Bug 14765 - Acer Aspire 5315 cooling problems
Summary: Acer Aspire 5315 cooling problems
Status: REJECTED WILL_NOT_FIX
Alias: None
Product: ACPI
Classification: Unclassified
Component: Power-Fan (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Zhang Rui
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-12-08 13:52 UTC by Eabe Kuik
Modified: 2010-01-22 22:07 UTC (History)
2 users (show)

See Also:
Kernel Version: 2.6.31.6 / 2.6.32
Subsystem:
Regression: No
Bisected commit-id:


Attachments
The dumps from the kernels (dmesg / acpidump / config) (190.39 KB, application/gzip)
2009-12-09 00:08 UTC, Eabe Kuik
Details
Test with acpi_enforce_resources=lax (65.71 KB, application/gzip)
2009-12-09 10:53 UTC, Eabe Kuik
Details
Small script that copies coretemp into acpi fan temp (10.31 KB, application/gzip)
2010-01-17 09:31 UTC, Eabe Kuik
Details

Description Eabe Kuik 2009-12-08 13:52:00 UTC
Hi,

I'm using the kernel 2.6.29.6-huge-smp that comes with Slackware 13.0, the installed distro. I noticed that newer kernels (2.6.31.6 & 2.6.32) don't handle the cooling fan correct on my Acer Aspire 5315 (Bios V1.29). The 2.6.31.6 kernel does not turn on the fan so that the laptop shuts itself down to prevent overheating and damaging the hardware. With the 2.6.32 kernel the fan is turn on sometimes, but not controled! E.g. it runs on one speed, not like the 2.6.29.6 kernel that matches the fan speed to the temperature.

Another thing noted is that when the fan is not turned on the temperature reports 40 C (always).

2.6.29.6
config   -> http://pastebin.com/m7e5c66b
dmesg    -> http://pastebin.com/m11abdbdf
acpidump -> http://pastebin.com/m62505a9d

2.6.31.6
config   -> http://pastebin.com/m4ec3e8a1
dmesg    -> http://pastebin.com/m22e68bc6
acpidump -> http://pastebin.com/m6d09356e

2.6.32
config   -> http://pastebin.com/m58669dc
dmesg    -> http://pastebin.com/m788be110
acpidump -> http://pastebin.com/m2d510e9b
Comment 1 Eabe Kuik 2009-12-09 00:08:06 UTC
Created attachment 24103 [details]
The dumps from the kernels (dmesg / acpidump / config)

The dumps from the kernels (dmesg / acpidump / config)
Comment 2 Zhang Rui 2009-12-09 01:30:00 UTC
how do you control the fan in 2.6.29 kernel?
Comment 3 Eabe Kuik 2009-12-09 09:15:32 UTC
(In reply to comment #2)
> how do you control the fan in 2.6.29 kernel?

It's done by the kernel. I don't use a special application to control the fan. From the first distro (slack 12.1) to the current (slack 13) installed on this laptop it always worked out of the box. Even with new build kernels < 2.6.29.6 no special actions where needed to make the fan working.

For performance actions related to the battery I use Klaptop.
Comment 4 Zhang Rui 2009-12-09 09:25:53 UTC
please try boot option "acpi_enforce_resources=lax" and see if it helps.
Comment 5 Eabe Kuik 2009-12-09 10:52:02 UTC
(In reply to comment #4)
> please try boot option "acpi_enforce_resources=lax" and see if it helps.

Does not make any difference.
Comment 6 Eabe Kuik 2009-12-09 10:53:59 UTC
Created attachment 24117 [details]
Test with acpi_enforce_resources=lax

Also included the /var/log/messages from this test.
Comment 7 Eabe Kuik 2009-12-09 14:33:17 UTC
Maybe this may lead to something.

---- SNIP ----
Dump from sensors with Kernel 2.6.31.6 and module coretemp loaded

acpitz-virtual-0
Adapter: Virtual device
temp1:       +40.0°C  (crit = +110.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:      +67.0°C  (high = +100.0°C, crit = +100.0°C)



Dump from sensors with Kernel 2.6.29.6 and module coretemp loaded

acpitz-virtual-0
Adapter: Virtual device
temp1:       +55.0°C  (crit = +110.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:      +54.0°C  (high = +100.0°C, crit = +100.0°C)
---- SNIP ----
Comment 8 Eabe Kuik 2009-12-10 08:57:32 UTC
A closer look at the logs (messages) shows the following:

