diff mbox

[3/3] doc: cpufreq: tests: Provide cpufreq tests README

Message ID 1422268884-26173-3-git-send-email-l.majewski@samsung.com (mailing list archive)
State Not Applicable, archived
Headers show

Commit Message

Lukasz Majewski Jan. 26, 2015, 10:41 a.m. UTC
This commit adds README to ./tools/testing/cpufreq directory.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
---
 tools/testing/cpufreq/README | 53 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100644 tools/testing/cpufreq/README

Comments

Viresh Kumar Jan. 27, 2015, 3:56 a.m. UTC | #1
On 26 January 2015 at 16:11, Lukasz Majewski <l.majewski@samsung.com> wrote:
> This commit adds README to ./tools/testing/cpufreq directory.

You might want to have a look at this..

https://git.linaro.org/people/viresh.kumar/cpufreq-tests.git


I had been working on this recently ..
--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Lukasz Majewski Jan. 27, 2015, 8:35 a.m. UTC | #2
Hi Viresh,

> On 26 January 2015 at 16:11, Lukasz Majewski <l.majewski@samsung.com>
> wrote:
> > This commit adds README to ./tools/testing/cpufreq directory.
> 
> You might want to have a look at this..
> 
> https://git.linaro.org/people/viresh.kumar/cpufreq-tests.git
> 
> 
> I had been working on this recently ..

Very nice pack of tests.

What I can say after a glimpse look on them is that your tests already
implement case for "governors switching".

However the 'cpufreq_freq_test.sh' do a bit more - it uses performance,
then forces 100% CPU usage and then switch from one single frequency to
all available (including BOOST modes). For that reason I think that it
would be valuable to add this test to your suite (of course on your
discretion).

Viresh, do you plan to add your tests to
kernel's ./tools/testing/cpufreq/ directory? It would enhance
recognition of your work and encourage others to use it before
submitting patches.
Viresh Kumar Jan. 27, 2015, 8:40 a.m. UTC | #3
On 27 January 2015 at 14:05, Lukasz Majewski <l.majewski@samsung.com> wrote:
> Very nice pack of tests.

Thanks.

> What I can say after a glimpse look on them is that your tests already
> implement case for "governors switching".

Correct and much more than that :)

> However the 'cpufreq_freq_test.sh' do a bit more - it uses performance,
> then forces 100% CPU usage and then switch from one single frequency to
> all available (including BOOST modes). For that reason I think that it
> would be valuable to add this test to your suite (of course on your
> discretion).

Will you be able to send a patch for that to me (*without* the lists)?

> Viresh, do you plan to add your tests to
> kernel's ./tools/testing/cpufreq/ directory? It would enhance
> recognition of your work and encourage others to use it before
> submitting patches.

Yeah, I do plan to. I have been doing some major reworks of cpufreq
core and so I built them. I haven't found time yet to get them in, but
they will surely be there in some time :)

--
viresh
--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Lukasz Majewski Jan. 27, 2015, 8:57 a.m. UTC | #4
Hi Viresh,

> On 27 January 2015 at 14:05, Lukasz Majewski <l.majewski@samsung.com>
> wrote:
> > Very nice pack of tests.
> 
> Thanks.
> 
> > What I can say after a glimpse look on them is that your tests
> > already implement case for "governors switching".
> 
> Correct and much more than that :)
> 
> > However the 'cpufreq_freq_test.sh' do a bit more - it uses
> > performance, then forces 100% CPU usage and then switch from one
> > single frequency to all available (including BOOST modes). For that
> > reason I think that it would be valuable to add this test to your
> > suite (of course on your discretion).
> 
> Will you be able to send a patch for that to me (*without* the lists)?

I rather thought about running this test as your run yours with runme.sh

> 
> > Viresh, do you plan to add your tests to
> > kernel's ./tools/testing/cpufreq/ directory? It would enhance
> > recognition of your work and encourage others to use it before
> > submitting patches.
> 
> Yeah, I do plan to. I have been doing some major reworks of cpufreq
> core and so I built them. I haven't found time yet to get them in, but
> they will surely be there in some time :)

I had similar problem with UDC/USB development in u-boot. I had a set
of "private/not ready to publish/etc" tests against regressions.

Only when I've posted those tests to the mailing list and they were
included to development tree, people started to use them before
submitting patches. I have less maintanence work and also developers
helped me a lot with improving the tests itself.

I'm a good example - I wasn't aware about your tests, and hence didn't
use them.
I believe that those tests shall land under ./tools/testing/cpufreq/
directory ASAP.

> 
> --
> viresh
diff mbox

Patch

diff --git a/tools/testing/cpufreq/README b/tools/testing/cpufreq/README
new file mode 100644
index 0000000..c934016
--- /dev/null
+++ b/tools/testing/cpufreq/README
@@ -0,0 +1,53 @@ 
+This file contains list of cpufreq's available regression tests with a short
+usage description.
+
+1. cpufreq_freq_test.sh
+
+Description:
+------------
+This script is supposed to test if cpufreq attributes exported by sysfs are
+exposing correct values.
+
+To achieve this goal it saves the current governor and changes it to
+"performance". Afterwards, it reads the "scaling_available_frequencies"
+property. With the list of supported frequencies it is able to enforce each of
+them by writing to "scaling_max_freq" attribute. To make the test more reliable
+a superfluous load with gzip is created to be sure that we are running with
+highest possible frequency. This high load is regulated with the 'sleep'
+duration. After this time the "cpufreq_cur_freq" is read and compared with the
+original value. As the last step the original governor is restored.
+
+This script can work with or without BOOST enabled and helps in spotting errors
+related to cpufreq and common clock framework.
+
+Used attributes:
+----------------
+- "scaling_available_frequencies"
+- "cpuinfo_cur_freq"
+- "scaling_governor"
+- "scaling_max_freq"
+
+Target devices:
+---------------
+
+All devices which exports mentioned above sysfs attributes.
+
+2. governors_switch.sh
+
+Description:
+------------
+This script is supposed to test if all governors are available and possible to
+set.
+
+Moreover, it allows to test if cpufreq is able to withstand multiple accesses to
+the same sysfs attribute without crash.
+
+Used attributes:
+----------------
+- "scaling_available_governors"
+- "scaling_governor"
+
+Target devices:
+---------------
+
+All devices which exports mentioned above sysfs attributes.