On Mon, 18 Aug 2003, David Howells wrote:
> > > (3) Modules from one version of the kernel shouldn't be changed...
> >
> > They were never changed. No modification is made to built modules.
>
> It invokes /sbin/insmod with a "-P prefix" flag. This constitutes a
> modification of the module.
Arguable. I agree with the point, but if the md5sums match it doesn't fit
with my world-view of modified. I'll buy that it's a correct change,
though.
> Yes, if you add the change to the spec file that I mentioned in response to
> Rudolph T Maceyko <[EMAIL PROTECTED]>.
>
> -%define krpmvers %(rpm -q --whatprovides kernel | sort | tail -1)
> +%define krpmvers_on_cmdline %{?krpmvers:1}%{!?krpmvers:0}
> +%if !%{krpmvers_on_cmdline}
> +%define krpmvers %(rpm -q --whatprovides kernel | sort | tail -1)
> +%endif
>
> And then:
>
> rpmbuild -bb \
> --target=i686 \
> --define "krpmvers kernel-2.4.20-1.1931.2.231.2.11.ent" \
> /usr/src/redhat/SPECS/openafs.spec
>
> Further command line define options could be made available to specify whether
> you desire the userspace and/or kernel modules to be built.
Really, this was the only fundamental problem I saw, (aside from the
unneeded code I mentioned separately) if Derek can deal I don't see why
any of it should be a problem (unless someone else has something we
missed?)
_______________________________________________
OpenAFS-devel mailing list
[EMAIL PROTECTED]
https://lists.openafs.org/mailman/listinfo/openafs-devel
