Bug 443089
Summary: | Cannot "Shutdown" system from menu | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Mathias Panzenböck <grosser.meister.morti> | ||||||
Component: | kdebase | Assignee: | Than Ngo <than> | ||||||
Status: | CLOSED CURRENTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||||
Severity: | medium | Docs Contact: | |||||||
Priority: | low | ||||||||
Version: | 8 | CC: | gchris, kevin, ltinkl, rdieter | ||||||
Target Milestone: | --- | ||||||||
Target Release: | --- | ||||||||
Hardware: | i386 | ||||||||
OS: | Linux | ||||||||
Whiteboard: | |||||||||
Fixed In Version: | F9 | Doc Type: | Bug Fix | ||||||
Doc Text: | Story Points: | --- | |||||||
Clone Of: | Environment: | ||||||||
Last Closed: | 2008-06-25 12:51:09 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: | |||||||||
Attachments: |
|
Description
Mathias Panzenböck
2008-04-18 16:05:23 UTC
Created attachment 302899 [details]
my /etc/kde/kdm/kdmrc
I'll take off desktop-bugs from the CC, those are GNOME folks who probably don't care about this KDE issue. ;-) Are you using gdm? If so, it's likely a dup of bug #429966 Nevermind, your initial comment mentions kdm. Fwiw, works fine here (using both f7, f8). Any hints on how to reproduce would be helpful. It only happens from time to time. I do not see any correlations about when it happens. Maybe it has something to do with multiple x sessions, but the last time it happened only one person was logged in. What video driver are you using? Standard xorg stuff or ati or nvidia? On different computers there are different symptoms. On my ThinkPad loggin off or trying to switch vt's end with black screen using ati drivers. On my desktop with nvidia drivers I get a long black screen (15-20 seconds) then kdm restarts. FYI I use nvidias drivers. It happens with both 171.06 and 169.12. I have an GeForce 8800 on a dual core intel. But under Ubuntu with a GeForce 4400 and 169.12 on an single core amd I never experienced any thing like this. I've run into this on two very different Intel systems, one a 865/P4 system, the other a X38/Core 2 duo system. The first system uses an Nvidia 6200 with the 169.12 driver, the second an Nvidia 9600 with the 173.14.05 driver. The failure shows up about one time out of 5 when shutting down from KDE - never from Gnome. The first system ran FC6/KDE for over a year with no failures. It started failing immediately when F8/KDE was installed. The common data here appears to be nVidia... If this cannot be reproduces without the proprietary nvidia driver, I'll have to blame that driver and close this bug. Just a minute! Two different Nvidia cards and two different Nvidia drivers?!? And it never fails with FC6/KDE but fails with both F7/F8 and KDE but never with Gnome?!? It seems a bit of a stretch to lay that on Nvidia. I could see looking hard at KDE or RH's implementation of KDE, but pointing at the most popular supplier of graphics for Linux seems a reach. They aren't really different drivers, just different versions of the same driver. NVidia doesn't rewrite their drivers from scratch for each release, nobody does that! So bugs can persist across several versions. The exact hardware used is also irrelevant in the case of a driver bug. So I think the fact that different versions of the driver and different cards are involved is not a good argument for claiming this is not a driver bug. The only way to prove that it is not a video driver bug would be to try reproducing it with a different graphics driver. Maybe you want to try nv or nouveau, just to see if this happens with those too? As for "the most popular supplier of graphics for Linux", that's an assertion you have to provide stats for. But in any case the nvidia driver is NOT supported by Fedora (and we CANNOT support it, because we have NO way to fix bugs in it, only nVidia can), and nVidia is a bad choice of a manufacturer, it's the only major graphics card producer which is NOT cooperating with Free Software. Some more evidence for a driver/X11 problem: here are the latest versions of xorg-x11-server in the updates for the different Fedora versions: FC6: xorg-x11-server-1.1.1-47.10.fc6 F7: xorg-x11-server-1.3.0.0-17.fc7 F8: xorg-x11-server-1.3.0.0-46.fc8 F9: xorg-x11-server-1.4.99.902-3.20080612.fc9 So my guess of why this worked in FC6 and not in F7 is that a bug was introduced in the nvidia driver when porting it to the xorg-x11-server-1.3. Regardless, only when/if someone can reproduce this with only fedora-supplied packages/drivers, can we *know* the bug lives here. Otherwise, there will always be uncertainty and doubt. Okay, I'll convert one of my systems back to nv and do some some testing with that tonight. Created attachment 309959 [details]
Using nv driver
This is the resulting xorg.conf after livna rpms removed.
Okay, after removing livna's Nvidia RPMs I've gotten several failures shutting down from KDE using the nv driver. The symptoms are that shutdown (or restart) is accepted from the gui which takes you to a black screen containing only an active mouse cursor and there it hangs. Hitting Ctrl-Alt-Backspace gets you a text mode login prompt and from there, logged in as root, a shutdown -r now takes you through a successful reboot to a KDE/nv session. So much for "uncertainty and doubt"? ;-) Over the weekend, a hardware failure forced me to install a new hard drive and re-install Fedora 8. Immediately after the install this bug re-occurred almost immediately (using nv). However, after installing the 324 outstanding updates I've not been able to recreate the problem. There must have been some magic in there somewhere, and unless I report something in the next day or two I'd be inclined to call this one history. I've not seen any re-occurrence of this bug so I'd have no objection to it being closed. BTW, I started running the Nvidia driver again yesterday and that had no adverse effects. |