On Sat, Jun 28, 2003 at 06:58:38PM -0400, Thor Lancelot Simon wrote:
> On Sat, Jun 28, 2003 at 06:52:11PM +0200, Manuel Bouyer wrote:
> > 
> > If noone object, I'll commit this in a few days.
> > Note that this doesn't address the other issue raised by Chuck: commit and
> > truncate ops needs to be serialised. I'll let peoples familiar with
> > vnode locking issues handle it.
> 
> They should not only be serialized, they should be _gathered_.  This is
> a path out of the "NFS commit murders LFS performance" swamp and could
> greatly help performance on FFS as well under certain conditions.

You're probably right on this. However, in the current state we have
a race condition which can lead to panic. This is different from the
performance problems, even if this affects the same part of code.
If there is a simple way to fix the race, when a more sophiscticated way
exists which would also improve performances exists but require more time,
we should first fix the race, and then work on the second solution.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--