[openib-general] Re: [PATCH] ipoib_mcast_restart_task

[Roland Dreier]

    Michael> Roland, Eli spotted the following race that might explain
    Michael> part of the crashes in sendonly complete.

Actually I don't see how this could explain it.  The oops seems to be
happening when the CPU follows the mcast pointer to get mcast->dev,
and even if mcast has been freed, it should still point to valid
kernel memory.

 - R.