[CPUFREQ] conservative: fix dbs_cpufreq_notifier so freq is not locked

When someone added the dbs_cpufreq_notifier section to the governor the
code ended up causing the frequency to only fall.  This is because
requested_freq is tinkered with and that should only modified if it has
an invlaid value due to changes in the available frequency ranges

This should fix #10055.

Signed-off-by: Alexander Clouter <alex@digriz.org.uk>
Signed-off-by: Dave Jones <davej@redhat.com>

diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c
index a18cfbf021b330e95ad818b6cf6d688829b8d676..a16a5b8c1dc56da7b7277fbd2c342654957b2342 100644 (file)
--- a/drivers/cpufreq/cpufreq_conservative.c
+++ b/drivers/cpufreq/cpufreq_conservative.c
@@ -137,10 +137,21 @@ dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
        struct cpu_dbs_info_s *this_dbs_info = &per_cpu(cpu_dbs_info,
                                                        freq->cpu);
 
+       struct cpufreq_policy *policy;
+
        if (!this_dbs_info->enable)
                return 0;
 
-       this_dbs_info->requested_freq = freq->new;
+       policy = this_dbs_info->cur_policy;
+
+       /*
+        * we only care if our internally tracked freq moves outside
+        * the 'valid' ranges of freqency available to us otherwise
+        * we do not change it
+       */
+       if (this_dbs_info->requested_freq > policy->max
+                       || this_dbs_info->requested_freq < policy->min)
+               this_dbs_info->requested_freq = freq->new;
 
        return 0;
 }