[ewg] Re: [PATCH] [RFC] IB/cache: Add ib_cache report for cache in process

[Michael S. Tsirkin]

> >> The new return code (-ESTALE) informs the callers to ib_get_cached_pkey 
> >>and ib_find_cached_pkey that the ib_cache is in process of updating itself and
> >>that the call should be retried if an up to date information is needed.
> >>
> >> Signed-off-by: Moni Levy <monil at voltaire.com>
> >
> >OK, but we still need the code to make ULPs retry failed cache queries,
> >right?
> 
> That does not seem trivial for all the ULPs. Can't we just assume that
> -ESTALE would fail the ULPs instead of misleading them ?

Not if it breaks for a user for reasons outside his control.

> That way we are not making anything behave worse.

Aren't we out to fix some issues?

Anyway, aren't you marking all cache "stale" while most pkeys might be still valid?
Can't this break valid usage in e.g. SRP?

-- 
MST