2003-06-16 21:34:02

by Dave Bentham

[permalink] [raw]
Subject: kernel 2.4.21 crash

Hello

I upgraded my kernel on a Mandrake 9.0 base from 2.4.20 to the new 2.4.21 tonight - built from source patches as I always do; followed by reinstalling the NVidia drivers and ALSA.

But there seems to be a major failure when the computer just stops with no warning. Two scenarios that seem to repeat it include starting Loki's Heretic2 off, and mounting the CDRW drive via WindowMaker dock app. I cannot do anything when this happens; can't hotkey out of X, can't telnet to it from my other networked PC. I have to power down and back up.

It seems to be a few seconds after the trigger that the lock up occurs, and also it starts flashing the keyboard Caps Lock and Scroll Lock LEDs in step at about 1 Hz. I'm sure its trying to tell me something...

Thanks in advance

Dave


2003-06-18 22:25:41

by Dave Bentham

[permalink] [raw]
Subject: Re: kernel 2.4.21 crash

Hi,

Regarding this issue I have been advised to setup a serial console to
capture the 'kernel panic' debug output. For the record, I can get the
panic from the BASH prompt _before_ starting X/WM just by mounting my
CDRW drive.

However, I think I need some advice on setting up the serial console. I
have attached another PC (WinXP with HyperTerminal) serially to my panic
Linux PC. Following the Remote Terminal HOWTO I have achieved some
success... but all I see on HyperTerminal is:

LILO 22.3.2 boot:
Loading Linux_2.4.21................
BIOS data check successful

Mandrake Linux release 9.0 (dolphin) for i586
Kernel 2.4.21 on an i686 / ttyS0
telekon.davesnet login:

It seems to start piping the stuff to ttyS0, but then gives up after
a few progress-dots: I'm missing the main kernel blurb (its all on the
attached monitor), and also the panic stuff appears only on the attached
monitor.

Excerpt of butchered lilo.conf

boot=/dev/hda
map=/boot/map
#vga=normal
default=Linux_2.4.21
keytable=/boot/uk.klt
prompt
nowarn
timeout=100
#message=/boot/message
#menu-scheme=wb:bw:wb:bw
ignore-table
# serial term bits
serial = 0,9600n8
image = /boot/vmlinuz-2.4.21
root = /dev/hda3
label = Linux_2.4.21
read-only
# vga=788
append = "devfs=mount hdd=ide-scsi console=tty0 console=ttyS0,9600n8"


Any help gratefully recieved!

Thanks

Dave

On Mon, 16 Jun 2003 22:48:33 +0100
[email protected] wrote:

> Hello
>
> I upgraded my kernel on a Mandrake 9.0 base from 2.4.20 to the new
> 2.4.21 tonight - built from source patches as I always do; followed by
> reinstalling the NVidia drivers and ALSA.
>
> But there seems to be a major failure when the computer just stops
> with no warning. Two scenarios that seem to repeat it include starting
> Loki's Heretic2 off, and mounting the CDRW drive via WindowMaker dock
> app. I cannot do anything when this happens; can't hotkey out of X,
> can't telnet to it from my other networked PC. I have to power down
> and back up.
>
> It seems to be a few seconds after the trigger that the lock up
> occurs, and also it starts flashing the keyboard Caps Lock and Scroll
> Lock LEDs in step at about 1 Hz. I'm sure its trying to tell me
> something...
>
> Thanks in advance
>
> Dave
>


--
A computer without Microsoft is like chocolate cake without mustard.

2003-06-19 00:43:33

by Adam Kropelin

[permalink] [raw]
Subject: Re: kernel 2.4.21 crash

On Wed, Jun 18, 2003 at 11:40:20PM +0100, Dave Bentham wrote:
> However, I think I need some advice on setting up the serial console. I
> have attached another PC (WinXP with HyperTerminal) serially to my panic
> Linux PC. Following the Remote Terminal HOWTO I have achieved some
> success... but all I see on HyperTerminal is:
>
> LILO 22.3.2 boot:
> Loading Linux_2.4.21................
> BIOS data check successful

That output is from LILO...

> Mandrake Linux release 9.0 (dolphin) for i586
> Kernel 2.4.21 on an i686 / ttyS0
> telekon.davesnet login:

...and that bit is from the getty launched in /etc/inittab.

> a few progress-dots: I'm missing the main kernel blurb (its all on the
> attached monitor), and also the panic stuff appears only on the attached
> monitor.

Yeah, the kernel's serial console isn't kicking in.

