Re: [patch] voluntary-preempt-2.6.9-rc2-mm3-S5

[K.R. Foley]

Ingo Molnar wrote:
> * K.R. Foley <kr@xxxxxxxxxx> wrote:
>
>
> > The following, on top of Ingo's patch above, fixes the problem with
> > dropping new connections and doesn't have any adverse affects that
> > I've seen:
> >
> > --- linux-2.6.9-rc2-pre-mm3/net/ipv4/tcp_output.c.orig  2004-09-23 
> > 22:16:42.249435870 -0500
> > +++ linux-2.6.9-rc2-pre-mm3/net/ipv4/tcp_output.c       2004-09-23 
> > 22:12:03.911811945 -0500
> > @@ -699,11 +699,6 @@
> >
> >                        tcp_minshall_update(tp, mss_now, skb);
> >                        sent_pkts = 1;
> > -                       /*
> > -                        * Break out early - we'll continue later:
> > -                        */
> > -                       if (softirq_need_resched())
> > -                               break;
>
>
> hm, ok, i'll revert this in my tree. I suspect we'll see some latencies
> resurfacing under high network load again, but correctness goes first
> obviously. If then we'll have to find some other method to break that
> critical path.
>
> 	Ingo
Ingo,

Maybe this wasn't the right way to fix the problem? I just looked at the 
S4 patch and it had the same change in it, but did not exhibit the same 
problem. Not knowing exactly what I was looking for, I just started 
looking for obvious changes that might affect dropping tcp connections 
and this one seemed reasonable. I made the change and the problem went 
away. Maybe this needs looking at a little closer.

kr
-
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/