]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
sunrpc: pin svc_xprt across the asynchronous TLS handshake callback
authorChris Mason <clm@meta.com>
Fri, 22 May 2026 13:39:06 +0000 (09:39 -0400)
committerChuck Lever <cel@kernel.org>
Tue, 9 Jun 2026 20:32:59 +0000 (16:32 -0400)
svc_tcp_handshake() stores the raw svc_xprt pointer in
tls_handshake_args.ta_data and submits the request through
tls_server_hello_x509(). The handshake core takes only
sock_hold(req->hr_sk); nothing references the embedding struct
svc_sock that svc_tcp_handshake_done() reaches via container_of().

Two close races leave the in-flight callback writing through a freed
svc_sock. svc_sock_free() calls tls_handshake_cancel() and discards
its return value: a false return means handshake_complete() has
already set HANDSHAKE_F_REQ_COMPLETED but hp_done() may not have
finished, yet svc_sock_free() proceeds to kfree(svsk). The
cancel-loser fall-through inside svc_tcp_handshake() itself produces
the same window: when wait_for_completion_interruptible_timeout()
returns <= 0 (timeout or signal) and tls_handshake_cancel() returns
false, the function does not drain, returns, and svc_handle_xprt()
calls svc_xprt_received(), which clears XPT_BUSY and can drop the
last reference. A concurrent close then runs svc_sock_free() while
svc_tcp_handshake_done() is still updating xpt_flags and walking
svsk->sk_handshake_done.

The corruption surfaces as set_bit/clear_bit RMW into the freed
xpt_flags slab slot and as complete_all() walking and writing the
freed wait_queue_head_t list embedded in sk_handshake_done -- a
slab-corruption primitive, not a benign read. The path is reachable
on any TLS-enabled NFS server whenever a connection close overlaps
the tlshd downcall delivery window; the interruptible wait means
signal delivery suffices, not just SVC_HANDSHAKE_TO expiry.

Take svc_xprt_get(xprt) immediately before tls_server_hello_x509()
so the in-flight callback owns its own reference. Release it on the
two edges where the callback is guaranteed not to fire -- submission
failure from tls_server_hello_x509() and a successful
tls_handshake_cancel() -- and at the tail of
svc_tcp_handshake_done() after complete_all().

Fixes: b3cbf98e2fdf ("SUNRPC: Support TLS handshake in the server-side TCP socket code")
Cc: stable@vger.kernel.org
Signed-off-by: Chris Mason <clm@meta.com>
Assisted-by: kres (claude-opus-4-7)
[cel: rewrote commit message to describe the actual change]
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
net/sunrpc/svcsock.c

index c434b6a6637d9b9dcb7febf7d951ad101358ee97..6a75ac4565db87d85a2a7edf71effd993db31c41 100644 (file)
@@ -471,6 +471,7 @@ static void svc_tcp_handshake_done(void *data, int status, key_serial_t peerid)
        }
        clear_bit(XPT_HANDSHAKE, &xprt->xpt_flags);
        complete_all(&svsk->sk_handshake_done);
+       svc_xprt_put(xprt);
 }
 
 /**
@@ -494,9 +495,13 @@ static void svc_tcp_handshake(struct svc_xprt *xprt)
        clear_bit(XPT_TLS_SESSION, &xprt->xpt_flags);
        init_completion(&svsk->sk_handshake_done);
 
+       /* Pin the transport across the asynchronous handshake callback. */
+       svc_xprt_get(xprt);
+
        ret = tls_server_hello_x509(&args, GFP_KERNEL);
        if (ret) {
                trace_svc_tls_not_started(xprt);
+               svc_xprt_put(xprt);
                goto out_failed;
        }
 
@@ -505,6 +510,7 @@ static void svc_tcp_handshake(struct svc_xprt *xprt)
        if (ret <= 0) {
                if (tls_handshake_cancel(sk)) {
                        trace_svc_tls_timed_out(xprt);
+                       svc_xprt_put(xprt);
                        goto out_close;
                }
        }