On Mon, Jul 24, 2006 at 11:27:37AM +1000, Nathan Scott wrote:
> On Mon, Jul 24, 2006 at 10:01:48AM +1000, Nathan Scott wrote:
> > On Sun, Jul 23, 2006 at 08:06:50PM +0100, Christoph Hellwig wrote:
> > > Shouldn't we make sure we clear all flags when reusing a log buffer?
> > > Relying on clearing individual flags seems rather fragile to me.
> >
> > *nod* - good idea. I'll rework xlog_sync, and resend later.
>
> After looking more, I'm less convinced. There's some flags we wont
> want to touch - the "internal" flags like PAGE_CACHE, etc (that one
> is obviously not relevent here, but still, at some point a flag may
> be introduced that we accidentally break by clearing all flags).
>
> There is a ZEROFLAGS macro, I've added ORDERED to that and used it
> instead. I also fixed the double barrier issue for the split log
> write case - here's an updated patch...
The flag clearing changes look good. But why is it okay to skip the
ordered flag on the first block? We want to make sure all previous I/O
is finished before even doing the first log block write, don't we?
|