0.9.70 loading kmods failes on ydl3

[Eddie Bindt]

On Mon, 1 Mar 2004, Samuel Rydh wrote:

> Yes, this problem has been fixed in later MOL versions
> (it was triggered when enough Mac OS X side KEXTs was
> put into the System/Library/Extension directory).

OK, tnx ...
 
> > Loading Mac-on-Linux kernel module:
> > /usr/lib/mol/0.9.70/modules/2.4.22-2g/mol.o: unresolved symbol force_sig
> > /usr/lib/mol/0.9.70/modules/2.4.22-2g/mol.o: unresolved symbol __up
> > /usr/lib/mol/0.9.70/modules/2.4.22-2g/mol.o: unresolved symbol find_vma
> > /usr/lib/mol/0.9.70/modules/2.4.22-2g/mol.o: unresolved symbol __down
> > /usr/lib/mol/0.9.70/modules/2.4.22-2g/mol.o: unresolved symbol 
> > misc_register
> > /usr/lib/mol/0.9.70/modules/2.4.22-2g/mol.o: unresolved symbol mem_map
> > ====================================================================
> >   Failed to load the module - try recompiling the MOL kernel
> 
> The problem is the kernel source/headers used while compiling
> mol-kmods. The above error typically occurs when the kernel headers
> used during the compilation has been configured with for
> instance CONFIG_SMP off while your running kernel has CONFIG_SMP on
> (or vice versa).

Hmmm, that is strange, I really have the non-smp modules and the non-smp 
config in he right places... Rechecked and rebuilded kmods, but still the 
same problem ...
 
> MOL looks for the kernel source in /lib/modules/`uname -r`/build.
> The main problem is that vendors tend to ship a single kernel 
> tree but multiple kernels. The source tree configuration will in general
> not match the kernel. By doing distribution specific stuff it may
> or may not be possible to recover working kernel headers.
> 
> My personally opinion is that /lib/modules/`uname -r` should always
> point to working kernel headers and the distro is broken if this
> is not the case...

hmmm, I am sure the right config is in the right source tree. I even did 
make oldconfig and make dep in the source tree ... Still no luck, I always 
was able to rebuild kmods, until now ...

> The easiest way to resolve this problem is probably to compile
> your own kernel.

There must be someway to solve this and still keep a stock YellowDog 
kernel and kernel-source ....

Eddie
 
> /Samuel
> _______________________________________________
> mol-general mailing list
> mol-general@lists.maconlinux.org
> http://lists.maconlinux.org/mailman/listinfo/mol-general