Revert "cpufreq: postfix policy directory with the first CPU in related_cpus"
authorViresh Kumar <viresh.kumar@linaro.org>
Thu, 3 Mar 2016 09:21:33 +0000 (14:51 +0530)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Wed, 9 Mar 2016 20:42:45 +0000 (21:42 +0100)
Revert commit 3510fac45492 (cpufreq: postfix policy directory with the
first CPU in related_cpus).

Earlier, the policy->kobj was added to the kobject core, before ->init()
callback was called for the cpufreq drivers. Which allowed those drivers
to add or remove, driver dependent, sysfs files/directories to the same
kobj from their ->init() and ->exit() callbacks.

That isn't possible anymore after commit 3510fac45492.

Now, there is no other clean alternative that people can adopt.

Its better to revert the earlier commit to allow cpufreq drivers to
create/remove sysfs files from ->init() and ->exit() callbacks.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/cpufreq/cpufreq.c

index bdf258ea09775e6511ada8ea7af8d93ae4368771..abca44c2e4e3ed899a42b8a5717c6f1fc590af5d 100644 (file)
@@ -985,6 +985,7 @@ static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
 {
        struct device *dev = get_cpu_device(cpu);
        struct cpufreq_policy *policy;
+       int ret;
 
        if (WARN_ON(!dev))
                return NULL;
@@ -1002,7 +1003,13 @@ static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
        if (!zalloc_cpumask_var(&policy->real_cpus, GFP_KERNEL))
                goto err_free_rcpumask;
 
-       kobject_init(&policy->kobj, &ktype_cpufreq);
+       ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
+                                  cpufreq_global_kobject, "policy%u", cpu);
+       if (ret) {
+               pr_err("%s: failed to init policy->kobj: %d\n", __func__, ret);
+               goto err_free_real_cpus;
+       }
+
        INIT_LIST_HEAD(&policy->policy_list);
        init_rwsem(&policy->rwsem);
        spin_lock_init(&policy->transition_lock);
@@ -1013,6 +1020,8 @@ static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
        policy->cpu = cpu;
        return policy;
 
+err_free_real_cpus:
+       free_cpumask_var(policy->real_cpus);
 err_free_rcpumask:
        free_cpumask_var(policy->related_cpus);
 err_free_cpumask:
@@ -1117,16 +1126,6 @@ static int cpufreq_online(unsigned int cpu)
                cpumask_copy(policy->related_cpus, policy->cpus);
                /* Remember CPUs present at the policy creation time. */
                cpumask_and(policy->real_cpus, policy->cpus, cpu_present_mask);
-
-               /* Name and add the kobject */
-               ret = kobject_add(&policy->kobj, cpufreq_global_kobject,
-                                 "policy%u",
-                                 cpumask_first(policy->related_cpus));
-               if (ret) {
-                       pr_err("%s: failed to add policy->kobj: %d\n", __func__,
-                              ret);
-                       goto out_exit_policy;
-               }
        }
 
        /*