Heads up 2.5.42 does not work with mkelfImage

I just did a quick check and the kernels 32bit entry point
has changed again.  In particular the segment numbers
that are used are different.

I am going to see what I can do but this could be a serious
problem.

Eric

[Ronald G Minnich]

On 16 Oct 2002, Eric W. Biederman wrote:
> I just did a quick check and the kernels 32bit entry point
> has changed again.  In particular the segment numbers
> that are used are different.

how different? Are there a different number of registers, or did they just
move around?

Also, doesn't the 32-bit entry point get handled transparently by the ELF
code?

We have the same problem with plan 9. We're going to need to figure out
how to parameterize this sooner or later.

kron

[Eric W Biederman]

Ronald G Minnich  writes:
> On 16 Oct 2002, Eric W. Biederman wrote:
> > I just did a quick check and the kernels 32bit entry point
> > has changed again.  In particular the segment numbers
> > that are used are different.
>
> how different?

The contents of the segment registers are different.  As 2.5.x
currently sits they only way to have a version of mkelfImage
that works for both 2.5.recent and 2.4.x is to explicitly test
the version number.
> Are there a different number of registers, or did they just
> move around?
>
> Also, doesn't the 32-bit entry point get handled transparently by the ELF
> code?

Yes so no BIOS's need to be rebuilt.  But tools like mkelfImage need
some help.  Hopefully I can get some native support into Linux and
then I can just skip the need to port mkelfImage, to newer kernels.
The feature freeze is on the 20th so I don't know if I will be able to
get anything in or not.
> We have the same problem with plan 9. We're going to need to figure out
> how to parameterize this sooner or later.

Getting support into the kernel proper is they only long term
maintainable way to go.

Eric