On Thu, 17 Jun 2010, Peter Zijlstra wrote:

quoted text
> On Wed, 2010-06-16 at 17:29 -0400, Steven Rostedt wrote:
> > On Wed, 2010-06-16 at 23:16 +0200, Peter Zijlstra wrote:
> > 
> > > Also, google doesn't seem to index msg-ids, so I've no idea what you're
> > > referring to.
> > 
> > But marc.info does:
> > 
> > http://marc.info/?i=4BCEAD7B0200005A0006513E@soto.provo.novell.com
> 
> Right, so that was before Yong Zhang's patch, if it still happens we
> need to again look at what is causing this. Again, blindly increasing
> the limit is not a good option.

Well, as I said, I'm testing with Yong Zhang's patch, and it seems to be 
doing the trick, so I am actually not pushing for my patch right now.
 
But please stop characterizing this as "blindly increasing the limit", 
because that is not at all what I or others do. We have a debug build with 
tons of options turned on in which case we increased it to the minimum 
that worked for us, and we have a tracing build in which case we left it 
at the default. Also as I pointed out, in Sven's case it sounds like they may 
have had a build where they even wanted to decrease it.

Your objection in the past was that it was another tunable that nobody understands, and I 
have more sympathy for that argument. My counterargument is that if we're 
all putting a version of this patch in our private builds, then it's a tad 
counterproductive. Let's leave things the way they are for now, unless 
this becomes a problem again.

Thanks.

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