Martin Bochnig wrote:
> On Tue, Mar 31, 2009 at 12:52 AM, Alan Coopersmith
> <Alan.Coopersmith at sun.com> wrote:
>>> Of course I will do this.
>>> Nobody EVER told me.
>>> If we are a team ... ? Why?
>> Because sw-porters just opened up the sourcejuicer process last week and
>> I hadn't gotten around to sending mail suggesting it yet.
>
>> I apologize for thinking of possible solutions and not immediately telling
>> you about
>> them.
>
> This sounds like sarcasm.
Maybe a little, but I was responding to something that sounded like unwarranted
complaints to me from someone who flames a lot. While I'd mentioned before the
possibility of putting the drivers in /contrib repo as unsupported, it really
was only a few days before that I found out the sourcejuicer project was now
open and accepting specs, and I thought it would be a good way to get the
drivers in.
> This looks promising in many aspects.
> In case of Xorg, however, at the minimum the UPA bus scanning patch
> (from last week) needs to go into the core Xserver.
Understood, but we're now past code freeze for 2009.06, and adding support
for hardware that wasn't planned to be supported isn't a stopper bug that
qualifies for a respin, so it's not going in until after that.
I'm assuming you're referring to this commit, btw:
# User bochnig at SunBlade-2000
# Date 1237998730 -3600
# Node ID 2060af1a1c60cf44b08c39cd48d82434203a3cc4
# Parent 1995881bfb926735fad522779ddb9e15362eb82b
This commit makes Sbus- (in theory) and UPA- (very much in practice, I am workin
g on a Blade 2000
with 2 Creator 3D UPA boards running Xorg 1.5.3 right now while typing in this m
essage) bus
scanning work again.
The changes in there that I understand (devPrivates fixes & getpagesize API
update) look good to me - the bulk of the patch is in the UPA & Sbus code and
I don't know anything about those busses, so can't comment on those. Some
of the stuff in sparc-probe.patch from earlier revisions looks wrong (like
commenting out _X_EXPORT declarations), and the gratuitous spacing changes
should be removed to lessen the upgrade pain, but that's mostly minor.
The prior commit, which just undoes the PCI bus scan disabling on SPARC is a
problem though - we can't take 2-3 minutes to scan pci busses on Xorg startup,
or gdm & dtlogin will kill us before we get going (they only wait about 30
seconds) - either the underlying libpciaccess problem needs to be fixed or
the scanning needs to be configurable and disabled by default.
> Only all the other
> stuff like ddx modules and mesa can go into an additional repo.
Right - that's what I was thinking of in /contrib - just the driver modules
and GL stuff, not a whole second copy of Xorg.
--
-Alan Coopersmith- alan.coopersmith at sun.com
Sun Microsystems, Inc. - X Window System Engineering
