Bug 708277

Created attachment 501252 [details]
Xorg log of system booting with Radeon HD 3200

On initial setup and boot of the system, there was no xorg.conf created.

don't suppose you have any way to ssh in?

another option could be to boot with nomodeset 3

then when it boots login,


modprobe drm
echo 4 > /sys/module/drm/parameters/debug
modprobe radeon modeset=1

it might go blank but may allow typing
dmesg > mymsgs
reboot

Created attachment 501301 [details]
dmesg after running modprobe (as requested)

Comment on attachment 501301 [details]
dmesg after running modprobe (as requested)

Didn't run commands exactly as requested.

Created attachment 501307 [details]
dmesg attachment after running modprobe on drm, turning debug mode to 4, and running modprobe on radeon

Created attachment 501310 [details]
/var/log/messages

Also, I was able to successfully run modprobe drm; echo 4 > /sys/module/drm/parameters/debug; modprobe radeon modeset=1  without the system blanking out the screen.

does reomving rhgb quiet from the kernel command line help?

I am able to see kernel messages go by as the system is coming up.  I see most of the system services come up, that should be coming up.  I can see the services until about s90, and I can's say for sure that I see s95atd or s99rc-local.  Mind you, it could be that I can't see s95 or s99 flying by on the screen.

After I see s90smolt startup is about when the screen goes blank, and there is no control of the keyboard (numlock or capslock).

okay so KMS is working and picking up the correct output, so you can do rhgb quiet 3 and have it boot to a hi-res text login? (see if radeon is loaded).

then doing init 5 makes it die?

Ok, I was able to successfully boot into runlevel 3, without using nomodeset; however, at this runlevel, it doesn't appear that drm or radeon is loaded (checking with lsmod).

I was then able to take it to runlevel 5 successfully, using init 5.  I verified that Gnome was still running in "Fallback" mode.  

Another issue, that I found, when launching into X this way (not using nomodeset, and booting initially into runlevel 3), was that when I attempted to reboot, the process failed, locking up my keyboard, and blackening out the screen.

The other thing that I attempted, I toggled the option in Gnome, via the System Info menu option, to force fallback mode, mind you, I just toggled it from off, to on, and then back to off.  But in the configuration window, it said "The next login attempt will attempt to use the standard experience".

After restarting, I booted into runlevel 3 again, without "nomodeset", and that was successful.  Then I attempted to boot into runlevel 5, and that had failed; and I got the black screen, and no keyboard response.

Created attachment 501617 [details]
/var/log/messages -- current

Created attachment 501618 [details]
Logrotated /var/log/messages

okay so its a bit wierd,

1. can you boot to runlevel 3, modprobe radeon modeset=1 and get a hires text console?

2. then init 5 and have X start or not start?

It seems like your initramfs is busted, did you ever install fglrx?

If 2. works, then do 1 again, and run /usr/libexec/plymouth/plymouth-update-initrd
and reboot. In theory it should load radeon them from the initramfs and stuff might work.

Created attachment 501705 [details]
Latest /var/log/messages -- current

Ok, I was able to successfully execute step 1, boot from runlevel 3, modprobe radeon modeset=1, and get a Hi Res text console.

When I init 5 to X, it starts up, in "failover" mode.

I then did as you asked, I started into runlevel 3, modprobe radeon modeset=1, and then executed:  /usr/libexec/plymouth/plymouth-update-initrd

I then rebooted, and X failed to launch.

I'm sorry, I didn't answer your questino about fglrx.  At one point, during this testing, I attempted to load the driver from ATI, which I assume is similar to fglrx, the driver loaded fine.  I have since uninstalled it, and I have also since rebuilt the initrd, using mkinitrd.

Okay so the ATI driver has screwed up something or left something in a place it shouldn't.

so when you rebuilt the initramfs does the hires console come up now when you boot to runlevel 3?

there might be some fglrx leftovers in /etc/modprobe.d or somewhere like that.

The only thing that I wasn't sure if it was out of place or not, there was a file there, named /etc/modprobe.d/blacklist-fglrx.conf

In blacklist-fglrx.conf, there is a blacklist option for radeon (ala  blacklist radeon)

I commented out that line that is in blacklist-fglrx.conf (so radeon is no longer blacklisted).  And then rebooted into runlevel 3.  Something interesting I noticed, is that when I booted into runlevel 3 without "nomodeset" it successfully loaded, and was loaded in Hi Res mode.

