Red Hat Bugzilla – Bug 162917
Radeon 9200 (Mac Ed) only works with fbdev
Last modified: 2008-08-02 19:40:33 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en) AppleWebKit/412 (KHTML, like Gecko) Safari/412
Description of problem:
Basically, I cannot use any of the capabilities of my Radeon 9200 (Mac Ed) graphics card under Rawhide
PPC. I can only use it as a fbdev. This was also true for FC3 and FC4 and is one of the main reasons I
tried Rawhide. However, I know the Radeon 9200 Mac Edition can be supported successfully under
xorg and Linux, since it works on a Ubuntu distro I run. I hope you can make it run under Fedora/
Redhat as well.
For the sake of bugzilla, I copy the content of the problem, as discussed at <http://lists.infradead.org/
I have a machine similar to J.M. Brough, a G4 PowerMac with an ATI
Radeon 9200 PCI card in it.
I updated my card's ROM through the software available at ATI so mine
reads (from the Mac OS X partition):
ATI 4.5.1 Radeon 9200
Card#: 109-actually gives numbers here-00
Card Model: ATY, RV280
ROM: ATY, Bugsy_A
FCode version: 1.94
NDRV version: 1.0.1f59
I installed the Xautoconfig tool from <http://www.bytebot.net/geekdocs/ibook/fedorappc.html> as
you suggest. I tried various
options, and got the following results:
with --safe option ==> an X cursor on a black screen. I can move the
cursor, but I when I right click, I get no menu. This is the best
I've got so far. X log at <http://www.pdc.kth.se/~mike/ppc/Bunny/Xorg_log_xautoconfig_safe>. It is
not clear to me that this
configuration is usable as is.
with --fbdev option ==> as above, log at <http://www.pdc.kth.se/~mike/ppc/Bunny/
with no options ==> dies with log as shown in <http://www.pdc.kth.se/~mike/ppc/Bunny/
I brought over a working xorg.conf from Ubuntu running on another G4
with the same card in it. See <http://www.pdc.kth.se/~mike/ppc/Monkey/xorg.conf> and the happy
log it produces on Ubuntu <http://www.pdc.kth.se/~mike/ppc/Monkey/Xorg.0.log>
When I try this xorg.conf under Fedora rawhide, it bombs see log at
careful analysis of the working X log on the one machine and the
failing log on the other should give a pretty good hint as to what is
wrong, I suspect. A simple diff, unfortunately, is a bit messy. I
haven't had a chance to look through the two logs extensively to get
at what the real differences are. Perhaps someone with more X
knowledge could give it a whirl?
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Using a working xorg.conf file from Ubuntu I try to startx using Fedora rawhide. X fails to start (with
messages as show at given URLs).
2. Using a X config file generated by Xautoconfig --safe or --fbdev, X works, though at a really poor
Actual Results: Error messages and failed X
Expected Results: Nice X11 starting.
Please see attached URLs which point to the important details. If you want these files inserted in their
entirety into bugzilla, let me know.
Other similar bugs I saw in bugzilla, but which were not quite close enough to simply add to, were:
121051 (radeon se card)
134967 (dual head--I am only trying to get one head working now)
162874 (radeon pro card)
The main problem IIRC, is that the IBM pSeries/iSeries machines need to
be configured one way, but Macs need to be configured a different way, and
you have to have things configured correct for the machine you're using.
Our PPC development started out for IBM PPC servers, and so our config
tools and whatnot assume IBM hardware and set it up accordingly as this
is what is desired for Red Hat Enterprise Linux.
However... now we are doing Fedora for PPC too, and this is not good for
Fedora/PPC users who are mostly if not entirely Mac users. We really do
need a solution that not only works for both systems, but preferably can
detect wether the given PPC box is IBM or Apple, and act accordingly.
I discussed this with someone before, perhaps Jeremy... I don't recall
exactly. I was pretty sure that someone had already changed our tools
to handle this properly, so that Macs get configured correctly. If not,
then we still have a config tool issue that needs to get ironed out.
Reassigning to system-config-display, as I believe that is where the
decision gets made. Please reassign to pyx*config or whatever if
that's more appropriate.
This ought to be working now, the main problem with the old method was that
pSeries wanted fbdev because the mga driver wouldn't work (due to 64-bit PCI bus
shenanigans). Please reopen if this issue still occurs for you on FC6 or newer.
NEEDINFO timeout, resolving as INSUFFICIENT_DATA. If you are still experiencing
this bug in FC6 or FC7, please reopen. Thanks for the report!