On Tue, Jun 28, 2005 at 07:00:34PM +0200, Manuel Bouyer wrote:
> On Tue, Jun 28, 2005 at 09:05:11AM -0700, Chuck Silvers wrote:
> > > Yes, but this requires an interface change in the vnode op, which may not
> > > be doable on the existing branches. In addition, this code doens't exist
> > > yet.
> > 
> > my point is that it does not seem correct to update timestamps as part of
> > cache-flushing.  whether or not the incorrect change is doable on a branch
> > is irrelevant.
> 
> It is if the "correct" fix can't be pulled up to branches.

after some thought I realized that the change you were proposing yesterday
had much worse problems.  every vnode that left the vnode cache or had pages
(clean or dirty) invalidated would have its mtime updated.  I tried it out
just to verify this, and the mtimes on /sbin/init, /lib/libc.so.*, etc on
my test machine were updated to the time I last rebooted the machine while
running a test kernel with that change.  so that particular change had
problems that were even worse than the issue it was attempting to fix.

I'll look more at the latest patch later today.

-Chuck