To anyone interested in support for various Alchemy processors or their
onboard peripherials (PCI, PCMCIA, etc.):

I have been working on platform support which greatly enhances these
platforms.   I have PCI working well, and have fixes to make USB work in
big-endian mode, added support for the Au1550 processor, and have
addressed some issues with portability.  I am adding support for a
(machine independent) radeon framebuffer, and will be shortly following
up with PCMCIA and Ac97 support on the Alchemy, followed by the other
various devices (hw crypto, GPIOs, possibly flash memories, etc.). 
There will probably be other bits that I fix and improve as I go, in MI
and MD code alike.  (E.g. recent fixes I made to make certain USB
network devices work on big-endian systems.)

My company and I would like to be able to give this all back to NetBSD. 
I have been going thru the "channels", but it seems like there has been
some impasse where I the changes that I've submitted now months ago have
not even been reviewed, apparently due to lack of resource to perform
the reviews.

We need to continue to make forward progress to make our schedules, and
the dependency tracking involved with keeping all these changes separate
is quickly becoming intractable.

SO, if *you* are interested in support for these parts, you need to
realize that unless someone steps forward with the time and ability to
help move the process along, it is quite likely that I will cease my
efforts to send incremental patches and PRs back to NetBSD.  This also
raises the question about whether those improvements will ever get
integrated, because the cost of reviewing the large set of changes that
I create is likely to be a little overwhelming to review and test all at
once.

At some point I may still (would like to) contribute my improvements
back to NetBSD, but it is likely to be very much later, in the form of a
mondo tarball instead of separately reviewed/tested changes.

I apologize for needing to proceed this way, but it has now been a
couple of months while I have waited to get even the very most simple,
trivial enhancements committed to the evbmips port, and the dependency
graph for my changes is starting to become unmanageable.

Again, if anyone with sufficient skill, willingness, time, and access to
the NetBSD CVS repository wants to help expedite these changes, then I'd
be more than willing to go thru the process.  Right now I just can't
seem to identify any such person and my understanding is that NetBSD
rules don't provide any alternative relief.  (At this point, I do not
have CVS access to NetBSD.)

Another option I'm willing to consider is to set up a public CVS clone
of the NetBSD tree, and folks who want to work from an unofficial copy
of the source can access my changes that way.  (Much like some of the
Linux ports have done business in the past, where a separate
port-specific version of the tree existed prior to integration into the
main Linux tree.)  If folks are interested in having me do this, then
please let me know and I will begin the inquiries to find out about
setting up a public anonymous CVS server.

-- 
Garrett D'Amore                          http://www.tadpolecomputer.com/
Sr. Staff Engineer          Extending the Power of 64-bit UNIX Computing
Tadpole Computer, Inc.                             Phone: (951) 325-2134