[netdev added to cc list: it looks like a module refcount bug with ipx]
David Liontooth wrote:
The natsemi oops is triggered in 2.6.0-test9 too.
kernel BUG at include/linux/module.h:296
That's BUG_ON(module_refcount(module) == 0) in __module_get. I doubt
that the natsemi driver has anything to do with the bug, it looks like a
bug in the ipx core.
Everything freezes.
Am I the only one to get this?
I don't know when it started, but 2.5.69 has no problems.
My guess: 2.5.69 has no bug check. It will oops with the right timing of
rmmod and a packet arrival.
Cheers,
David
----- Original Message -----
From: "David Liontooth" <liontooth@xxxxxxxx>
Date: Mon, 20 Oct 2003 21:05:13 -0500
To: linux-kernel@xxxxxxxxxxxxxxx
Subject: Re: [2.6.0-test-7] natsemi oops
Correction: I get the oops also when natsemi is compiled as a module.
It is triggered not when the module is loaded, but when it is used
the first time. Oops (some fragments below) followed by a total freeze;
nothing gets logged.
Is this a known problem?
Is there a workaround?
Cheers,
David
----- Original Message -----
From: David Liontooth
Date: Mon, 20 Oct 2003 05:24:52 -0500
To: linux-kernel@xxxxxxxxxxxxxxx
Subject: [2.6.0-test-7] natsemi oops
The 2.6.0-test-7 boots fine and works great -- until I plug
in the ethernet cable. Within a second I get an oops and
everything freezes. Booting with "acpi=off" makes no difference.
If I boot with the ethernet cable plugged in, I get to the
login prompt, and it oopses within a second. If I time it right,
I can log into the machine remotely for one second before it
oopses (so the natsemi driver is working). Very reproducible!
/proc/kmsg is empty.
If I compile natsemi as a module, I don't get the oops.
However, now the driver is not working -- I can't ping out.
Everything works fine in 2.5.69, which I've been running
since early July.
Here's some of the oops, taken by hand:
Process swapper (pid: 0, threadinfo=c042a000 task c03a47a0)
Stack
Call trace:
ipxitf_auto_create
ipx_rcv
netif_receive_skb
process_backlog
net_rx_action
do_softirq
do_IRQ
_stext
common_interrupt
acpi_processor_idle
cpu_idle
start_kernel
unknown_bootoption
Kernel panic: Fatal exception in interrupt
In interrupt handler -- not syncing
Configuration, lspci, and dmesg attached.
Cheers,
David
<< config-2.6.0-test7-3 >>
<< dmesg-2.6.0-test7-7 >>
<< lspci-2.6.0-test7 >>
--
__________________________________________________________
Sign-up for your own personalized E-mail at Mail.com
http://www.mail.com/?sr=signup
CareerBuilder.com has over 400,000 jobs. Be smarter about your job search
http://corp.mail.com/careers
--
Manfred
|