[openib-general] Re: RFC: ib_cache_event problems

[Michael S. Tsirkin]

Quoting r. Roland Dreier <rdreier at cisco.com>:
> Subject: Re: RFC: ib_cache_event problems
> 
>  > 1. Cache should create ib_update_work objects statically upon hotplug event.
> 
> Seems reasonable, since multiple pending cache update events can be
> safely coalesced into one.
> 
>  > 2. Need a mechanism for cache to consume events which trigger cache updates,
>  > and delay reporting them to ULPs until after cache is updated.
> 
> This seems like overkill to me.

How then can we solve the problem of IPoIB querying the cache as a result of
an event, and getting a stale value? Note we are actually seeing this in
practice when changing pkeys.

> And I don't see how to avoid
> GFP_ATOMIC allocations that might fail, since the cache module would
> need to maintain an arbitrary-length queue of pending events.

IMO order of events is typically not important, so 
we only need to handle up to 6 different events in some kind of bitmask.

-- 
MST