Re: ppc64 pseries 2.6.11-2 .deb kernel available
On Thu, Jun 02, 2005 at 11:25:02PM +0100, Paul Brossier wrote:
> On Sat, May 28, 2005 at 04:15:39PM +0200, Sven Luther wrote:
> > Hello all,
> >
> > Thanks to the great job of Jeff Bailey and Matthias Klose, who provided a set
> > of gcc-3.4/glibc-2.3.5 based biarch toolchain, i have managed to build a first
> > ppc64 kernel package, which is available at :
> >
> > http://people.debian.org/~luther/ppc64
> >
> > Altough obviously untested, as i don't have a ppc64 box, it should install
> > over a normal sarge system, and run on all ppc64 boxes, from power3 RS6K ones
> > to newer iseries, passing by all apple G5 macs, except for the older
> > pre-power5 iseries. It is built without power4 optimization and in SMP mode
> > only for now, since i want to get this flavour tested and investigated before
> > multiplicating the different kernel images.
> >
> > This kernel is *not* yet for production, since at least it will want a 64bit
> > compiled procutils, to take care of 64bit process ids in /proc.
> >
> > Future plans include integrating these kernels in etch, once the biarch
> > toolchain makes it to it, and working on a ppc64 d-i, but first please test
> > this kernel on as much as possible 64bit powerpc machines, and provide
> > feedback to me.
>
> i have tried this kernel on a powermac g5. it works fine, except the fan
> goes to full speed. i didn't stay a long time on it as it is really
> loud, but boot messages were looking good.
Ok, thanks, good to know. I don't know about fan control, BenH, can you say
something about this ? I will have a look at the code also this WE.
> there are quite a few differences comparing to the -power4-smp config.
> is it normal that CONFIG_ADB, CONFIG_DMASOUND_PMAC, CONFIG_SND_POWERMAC
These refused to compile, so i disabled it, needs fixing, now that i have a
success report about this.
> and CONFIG_SND_BIT32_EMUL are not set?
Same here, i think. Mmm, will need to investigate this, maybe i just missed
it, since i took the power4 config as basis, and this one obviously didn't
need this.
> the early_printk option seems to work, but doesn't look very nice, as
> the black lines are drawn on top of the OF gray screen. it gets better
> when the logos appear.
Ok,
Thanks for testing.
Friendly,
Sven Luther
Reply to: