> I tested with your config and mainline kernel 4.0 (commit below) but I
> cannot reproduce the crash. Do you have more data, crash dump?
Unfortunately, I can't. For some weird trick of fate, I typed `rm -rfv
/usr/lib` on my UBUNTU 15.04 and should I say what happened? I had to
reinstall my whole environment, and since I was getting all kinds of
glitches (HDD, all of the sudden it gave up with S.M.A.R.T. errors
forcing me to buy another one, WIRELESS B43 driver was restarting my
system even before the Intel_power_clamp started screaming, the
SYSTEMD wasn't starting the OpenVSwitch in correct boot order, so I
couldn't start my firewall VM automatically), well, I decided to give
another distro a try, and guess what? No problems at all... With the
intel_power_clamp running on kernel 4.0, so far, so good, and for
whatever reason it was locking my system it appears to be only
affecting UBUNTU 15.04. Right now I am using Linux MINT 17.1 which is
a distro based on UBUNTU 14.04, perhaps in a future time I may have to
call you back, but until then I hope you have sorted what is breaking
If there is anything I can do to help you with this 'older' version
that I can try, please ask me, but no promises on finding the glitch
since I won't go into UBUNTU 15.04 so soon again, after all this
machine is my only network server. Anyway, thanks for your time and
help, and I hope you can find that pesky bug.
Looking inside the Ubuntu page about their mainline kernel build:
They clearly say that those builds I am using are the same as from linus
mainline. I guess I am facing an undocumented bug. Is there anything I can
do to track it inside the driver?
Em 21/04/2015 10:09, "Jacob Pan" <jacob.jun.pan(a)linux.intel.com> escreveu:
> On Mon, 20 Apr 2015 20:40:43 -0300
> Carlos Gustavo Ramirez Rodriguez <carlosgrr(a)gmail.com> wrote:
> > I didn't change any of the drivers, I am using a "vanilla" kernel, no
> > drivers changed... if that is what you are asking...
> That is part of it, but distro has their own patches sometimes. So i
> was asking if the distro kernel is any different than the mainline. Let
> me dig into that based on your links.
> Jacob Pan
2015-04-16 12:59 GMT-03:00 Pandruvada, Srinivas <srinivas.pandruvada(a)intel.com>:
> On Thu, 2015-04-16 at 01:18 -0300, Carlos Gustavo Ramirez Rodriguez
>> Hello Srinivas,
>> >> My screen, keyboard and mouse all of the sudden freezes and my dmesg
>> >> outputs the error messages below, can you help me or is there any
>> >> other mailing list I should ask for help, since it is a problem with
>> >> the Intel Power Clamp Driver?
>> > Which kernel version you are using? There was a problem but, it was
>> > fixed. It is in later version, we need to take a look.
>> I am using the kernel version 4.0.0. Do you recommend updating my
>> kernel to a newer version, to fix this problem once and for all?
> Copied to power clamp developer.
> Can you confirm that it is 4.0.0. He tried on 4.0.0 and can't reproduce.
Yes, just to confirm I am using 4.0.0 Kernel downloaded from
Linux version 4.0.0-040000-generic (kernel@gomeisa) (gcc version 4.6.3
(Ubuntu/Linaro 4.6.3-1ubuntu5) ) #201504121935 SMP Sun Apr 12 23:36:33
>> > You can remove powerclamp entry from thermal daemon usage meanwhile by
>> > removing
>> > thermal-cpu-cdev-order.xml (Based on your build it will
>> > be ../etc/thermald/..)
>> Thanks for the workaround, I will try it and in case something else
>> appears I will most surely ask for help.
> Let me know if this works, then we can know if power clamp caused this
> issue or not.
So far no issues, but it is kinda sporadic I don't know exactly how to
trigger this issue, but a common ground is that I was using my
computer, just leaving it idling never made it freeze. But usually it
never been more than a day to freeze, and so far it is working without
>> My screen, keyboard and mouse all of the sudden freezes and my dmesg
>> outputs the error messages below, can you help me or is there any
>> other mailing list I should ask for help, since it is a problem with
>> the Intel Power Clamp Driver?
> Which kernel version you are using? There was a problem but, it was
> fixed. It is in later version, we need to take a look.
I am using the kernel version 4.0.0. Do you recommend updating my
kernel to a newer version, to fix this problem once and for all?
> You can remove powerclamp entry from thermal daemon usage meanwhile by
> thermal-cpu-cdev-order.xml (Based on your build it will
> be ../etc/thermald/..)
Thanks for the workaround, I will try it and in case something else
appears I will most surely ask for help.
Meanwhile, is there anything else I can do, to help to fix this
problem, so that nobody else has to deal with it?
.TP gives a paragraph with a hanging indentation. That looks odd, so
let's use the non-indented .LP paragraph instead.
