FreeBSD/xen structure

[M. Warner Losh]

In message: <20070528130101.GD48357 at rink.nu>
            Rink Springer <rink at freebsd.org> writes:
: Hi everyone,
: 
: As I've just mailed to current@, work is well underway on the Xen
: porting effort. However, as not only I but a lot of people will want to
: see this work integrated into CURRENT at some point, I'd like to raise a
: discussion on the directory layout I'm using. It has not changed from
: Kip Macy's perforce tree, but I want to ensure that this will be
: suitable for inclusion in the tree.
: 
: Basically, i386-xen (it's i386 only for now) is a sub-architecture just
: like pc98. The layout is the following:
: 
:  i386-xen/		Xen main tree
: 	compile/	Compile tree
: 	conf/		Kernel configs
:  	i386-xen/	Low-level code, comparable to i386/i386/
: 	include/	Include files - most include their i386/include/
: 			version, but some are different or extended.
: 
: This is basically the machine-dependant stuff. Should a port of amd64-xen
: happen in the future, it would go using a simular directory layout.
: 
: Xen-dependant but architecture-independant drivers (such as the Xen
: block device drivers) are put in the dev/xen directory. It should be
: possible to use these drivers in a amd64-xen version as well.
: 
: I'd prefer to keep Xen in a i386-xen tree, as there are quite a lot of
: changes, comparable to the amd64 <-> i386 split. And I am sure we are
: not really in favour for douzens of #ifdef XEN's in the tree.

How pervasive are these diffs?  And to what extent are you able to
reuse the i386 stuff?  Can you characterize, say, the number of lines
and/or files that would be affected if you went the ifdef route?  It
sounds like the tree is right, but that has its own cost and shouldn't
be undertaken lightly.

: Are there any questions, comments, remarks etc. on this layout? You can
: inspect the work in perforce (//projects/xen3); currently, none of my
: changes have been committed, but the layout is the same.

I like the layout, assuming it is needed.

Warner