Mailing List Archive

David Cummings wrote:
> So after much googling and some work, the O2k is running busybox.
> Hooray! I've gone through the install guide up to the point of kernel
> compilation. What's the deal with that? I'm going to list my
> assumptions and/or questions and I'd really appreciate any help you
> all can provide.
>
> Assumptions:
> 1 arcload will load a 64-bit kernel on ip27

I think the latest version does, but the latest version may not be in
portage yet.

> 2 arcboot requires a 64-bit kernel on ip27

s/requires a 64-bit kernel/doesn't work/

> 3 gcc-mips64 is required to compile a 64-bit kernel

for your purposes, yes

> 4 -mips64 -mabi=32 -mtune=r10000 should all work and be correct for make.conf

Those CFLAGS are wrong. -mips64 is equivalent to -march=mips64, which
is actually a processor type. Therefore, you will be breaking things
badly with this. AReally, you should probably be using something more
like "-O2 -mips4 -mtune=r10k -mabi=32", or perhaps "-O2 -march=r10k
-mabi=32"

-Steve
--
gentoo-mips@gentoo.org mailing list

sorry, CFLAGS were wrong, should have been
4 -O2 -mips4 -mabi=32 -mtune=r10000
sorry for trying to recall from memory something I should've just looked at!

and i forgot that arcboot was also an older package that was
replaced(?) by arcload, so that should've been, arcload requires a
64-bit kernel?
since, after 0.41 arcload supports ip27...

Also, what kernel version are people using?

On 6/28/05, Stephen P. Becker <geoman@gentoo.org> wrote:
> David Cummings wrote:
> > So after much googling and some work, the O2k is running busybox.
> > Hooray! I've gone through the install guide up to the point of kernel
> > compilation. What's the deal with that? I'm going to list my
> > assumptions and/or questions and I'd really appreciate any help you
> > all can provide.
> >
> > Assumptions:
> > 1 arcload will load a 64-bit kernel on ip27
>
> I think the latest version does, but the latest version may not be in
> portage yet.
>
> > 2 arcboot requires a 64-bit kernel on ip27
>
> s/requires a 64-bit kernel/doesn't work/
>
> > 3 gcc-mips64 is required to compile a 64-bit kernel
>
> for your purposes, yes
>
> > 4 -mips64 -mabi=32 -mtune=r10000 should all work and be correct for make.conf
>
> Those CFLAGS are wrong. -mips64 is equivalent to -march=mips64, which
> is actually a processor type. Therefore, you will be breaking things
> badly with this. AReally, you should probably be using something more
> like "-O2 -mips4 -mtune=r10k -mabi=32", or perhaps "-O2 -march=r10k
> -mabi=32"
>
> -Steve
> --
> gentoo-mips@gentoo.org mailing list
>
>


--
The way that can be named is not the Way.

--
gentoo-mips@gentoo.org mailing list

David Cummings wrote:
> Also, what kernel version are people using?

There really isn't a truely working kernel kernel for ip27 at the moment. The
latest kernel from linux-mips.org is probably your best bet, except that there
isn't currently a bug that surfaced which causes the network not to work. Kumba
has a work around patch for it, but I don't believe it's in mips-sources yet.
There are also issues with multiple nodes, 2.6.12 will not boot on my two O200's
if they are craylinked together, however it works fine with just one node. We
are not sure how this affects O2000's since even in one physical node block,
each module is technically a seperate node. If you're going to build your own
kernel just do a `make ip27_defconfig`, then run make menuconfig to tweak to taste.

Also if you end up using a prebuilt 2.6.6 kernel and are running ~mips make sure
to add >sys-kernel/mips-headers-2.5 to /etc/portage/package.mask.
mips-headers-2.6.11 does not play well at all with older kernels, I found this
out the hard way.

Hardave

hmm, alright. I'll play with it.
Am I missing a patch if i keep getting an error

arch/mips/sgi-ip27/ip27-klnuma.c: In function `copy_kernel':
arch/mips/sgi-ip27/ip27-klnuma.c:90: warning: implicit declaration of
function `PLAT_NODE_DATA'
arch/mips/sgi-ip27/ip27-klnuma.c:90: error: invalid type argument of `->'
make[1]: *** [arch/mips/sgi-ip27/ip27-klnuma.o] Error 1

??? This is after I changed include/asm/sn/sn_private.h to not define
functions because they were barfing. Lines 15/16 if anyone wants to
know what i'm talking about.
thanks -dave
On 6/28/05, Hardave Riar <hardave@gentoo.org> wrote:
> David Cummings wrote:
> > Also, what kernel version are people using?
>
> There really isn't a truely working kernel kernel for ip27 at the moment. The
> latest kernel from linux-mips.org is probably your best bet, except that there
> isn't currently a bug that surfaced which causes the network not to work. Kumba
> has a work around patch for it, but I don't believe it's in mips-sources yet.
> There are also issues with multiple nodes, 2.6.12 will not boot on my two O200's
> if they are craylinked together, however it works fine with just one node. We
> are not sure how this affects O2000's since even in one physical node block,
> each module is technically a seperate node. If you're going to build your own
> kernel just do a `make ip27_defconfig`, then run make menuconfig to tweak to taste.
>
> Also if you end up using a prebuilt 2.6.6 kernel and are running ~mips make sure
> to add >sys-kernel/mips-headers-2.5 to /etc/portage/package.mask.
> mips-headers-2.6.11 does not play well at all with older kernels, I found this
> out the hard way.
>
> Hardave
>
>
>


--
The way that can be named is not the Way.

--
gentoo-mips@gentoo.org mailing list

David Cummings wrote:
> hmm, alright. I'll play with it.
> Am I missing a patch if i keep getting an error
>
> arch/mips/sgi-ip27/ip27-klnuma.c: In function `copy_kernel':
> arch/mips/sgi-ip27/ip27-klnuma.c:90: warning: implicit declaration of
> function `PLAT_NODE_DATA'
> arch/mips/sgi-ip27/ip27-klnuma.c:90: error: invalid type argument of `->'
> make[1]: *** [arch/mips/sgi-ip27/ip27-klnuma.o] Error 1
>
> ??? This is after I changed include/asm/sn/sn_private.h to not define
> functions because they were barfing. Lines 15/16 if anyone wants to
> know what i'm talking about.
> thanks -dave

I have my old patches for the ioc3 onewire problem at
http://www.ldsys.net/~cgs/linux-mips/ as well as a working, older kernel
or two or three. I don't know. I haven't had any time to do anything
with the origin in almost a year. :(

The patches are against CVS, so you'll have to fuck with 'em. If you
know what you're doing, it really only shuts off interrupts while the
onewire read happens. Not bad, but not really good, either. :)

Regarding the NUMA thing... I don't think you'll need that. Is that in
defconfig now?
--
gentoo-mips@gentoo.org mailing list

Christopher G. Stach II wrote:
>
> I have my old patches for the ioc3 onewire problem at
> http://www.ldsys.net/~cgs/linux-mips/ as well as a working, older kernel
> or two or three. I don't know. I haven't had any time to do anything
> with the origin in almost a year. :(
>
> The patches are against CVS, so you'll have to fuck with 'em. If you
> know what you're doing, it really only shuts off interrupts while the
> onewire read happens. Not bad, but not really good, either. :)
>
> Regarding the NUMA thing... I don't think you'll need that. Is that in
> defconfig now?

Skylark just mentioned about 15mins ago in #mipslinux that he's working on a
partial re-write of IOC3 based on the recent commits to mainline of IOC4
material for sn2/altix, so it's possible (based on what he is re-writing) that a
number of IOC3 problems may be solved with this. I'm not sure when he'll finish
it, could be next week, so when it's finished, I'll have to put it through some
rounds to see if it fixes anything.


--Kumba

--
Gentoo/MIPS Team Lead
Gentoo Foundation Board of Trustees

"Such is oft the course of deeds that move the wheels of the world: small hands
do them because they must, while the eyes of the great are elsewhere." --Elrond
--
gentoo-mips@gentoo.org mailing list