On Fri, Sep 09, 2016 at 11:51:48AM +0200, Peter Zijlstra wrote:
> Completions and semaphores don't work? And yes, I need to look at that
> cross-release muck, but as is that stuff sets my teeth on edge.
Completions can be used as hacks for some of it - we have two or three
places where we do that in XFS. Using semaphores doesn't seem very
popular. Also I'd much prefer to have a proper lock instead of working
around it, most importantly to get good lockdep support.
And none of that addresses the fact that we're talking about a
shared/exclusive lock here.
> > I think everyone would be better server by accepting
> > that this case exists and finding a place for it in the framework.
> > E.g. for RT trying to boost something that is fully under control
> > of hardware is pointless, but if we have a way to transfer a lock
> > from an owner to a hardware owned state we could at least boost
> > until that handoff happened.
>
> Could be worse than pointless, could indicate borkage.
Yes - pointless is still the best case.
> But yes, once you
> have that event you could propagate it up the PI chain and notify
> things.
>
> IO rarely is deterministic, so having RT tasks in a blocked-on chain
> with it is fail. And yes, there's exceptions etc..
That's often true, but not always. There is things like battery backed
DRAM which is very deterministic, and there is a lot of work going on
to provide relatively deterministic ways of using flash storage.
|