Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: ipoib oops
>
> Michael> Looks like send_wc is NULL. And given that the send
> Michael> handler seems to be always called with wc on the stack,
> Michael> it now appears that it was actually ipoib that triggered
> Michael> some data corruption for umad.
>
> If send_wc is NULL how could we access send_wc->send_buf->context[0]?
Right, I wasnt thinking straight.
static void send_handler(struct ib_mad_agent *agent,
struct ib_mad_send_wc *send_wc)
{
struct ib_umad_file *file = agent->context;
struct ib_umad_packet *timeout;
struct ib_umad_packet *packet = send_wc->send_buf->context[0];
ib_destroy_ah(packet->msg->ah); <----------------------------- here
ib_free_send_mad(packet->msg);
Means that packet is NULL. Hmm.
--
MST
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
