[openib-general] Re: ibv_get_async_event

[Roland Dreier]

    Fab> Couldn't an event be "in flight" when the user destroys an
    Fab> object, causing the event to be delivered post-destruction?

Not sure I follow you -- if we sweep the list of pending events and
remove any that relate to the object being destroyed before we return
from the destroy call, then I don't see how userspace can see any
stale events after the destroy call returns.

 - R.