2003-06-16 18:53:53

by Robert Grzelak

[permalink] [raw]
Subject: 2.4.21 oops second time

Welcome!
My colege Andrzej Sosnowski in post "2.4.21 oops"
has write about error in kernel 2.4.21 in time of using this script
"#!/bin/sh
for IP in `/usr/bin/seq 3 500`; do
ip addr add 3ffe:80ee:c1d::$IP/48 dev eth0
ip addr add 3ffe:80ee:c1d::a:$IP/48 dev eth0
done"

I've made tests on clear kernel 2.4.21 in distributions:
debian 3.0r1 woody & Red Hat Linux release 8.0 (Psyche)
error kernel BUG sched.c 564! - is showing while making this script.

At the latest version of kernel 2.4.20 this error does'nt exists.

I have been tested this kernel (2.4.21) also with patch
grsecurity-1.9.10-2.4.21.patch.

It makes no difference - error shows again.

With best regards!
Robert Grzelak



2003-06-17 09:07:31

by Andre Tomt

[permalink] [raw]
Subject: Re: 2.4.21 oops second time

(Added CC to netdev)

On man, 2003-06-16 at 21:07, Robert Grzelak wrote:
> Welcome!
> My colege Andrzej Sosnowski in post "2.4.21 oops"
> has write about error in kernel 2.4.21 in time of using this script
> "#!/bin/sh
> for IP in `/usr/bin/seq 3 500`; do
> ip addr add 3ffe:80ee:c1d::$IP/48 dev eth0
> ip addr add 3ffe:80ee:c1d::a:$IP/48 dev eth0
> done"

I adapted this script a bit, and got it reproduced. It _seems_
(unverified) to not trigger as fast over ssh, as on local console, for
some reason.

You can wonder about the legitimacy of adding almost 1000 addresses to
one interface, however, it should not crash, rather return something
more useful when it happens and the kernel can not handle it.

Captured the OOPS over serial console, and decoded it, attached.

--
Andr? Tomt
[email protected]


Attachments:
oops.decode (4.17 kB)