On Fri, Sep 26, 2014 at 03:00:11PM +0300, Mikko Perttunen wrote:

On 09/26/2014 02:48 PM, Thierry Reding wrote:

quoted

On Fri, Sep 26, 2014 at 01:22:52PM +0300, Mikko Perttunen wrote:

quoted

On 09/26/2014 01:19 PM, Thierry Reding wrote:

quoted

On Fri, Sep 26, 2014 at 12:43:09PM +0300, Mikko Perttunen wrote:

quoted

Hi,

this series adds support for the thermal monitoring features of the
soctherm unit on the Tegra124 SoC.

The branch is also available in my github repo,
  git://github.com/cyndis/linux.git soctherm-v6

Thanks,
Mikko

Mikko Perttunen (4):
  of: Add bindings for nvidia,tegra124-soctherm
  ARM: tegra: Add soctherm and thermal zones to Tegra124 device tree
  ARM: tegra: Add thermal trip points for Jetson TK1
  thermal: Add Tegra SOCTHERM thermal management driver

 .../devicetree/bindings/thermal/tegra-soctherm.txt |  53 +++
 arch/arm/boot/dts/tegra124-jetson-tk1.dts          |  44 ++
 arch/arm/boot/dts/tegra124.dtsi                    |  47 ++
 drivers/thermal/Kconfig                            |  10 +
 drivers/thermal/Makefile                           |   1 +
 drivers/thermal/tegra_soctherm.c                   | 471 +++++++++++++++++++++
 include/dt-bindings/thermal/tegra124-soctherm.h    |  13 +
 7 files changed, 639 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/thermal/tegra-soctherm.txt
 create mode 100644 drivers/thermal/tegra_soctherm.c
 create mode 100644 include/dt-bindings/thermal/tegra124-soctherm.h

One thing that I've wanted to start doing for a while now is request
patch submissions like this to come accompanied with a way on how to
test them. Ideally this would be in a scripted way that can test for
success programatically, but it doesn't necessarily have to be if it
turns out too difficult or impractical to do.

Indeed, that would be very useful.

quoted

The goal is to eventually come up with a test suite that can run the
majority of test cases automatically to make it easy to test for any
regressions. And even if tests can't be run automatically it'd still
be an advantage to have them all collected in some repository, since
it saves a lot of typing and time to run tests, and it will give us
a standard set of tests that everybody can verify changes against.

I realize that it's somewhat unfair to start requesting this from you
now, but we've got to start somewhere. Could you give a short summary
of how you test this? What are the interfaces that the kernel exposes
for these thermal drivers?

You need to enable the driver in Device Drivers -> Generic Thermal sysfs
driver -> Tegra SOCTHERM thermal management. Then, you should see
directories appear in /sys/class/thermal. You can also use the `tmon' tool
included in the kernel tree to quickly see values; that's what I use for
testing.

Okay. So what are expected values for these temperatures? It's going to
be pretty much impossible to say what the correct value is on a given
board at any time, but perhaps a "test" could consist of checking that
all temperatures are within a reasonable range.

On Jetson TK1, at least without the CL-DVFS series, I get around 32 Celsius.
If you want to account for cpu/gpufreq then I guess something like 25-70
would be a good range.

Okay, thanks. Can you remind me how this relates to the thermal tripping
support?

Thierry