On 07/23/2013 04:42 PM, Chris Leech wrote:
> It seems some iSCSI targets (including the Linux kernel target) close
> the TCP connection from the target side immediately after processing a
> session logout.
>
> When a TCP FIN comes in right after the iSCSI logout response,
> iscsi_sw_sk_state_check sees the local socket as not yet being in
> CLOSE_WAIT or CLOSE and logs an error. But the initiator would close
> the connection right after processing the logout response anyway, and
> the error is confusing to admins who just requested that the session be
> shut down.
>
> This adds a check of the session state, and suppresses the error if we
> are in the process of logging out.
>
> Signed-off-by: Chris Leech <cleech@xxxxxxxxxx>
> ---
> drivers/scsi/iscsi_tcp.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/scsi/iscsi_tcp.c b/drivers/scsi/iscsi_tcp.c
> index 9e2588a..add6d15 100644
> --- a/drivers/scsi/iscsi_tcp.c
> +++ b/drivers/scsi/iscsi_tcp.c
> @@ -116,6 +116,7 @@ static inline int iscsi_sw_sk_state_check(struct sock *sk)
> struct iscsi_conn *conn = sk->sk_user_data;
>
> if ((sk->sk_state == TCP_CLOSE_WAIT || sk->sk_state == TCP_CLOSE) &&
> + (conn->session->state != ISCSI_STATE_LOGGING_OUT) &&
> !atomic_read(&sk->sk_rmem_alloc)) {
> ISCSI_SW_TCP_DBG(conn, "TCP_CLOSE|TCP_CLOSE_WAIT\n");
> iscsi_conn_failure(conn, ISCSI_ERR_TCP_CONN_CLOSE);
>
My only concern with the patch is what happens if the target closes
connection due to a logout processing error. The logout timeout will
kick in so we do not hang, but the patch makes it harder to debug those
type of problems. I think the annoyance that this error message causes
is higher than those concerns though since that happens so rarely now.
So patch is ok with me.
Reviewed-by: Mike Christie <michaelc@xxxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
