cpufreq: Skip current frequency initialization for ->setpolicy drivers
authorRafael J. Wysocki <rafael.j.wysocki@intel.com>
Wed, 12 Mar 2014 20:49:33 +0000 (21:49 +0100)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Wed, 12 Mar 2014 23:37:16 +0000 (00:37 +0100)
After commit da60ce9f2fac (cpufreq: call cpufreq_driver->get() after
calling ->init()) __cpufreq_add_dev() sometimes fails for CPUs handled
by intel_pstate, because that driver may return 0 from its ->get()
callback if it has not run long enough to collect enough samples on the
given CPU.  That didn't happen before commit da60ce9f2fac which added
policy->cur initialization to __cpufreq_add_dev() to help reduce code
duplication in other cpufreq drivers.

However, the code added by commit da60ce9f2fac need not be executed
for cpufreq drivers having the ->setpolicy callback defined, because
the subsequent invocation of cpufreq_set_policy() will use that
callback to initialize the policy anyway and it doesn't need
policy->cur to be initialized upfront.  The analogous code in
cpufreq_update_policy() is also unnecessary for cpufreq drivers
having ->setpolicy set and may be skipped for them as well.

Since intel_pstate provides ->setpolicy, skipping the upfront
policy->cur initialization for cpufreq drivers with that callback
set will cover intel_pstate and the problem it's been having after
commit da60ce9f2fac will be addressed.

Fixes: da60ce9f2fac (cpufreq: call cpufreq_driver->get() after calling ->init())
References: https://bugzilla.kernel.org/show_bug.cgi?id=71931
Reported-and-tested-by: Patrik Lundquist <patrik.lundquist@gmail.com>
Acked-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Cc: 3.13+ <stable@vger.kernel.org> # 3.13+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/cpufreq/cpufreq.c

index cf485d9289035fe58ed9c84223ccd9f412a4b6cd..199b52b7c3e1ad6e9d00102905a215137901b71d 100644 (file)
@@ -1129,7 +1129,7 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
                per_cpu(cpufreq_cpu_data, j) = policy;
        write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-       if (cpufreq_driver->get) {
+       if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
                policy->cur = cpufreq_driver->get(policy->cpu);
                if (!policy->cur) {
                        pr_err("%s: ->get() failed\n", __func__);
@@ -2143,7 +2143,7 @@ int cpufreq_update_policy(unsigned int cpu)
         * BIOS might change freq behind our back
         * -> ask driver for current freq and notify governors about a change
         */
-       if (cpufreq_driver->get) {
+       if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
                new_policy.cur = cpufreq_driver->get(cpu);
                if (!policy->cur) {
                        pr_debug("Driver did not initialize current freq");