Re: [PATCH v2 4/7] SUNRPC: Mark connection-oriented transports on svc_xprt

[Jeff Layton <jlayton@kernel.org>]

On Fri, 2026-08-28 at 12:17 -0400, Chuck Lever wrote:
> A TCP or RDMA svc_xprt carries one client's request stream, delivered
> reliably and in order, so a later request on it indicates that earlier
> replies reached the client. A UDP svc_xprt is shared by every peer and
> offers no such signal.
> 
> Add an XPT_ORDERED flag, set by the TCP and RDMA init paths and clear
> for UDP, so the NFSD duplicate reply cache can confine implied-ACK
> eviction to transports where the signal holds.
> 
> The flag has a cost on svcrdma. handle_connect_req() zeroes the remote
> port in the DRC key so a cached reply survives a reconnect, and
> implied-ACK eviction can retire such an entry while the original
> connection is still up. TCP gives up the same thing whenever a client
> re-binds its reserved port, so this follows from the eviction policy
> rather than from RDMA.
> 
> Signed-off-by: Chuck Lever <cel@kernel.org>
> ---
>  include/linux/sunrpc/svc_xprt.h          | 1 +
>  include/trace/events/sunrpc.h            | 1 +
>  net/sunrpc/svcsock.c                     | 1 +
>  net/sunrpc/xprtrdma/svc_rdma_transport.c | 1 +
>  4 files changed, 4 insertions(+)
> 
> diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
> index 53ed073640a6..ebab18e403b2 100644
> --- a/include/linux/sunrpc/svc_xprt.h
> +++ b/include/linux/sunrpc/svc_xprt.h
> @@ -99,6 +99,7 @@ enum {
>  	XPT_LOCAL,		/* connection from loopback interface */
>  	XPT_KILL_TEMP,		/* call xpo_kill_temp_xprt before closing */
>  	XPT_CONG_CTRL,		/* has congestion control */
> +	XPT_ORDERED,		/* connection-oriented, in-order delivery */
>  	XPT_HANDSHAKE,		/* xprt requests a handshake */
>  	XPT_TLS_SESSION,	/* transport-layer security established */
>  	XPT_PEER_AUTH,		/* peer has been authenticated */
> diff --git a/include/trace/events/sunrpc.h b/include/trace/events/sunrpc.h
> index 180346e520ff..2154fbe69945 100644
> --- a/include/trace/events/sunrpc.h
> +++ b/include/trace/events/sunrpc.h
> @@ -1931,6 +1931,7 @@ TRACE_EVENT(svc_stats_latency,
>  	svc_xprt_flag(LOCAL)						\
>  	svc_xprt_flag(KILL_TEMP)					\
>  	svc_xprt_flag(CONG_CTRL)					\
> +	svc_xprt_flag(ORDERED)						\
>  	svc_xprt_flag(HANDSHAKE)					\
>  	svc_xprt_flag(TLS_SESSION)					\
>  	svc_xprt_flag(PEER_AUTH)					\
> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
> index 840bec53e1ee..97a592ddeb48 100644
> --- a/net/sunrpc/svcsock.c
> +++ b/net/sunrpc/svcsock.c
> @@ -1486,6 +1486,7 @@ static bool svc_tcp_init(struct svc_sock *svsk, struct svc_serv *serv)
>  		return false;
>  	set_bit(XPT_CACHE_AUTH, &svsk->sk_xprt.xpt_flags);
>  	set_bit(XPT_CONG_CTRL, &svsk->sk_xprt.xpt_flags);
> +	set_bit(XPT_ORDERED, &svsk->sk_xprt.xpt_flags);
>  	if (sk->sk_state == TCP_LISTEN) {
>  		strcpy(svsk->sk_xprt.xpt_remotebuf, "listener");
>  		set_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags);
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
> index fabdfd891a41..5d39fce231a3 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
> @@ -218,6 +218,7 @@ static struct svcxprt_rdma *svc_rdma_create_xprt(struct svc_serv *serv,
>  	 * transports are suitable here.
>  	 */
>  	set_bit(XPT_CONG_CTRL, &cma_xprt->sc_xprt.xpt_flags);
> +	set_bit(XPT_ORDERED, &cma_xprt->sc_xprt.xpt_flags);
>  
>  	return cma_xprt;
>  }

It would be helpful to see how this new flag will be used other than in
the tracepoint. Maybe this patch should be folded into the following
one?
-- 
Jeff Layton <jlayton@kernel.org>