa47ba.camel@kernel.org>
Subject: Re: [PATCH] SUNRPC: Fix potential memory leak in
 xs_udp_send_request()
From:   Jeff Layton <jlayton@kernel.org>
To:     Jianglei Nie <niejianglei2021@163.com>, chuck.lever@oracle.com,
        trond.myklebust@hammerspace.com, anna@kernel.org,
        davem@davemloft.net, edumazet@google.com, kuba@kernel.org,
        pabeni@redhat.com
Cc:     linux-nfs@vger.kernel.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
Date:   Wed, 07 Sep 2022 06:08:25 -0400
In-Reply-To: <20220907071338.56969-1-niejianglei2021@163.com>
References: <20220907071338.56969-1-niejianglei2021@163.com>
Content-Type: text/plain; charset="ISO-8859-15"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4 (3.44.4-1.fc36) 
MIME-Version: 1.0
Precedence: bulk

On Wed, 2022-09-07 at 15:13 +0800, Jianglei Nie wrote:
> xs_udp_send_request() allocates a memory chunk for xdr->bvec with
> xdr_alloc_bvec(). When xprt_sock_sendmsg() finishs, xdr->bvec is not
> released, which will lead to a memory leak.
>=20
> we should release the xdr->bvec with xdr_free_bvec() after
> xprt_sock_sendmsg() like bc_sendto() does.
>=20
> Signed-off-by: Jianglei Nie <niejianglei2021@163.com>
> ---
>  net/sunrpc/xprtsock.c | 1 +
>  1 file changed, 1 insertion(+)
>=20
> diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
> index e976007f4fd0..298182a3c168 100644
> --- a/net/sunrpc/xprtsock.c
> +++ b/net/sunrpc/xprtsock.c
> @@ -958,6 +958,7 @@ static int xs_udp_send_request(struct rpc_rqst *req)
>  		return status;
>  	req->rq_xtime =3D ktime_get();
>  	status =3D xprt_sock_sendmsg(transport->sock, &msg, xdr, 0, 0, &sent);
> +	xdr_free_bvec(xdr);
> =20
>  	dprintk("RPC:       xs_udp_send_request(%u) =3D %d\n",
>  			xdr->len, status);

I think you're probably correct here.

I was thinking we might have a similar bug in svc_tcp_sendmsg, but it
looks like that one gets freed in svc_tcp_sendto.

Reviewed-by: Jeff Layton <jlayton@kernel.org>