Re: xen dom0 debugging

[Horms <horms@verge.net.au>]

Robin van Leeuwen <rvl@rldsoftware.nl> wrote:
> a question, maybe not very xen specific.
> I want to boot my newly created (fresh compiled vmlinuz)
> dom0 but it gives an error during boot. One that would
> show up in dmesg. But it hangs the booting process
> so i have to boot with another vmlinuz.
> 
> When i boot with another vmlinuz however my dmesg
> with the error is overwritten with the one that boots
> correctly. How can i save the one with the error?
> 
> The error occures before i can execute a custom command.

Generally speaking, kernel logs can go to three places.

1. The console, which is what you are looking at. 

2. A ring-buffer in the kernel which is what you see
   if you type dmesg (on a running system). 
   Being a memory buffer, it is not persistant across reboots.
   And being a ring-buffer, it only stores the most recent log
   messages, though usually enough to be useful.

3. To disk, usually to /var/log/kern.log or something like that. 
   Obviously this won't happen before the disks are available, which is
   almost certainly the case here.

So, 2) and 3) are pretty much useless to you, which leaves you
with 1). You could just transcribe the messages by hand. Or if
you have another machine available, you could try setting
xen to boot using a serial console and capture the output
on the other end.

-- 
Horms