Re: [patch 1/4] mmu_notifier: Core code

[Christoph Lameter]

On Sat, 26 Jan 2008, Robin Holt wrote:

> > No you cannot do that because there are still callbacks that come later. 
> > The invalidate_all may lead to invalidate_range() doing nothing for this 
> > mm. The ops notifier and the freeing of the structure has to wait until 
> > release().
> 
> Could you be a little more clear here?  If you are saying that the other
> callbacks will need to do work?  I can assure you we will clean up those
> pages and raise memory protections.  It will also be done in a much more
> efficient fashion than the individual callouts.

No the other callbacks need to work in the sense that they can be called. 
You could have them do nothing after an invalidate_all().
But you cannot release the allocated structs needed for list traversal 
etc.

> If, on the other hand, you are saying we can not because of the way
> we traverse the list, can we return a result indicating to the caller
> we would like to be unregistered and then the mmu_notifier code do the
> remove followed by a call to the release notifier?

You would need to release the resources when the release notifier is 
called.

> > That does not sync with the current scheme of the invalidate_range() 
> > hooks. We would have to do a global invalidate early and then place the 
> > other invalidate_range hooks in such a way that none is called in later in 
> > process exit handling.
> 
> But if the notifier is removed from the list following the invalidate_all
> callout, there would be no additional callouts.

Hmmm.... Okay did not think about that. Then you would need to do a 
synchronize_rcu() in invalidate_all()?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/