Description of problem: Anaconda, before formatting the hard disk, before installing packages, decides its necessary to add rainbow colors to the screen. Version-Release number of selected component (if applicable): /export/fedorappc/Fedora/RPMS/anaconda-10.0.2-3.ppc.rpm (rawhide 9/11) How reproducible: 100% (okay, I lied, I tried this twice) Steps to Reproduce: 1. Start installation on Mac with Radeon 7500 2. Nice graphical boot (happy!) 3. Suddenly screen becomes rainbow colored when we reach the partitioning stage Actual results: Rainbow colored screen Expected results: Nice normal colored screen Additional info: Which you get back by going to a console and back to the graphical installer. Its a low priority, if we don't fix it, we'll just clog the release notes about it as a quick workaround This is a complaint on YDL's site as well (http://cvs.terraplex.com/bugzilla/show_bug.cgi?id=20). Guess they should complain upstream!
*** Bug 132381 has been marked as a duplicate of this bug. ***
I don't have hardware, so there's not really much I can do as this is completely hardware specific (doesn't happen on !ppc and doesn't happen on pseries either).
I don't see how this isn't an X bug (or, potentially, a kernel FB bug.)
Confirmed. Same problem on my ibook G4.
Discussion with benh: the VT/fbdev code is bogus and sets it's local 8 bits palette even when it's not the frontmost console The workaround is to open the new VT before launching X (by just echoing something to it) It's whatever VT you open at the moment the problem appearzs But you can echo ... to all VTs from 0 to 7 just before launching X, that will work too I suppose. So workaround on PowerMac would be to initialise all VT's used before X
I see this behavior on my x86 machines.
This also happens on my Titanium Powerbook with latest rawhide. My video chip is ATI Radeon Mobility M9.
As an additional note, the machines I have seen this on have Matrox G400 video cards.
Does anyone in the Westford or Raleigh office have this problem that could make their hardware available to someone local to that office to investigate? Also, when using rawhide, please indicate the xorg-x11 version-release that was used to test, as rawhide changes daily. ;o)
Confirmed, installed todays rawhide on version xorg-x11-6.8.1-12. Maybe this is necessary, maybe it isnt. But the first initial splash screen is fine. Normal colors. After pressing the "next" button, the colors go all off. LIke it has been reduced to 16 colors, maybe less. I'm in the Brisbane office, dont know if that helps any.
Confirmed weirdness with Radeon Mobility 9200 as well, with today's rawhide xorg-x11-6.8.1-12.ppc and anaconda-10.1.0.0-1.ppc. Funny thing is, I don't remember this "bug" being around during the FC2 days...
Still seeing the problem with rawhide-nightly as of Oct. 27. (xorg-x11-6.8.1-12). Again, this is on an x86 Dell Precision box with a Matrox G400 video card. Mike, I'd be happy to try any kind of testing that you'd like me to do.
Changing platform to "All"
Some more information. I have 3 Dell Precision machines: 1) Matrox G400 - this one shows the problem 2) Matrox G200 - this one shows the probelm too 3) nVidia TNT2 - this one is ok I also have an Intel pre-production machine with a G400 that also shows the problem.
Brent - does it occur if you pass nofb at the boot prompt?
No, it does not occur if I pass nofb. Perhaps something has changed with the kernel's framebuffer handling?
I'm going to go out on a limb and chalk this up to something in the framebuffer in the kernel. Please correct me if I'm wrong. Changing component to the kernel.
Since I got the imac installed (#121265), I got the same rainbow-ish colormap issues. Twice, as per normal. Video card there is a r128.
Colin if you can bear to do another install I have a small hack to anaconda which may circumvent this.
Paul, does your anaconda hack just pass 'nofb' to the kernel?
no it loops through tty 2-6 and prints \n
mharris: Why did you change the platform to powerpc? I'm seeing this on my x86 Dell boxes.
Brent: When I skimmed over the bug report, I saw it was filed against "All" platforms, but noticed ".ppc.rpm" in the first comment, and skimming further saw ppc mentioned again a few times. I didn't notice any other architectures mentioned anywhere, but didn't read the entire bug report top to bottom either. Since I assumed it was PPC specific, I updated the platform to show PPC to try to better categorize the bug for investigation. Thanks for spotting this and updating the report to highlight x86 also.
Brent/Colin can you try with a ks install that has the following pre scriptlet and see if that fixes tested here on imac DV passing resolution=1024x768 %pre chvt 6 for in in 2 3 4 5 6; do echo "." > /dev/tty${i} ; done chvt 7
Paul, which tree would you like me to test with?
FC3 is fine
I think this is a dupe of #137849 . If so, it should be fixed in the latest build (assuming Jeremy did one on friday afternoon or since then).
Current rawhide should have the latest changes. Setting to modified, reopen if you still see the problem with rawhide.
Confirmed working on iMac