cpufreq: Reorder cpufreq_online() error code path
authorViresh Kumar <viresh.kumar@linaro.org>
Thu, 22 Feb 2018 05:59:43 +0000 (11:29 +0530)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 30 May 2018 05:50:49 +0000 (07:50 +0200)
[ Upstream commit b24b6478e65f140610ab1ffaadc7bc6bf0be8aad ]

Ideally the de-allocation of resources should happen in the exact
opposite order in which they were allocated. It helps maintain the code
in long term, even if nothing really breaks with incorrect ordering.

That wasn't followed in cpufreq_online() and it has some
inconsistencies.  For example, the symlinks were created from within
the locked region while they are removed only after putting the locks.
Also ->exit() should have been called only after the symlinks are
removed and the lock is dropped, as that was the case when ->init()
was first called.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
[ rjw: Subject ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/cpufreq/cpufreq.c

index 35e34c0e0429a5e693e2c77d99afa182175cace9..7523929becdca614f6a383b78ac4ab7939bbb230 100644 (file)
@@ -1288,14 +1288,14 @@ static int cpufreq_online(unsigned int cpu)
        return 0;
 
 out_exit_policy:
+       for_each_cpu(j, policy->real_cpus)
+               remove_cpu_dev_symlink(policy, get_cpu_device(j));
+
        up_write(&policy->rwsem);
 
        if (cpufreq_driver->exit)
                cpufreq_driver->exit(policy);
 
-       for_each_cpu(j, policy->real_cpus)
-               remove_cpu_dev_symlink(policy, get_cpu_device(j));
-
 out_free_policy:
        cpufreq_policy_free(policy, !new_policy);
        return ret;