> Excerpt of butchered lilo.conf

<snip>

> image = /boot/vmlinuz-2.4.21
> root = /dev/hda3
> label = Linux_2.4.21
> read-only
> # vga=788
> append = "devfs=mount hdd=ide-scsi console=tty0 console=ttyS0,9600n8"

That looks basically right. The kernel you're running may not be
compiled with serial console support. Also, you might want to swap the
order of the two console= options so that /dev/console refers to your
real virtual console, but that shouldn't keep the serial console from
working. You might try dropping the 'n8'...I don't use it here although
it is documented. I vaguely recall having problems that went away when
I stopped (redundantly) specifying the character format --but that could
be hogwash. My bet is the kernel not being built with serial console
support.

--Adam

2003-06-22 03:36:02

by Adam Majer

[permalink] [raw]
Subject: Re: kernel 2.4.21 crash

On Mon, Jun 16, 2003 at 10:48:33PM +0100, [email protected] wrote:
> Hello
>
> But there seems to be a major failure when the computer just stops with no warning. Two scenarios that seem to repeat it include starting Loki's Heretic2 off, and mounting the CDRW drive via WindowMaker dock app. I cannot do anything when this happens; can't hotkey out of X, can't telnet to it from my other networked PC. I have to power down and back up.

There was something like this posted on the list a few days ago. Someone
said that it has to do with IDE-SCSI timing or what not. That is, try if you can
reproduce it without the ide-scsi driver in the kernel..

> It seems to be a few seconds after the trigger that the lock up occurs, and also it starts flashing the keyboard Caps Lock and Scroll Lock LEDs in step at about 1 Hz. I'm sure its trying to tell me something...

That means the kernel detected something evil (oops caused by null pointer access,
etc...). Sicne the leds are still flashing, at least the kernel is not totally dead. :)

hope this helps a bit,
- Adam

2003-06-22 08:47:29

by Dave Bentham

[permalink] [raw]
Subject: Re: kernel 2.4.21 crash

On Sat, 21 Jun 2003 22:41:32 -0500
Adam Majer <[email protected]> wrote:

> On Mon, Jun 16, 2003 at 10:48:33PM +0100, [email protected]
> wrote:
> > Hello
> >
> > But there seems to be a major failure when the computer just stops
> > with no warning. Two scenarios that seem to repeat it include
> > starting Loki's Heretic2 off, and mounting the CDRW drive via
> > WindowMaker dock app. I cannot do anything when this happens; can't
> > hotkey out of X, can't telnet to it from my other networked PC. I
> > have to power down and back up.
>
> There was something like this posted on the list a few days ago.
> Someone said that it has to do with IDE-SCSI timing or what not. That
> is, try if you can reproduce it without the ide-scsi driver in the
> kernel..

You may be right - I turned off SCSI support in the kernel and removed
the'hdd=ide-scsi' boot appendage and I could mount the CDRW ok.

I'll try and find the history of this known bug.

>
> > It seems to be a few seconds after the trigger that the lock up
> > occurs, and also it starts flashing the keyboard Caps Lock and
> > Scroll Lock LEDs in step at about 1 Hz. I'm sure its trying to tell
> > me something...
>
> That means the kernel detected something evil (oops caused by null
> pointer access, etc...). Sicne the leds are still flashing, at least
> the kernel is not totally dead. :)
>

It may as well be totally dead!!!

Thanks
Dave

2003-08-27 22:18:24

by Dave Bentham

[permalink] [raw]
Subject: kernel 2.4.22 IDE/SCSI CDRW crashing still

I am still getting the kernel panic when mounting my CD-RW
drive (ide-scsi) with 2.4.22. This is the one that appeared with 2.4.21.
I'm not an expert on these things but I got the impression this was
being addressed but I have been off the list since reporting it with
2.4.21. Am I mistaken? Or has my 2.4.22 build perhaps gone awry?

Dave


On Mon, 16 Jun 2003 22:48:33 +0100
[email protected] wrote:

> Hello
>
> I upgraded my kernel on a Mandrake 9.0 base from 2.4.20 to the new
> 2.4.21 tonight - built from source patches as I always do; followed by
> reinstalling the NVidia drivers and ALSA.
>
> But there seems to be a major failure when the computer just stops
> with no warning. Two scenarios that seem to repeat it include starting
> Loki's Heretic2 off, and mounting the CDRW drive via WindowMaker dock
> app. I cannot do anything when this happens; can't hotkey out of X,
> can't telnet to it from my other networked PC. I have to power down
> and back up.
>
> It seems to be a few seconds after the trigger that the lock up
> occurs, and also it starts flashing the keyboard Caps Lock and Scroll
> Lock LEDs in step at about 1 Hz. I'm sure its trying to tell me
> something...
>
> Thanks in advance
>
> Dave
>


--
A computer without Microsoft is like chocolate cake without mustard.