From: Dominique Martinet <dominique.martinet@xxxxxx>
9p/rdma would sometimes drop the connection and display errors in
recv_done when the user does ^C.
The errors were IB_WC_WR_FLUSH_ERR caused by recv buffers that were
posted at the time of disconnect, and we just do not want to disconnect
when down_interruptible is... interrupted.
As a bonus, re-evaluate the log importance of post_recv() failing: we
want to see that one if it ever happens.
Signed-off-by: Dominique Martinet <dominique.martinet@xxxxxx>
---
this had actually been bugging me forever, but I couldn't test the refcount
stuff properly around flush with that so I had to finally figure it out..
net/9p/trans_rdma.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c
index 9cc9b3a19ee7..9719bc4d9424 100644
--- a/net/9p/trans_rdma.c
+++ b/net/9p/trans_rdma.c
@@ -477,7 +477,7 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req)
err = post_recv(client, rpl_context);
if (err) {
- p9_debug(P9_DEBUG_FCALL, "POST RECV failed\n");
+ p9_debug(P9_DEBUG_ERROR, "POST RECV failed: %d\n", err);
goto recv_error;
}
/* remove posted receive buffer from request structure */
@@ -546,7 +546,7 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req)
recv_error:
kfree(rpl_context);
spin_lock_irqsave(&rdma->req_lock, flags);
- if (rdma->state < P9_RDMA_CLOSING) {
+ if (err != -EINTR && rdma->state < P9_RDMA_CLOSING) {
rdma->state = P9_RDMA_CLOSING;
spin_unlock_irqrestore(&rdma->req_lock, flags);
rdma_disconnect(rdma->cm_id);
--
2.17.1