2.9.29.6:
Dec 10 09:18:07 eabeacer kernel: ACPI: EC: GPE storm detected, transactions will use polling mode
Dec 10 09:18:07 eabeacer kernel: ACPI: Interpreter enabled
Dec 10 09:18:07 eabeacer kernel: ACPI: (supports S0 S3 S4 S5)
Dec 10 09:18:07 eabeacer kernel: ACPI: Using IOAPIC for interrupt routing
Dec 10 09:18:07 eabeacer kernel: ACPI: EC: GPE = 0x1c, I/O: command/status = 0x66, data = 0x62
Dec 10 09:18:07 eabeacer kernel: ACPI: EC: driver started in interrupt mode

2.6.31.6:
Dec 10 09:14:17 eabeacer kernel: ACPI: EC: GPE storm detected, transactions will use polling mode
Dec 10 09:14:17 eabeacer kernel: ACPI: Interpreter enabled
Dec 10 09:14:17 eabeacer kernel: ACPI: (supports S0 S3 S4 S5)
Dec 10 09:14:17 eabeacer kernel: ACPI: Using IOAPIC for interrupt routing
Dec 10 09:14:17 eabeacer kernel: ACPI: EC: GPE = 0x1c, I/O: command/status = 0x66, data = 0x62
Dec 10 09:14:17 eabeacer kernel: ACPI: EC: driver started in poll mode

Is it possible to start the driver in "interrupt mode" in 2.6.31.6? Just to see if this causes the problem?
Comment 9 Zhang Rui 2009-12-22 03:06:48 UTC
Alexey, can you help on this?
Comment 10 Eabe Kuik 2010-01-11 22:52:34 UTC
It's seems to be a problem that can be solved with the fan control script from http://bugzilla.kernel.org/show_bug.cgi?id=9167

With this script I'm now running 2.6.31.7 without heating problems.

More testing learned that with the 2.6.29.6 the heating problem appears after a suspent to sleep just like all other Acer's (http://ubuntuforums.org/archive/index.php/t-604158.html)
Comment 11 Zhang Rui 2010-01-13 08:10:53 UTC
which comment is the script in?
Comment 13 Eabe Kuik 2010-01-17 09:31:58 UTC
Created attachment 24601 [details]
Small script that copies coretemp into acpi fan temp

Small script based on the script from David Edwards that copies the temperature from the coretemp module in the fan temperature register.
Comment 14 Alexey Starikovskiy 2010-01-22 16:08:00 UTC
could you please check if patches from bug 14747 help?
Comment 15 Alexey Starikovskiy 2010-01-22 16:17:43 UTC
Ok, this one is not related to EC driver. Temperature is read from DTS1/DTS2 fields of NVS system memory region, and the script you provide just writes it there.
Comment 16 Alexey Starikovskiy 2010-01-22 16:24:26 UTC
Quote from the above ubuntu forum. The bug is in hardware, not in Linux or Windows for that matter... You should either use your script, don't use suspend, or return this notebook back to Acer for repair/refund...

"Addenda: A forum member has contacted Acer about the overheat and shut down issues and Acer has confirmed that when the ePower software is not active, the fan will not function properly (overheating after sleep or hibernation mode). This applies to ALL of Acer's notebooks sold with Windows Vista (ePower is only available for Windows Vista). Acer has NO INTENTION to fix this problem because they do not consider this a fault. ePower has also a massive impact on the speed of the computer especially on low spec Celeron machines. What this means is that if you want to run your Acer notebook faster by disabling ePower and other bloat ware or if you want to install Ubuntu or Windows XP, you need to live with those shortcomings. There maybe other issues but I have tested my 5315 extensively with Windows XP and have not come across any so far. Drivers for Windows XP are all available, albeit you need to search for them from the manufactures web sites (when I have time over the holidays I will upload all drivers and publish the link for people to download). Conclusion: If you want to run Ubuntu or Windows XP on your notebook or want to run your notebook faster, DON’T BUY ACER! However, I will probably not return my 5315 because for the same price I cannot find one with all the specs important to me (1x1GB DDR, S-Video, 2.5hrs battery life). To avoid overheating I set the power setting to “shut down” when lid is closed and NEVER use sleep mode or hibernation mode. I also urge everybody to spread the message about the shortcomings of Acer notebooks on forums and especially on user feedback sites such as ciao, yahoo, etc., because I believe this may help Acer to change their mind and come up with a fix."
Comment 17 Eabe Kuik 2010-01-22 22:07:54 UTC
Agreed, not a kernel bug, but a bug in the Acer hardware.

Note You need to log in before you can comment on or make changes to this bug.