[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Kernel 2.6 on a PegasOS?



On Sat, Jul 01, 2006 at 06:06:58PM +0200, Turbo Fredriksson wrote:
> [I just haven't had the time or will to work on this since I was reasonably
>  happy with the current/working 2.4.27-powerpc kernel. Now I've 're-opened'
>  this, and are trying to resolve my 2.6 booting problems.]
> 
> 
> >>>>> "Sven" == Sven Luther <sven.luther@wanadoo.fr> writes:
> 
>     Sven> On Tue, Sep 27, 2005 at 09:56:12PM +0200, Turbo Fredriksson
>     Sven> wrote:
>     >> > This is not the case, what kind of pegasos board do you have
>     >> exactly, and may > you perchance have faulty memory ? When did
>     >> you buy the board, and is it a > pegasos 2, or one of those
>     >> pre-april boards floating around.
>     >> 
>     >> No idea. It's a (long) loaner. Is there any commands that can
>     >> tell you this? Neither /proc/pci nor dmesg tell me anything
>     >> that I deem interesting.
> 
>     Sven> the content of /proc/cpuinfo would be a first help, if it is
>     Sven> pegasos 1
> 
> The /proc/cpuinfo tell me:
> 
> ----- s n i p -----
> processor       : 0
> cpu             : 7457, altivec supported
> clock           : 999MHz
> revision        : 1.1 (pvr 8002 0101)
> bogomips        : 665.19
> machine         : CHRP Pegasos2
> ----- s n i p -----
> 
>     Sven> then you would need to open it if it is
>     Sven> pre-april, or april, don't know how to distinguish april1
>     Sven> from april2 though, maybe the april1 don't have mac address
>     Sven> in the VIA chip.
> 
> The marking on the MB tell me that it's at revision 283 (if that tell
> you anything).
> 
> The OBP (OF?) tell me it's an PegasOS 2, v1.2 (20040810).
> 
>     >> >> seconds, and my own, well can't remember exactly, but they
>     >> >> hang after starting to boot..
>     >> >
>     >> > you need to create a zImage, not use the plain vmlinux.
>     >> 
>     >> That much I actually figured out as well :). Didn't help...
> 
>     Sven> Please get the exact error message, or maybe a serial
>     Sven> console will tell you more ?
> 
> I couldn't get a serial console to work (I think the serial port is
> fried).
> 
> But I've happened to stumble across a reference to the 'load-base'
> today (which renewed my interest in getting this resolved :) environment
> variable and have tried a number of number, to no avail.
> 
> 
> The kernel is the default Debian GNU/Linux (sarge) 'kernel-image-2.6.8-3-powerpc'
> (version 2.6.8-12sarge2), and I've tried the following values for
> my load-base (neither managed to boot/load the kernel):

Ok, pre-2.6.16, the boot-wrapper was not compiled with -fPIC, and thus after
being loaded to load-base, the kernel was elff-relocated to his link address,
which used to be either 4MB or 8MB., which is why setting load-base did not
work.

Doing the same with a 2.6.1[67] arch=powerpc kernel (like the dbeian 2.6.16 or
2.6.17 kernels) and setting load-bvase beyond 24MB (ideally 32MB) should work.
PRoblem is if you use a huge ramdisk which is loaded down from 64MB though, so
setting load-base to 64MB should be fine.

>        0x20000000
>        0x06000000
>        0x04000000
>        0x02000000
>        0x01000000
>        0x00800000
>        0x00600000
>        0x00400000       Original value, only loads+boots 2.4 kernel
>        0x00200000
>        0x00100000
> 
> In almost all instances, the load just freezes (or in some cases
> finish loading, but kernel refuses to boot).
> 
> But the differences in which these values behave, leads me to believe
> that I'm on the right track... Some values work better than others,
> but NEITHER loads AND boots the kernel...
> 
> I have 512Mb memory in the machine, which I believe to be ok (have
> tried other memory).

Indeed, but the default debian kernel should have worked just fine, but forget
about the sarge kernels, and give 2.6.16 a try.

Friendly,

Sven Luther



Reply to: