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

Re: [RFH] Mac-on-Linux for Debian lenny

Hi Rogério

On Mon, Jun 09, 2008 at 07:22:55PM -0300, Rogério Brito wrote:
> > Debian. 
> > 
> > -> Unusable kernel modules: This is caused by the removal of
> >    handle_mm_fault in recent kernels. To fix this we have to talk to the
> >    kernel team to see if they accept the attached patch which reenables
> >    the export.
> You forgot to attach the patch.

Ok, it's attached now. Sorry.

> >    Additionaly the modules as shipped in mol-source currently don't
> >    compile with recent kernels. This can be fixed by updating mol-source
> >    to the most recent version in the upstream SVN.
> >    This issue is Debian bug #445359.
> >    José: What's the status of the upstream integration of the MOL kernel
> >    modules. Is this expected to happen anytime soon or is the work on it
> >    abandoned?
> It would be quite a good thing to have the mol modules integrated on the
> mainline kernel from kernel.org, as we already have many emulation
> things there already (uml, kvm, lguest, xen etc).
> On the other hand, mol may be rejected for exactly this reason, if we
> don't argue well enough about what would make mol have a differential.
> Perhaps preparing a text and having it reviewed by enough people would
> I don't know very much about kernel hacking, but I have some books about
> the Linux kernel development. It may, perhaps, be the right time to get
> the dust off of them.

If you are interested in this, great. I think this would be a long term
project and certainly not ready for Lenny. Joseph Jezak (the current mol
upstream) did some work on this already. See
mac-on-linux.sourceforge.net. There is a directory mol-kernel-merge in
the upstream SVN. Joe: Probably you can better comment on this.

> > -> MOL SVN segfaults or hangs: The current version of MOL in unstable
> >    just hangs after startup if used with the latest kernel modules
> >    compiled from SVN. The latest SVN version segfaults at the same point
> >    where the other version hangs.
> Which other version hangs? Which one doesn't? That is, at which point
> things have stopped working?

OK, I have now a bit more information about this: 

mol and mol-modules from stable -> OK
mol from stable and latest upstream modules -> hang
latest upstream mol and modules -> mol (userspace) segfaults
mol from stable and modules from SuSE patches -> works
latest upstream mol and modules from SuSE patches -> mol (userspace) segfaults

The SuSE patches can be found here: 
I used all pachtes except suse-ppc32-mol-handle-mm-fault plus the
attached mol-mm.patch.

The SuSE patches can not be directly used in Debian. I think the kernel
team would reject them. But we could build an updated mol-source package
from them. 

> And one thing: why are the mol drivers for MacOS X in non-free while the
> ones for MacOS are in main?
> I think that I seem to recall that Benjaming H once told us that he
> didn't understand why this was sorted this way, because the drivers were
> developed in a similar fashion...

I don't know. This was done by the former maintainer Jens Schmalzing. As
I was never really sure if they belong to contrib or non-free I kept it
like that. If you have the time to investigate the current licensing
status and to propose a proper solution I would really appreciate it. 

> >    This segfault was also reported on the upstream mailinglist. AFAIK
> >    there is no fix for it atm. José: are you planning to work on this
> >    in the near future?
> By segfault I think that you may be referring to the userspace thing,
> right?


> > Unfortunaltely I have very little free time at the moment and I'm
> > unable to fix this two issues without help. Is there anyone else on
> > the MOL Packaging Team or the Debian PowerPC Mailinglist who could
> > take care of one of these issues. I wold be glad to assist if problems
> > arise. Solving the first problem probably only needs communication to
> > the kernel team.
> I think that I can lend a hand here, but I'm currently not sure of how
> much help I can be.

I think the first step would be to talk to the kernel team about the mm
symbol export patch. The next step would be to update mol-source. 

I would really appreciate your work and I can also add you to the Alioth
project if needed. Please base all your work on the latest version in
SVN: http://svn.debian.org/wsvn/pkg-mol


Ever tried. Ever failed. No matter.
Try again. Fail again. Fail better.
~ Samuel Beckett ~
diff --git a/arch/powerpc/kernel/ppc_ksyms.c b/arch/powerpc/kernel/ppc_ksyms.c
index c6b1aa3..455fb38 100644
--- a/arch/powerpc/kernel/ppc_ksyms.c
+++ b/arch/powerpc/kernel/ppc_ksyms.c
@@ -189,6 +189,7 @@ EXPORT_SYMBOL(set_context);
 extern long mol_trampoline;
 EXPORT_SYMBOL(mol_trampoline); /* For MOL */
 EXPORT_SYMBOL(flush_hash_pages); /* For MOL */
+EXPORT_SYMBOL(handle_mm_fault); /* For MOL */
 #ifdef CONFIG_SMP
 extern int mmu_hash_lock;
 EXPORT_SYMBOL(mmu_hash_lock); /* For MOL */

Reply to: