cpufreq: suspend governors on system suspend/hibernate
authorViresh Kumar <viresh.kumar@linaro.org>
Tue, 4 Mar 2014 03:00:26 +0000 (11:00 +0800)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Thu, 6 Mar 2014 14:04:12 +0000 (15:04 +0100)
commit2f0aea9363609433b38ab13f0f86f38372668c68
tree459115b73ec6e5ece6ee73a711fa95dac8a98da0
parent6e2c89d16d987e6e11c531b039a42d3f5f1d7c32
cpufreq: suspend governors on system suspend/hibernate

This patch adds cpufreq suspend/resume calls to dpm_{suspend|resume}()
for handling suspend/resume of cpufreq governors.

Lan Tianyu (Intel) & Jinhyuk Choi (Broadcom) found an issue where the
tunables configuration for clusters/sockets with non-boot CPUs was
lost after system suspend/resume, as we were notifying governors with
CPUFREQ_GOV_POLICY_EXIT on removal of the last CPU for that policy
which caused the tunables memory to be freed.

This is fixed by preventing any governor operations from being
carried out between the device suspend and device resume stages of
system suspend and resume, respectively.

We could have added these callbacks at dpm_{suspend|resume}_noirq()
level, but there is an additional problem that the majority of I/O
devices is already suspended at that point and if cpufreq drivers
want to change the frequency before suspending, then that not be
possible on some platforms (which depend on peripherals like i2c,
regulators, etc).

Reported-and-tested-by: Lan Tianyu <tianyu.lan@intel.com>
Reported-by: Jinhyuk Choi <jinchoi@broadcom.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
[rjw: Changelog]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/base/power/main.c
drivers/cpufreq/cpufreq.c
include/linux/cpufreq.h