Dean, I am able to build dmapi directly into the kernel, but it doesn't
build as a module. Just my 2 cents. :-)
** Derek J Witt **
* Email: mailto:djw@xxxxxxxxxxxxxx *
* Home Page: http://www.flinthills.com/~djw/ *
*** "...and on the eighth day, God met Bill Gates." - Unknown **
On Mon, 3 Jun 2002, Dean Roehrich wrote:
> Date: Mon, 03 Jun 2002 10:04:22 -0500
> From: Dean Roehrich <roehrich@xxxxxxx>
> To: Adrian Head <ahead@xxxxxxxxxxxxxx>
> Cc: linux-xfs@xxxxxxxxxxx
> Subject: Re: Problems with 2.4.19-pre9 & XFS/DMAPI kernel config logic
> >From: Adrian Head <ahead@xxxxxxxxxxxxxx>
> >What I did come across though is that the kernel config logic for XFS & DMAPI
> >doesn't really do what it should (AFAICT) from the TAKE message from a while
> >ago. This problem I discussed with Eric & Chris Pascoe on #xfs but I'm not
> >sure what the outcome was.
> >What happens is that make oldconfig doesn't pick up on the fact that DMAPI
> >should not be compiled as a module when XFS is compiled into the kernel.
> >Everything compiles correctly - however when a depmod is run it fails because
> >of missing symbols in the dmapi module.
> When xfs_support was merged back into the xfs directory, the dmapi module
> build was broken. You're missing symbols for things like sv_wait and kmem_*
> I'm working on it from time-to-time.
> At the very least, this pointed out a few more dependencies between XFS and
> DMAPI that I can work on. Someday DMAPI won't need XFS...