Skip to contentSkip to conversation

Alternatives to serial console for oops.

DISCUSSION James Courtier-Dutton 9 participants 9 messages Nov 27, 2005 – Nov 28, 2005

lore
Share & tools
lore
#0 lore raw

Hi,

I wish to view the oops that are normally output on the serial port of the PC. The problem I have, is that my PC does not have a serial port. Are there any alternatives for getting that vital oops from the kernel just as it crashes apart from the serial console. Could I get it to use some other interface? e.g. Network interface.
Parallel port is also not an option.

James

#1 lore raw

Hi,

Quoted 8 lines · James Courtier-Dutton

Hi,

I wish to view the oops that are normally output on the serial port of the PC. The problem I have, is that my PC does not have a serial port. Are there any alternatives for getting that vital oops from the kernel just as it crashes apart from the serial console.
Could I get it to use some other interface? e.g. Network interface. Parallel port is also not an option.

The netconsole patch should let you send it over a network interface, but I've not used it much so I won't be able to help. kmsgdump will let you write it on a floppy disk, but I bet that you don't have any if you don't have any serial nor parallel ports.

VMWare may help if you are mainly trying to debug a kernel and not using it for production.

Regards,
Willy

#2 lore raw
In reply to James Courtier-Dutton
Quoted 9 lines · James Courtier-Dutton

Hi,

I wish to view the oops that are normally output on the serial port of the PC. The problem I have, is that my PC does not have a serial port. Are there any alternatives for getting that vital oops from the kernel just as it crashes apart from the serial console.
Could I get it to use some other interface? e.g. Network interface. Parallel port is also not an option.

Netconsole (Documentation/networking/netconsole.txt) Digital camera to take a photo of the screen. Pen & paper to manually write down the Oops.

--
Jesper Juhl <[email protected]>
Don't top-post http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please http://www.expita.com/nomime.html

#3 lore raw
Quoted 1 line · Jesper Juhl

Pen & paper to manually write down the Oops.

you really need to be _desperate_ :-) I tried this literally _once_ .

#6 lore raw
In reply to Jesper Juhl
Quoted 4 lines · Jesper Juhl
Quoted 2 lines · James Courtier-Dutton

Hi,

Quoted 8 lines
Quoted 6 lines

I wish to view the oops that are normally output on the serial port of the PC. The problem I have, is that my PC does not have a serial port. Are there any alternatives for getting that vital oops from the kernel just as it crashes apart from the serial console.
Could I get it to use some other interface? e.g. Network interface. Parallel port is also not an option.

Netconsole (Documentation/networking/netconsole.txt)

I wasn't able to capture an OOPS using netconsole from my arm-based firewall box -- the output comes too fast and receiving end only captures some random parts of the OOPS. I doubt the receiving box is too slow (2.4GHz Pentium-IV, Intel GigE ethernet), so it must be the sending end which is trying to generate packets faster than it's own NIC can transmit them... or something else.

Even "normal" serial console does not work here for this device:
serial port on the box has no flow control, and at 9600 it still loses some parts of the OOPS, or especially a sysrq-t output (which is much larger than a typical OOPS).

Quoted 1 line

Digital camera to take a photo of the screen.

Not all systems has a screen (my router certainly does not). Also, even for "usual" PC, amount of information in an OOPS is larger than a screen, so most important part of the OOPS (several lines on the top) is lost.

Quoted 1 line

Pen & paper to manually write down the Oops.

The same problem as above with a camera.

This stuff worked perfectly in 2.4 kernel, where it was possible to use keyboard to scroll to the beginning of an OOPS - only half a screen and we're here (well, sometimes it is larger, but Shift-PgUp still worked). Now, when keyboard does not work anymore, it's not possible. As is not possible to reboot the system using "Three Finger Solution" (Ctrl-Alt-Del) -- some modern systems does not have "Reset" button, so in order to reboot after an OOPS, one have to power-cycle the box...

It'd be really nice to have keyboard back in "OOPS mode", as 2.4 had...

Oh well.

/mjt

#7 lore raw
In reply to James Courtier-Dutton
Quoted 11 lines · James Courtier-Dutton

Hi,

I wish to view the oops that are normally output on the serial port of the PC. The problem I have, is that my PC does not have a serial port. Are there any alternatives for getting that vital oops from the kernel just as it crashes apart from the serial console. Could I get it to use some other interface? e.g. Network interface. Parallel port is also not an option.

James

I use Netconsole for routine monitoring.

See Documentation/networking/netconsole.txt

Example for kernel command line [email protected]/eth0,[email protected]/00:00:0E:5F:15:64

You don't want to hassle with netcat.

Many sysloggers can log udp

I use syslog-ng. Excerpts from configuration:

source netlog { 
	udp(ip(0.0.0.0) port(4450)); 
};

destination netlog

{ 
	file("/var/log/netlog"); 
};

log

{ 
	source(netlog); 
	destination("netlog"); 
	flags(final); 
};
#8 lore raw
In reply to James Courtier-Dutton
Quoted 8 lines · On Sun, 27 Nov 2005 15:52:27 +0000 James Courtier-Dutton wrote:

Hi,

I wish to view the oops that are normally output on the serial port of the PC. The problem I have, is that my PC does not have a serial port. Are there any alternatives for getting that vital oops from the kernel just as it crashes apart from the serial console.
Could I get it to use some other interface? e.g. Network interface. Parallel port is also not an option.

If the oops occurs after booting (e.g., it's in a module that you can load after boot to cause the oops) and if you have USB ports and a usb-serial converter device, you should be able to use usb-console output to capture the oops.

---
~Randy

#4 lore raw
In reply to Mateusz Berezecki
Quoted 5 lines · Mateusz Berezecki
Quoted 1 line · Jesper Juhl

Pen & paper to manually write down the Oops.

you really need to be _desperate_ :-) I tried this literally _once_ .

Obviously you never flashed the screen using 8 colors (3 bits of information), to transfer a crash dump message...

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#5 lore raw
Quoted 9 lines · Geert Uytterhoeven
Quoted 5 lines · Mateusz Berezecki
Quoted 1 line · Jesper Juhl

Pen & paper to manually write down the Oops.

you really need to be _desperate_ :-) I tried this literally _once_ .

Obviously you never flashed the screen using 8 colors (3 bits of information), to transfer a crash dump message...

In 1998, we used a video camera to capture a long-scrolling Oops at the CLOWN cluster project in the University of Paderborn, Germany... So we could finally fix the problem and make this Cluster event a real success 8^P

MfG, JBG

Signature · 4 lines
Jan-Benedict Glaw       [email protected]    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 für einen Freien Staat voll Freier Bürger"  | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));