| Summary: | kerneloops using akmod-fglrx-9.3-1.fc10.x86_64 | ||
|---|---|---|---|
| Product: | Fedora | Reporter: | Lars Hamann <lars> |
| Component: | fglrx-kmod | Assignee: | Stewart Adam <s.adam> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | fedora, kwizart |
| Priority: | P5 | ||
| Version: | 10 | ||
| Hardware: | x86_64 | ||
| OS: | GNU/Linux | ||
| namespace: | |||
| Attachments: |
kernel-message
lspci -vv xorg.conf |
||
Created attachment 145 [details]
lspci -vv
Can you attach your /etc/X11/xorg.conf file? Created attachment 146 [details]
xorg.conf
(In reply to comment #3) > Section "Files" > EndSection That's the problem - I'm not sure what's causing it, but many users have been reporting problems where the cause is that the ModulePath lines are missing... Here's how to re-add them automatically: # fglrx-config-display disable # fglrx-config-display enable Yes, running fglrx-config-display fixes the problem. Many thanks for your help. My xorg.conf never contained the ModulePath lines. I don't know which version of livna-config-display created that conf, but obviously 9.2 worked without those lines and rpm update to 9.3 does not run fglrx-config-display(?). They were definitely there in 9.2 - I forget the exact release, but I think it was 8.12 & up that required the extra ModulePath lines, otherwise X would crash or just present a black screen. fglrx-config-display should run on updates, but if it finds that it's already been run (the ld override configuration file is found), then it skips configuration - so another update or a bug in the script has been accidentally removing the ModulePath lines without us knowing about it. Yes, make sense to me. My xorg.conf was created 2008-11-14 (fglrx 8.11). I was unable to get 8.12+ working. But didn't file any bug-reports. Anyway 9.2 somehow worked (without the ModulePath lines), although I got problems using googleearth. So I guess the 8.11 -> 8.12 update did not run fglrx-config-display, because it found the configuration file. Maybe there should be a better test whether fglrx-config-display is needed or not. (In reply to comment #7) > So I guess the 8.11 -> 8.12 update did not run fglrx-config-display, > because it found the configuration file. Maybe there should be a better > test whether fglrx-config-display is needed or not. A new RPM (xorg-x11-drv-fglrx-9.3-2) should be available soon in updates-testing which resolves this issue. Instead of skipping all of the auto-configuration when the fglrx ld configuration file is found, it now only skips the 'ldconfig' call. |
Created attachment 144 [details] kernel-message fglrx 9.3 fails to start X11, while 9.2 worked fine.