rpcrdma_conn_upcall() converts RDMA CM connection events into
errnos:
RDMA_CM_EVENT_ADDR_ERROR -> EHOSTUNREACH
RDMA_CM_EVENT_ROUTE_ERROR -> ENETUNREACH
RDMA_CM_EVENT_UNREACHABLE -> ENETDOWN
mount.nfs seems to want to retry EHOSTUNREACH and friends forever.
This makes mount.nfs appear to hang if I try to mount with
proto=rdma via a conventional Ethernet device.
If the admin has specified proto=rdma along with a server IP address
that requires a local interface or network path that does not
support RDMA, fail immediately with a permanent error during
transport set up.
Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx>
---
net/sunrpc/xprtrdma/transport.c | 14 ++++++++++----
1 file changed, 10 insertions(+), 4 deletions(-)
Hi all-
I'm probably confused about when certain connection events fire.
I'd like to ensure that the RPC-over-RDMA client implementation
can very quickly detect when RDMA can't be supported so that it
can fall back to trying TCP (IPoIB) or reporting a problem.
diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c
index 68ab522..5f534ba 100644
--- a/net/sunrpc/xprtrdma/transport.c
+++ b/net/sunrpc/xprtrdma/transport.c
@@ -396,12 +396,18 @@
new_xprt = rpcx_to_rdmax(xprt);
rc = rpcrdma_ia_open(new_xprt, sap);
- if (rc)
+ switch (rc) {
+ case 0:
+ break;
+ case -ENETDOWN:
+ case -EHOSTUNREACH:
+ case -ENETUNREACH:
+ rc = -EINVAL;
+ /* fall through */
+ default:
goto out1;
+ }
- /*
- * initialize and create ep
- */
new_xprt->rx_data = cdata;
new_ep = &new_xprt->rx_ep;
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
