[OpenAFS] naive questions about openafs

Jeffrey Hutzelman jhutz@cmu.edu
Fri, 21 Jan 2005 16:51:18 -0500


On Friday, January 21, 2005 16:24:43 +0100 Guillaume Rousse 
<rousse@ccr.jussieu.fr> wrote:

> I also tried the libafs_tree, and found a minor bug: the configure_libafs
> must be renamed to configure, otherwise the first configure test (ls -t)
> will fail. Anyway, it doesn't solve my problem, as the configure doesn't
> generate the final Makefile for creating the module, only a primary
> Makefile in src/libafs that will create AND execute in the same run final
> makefile for two hardcoded target modules.


I think you might be confused about how libafs_tree is supposed to work.
If you configure openafs normally and then 'make only_libafs_tree', you 
will get a directory named 'libafs_tree' which is a standalone package that 
can be used to build OpenAFS kernel modules.  You can pick it up and move 
it elsewhere, copy it, rename it, whatever.  It is completely 
self-contained and does not depend on anything in the source directory 
where you made it.

To build a kernel module, you will need to run configure and make in the 
libafs_tree package.  The module it builds will depend on the kernel you 
configure it against.  This mechanism _does_ work; I've used it locally for 
quite some time, and it is the basis for the openafs-kernel-source package 
in at least one of the major Linux distributions we support.

It looks like there is indeed a bug -- the configure-libafs script should 
have been renamed as it was copied in, but apparently someone committed a 
patch a while back that broke this.

-- Jeffrey T. Hutzelman (N3NHS) <jhutz+@cmu.edu>
   Sr. Research Systems Programmer
   School of Computer Science - Research Computing Facility
   Carnegie Mellon University - Pittsburgh, PA