I have a cheap Radeon 9200 with chip ID 5964 which is not detected by XFree86-4.3.0-42 (rebuilt for Red Hat 9). Everything works if I force a ChipID of 0x5963. Looking at XFree86-4.3.0-radeon-support-backport-from-CVS-v2.patch, I would assume that this card would be detected fine because that chip ID is explicitly added in several places in that patch. Just for grins I tried forcing a ChipID of 0x5964 to make sure it wasn't a detection issue but the server errors out with: (WW) RADEON: ChipID 0x5964 in Device section "Videocard0" isn't valid for this driver I will attach lspci -vvv and XFree logs both with and without the ChipID specified as 0x5963.
Created attachment 95532 [details] lspci output
Created attachment 95533 [details] log without ChipID 0x5963
Created attachment 95534 [details] log with ChipID 0x5963
Just a quick note that after more experimentation I found that any 3D application would hang immediately upon starting if I used that ChipID. My guess is that this card is a cheap or slow version of a regular Radeon 9200 and can't tolerate the same timings. Right now I'm using ChipID 0x4967 and everything is working great.
The problem is very unlikely to be due to cheap hardware. The support for Radeon 9200 is semi-experimental, but works for some people. I am collecting bug reports on various non-working Radeon hardware and trying to find commonalities between reported problems. I will be working on the various reported problems sometime within the next month or so, and hopefully can work out any glitches that have been reported. Your newly supplied information should be useful when I get to tackling this also. If you figure out anything else new that you think might be useful, please add it to the report too. TIA
Jason Tibbits, any newer test results with FC1 or FC2 test?
I haven't upgraded a machine with this card yet, but I have some spare cards around so I'll try to build up a machine and test it out this week.
Just had occasion to try FC1 on a machine with this card; it was not detected and the box came up in VESA mode as RH9 did. XFree version is 4.3.0-55. I haven't started to move on FC2 as of yet, but I'll report back when I have a chance to try it. Alternately I can ship a card to someone; I have several in the spares drawer.
You can give the XFree86 in rawhide a try on FC1. I think it is kept in sync across both distributions.
I updated XFree86* from my local Fedora mirror and edited /etc/X11/XF86Config to change "vesa" to "radeon"; everything seems to come up OK. OpenGL apps are stableas well. I'm not sure how to test that a fresh install will properly detect the chipset, however.
rawhide XFree86 is meant more to be used with Fedora Core development, and may or may not work at all in Fedora Core 1. The XFree86 spec file inside the src.rpm is kept in a state that is used to generate rpms for as far back as Red Hat Linux 8.0, however there are various aspects of the spec file which are OS release independant and may end up breaking your OS in subtle ways, if you install binaries generated for one release on another, or in other cases it may make your system inoperable at all due to enhancements that require other OS improvements only present in newer OS releases. I do not recommend installing Fedora Core developmental XFree86 binary rpms on a Fedora Core 1 system. I recommend instead if one wishes to do that, to upgrade to the latest Fedora Core test relesae and then upgrade to current rawhide bits, that way the package is running in the environment it was compiled for. That said, you may find that the binaries do actually work in Fedora Core 1 perhaps, but if so, it is just luck and not something intentionally or guaranteed to always happen. ;o) Nothing has changed in the radeon driver for quite a while, and since I'm replacing 4.3.0 with Xorg X11 in Fedora Core devel soon, there are no planned radeon updates for 4.3.0 for quite some time in the future at least at the current time. Hope this helps.
No problem at all with these cards in FC2 and beyond, so I'll close this bug out.