I rebuilt initrd using mkinitrd, followed that up, by executing the plymouth-update-initrd.

Bug appears to be fixed with this latest round of updates.  I installed the following updates this morning, and restarted:

Jun 07 06:49:40 Updated: libgcc-4.6.0-9.fc15.x86_64
Jun 07 06:49:54 Updated: glibc-common-2.14-2.x86_64
Jun 07 06:50:01 Updated: glibc-2.14-2.x86_64
Jun 07 06:50:02 Updated: libstdc++-4.6.0-9.fc15.x86_64
Jun 07 06:50:04 Updated: 1:libreoffice-ure-3.3.2.2-9.fc15.x86_64
Jun 07 06:50:04 Updated: ibus-libs-1.3.99.20110419-1.fc15.x86_64
Jun 07 06:50:05 Updated: ibus-gtk2-1.3.99.20110419-1.fc15.x86_64
Jun 07 06:50:16 Updated: ibus-1.3.99.20110419-1.fc15.x86_64
Jun 07 06:50:16 Updated: ibus-gtk3-1.3.99.20110419-1.fc15.x86_64
Jun 07 06:50:17 Updated: libquadmath-4.6.0-9.fc15.x86_64
Jun 07 06:50:24 Updated: libgcj-4.6.0-9.fc15.x86_64
Jun 07 06:50:27 Updated: 1:gdm-3.0.4-1.fc15.x86_64
Jun 07 06:50:29 Updated: 1:libreoffice-opensymbol-fonts-3.3.2.2-9.fc15.noarch
Jun 07 06:50:30 Updated: 1:autocorr-en-3.3.2.2-9.fc15.noarch
Jun 07 06:51:04 Updated: 1:libreoffice-core-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:05 Updated: 1:libreoffice-impress-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:09 Updated: 1:libreoffice-presenter-screen-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:10 Updated: 1:libreoffice-graphicfilter-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:11 Updated: 1:libreoffice-pdfimport-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:11 Updated: 1:libreoffice-draw-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:12 Updated: 1:libreoffice-xsltfilter-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:15 Updated: 1:libreoffice-writer-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:18 Updated: 1:libreoffice-calc-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:18 Updated: 1:libreoffice-langpack-en-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:19 Updated: 1:libreoffice-math-3.3.2.2-9.fc15.x86_64
Jun 07 06:51:19 Updated: 1:gdm-plugin-fingerprint-3.0.4-1.fc15.x86_64
Jun 07 06:51:20 Updated: libgfortran-4.6.0-9.fc15.x86_64
Jun 07 06:51:21 Updated: perf-2.6.38.7-30.fc15.x86_64
Jun 07 06:51:22 Updated: cifs-utils-5.0-1.fc15.x86_64
Jun 07 06:51:22 Updated: libgomp-4.6.0-9.fc15.x86_64
Jun 07 06:51:23 Updated: jwhois-4.0-28.fc15.x86_64
Jun 07 06:51:24 Updated: libraw1394-2.0.7-1.fc15.x86_64
Jun 07 06:51:31 Updated: orca-3.0.1-1.fc15.x86_64
Jun 07 06:51:33 Updated: fedora-release-notes-15.1.0-1.fc15.noarch
Jun 07 06:51:37 Updated: glibc-2.14-2.i686
Jun 07 06:51:37 Updated: libgcc-4.6.0-9.fc15.i686
Jun 07 06:51:48 Installed: kernel-2.6.38.7-30.fc15.x86_64
Jun 07 06:51:49 Updated: libstdc++-4.6.0-9.fc15.i686


I'm assuming the kernel update fixed this issue.  

Now, when I go to System-Settings --> System Info, I get the following:

 Driver:  Gallium 0.4 on AMD RS780
Experience:  Standard
Fallback Mode: Off

    Technical note added. If any revisions are required, please edit the "Technical Notes" field
    accordingly. All revisions will be proofread by the Engineering Content Services team.
    
    New Contents:
Appeared to be fixed by kernel update:  kernel-2.6.38.7-30.fc15.x86_64