Thanks, I applied the version below.
> This is an initial patch. Maybe it will be more efficient to initiate a
> reconnect
> in case we get IB_CM_DREQ_RECEIVED. What do you think?
I guess it depends on when real targets would send DREQs. In general
if a target is saying it wants to close a connection, it seems sort of
rude to try and reconnect immediately...
- R.
Index: infiniband/ulp/srp/ib_srp.c
===================================================================
--- infiniband/ulp/srp/ib_srp.c (revision 7398)
+++ infiniband/ulp/srp/ib_srp.c (working copy)
@@ -1201,11 +1201,10 @@ static int srp_cm_handler(struct ib_cm_i
srp_cm_rej_handler(cm_id, event, target);
break;
- case IB_CM_MRA_RECEIVED:
- printk(KERN_ERR PFX "MRA received\n");
- break;
-
- case IB_CM_DREP_RECEIVED:
+ case IB_CM_DREQ_RECEIVED:
+ printk(KERN_WARNING PFX "DREQ received - connection closed\n");
+ if (ib_send_cm_drep(cm_id, NULL, 0))
+ printk(KERN_ERR PFX "Sending CM DREP failed\n");
break;
case IB_CM_TIMEWAIT_EXIT:
@@ -1215,6 +1214,11 @@ static int srp_cm_handler(struct ib_cm_i
target->status = 0;
break;
+ case IB_CM_MRA_RECEIVED:
+ case IB_CM_DREQ_ERROR:
+ case IB_CM_DREP_RECEIVED:
+ break;
+
default:
printk(KERN_WARNING PFX "Unhandled CM event %d\n",
event->event);
break;
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
