Subject: Re: DRI on NetBSD
To: None <current-users@netbsd.org>
From: Eric Anholt <eanholt@gladstone.uoregon.edu>
List: current-users
Date: 06/17/2002 02:05:49
It seems there has been some confusion on what I'm doing with the DRI on
NetBSD.  I've had it in my TODO list as something I've wished I'd worked
on for a while, but only recently started really poking at.  Part of the
problem is that right now my disks are mostly dedicated to FreeBSD, and
my only Net installation is a 1G emulated one on linux on the laptop.  I
hope to correct this soon.

So far I've managed to hook the code up to the kernel build locally and
apply fixes to get it almost compiling. I understand Erik Reid has taken
the old sysdrm tarball from 2002-04-17 (when I was originally working on
getting the DRM into FreeBSD's base) and got it compiling but not
working.  I can't remember the state of the 2002-04-17 stuff, so it
might not be his problem.

BSD developers have a branch of DRI CVS, bsd-3-0-0-branch.  It's where
I've been trying to bring DRI CVS up to date with FreeBSD and hopefully
soon get some of the driver-specific files (ones starting with a card
name) to be os-independent.  I think that NetBSD work that doesn't break
FreeBSD should be fine, and I hope to get the code I've done so far
committed tonight or at least soon.

Any diffs to bsd-3-0-0-branch are greatly appreciated, and particularly
any help on how to LKM-ify the code if possible (so it could be
compilable directly from DRI CVS).

To respond to someone else's question I saw in the archives: The sysctl
is for debugging purposes and replaces the /proc/dri (iirc) information
outlet on linux.  It has proved useful for some people, but is not a
requirement for a working DRM.

(I'm on current-users now, so I'll be able to get replies there)
-- 
Eric Anholt <eanholt@gladstone.uoregon.edu>
http://gladstone.uoregon.edu/~eanholt/dri/