From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4.1) Gecko/20031009 Description of problem: If I use a DRI-enabled program (for example, any 3D screensaver or 3D games), after a little while (about 1 minute) my computer just shutdowns by itself, as if I had unplugged it. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1.Launch a program using DRI 2.Wait a little (about 1 minute) Actual Results: Power off :( Expected Results: Go on using my computer ! Additional info: My computer is a Fujitsu/Siemens Amilo D laptop. It features an ATI Radeon Mobility M6 video card. I can reproduce this bug with or without monitor power management enabled, with or without ACPId and with or without APMd.
Please attach your /var/log/XFree86.* and /etc/X11/XF86Config, and /var/log/messages to the bug report as single file attachments uncompressed using the link below. Also, you using 2.4.x or 2.6.x kernel when this occurs?
Created attachment 95462 [details] /var/log/XFRee86.0.log
Created attachment 95463 [details] /etc/X11/XF86Config
Here you are.
Created attachment 95464 [details] /var/log/messages I've cut it to show only messages for today
Sorry, forgot to say that I'm using kernel 2.4.x (the one coming with Fedora Core Test 3 by default).
To be more precise, it's the 2.4.22-1.2088.nptl.
Oct 24 19:40:53 localhost smartd[2576]: Device: /dev/hda, SMART Usage Attribute: 194 Temperature_Celsius changed from 137 to 144 Um, 144 degrees celsius seems a tad high for normal operation. Are you sure the machine isn't overheating?
I found out that the problem is not DRI-related as I can reproduce it using applications not using DRI. It seems it shuts down when an application is using the CPU heavily. More over, when I use my laptop on battery mode, this bug disappears...
You're using Radeon hardware, which means that if DRI is enabled in your config file, DRI is used for both 2D and 3D acceleration. In order to exclude DRI completely, please edit your X config file and comment out the 'Load "dri"' line, so that DRI is totally disabled. This will cause the Radeon driver to fall back to MMIO 2D acceleration instead of using the CP. If the hangs still occur with DRI totally disabled, but go away when on battery, most likely while on battery your system is going into low power mode and thus wont generate as much heat. Can you post a kernel log from being on battery power also, which shows the temperature?
I've removed Load "DRI" from my config file nd the problem remains. I'd say it's even worse as the laptop shuts down even faster (about 10 seconds after I launche the 3D-enabled game). I guess that's because the CPU now has to handle all the 3D instead of the video card. So, IMO, that goes in the direction of CPU over-heating... However, I can't get a new log file showing the temperature. But I did the following (disclaimer: I don't know what I'm doing...) In battery mode without the 3D app running: [root@dhcppc0 julo]# cat /proc/acpi/thermal_zone/THR1/temperature temperature: 63 C [root@dhcppc0 julo]# cat /proc/acpi/thermal_zone/THR2/temperature temperature: 62 C In battery mode with the 3D app running: [root@dhcppc0 julo]# cat /proc/acpi/thermal_zone/THR1/temperature temperature: 74 C [root@dhcppc0 julo]# cat /proc/acpi/thermal_zone/THR2/temperature temperature: 65 C Is it interesting to you ?
When not on battery mode, cat /proc/acpi/thermal_zone/THR1/temperature gives more something like 86 when running this 3D app, just before the laptop shuts down... So I guess, it's over-heating. More over, I changed my BIOS setup to "smart throttle". Now, I have a brighter screen and a slower laptop. But it doesn't shut down anymore. Instead of shutting down, it becomes very slow. what troubles me, though, is that I didn't have those problems with Redhat 9. Any idea ?
Sounds to me like this is either a hardware malfunction issue, or it is power management related problem. If it doesn't occur in RHL 9, I would assume perhaps something in the kernel changed, although I can't really say for sure. I've carbon copied several kernel people who might be aware of any relevant changes or issues WRT power management.
I'm totally confused about this bug... I've stopped using 3D apps for now (which isn't a big deal because most of them are games/screensavers) and I'm waiting for ideas from kernel people. But wouldn't it be possible that one of my fans may have broken one day near the installation of Fedora Test3 ? Is it normal to have a big difference between THR1's temperature and THR2's temperature ?
Good news: today, after having switched my laptop on, a cloud of dust went out from its fan... And, from now on, it doesn't overheats anymore. So I guess the fan was stuck by something and didn't do its job. In conclusion, it was a hardware problem... Thanks and sorry for the noise, I'm closing this bug.