cpufreq: tegra: don't error target() when suspended
authorStephen Warren <swarren@nvidia.com>
Wed, 20 Nov 2013 21:06:22 +0000 (14:06 -0700)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Wed, 20 Nov 2013 22:28:32 +0000 (23:28 +0100)
d4019f0a92ab "cpufreq: move freq change notifications to cpufreq core"
added code to the cpufreq core to print an error if a cpufreq driver's
.target() function returned an error. This exposed the fact that Tegra's
cpufreq driver returns an error when it is ignoring requests due to the
system being suspended.

Modify Tegra's .target() function not to return an error in this case;
this prevents the error prints. The argument is that since the suspend
hook can't and doesn't inform the cpufreq core when its requests will
be ignored, there's no way for the cpufreq core to squelch them, so it's
not an error for the requests to keep coming. This change make the Tegra
driver consistent with how the Exynos handles the same situation. Note
that s5pv210-cpufreq.c probably suffers from this same issue though.

Fixes: d4019f0a92ab (cpufreq: move freq change notifications to cpufreq core)
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/cpufreq/tegra-cpufreq.c

index f42df7ec03c53a00838839861567de2e0d8060fa..b7309c37033d57b64d37ca3b93354f0b2848102d 100644 (file)
@@ -142,10 +142,8 @@ static int tegra_target(struct cpufreq_policy *policy, unsigned int index)
 
        mutex_lock(&tegra_cpu_lock);
 
-       if (is_suspended) {
-               ret = -EBUSY;
+       if (is_suspended)
                goto out;
-       }
 
        freq = freq_table[index].frequency;