[PATCH AUTOSEL 5.14 36/99] xprtrdma: Disconnect after an ib_post_send() immediate error

From: Sasha Levin
Date: Thu Sep 09 2021 - 20:41:57 EST


From: Chuck Lever <chuck.lever@xxxxxxxxxx>

[ Upstream commit 1143129e4d0d27740ce680d2fb0161ad4f27aa7e ]

ib_post_send() does not disconnect the QP when it returns an
immediate error. Thus, the code that posts LocalInv has to
explicitly disconnect after an immediate error. This is just
like the frwr_send() callers handle it.

If a disconnect isn't done here, the transport deadlocks.

Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx>
Signed-off-by: Anna Schumaker <Anna.Schumaker@xxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
net/sunrpc/xprtrdma/frwr_ops.c | 8 ++++++++
net/sunrpc/xprtrdma/verbs.c | 2 +-
net/sunrpc/xprtrdma/xprt_rdma.h | 1 +
3 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c
index 229fcc9a9064..754c5dffe127 100644
--- a/net/sunrpc/xprtrdma/frwr_ops.c
+++ b/net/sunrpc/xprtrdma/frwr_ops.c
@@ -557,6 +557,10 @@ void frwr_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)

/* On error, the MRs get destroyed once the QP has drained. */
trace_xprtrdma_post_linv_err(req, rc);
+
+ /* Force a connection loss to ensure complete recovery.
+ */
+ rpcrdma_force_disconnect(ep);
}

/**
@@ -653,4 +657,8 @@ void frwr_unmap_async(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
* retransmission.
*/
rpcrdma_unpin_rqst(req->rl_reply);
+
+ /* Force a connection loss to ensure complete recovery.
+ */
+ rpcrdma_force_disconnect(ep);
}
diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index 649c23518ec0..c1797ea19418 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -124,7 +124,7 @@ static void rpcrdma_xprt_drain(struct rpcrdma_xprt *r_xprt)
* connection is closed or lost. (The important thing is it needs
* to be invoked "at least" once).
*/
-static void rpcrdma_force_disconnect(struct rpcrdma_ep *ep)
+void rpcrdma_force_disconnect(struct rpcrdma_ep *ep)
{
if (atomic_add_unless(&ep->re_force_disconnect, 1, 1))
xprt_force_disconnect(ep->re_xprt);
diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
index 5d231d94e944..927e20a2c04e 100644
--- a/net/sunrpc/xprtrdma/xprt_rdma.h
+++ b/net/sunrpc/xprtrdma/xprt_rdma.h
@@ -454,6 +454,7 @@ extern unsigned int xprt_rdma_memreg_strategy;
/*
* Endpoint calls - xprtrdma/verbs.c
*/
+void rpcrdma_force_disconnect(struct rpcrdma_ep *ep);
void rpcrdma_flush_disconnect(struct rpcrdma_xprt *r_xprt, struct ib_wc *wc);
int rpcrdma_xprt_connect(struct rpcrdma_xprt *r_xprt);
void rpcrdma_xprt_disconnect(struct rpcrdma_xprt *r_xprt);
--
2.30.2