Work can continue.
On Thu, Jun 10, 2010 at 4:41 AM, Alan Coopersmith
<[email protected]> wrote:
Frank Middleton wrote:
This is my understanding of the issues
Frank just reminded me I hadn't confirmed this list was correct - it's
close, but not quite.
o The package of proprietary drivers for the legacy frame buffer
hardware will install on Open Solaris, but they won't work
because the PCI scanning software isn't compatible.
It is a longer story.
Xsun vs. Xorg. And in case of Xorg, then prelibpciaccess vs.
libpciaccess. And in case of pciaccess-Xorg it gets really nasty
(solved after 2 years with external help from an otherwise not very
helpful person).
If you're referring to:
http://hub.opensolaris.org/bin/view/downloads/sparc_graphics
The kernel drivers should work fine. They won't do anything really
useful though because they only include X server modules for the Xsun
server, not Xorg, and the two X servers use different interfaces for
their loadable driver modules for hardware.
The kernel drivers initiate the hardware, permitted to change
resolution during Xsun's times and (with Sun-internal knowledge) allow
to mmap() a frame buffer and its hardware registers without xsvc or
aperture.
Not too useful for the open src ddx modules available from www.x.org,
because they do not have this knowledge. Only exeption is Sun's
patched ast-ddx, which was open sourced thanks to Alan's engagement
and lives in the fox-gate and x-consolidation (not sure, must look
again).
The only "PCI scanning software" issue I know of is the issues between
the Xorg server we do ship, and the Xorg open source modules that Martin
ported from other OS'es for hardware we don't ship drivers for.
o Xorg won't work with the legacy drivers.
What do you mean with "legacy drivers"?
The Xsun ddx's? Then yes: NO WAY.
Because they don't include any Xorg driver modules.
Ah, right.
o XSun will work with the legacy drivers, but it needs to be
Open Sourced or installed from old SVR4 packages. Open
Sourcing it can't even begin until this project starts, and
even when it is, it will be low priority and hence could take
a long time. Once available, it will take considerable effort
to get it to build in the Open Solaris environment.
Technically, I could throw a tarball up on the website without a project,
and claim I'm done,
Oh, yes!
Please start by doing this without claiming you were done :-)
Please please ))
leaving any further action up to the community to run
with. Setting up a project is required to get an hg source code repository
setup on opensolaris.org - you could find some other place to host one instead,
github, sourceforge, google code, genunix, etc. - I just chose that as a
simple task to show that there was enough interest that I wasn't wasting
my time to do the code scrub and produce a tarball no one would ever use.
It will not stay unused.
Be sure of it.
I'm open to discussion of alternatives if you want, though I only have
permission to release code to opensolaris.org, so if you chose some other
site, I'd push the tarball to opensolaris, and you'd take it from there
and put it into whatever site you wanted.
Thanks to the pro-active community interest, this will be a vibrant
project with several core contributors.
No fear. Even though my responses were a bit sluggish due to some OT problems.
o In theory but probably not in practise, XSun could track
Xorg in features but realistically it won't happen, for
example, RandR.
In theory the server could add those features, but I don't think you
could get any usable RandR support without modifying the drivers to
handle the requests to change resolution/modes on the fly, and that
will be quite difficult without the sources for the drivers, which
you won't get.
Wanting to add features like RandR to Xsun would be difficult enough
by itself. But the lack of src code availability of Xsun PCI-ddx's
makes this completely impossible, rather than just complicated and
unrealistic.
I think the most important objective is to revive hardware support for
all legacy PCI frame buffers.
o There may be some features in XSun that might be useful
in Xorg
True.
that's the main reason for Sun/Oracle's interest.
Not really - Sun/Oracle has no official interest here. Management isn't
stopping me from doing it, but it's not on my official task list that they
expect me to complete.
I am glad they aren't stopping us.
And we should use the chance before they change their mind.
My main interest is in getting fewer people whining at me that I'm not
letting their 2010 OS run on their 2001 hardware (not that it was ever up to
me, I'm just the messenger).
This is 100% correct.
^I could also achieve that by just unsubscribing
from a ton of opensolaris.org mailing lists though...
--
-Alan Coopersmith- [email protected]
Oracle Solaris Platform Engineering: X Window System
Please don't do that.
You are one of the last Oracle employees who believe in the original
20050614 vision of OpenSolaris and work for it!
%martin
_______________________________________________
xwin-discuss mailing list
[email protected]
List info: http://mail.opensolaris.org/mailman/listinfo/xwin-discuss
Unsubscribe: http://mail.opensolaris.org/mailman/options/xwin-discuss
