Bug 484859
Summary: | xorg.conf is modified during each system upgrade | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Zdenek Kabelac <zkabelac> |
Component: | xorg-x11-server | Assignee: | Adam Jackson <ajax> |
Status: | CLOSED RAWHIDE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | medium | Docs Contact: | |
Priority: | low | ||
Version: | rawhide | CC: | patrick133t, xgl-maint |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2009-05-08 18:41:36 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Zdenek Kabelac
2009-02-10 12:12:56 UTC
If you lose power during a package upgrade you're already boned. I mean, we're replacing the _binary_. The config file is the least of your worries. What are you modifying ModulePath for? I've started to use ModulePath for my git-compiled libdrm libs and intel driver to be able to use UXA acceleration - as the current packages in Fedora are crashing my box mainly in suspend/resume or console switching. IMHO touching such configure file without giving the user some option to avoid it and without providing backup file is not a good idea. I have the exact same problem. I'm not trying to use any advanced features of the driver, but the current Intel driver supplied by Fedora 10 causes X to lockup right after the X server starts. I had to compile a newer version from git and put that into /usr/local/lib/xorg/modules. Then I had to add an entry to my ModulePath in xorg.conf. When I upgrade xorg-x11-server-Xorg, the ModulePath line is completely removed, apparently as a hack for the nvidia driver. That means that when I reboot after an update, my X server is going to lock (this is my only notification that something went wrong), I have to use magic sysrq hotkeys to reboot. Then I have to modify the grub boot to add a "3" to the end of the kernel command line so I can repair xorg.conf before starting the X server. Now that I understand what's happening and why, I can do this pretty quickly, but there has to be a better test for nvidia use than the one currently in there: # lame, the nvidia driver needs to override this if ! grep -q 'ModulePath.*nvidia' xorg.conf ; then sed -i 's#^\s*ModulePath.*$##gi' xorg.conf fi For reference, my line looks like: ModulePath "/usr/local/lib/xorg/modules,/usr/lib/xorg/modules" So, the RPM wipes out the ModulePath line if the text "nvidia" does not appear in it. It doesn't appear in mine, but I also don't need any hacks for nvidia, so the logic is clearly wrong. I assume this is for the proprietary nvidia driver, so we're doing this because one cannot tell in advance if a user wants to install that driver, and installation will fail if a ModulePath already exists? I hate to see a workaround for proprietary drivers break things for users that are sticking to free software. I would be willing to deal with creating a file named /etc/X11/i-hate-nvidia so that the preinstall scriptlet could do this: # lame, the nvidia driver needs to override this if ! -e /etc/X11/i-hate-nvidia ; then if ! grep -q 'ModulePath.*nvidia' xorg.conf ; then sed -i 's#^\s*ModulePath.*$##gi' xorg.conf fi fi Fixed in 1.6.0-16 and later |