Most recent kernel where this bug did not occur: Distribution:Ubuntu gutsy ( same bugg on ubuntu feisty and pardus ) Hardware Environment: Desktop HP pavillon a629.fr / AMD Athlon(tm) XP2600+ / Bios 3.08 Software Environment: Problem Description: Trip points are wrong : cat /proc/acpi/thermal_zone/*/* 0 - Active; 1 - Passive <polling disabled> state: passive temperature: 39 C critical (S5): 100 C passive: -248 C: tc1=4 tc2=3 tsp=60 devices=CPU0 active[0]: -266 C: devices= FAN and an error is printed in the log files every 6 seconds : ACPI: Unable to turn cooling device [debcbf18] 'on' With previous kernell I "fix" the issue by overwriting the trip points : echo -n "100:0:90:60:0" > /proc/acpi/thermal_zone/THRM/trip_points but it is no more possible with this kernell. I can run kernell with acpi=off, but then the power shut down is not working. The message disappear if I kill thermal: sudo rmmod thermal. I will provide results of acpidump and dmidecode ( running with acpi=off ) ... when I'll found how to do it. Steps to reproduce:
Created attachment 13280 [details] result of acpidump, running kernel with acpi=off Following my bugg report ...
Created attachment 13281 [details] result of dmidecode , running kernell with acpi=off following my bugg report
Please try to boot with "thermal.act=xxx" and "thermal.psv=yyy" to override the active and passive trip point.
I tried to boot with Kernel parameters : ro quiet splash thermal.act=60 thermal.psv=50 and get : Unknown boot option `thermal.act=60': ignoring Unknown boot option `thermal.psv=50': ignoring
That feature is merged recently. Please try a newer kernel, eg. 2.6.23.
resolve this bug due to no response from bug reporter...
(In reply to comment #6) > resolve this bug due to no response from bug reporter... > Hi. Sorry to not have been able to test with kernel 2.6.23, I'm running > ubuntu gutsy which is still using kernell 2.2.22 .
please re-open if this is a problem in 2.6.23 or later. also, i'm curious if you've run windows on this box and if/why it doesn't run into the same bogus trip points.
I'm now using Ubuntu 8.04, Kernel 2.6.24-18-generic I tried to boot with Kernel parameters : ro quiet splash thermal.act=60 thermal.psv=50 and get : Unknown boot option `thermal.act=60': ignoring Unknown boot option `thermal.psv=50': ignoring For the timebeing,I added "modprobe -r thermal" in file /etc/rc.local to kill thermal after boot and avoid to get the message "unable to turn on cooling device" every 6 sec. This Desktop never heat-up so it is OK. I cannot say if it worked better under windows, I never check the trip points under XP, may be there was simply no active thermal management ? Looks like new kernell still not works properly with the mother board of this HP computer ( ASUS A7V8X-LA ).
(In reply to comment #9) > I'm now using Ubuntu 8.04, Kernel 2.6.24-18-generic > > I tried to boot with Kernel parameters : > ro quiet splash thermal.act=60 thermal.psv=50 > and get : > Unknown boot option `thermal.act=60': ignoring > Unknown boot option `thermal.psv=50': ignoring > weird, we didn't change this piece of code, could you make a double check? Look at this AML code. the active[0] and passive trip point is just gotten by reading a specific system memory address. Method (_AC0, 0, NotSerialized) { If (Or (PLCY, PLCY, Local7)) { Return (TP2H) } Else { Return (TP1H) } } and OperationRegion (TEMM, SystemMemory, 0x000FF810, 0x10) Field (TEMM, WordAcc, NoLock, Preserve) { TP1H, 16, TP1L, 16, TP2H, 16, TP2L, 16, TRPC, 16, SENF, 16, TP3H, 16, TP3L, 16 } so could you please do this test: 1.boot with ACPI enabled, 2.BEFORE loading the thermal driver, do "acpidump --addr 0x000FF810 --length 0x10 > temm-1" 3.load thermal driver 4.do "acpidump > --addr 0x000FF810 --length 0x10 > temm-2" and attach the test result here.
Created attachment 16547 [details] (before thermal) acpidump --addr 0x000FF810 --length 0x10 > temm-1
Created attachment 16548 [details] (after thermal) acpidump > --addr 0x000FF810 --length 0x10 > temm-2
oops. sorry. it should be 4. do "acpidump --addr 0x000ff810 --length 0x10 > temm-2" please re-attach the test result after thermal driver is loaded,
No poblem. see below.
Created attachment 16569 [details] ( thermal loaded ) acpidump --addr 0x000ff810 --length 0x10 > temm-2
Well, AML read the invalid data from OperationRegion (TEMM, SystemMemory, 0x000FF810, 0x10) And this is not a Linux/ACPI bug to me. Please check the BIOS to see if there are any thermal related options. Please try to upgrade the BIOS to see if it's fixed. Anyway, this is a BIOS problem and we will/can not fix it in Linux kernel. Reject this bug as it's INVALID. Please re-open it if you still have some questions.
Up to now, I "solved" the problem by removing the thermal module : modprobe -r thermal in /etc/rc.local. Just updated my system to ubuntu 9.04 (Kernel 2.6.28-11) and the fix doesn't work anymore : No more "thermal" module and message "ACPI: Unable to turn cooling device" added any 6 seconds in the logs..... I understand that the origin of the problem comes from my old bios and that it is not a Linux/ACPI bug, but I cannot change the bios and would appreciate any help to stop the warning message.
please try boot option acpi.power_nocheck=1
I tried successively the following boot options without any success : 1) acpi.power_nocheck=1 2) thermal.nocrt=1 3) thermal.crt=-1 In any case the warning is still printed in the logs every 6 seconds. Some more informations : Computer = Desktop HP Pavilion a629.fr Motherboard = ASUS A7V8X-LA. From kernell.log : [ 1.351468] fan PNP0C0B:00: registered as cooling_device0 [ 1.351475] ACPI: Fan [FAN] (on) [ 1.351759] processor ACPI_CPU:00: registered as cooling_device1 [ 1.356556] thermal LNXTHERM:01: registered as thermal_zone0 [ 1.358249] ACPI: Unable to turn cooling device [df42cf18] 'on' [ 1.358255] ACPI: Thermal Zone [THRM] (56 C)
I finally be succesfull by adding the boot parameter thermal.act=80 to overwrite the wrong trip point value. No more warning in the log.