Hi,
Today, I've found a weird one. I had my IAS server LASP which
was stuck in CONN_PEND state. Nothing I could do could unstuck the
socket (short of restarting IrDA). Of course, beeeing the IAS server,
I could not connect to this box.
Looking in the code, I don't really understand how it
happened, and it might be a very rare and weird race
condition. However I made a little patch that should unstuck to
socket, should that happen again.
This patch is untested. I mean, with the patch, my IrDA stack
still work perfectly, but I could not reproduce the conditions leading
to that situation.
That's it...
Jean
diff -u -p linux/net/irda/irlmp_event.d7.c linux/net/irda/irlmp_event.c
--- linux/net/irda/irlmp_event.d7.c Tue Feb 20 14:14:33 2001
+++ linux/net/irda/irlmp_event.c Tue Feb 20 14:41:31 2001
@@ -472,8 +472,6 @@ static int irlmp_state_disconnected(stru
irlmp_start_watchdog_timer(self, 5*HZ);
break;
case LM_CONNECT_INDICATION:
- irlmp_next_lsap_state(self, LSAP_CONNECT_PEND);
-
if (self->conn_skb) {
WARNING(__FUNCTION__
"(), busy with another request!\n");
@@ -481,6 +479,8 @@ static int irlmp_state_disconnected(stru
}
self->conn_skb = skb;
+ irlmp_next_lsap_state(self, LSAP_CONNECT_PEND);
+
irlmp_do_lap_event(self->lap, LM_LAP_CONNECT_REQUEST, NULL);
break;
default:
@@ -562,6 +562,15 @@ static int irlmp_state_connect_pend(stru
switch (event) {
case LM_CONNECT_REQUEST:
/* Keep state */
+ break;
+ case LM_CONNECT_INDICATION:
+ /* Will happen in some rare cases when the socket get stuck,
+ * the other side retries the connect request.
+ * We just unstuck the socket - Jean II */
+ IRDA_DEBUG(0, __FUNCTION__ "(), LM_CONNECT_INDICATION, "
+ "LSAP stuck in CONNECT_PEND state...\n");
+ /* Keep state */
+ irlmp_do_lap_event(self->lap, LM_LAP_CONNECT_REQUEST, NULL);
break;
case LM_CONNECT_RESPONSE:
IRDA_DEBUG(0, __FUNCTION__ "(), LM_CONNECT_RESPONSE, "
