Venkat Devarasetty 4edc766a41 msm: cpufreq: make the gov stop part fully atomic ...

In the function cpufreq_governor_dbs, the case CPUFREQ_GOV_STOP
is not completely atomic. The removal of sysfs is outside the
critical path. If a START/STOP done frequently from user space
then there is chance of adding the sysfs interface before removal.
This creates a warning from sysfs saying creating an already
existing entry.

Also once it fails to create, cpufreq_governor_dbs returns with
error code. The dbs_enable value is incremented which is not
rolled back and cpufreq core tries to set the old
governor. This might create further problems.

Hence the removal of sysfs entry should happen in the critical
path.

CRs-fixed: 440969
Change-Id: I98891756caf2ce0b85e39e9b4f58034af1ed97ce
Signed-off-by: Venkat Devarasetty <vdevaras@codeaurora.org>

2013-02-16 10:16:21 +05:30

..

acpi-cpufreq.c

…

cpufreq_conservative.c

…

cpufreq_gov_msm.c

msm: dcvs: gpu minimum frequency levels

2012-11-14 14:33:50 -08:00

cpufreq_interactive.c

cpufreq: interactive: Fix blocked task warnings

2012-11-01 17:14:04 -07:00

cpufreq_ondemand.c

msm: cpufreq: make the gov stop part fully atomic

2013-02-16 10:16:21 +05:30

cpufreq_performance.c

…

cpufreq_powersave.c

…

cpufreq_stats.c

…

cpufreq_userspace.c

…

cpufreq-nforce2.c

…

cpufreq.c

…

db8500-cpufreq.c

…

e_powersaver.c

…

elanfreq.c

…

exynos4x12-cpufreq.c

…

exynos4210-cpufreq.c

…

exynos5250-cpufreq.c

…

exynos-cpufreq.c

…

freq_table.c

…

gx-suspmod.c

…

Kconfig

…

Kconfig.arm

…

Kconfig.powerpc

…

Kconfig.x86

…

longhaul.c

…

longhaul.h

…

longrun.c

…

Makefile

…

maple-cpufreq.c

…

mperf.c

…

mperf.h

…

omap-cpufreq.c

…

p4-clockmod.c

…

pcc-cpufreq.c

…

powernow-k6.c

…

powernow-k7.c

…

powernow-k7.h

…

powernow-k8.c

…

powernow-k8.h

…

s3c64xx-cpufreq.c

…

s3c2416-cpufreq.c

…

s5pv210-cpufreq.c

…

sc520_freq.c

…

speedstep-centrino.c

…

speedstep-ich.c

…

speedstep-lib.c

…

speedstep-lib.h

…

speedstep-smi.c

…