Re: [PATCH 10/11] tracing/perf: Fix lock events recursions in the fast path

[Frederic Weisbecker <fweisbec@gmail.com>]

On Sun, Feb 07, 2010 at 10:45:55AM +0100, Peter Zijlstra wrote:
> On Sat, 2010-02-06 at 17:10 +0100, Frederic Weisbecker wrote:
> > 
> > > > Looks pretty what I'm looking for. Except that it still continues
> > > > to fill and keep track of the locks held by the current thread,
> > > > namely the copies in curr->held_locks.
> > > 
> > > Which is exactly what you need for that lock hierarchy recording you
> > > wanted :-)
> > 
> > 
> > Well, the lock hierarchy should probably be retrieved from the traces,
> > using state machines.
> > Otherwise we would need yet other lock events for that, which is going
> > to add even more overhead.
> 
> 
> Right, well you could look at adding a mode that also strips out the
> held_lock tracing, but since you really need the register class stuff to
> re-generate the class mapping, avoiding the held_lock tracing doesn't
> look like it's going to save you much, its all thread local storage.


But if we draw a tree based representation (per-instances or per-class)
from perf lock, we'll already need to get the locking scenarios from
post processing, which induce the dependencies.

I don't think it will help much.

Thanks.