Re: [PATCH] cpufreq, store_scaling_governor requires policy->rwsem to be held for duration of changing governors [v2]

[Viresh Kumar <viresh.kumar@linaro.org>]

On 5 August 2014 01:46, Saravana Kannan <skannan@codeaurora.org> wrote:
> The problem is when between one thread trying to cat a governor's file (say,
> sampling_rate) vs the governor getting a POLICY_EXIT.

I don't see two threads racing against each other here. Simply changing
the governor from conservative->ondemand creates this.

Or is it that the kernel is detecting two different orders of taking lock?

But during governor change, isn't the sysfs lock taken first as we are
storing a value to "scaling_governor"? So, isn't this a sysfs lock first
in all cases?

In short, I am still unclear about the *exact* problem here.

> Could you please look at my policy free/remove patches? If you can do that,
> I can work on a fix for this. It might also be simpler to fix after my patch
> series (not sure, might be).

I had an overall look of those on the day you posted them, but haven't
commented yet as was going away..

There is no way those can land in 3.17-rc1 atleast and so we still have
some time to get them pushed..

Anyway, they are my number two priority and the number one is this bug,
which we have to fix in stable kernels as well. So, a dependency on your
series wouldn't work..

Let me see if I can get some solution to this problem first.