|
|
Overview
The following page documents some loose notes I have on running Linux, and specifically Fedora 8, on my HP a1424n Desktop. I've seen this box refered to as both a1400 series and a1000 series for generic model numbers. I think the a1000 is more correct.
Devices
- DVD +/- Lightscribe - WORKS (haven't tried Lightscribe)
- Firewire - UNKNOWN
- SATA Controller - WORKS
- Multimedia Keyboard - WORKS (haven't tried multimedia keys)
- Ethernet Jack - WORKS
- Sound Card - WORKS
- USB Ports - WORKS
- Intel 915 Graphics - WORKS including 3D accel
ACPI Issues
I really wouldn't have bothered with writing this page if it wasn't for issues with ACPI because everthing else "just worked".
I noticed that in Fedora's daily log email I was getting up to 50000 ACPI Error reports. This was really overwhelming my system. Example outputs found in /var/log/messages:
Feb 3 05:28:59 hostname kernel: ACPI Error (psargs-0355): [\_TZ_.THRM] Namespace lookup failure, AE_NOT_FOUND Feb 3 05:28:59 hostname kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_GPE._L1C] (Node f7d02450), AE_NOT_FOUND Feb 3 05:28:59 hostname kernel: ACPI Exception (evgpe-0576): AE_NOT_FOUND, while evaluating GPE method [_L1C] [20070126]
Doing a little googling I found out that ACPI DSDT has some bugs in in. Its referencing the above symbols at certain points without them existing/defined.
Well, one solution is to patch to DSDT I guess but the problem is somewhat showing that thermal related issues are occuring. The CPU is probably running a little hot.
When I first got the computer, there was no driver in Linux that supported is FAN controller. Over time the dme1737 was added to the Linux kernel and I could then use lm-sensors to verify that the CPU was running hot.
Once this driver existed, it was also possible to tweak the device to run the FAN's faster. I added the following two lines to my /etc/rc.local file and sure enough this alone reduced the frequency of the ACPI Error messages. I went from counts in 10000 multiples to counts in 100 multiples.
echo "2" > /sys/bus/i2c/drivers/dme1737/0-002e/pwm1_enable echo "2" > /sys/bus/i2c/drivers/dme1737/0-002e/pwm2_enable
What those lines do is change the device from manual mode to auto mode so that the FAN speed will automatically be adjusted as it gets hotter.
Something has recently changed after updating my software and the number of messages has increased some more. Looking into it, I've noticed that the default settings only cause the FAN to go into full speed mode as the CPU and internal temperatures reach 128 degrees C. Much to high for my liking. I'm currently using the following settings:
echo "60000" > /sys/bus/i2c/drivers/dme1737/0-002e/zone1_auto_point1_temp echo "65000" > /sys/bus/i2c/drivers/dme1737/0-002e/zone1_auto_point2_temp echo "71000" > /sys/bus/i2c/drivers/dme1737/0-002e/zone1_auto_point3_temp echo "35000" > /sys/bus/i2c/drivers/dme1737/0-002e/zone2_auto_point1_temp echo "39000" > /sys/bus/i2c/drivers/dme1737/0-002e/zone2_auto_point2_temp echo "42000" > /sys/bus/i2c/drivers/dme1737/0-002e/zone2_auto_point3_temp
Those values are loosely based off information from Intel's Thermal Management page and me just watching what my system appeared to prefer running at under both Windows and Linux. Basically, it prefers CPU of 55 C and max of 70 C and internal temp of 35 C and max of 40 C.
Those numbers are probably to conservative (which means excessive noise in this case). I notice Windows XP seems to let it run at 70 C alot. I'm just experimenting for now to see what it takes to stop the ACPI Error messages. If I'd be willing to live with some ACPI messages then I could have a quite (FAN wise) desktop.
With these settings, I can hear the fan speed up close to full speed for around 1 second under light web browsing loads but nothing to bad.