Also, switch to .IP for bulleting instead of hand-drawn ASCII bullets.
Signed-off-by: Brian Norris <computersforpeace(a)gmail.com>
man/thermal-conf.xml.5 | 34 +++++++++++++++++-----------------
man/thermald.8 | 12 +++++-------
2 files changed, 22 insertions(+), 24 deletions(-)
diff --git a/man/thermal-conf.xml.5 b/man/thermal-conf.xml.5
index b59d86c32e93..5205d1dfcb88 100644
@@ -31,28 +31,28 @@ $(TDCONFDIR)/etc/thermald/thermal-conf.xml
is a configuration file for the thermal daemon. It is used to configure thermal sensors, zone and cooling devices.The location of this file depends on the configuration option used during build time.
The terminology used in this file confirms to "Advanced Configuration and Power Interface Specification". The ACPI thermal model is based around conceptual platform regions called thermal zones that physically contain devices, thermal sensors, and cooling controls. For example of a thermal zone can be a CPU or a laptop cover. A zone can contain multiple sensors for monitoring temperature. A cooling device provides interface to reduce the temperature of a source device, which causes increase in the temperature. An example of a cooling device is a FAN or some Linux driver which can throttle the source device.
A thermal zone configuration includes one or more trip points. A trip point is a temperature at which a cooling device needs to be activated.
A cooling device can be either active or passive. An example of an active device is a FAN, which will not reduce performance at the cost of consuming more power and noise. A passive device uses performance throttling to control temperature. In addition to cooling devices present in the thermal sysfs, the following cooling devices are built into the thermald, which can be used as valid cooling device type:
- - rapl_controller
- - intel_pstate
- - cpufreq
+.IP \[bu] 2
The thermal sysfs under Linux (/sys/class/thermal) provides a way to represent per platform ACPI configuration. The kernel thermal governor uses this data to keep the platform thermals under control. But there are some limitations, which thermald tries to resolve. For example:
-- If the ACPI data is not optimized or buggy. In this case thermal-conf.xml can be used to correct the behavior without change in BIOS.
-- There may be thermal zones exposed by the thermal sysfs without associated cooling actions. In this case thermal conf.xml can be used to tie the cooling devices to those zones.
-- The best cooling method may not be in the thermal sysfs. In this case thermal-conf.xml can be used to bind a zone to an external cooling device.
-- Specify thermal relationships. A zone can be influenced by multiple source devices with varying degrees. In this case thermal-conf.xml can be used to define the relative influence for apply compensation.
+If the ACPI data is not optimized or buggy. In this case thermal-conf.xml can be used to correct the behavior without change in BIOS.
+There may be thermal zones exposed by the thermal sysfs without associated cooling actions. In this case thermal conf.xml can be used to tie the cooling devices to those zones.
+The best cooling method may not be in the thermal sysfs. In this case thermal-conf.xml can be used to bind a zone to an external cooling device.
+Specify thermal relationships. A zone can be influenced by multiple source devices with varying degrees. In this case thermal-conf.xml can be used to define the relative influence for apply compensation.
.SH FILE FORMAT
The configuration file format confirms to XML specifications. A set of tags defined to define platform, sensors, zones, cooling devices and trip points.
diff --git a/man/thermald.8 b/man/thermald.8
index 9b49334c7b92..93dc5c35c072 100644
@@ -33,26 +33,24 @@ thermald \- start Linux thermal daemon
is a Linux daemon used to prevent the overheating of platforms. This daemon monitors
temperature and applies compensation using available cooling methods.
By default, it monitors CPU temperature using available CPU digital temperature sensors and maintains CPU temperature under control, before HW takes aggressive correction action.
Thermal daemon looks for thermal sensors and thermal cooling drivers in the Linux thermal sysfs (/sys/class/thermal) and builds a
list of sensors and cooling drivers. Each of the thermal sensors can optionally be binded to a cooling drivers by the in kernel
drivers. In this case the Linux kernel thermal core can directly take actions based on the temperature trip points, for each sensor
and associated cooling device. For example a trip temperature X in a sensor can be associates a cooling driver Y. So when
the sensor temperature = X, the cooling driver "Y" is activated.
Thermal daemon allows to change this relationship or add new one via a thermal configuration file (thermal-conf.xml). This
file is automatically created and used, if the platform has ACPI thermal relationship table. If not this needs to be
When there is a sensor, which has no associate cooling device, via configuration file or thermal relationship table, then
this sensor is tested for relationship with CPU load dynamically upto maximum 3 times. If there is no relationship, then
it is added to a black list of unbinded sensors and not tried again.
Optionally thermal daemon can act as an exclusive thermal controller by using thermal sysfs and acting as a user space governor.
In this case kernel thermal core is not active and decision is taken by thermal daemon only.