On Mon, Oct 09, 2006 at 06:57:06PM +1000, matthew green wrote:
> 
>    On Sun, Oct 08, 2006 at 08:33:01PM -0700, Jason Thorpe wrote:
>    > 
>    > On Oct 8, 2006, at 6:53 PM, matthew green wrote:
>    > 
>    > >i agree.  we should move away from link sets even tho they are really
>    > >quite handy...
>    > 
>    > I am ok with keeping them around for kernel-private stuff, i.e. things  
>    > that will never be able to load as an LKM.  But eventually, I'd like  
>    > the "load as LKM" and "statically compiled into the kernel" code paths  
>    > to be identical, including the invocation of init and fini routines.
>    
>    Agreed.  This is what I need them for.  The file system transactions are
>    helper functions and will probably never be loadable as an LKM.
> 
> 
> hmm?  almost all our file systems work as LKM's currently...

Yes -- but these functions may be called outside of file systems.

-- 
Juergen Hannken-Illjes - hannken@eis.cs.tu-bs.de - TU Braunschweig (Germany)