Re: isdn performance problem in 2.1[19]-[22]

David Woodhouse (David.Woodhouse@mvhi.com)
Mon, 05 Oct 1998 11:27:48 +0100


(Aha - a post made before the Reply-To brain damage. I don't have to muck
about with the addresses)

This patch, unfortunately, doesn't help. Our proxy machine, running 2.1.123,
started showing precisely the same behaviour that I reported some time ago -
( http://www.linuxhq.com/lnxlists/linux-kernel/lk_9809_02/msg00075.html ) -
packets are being held on the queue and not being actually put on the wire
until something else is added to the end of the outgoing queue.

We also have a lot of TCP checksum failures, but that's a minor problem
compared with this one.

ISDN in 2.1.1xx is virtually useless to us when it does this.

eis@baty.hanse.de said:
> Hi,
>
> Richard Schiffelers <ricrjhl@xs4all.nl> writes:
>
> >
> > At 11:58 24-9-98 +0200 Markus Brischke wrote:
> > >I noticed that the throughput of my isdn is really bad with 2.1
> > >beginning with 1 kb/s and 5 sec later 400 b/s. I am using a teles hisax 16.3
> > >Card. First i thought that it could be a problem of my provider
> > >, so i testet the same things with my stable 2.0.33 kernel and got
> > >ca. 8kb/s. I am using predictor compression. My pppd 2.3pl3.
> > >
> > >I tested this on different days at different times.
> > >
> > >Does anybody notice same problems?
> >
> > Yep, same problems. Also using Teles card. I noticed the slowdown in speed
> > when i started to run 2.1.122 (didn't run any earlyer 2.1 version together
> > with ISDN). I decided to check out what was going on with IPTraf. It showed
> > that on de ippp0 interface around 50% of all packets had a wrong checksum
> > (and wich are therefore resend). I then switched back to 2.0.35 and again
> > checked with IPTraf. No problem and no bad packets. So it seems 2.1 is not
> > 100%. Ok, that probably why is it experimental *8^>
> >
> > >Does anybody have a hint how to fix it.
>
> Maybe this could also indicate a low level driver problem. Maybe upgrading
> to a recent snapshot from the i4l CVS tree might help against driver problem.
>
> However, I've also sometimes observed such problems (not reproducible,
> but with current i4l cvs versions). Could those of you observing
> performence problems with isdn on 2.1.x please try the following
> patch? It's against the current cvs version, but should also apply
> against stock 2.1.122 kernel.
>
> Henner
>
> --- 2.1.122-i4ldev/drivers/isdn/isdn_net.c Mon Sep 7 23:23:03 1998
> +++ 2.1.122-ix25/drivers/isdn/isdn_net.c Fri Sep 25 21:47:41 1998
> @@ -1225,12 +1226,14 @@
> if (ret == len) {
> lp->transcount += len;
> clear_bit(0, (void *) &(ndev->tbusy));
> + mark_bh(NET_BH);
> return 0;
> }
> if (ret < 0) {
> dev_kfree_skb(skb);
> lp->stats.tx_errors++;
> clear_bit(0, (void *) &(ndev->tbusy));
> + mark_bh(NET_BH);
> return 0;
> }
> return 1;
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
>

---- ---- ----
David Woodhouse David.Woodhouse@mvhi.com Office: (+44) 1223 810302
Project Leader, Process Information Systems Mobile: (+44) 976 658355
Axiom (Cambridge) Ltd., Swaffham Bulbeck, Cambridge, CB5 0NA, UK.
finger dwmw2@ferret.lmh.ox.ac.uk for PGP key.

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