[OpenAFS-devel] Linux Kernel Module build question

[Derek Atkins]

Elliot Lee <sopwith@redhat.com> writes:

> Hmm, have you considered a solution like the following:
> 
> #include "/boot/kernel.h"
> 
> #undef __MODULE_KERNEL_subarch1
> #undef __MODULE_KERNEL_subarch2
> ...
> #undef __MODULE_KERNEL_subarchN
> #undef __BOOT_KERNEL_SMP
> #undef __BOOT_KERNEL_UP
> 
> #define __MODULE_KERNEL_desiredsubarch 1
> #define __BOOT_KERNEL_SMP 1

Well, this is a _VERY_ Red Hat specific workaround.  I would certainly
NEVER want to put anything like this into the OpenAFS source code at
large.  I suppose I could supply my own patch to the sources to do
this, but that would require maintaining that patch throughout time.
Part of that maintenence would include tracking the Red Hat kernel
sources in case they add any new special architecture targets.

Why can't Red Hat just fix the bug.  It's code _they_ added to the
kernel anyways.

> It would seem to be preferable to just let the kernel have modprobe load
> the module as needed. But, probably not my problem. :)

Well, tell me how to differentiate i386/i586/i686/athlon in a Requires
statement within the RPM and then I'll consider it.  Until then, there
just isn't a good way for me to know _which_ to install, and I
certainly don't expect the user to know which kernel they have
running, either.  At least if I can put a reasonable Requires
statement in the openafs-kernel package, then a user wont shoot
themselves in the foot by installing the wrong module.

> -- Elliot

-derek

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord@MIT.EDU                        PGP key available