On Sat, 2010-03-06 at 11:46 +0800, Bian Naimeng wrote:
> If NFSv4 client send a request before connect, or the old connection was broken
> because a ETIMEOUT error catched by call_status, ->send_request will return
> ENOSOCK, but rpc layer can not deal with it, so translate it into ENOCONN.
>
> Signed-off-by: Bian Naimeng <biannm@xxxxxxxxxxxxxx>
>
> ---
> net/sunrpc/xprtsock.c | 18 ++++++++----------
> 1 files changed, 8 insertions(+), 10 deletions(-)
>
> diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
> index 3d739e5..45e5fd8 100644
> --- a/net/sunrpc/xprtsock.c
> +++ b/net/sunrpc/xprtsock.c
> @@ -549,14 +549,13 @@ static int xs_udp_send_request(struct rpc_task *task)
> /* Still some bytes left; set up for a retry later. */
> status = -EAGAIN;
> }
> - if (!transport->sock)
> + if (!transport->sock || status == -ENOTSOCK) {
> + /* Should we call xs_close() here? */
> + status = -ENOTCONN;
> goto out;
> + }
>
> switch (status) {
> - case -ENOTSOCK:
> - status = -ENOTCONN;
> - /* Should we call xs_close() here? */
> - break;
> case -EAGAIN:
> status = xs_nospace(task);
> break;
> @@ -652,14 +651,13 @@ static int xs_tcp_send_request(struct rpc_task *task)
> status = -EAGAIN;
> break;
> }
> - if (!transport->sock)
> + if (!transport->sock || status == -ENOTSOCK) {
> + /* Should we call xs_close() here? */
> + status = -ENOTCONN;
> goto out;
> + }
>
> switch (status) {
> - case -ENOTSOCK:
> - status = -ENOTCONN;
> - /* Should we call xs_close() here? */
> - break;
> case -EAGAIN:
> status = xs_nospace(task);
> break;
The only case where xs_sendpages() will return ENOTSOCK is if
transport->sock == NULL, so the correct fix here is to just remove that
redundant test. There is no need to move the test for ENOTSOCK out of
the switch statement.
Cheers
Trond
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
