cpufreq: governor: remove copy_prev_load from 'struct cpu_dbs_common_info'
authorViresh Kumar <viresh.kumar@linaro.org>
Mon, 9 Jun 2014 08:51:24 +0000 (14:21 +0530)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Mon, 9 Jun 2014 10:58:21 +0000 (12:58 +0200)
commitc8ae481b9a12f5cea080651ea87736104b111f8e
tree72fd7b221acfa258ae47df42a930d6fe21c38878
parent18b46abd0009516c1973a57ccf4d01b9eaa3422a
cpufreq: governor: remove copy_prev_load from 'struct cpu_dbs_common_info'

'copy_prev_load' was recently added by commit: 18b46ab (cpufreq: governor: Be
friendly towards latency-sensitive bursty workloads).

It actually is a bit redundant as we also have 'prev_load' which can store any
integer value and can be used instead of 'copy_prev_load' by setting it zero.

True load can also turn out to be zero during long idle intervals (and hence the
actual value of 'prev_load' and the overloaded value can clash). However this is
not a problem because, if the true load was really zero in the previous
interval, it makes sense to evaluate the load afresh for the current interval
rather than copying the previous load.

So, drop 'copy_prev_load' and use 'prev_load' instead.

Update comments as well to make it more clear.

There is another change here which was probably missed by Srivatsa during the
last version of updates he made. The unlikely in the 'if' statement was covering
only half of the condition and the whole line should actually come under it.

Also checkpatch is made more silent as it was reporting this (--strict option):

CHECK: Alignment should match open parenthesis
+ if (unlikely(wall_time > (2 * sampling_rate) &&
+ j_cdbs->prev_load)) {

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/cpufreq/cpufreq_governor.c
drivers/cpufreq/cpufreq_governor.h
This page took 0.03081 seconds and 5 git commands to generate.