From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Wed 03 Dec 2003 - 08:17:18 GMT
On Wed, Dec 03, 2003 at 01:42:14PM +1100, Charles Dale wrote:
> > okay, here you go .. this is a fast hack, there are
> > some things which need a closer look, and the scheduler
> > stuff was removed, but hey, it compiles ...
> > http://vserver.13thfloor.at/Experimental/split-2.4.21-4.0.1.EL-vs1.1.6/
> > you have to check each // commented part in the patches
> > I had to remove some stuff, as I don't know this kernel
> > yet ...
> Herbert you're a legend! Thanks so much.
> The patched kernel compiled fine. I couldn't really find any commented parts
> to check though. Well there were two lines that one of the patches commented
> out, but that didn't seem like much.
well, as I said, there where some things I had to remove ;)
but I guess we will fix them step by step ...
> Actually the kernel works alright with the patches, only problem I've found
> is a kernel panic when running groupadd. Not sure which bits of the panic
> are relevant - it said "EIP in kill_proc_info" which seems to be the most
just let it panic, capture the entire oops trace (for example
via serial console, or by running bochs as emulator ;) and
feed it through ksymoops, which returns a human readable
version ... ( ksymoops -v vmlinuz -m System.map -K )
best to recompile the kernel first, after changing the
KERNEL_CFLAGS = to KERNEL_CFLAGS = -g, this way you can
later use addr2line to convert the stack addresses to a file
name and line number ...
> likely culprit as there is a 20 line block of code added to kill_proc_info
> by one of the patches. I didn't have much luck working out exactly what's
> wrong with it though.
> If you have any time I'd be mighty happy if you can have a bit more of a
> look, otherwise I'll keep plodding away in my kernel newbie manner and keep
> you updated =).
we'll have to add the scheduler stuff, and fix the broken
uptime patches and the 'missing' network connect() code ...
> Vserver mailing list
Vserver mailing list