From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751719Ab3GaX6H (ORCPT ); Wed, 31 Jul 2013 19:58:07 -0400 Received: from hydra.sisk.pl ([212.160.235.94]:49614 "EHLO hydra.sisk.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751305Ab3GaX6F (ORCPT ); Wed, 31 Jul 2013 19:58:05 -0400 From: "Rafael J. Wysocki" To: Viresh Kumar , "Srivatsa S. Bhat" Cc: Linux PM list , LKML , cpufreq@vger.kernel.org Subject: [RFC][PATCH] cpufreq: Do not hold driver module references for additional policy CPUs Date: Thu, 01 Aug 2013 02:08:15 +0200 Message-ID: <2362640.pUofnXyzOi@vostro.rjw.lan> User-Agent: KMail/4.9.5 (Linux/3.10.0+; KDE/4.9.5; x86_64; ; ) MIME-Version: 1.0 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="utf-8" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: Rafael J. Wysocki The cpufreq core is a little inconsistent in the way it uses the driver module refcount. Namely, if __cpufreq_add_dev() is called for a CPU without siblings or generally a CPU for which a new policy object has to be created, it grabs a reference to the driver module to start with, but drops that reference before returning. As a result, the driver module refcount is then equal to 0 after __cpufreq_add_dev() has returned. On the other hand, if the given CPU is a sibling of some other CPU already having a policy, cpufreq_add_policy_cpu() is called to link the new CPU to the existing policy. In that case, cpufreq_cpu_get() is called to obtain that policy and grabs a reference to the driver module, but that reference is not released and the module refcount will be different from 0 after __cpufreq_add_dev() returns (unless there is an error). That prevents the driver module from being unloaded until __cpufreq_remove_dev() is called for all the CPUs that cpufreq_add_policy_cpu() was called for previously. To remove that inconsistency make cpufreq_add_policy_cpu() execute cpufreq_cpu_put() for the given policy before returning, which decrements the driver module refcount so that it will be 0 after __cpufreq_add_dev() returns, but also make it take a reference to the policy itself using kobject_get() and do not release that reference (unless there's an error or system resume is under way), which again is consistent with the "raw" __cpufreq_add_dev() behavior. Accordingly, modify __cpufreq_remove_dev() to use kobject_put() to drop policy references taken by cpufreq_add_policy_cpu(). Signed-off-by: Rafael J. Wysocki --- On top of current linux-pm.git/linux-next. Thanks, Rafael --- drivers/cpufreq/cpufreq.c | 24 +++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-) Index: linux-pm/drivers/cpufreq/cpufreq.c =================================================================== --- linux-pm.orig/drivers/cpufreq/cpufreq.c +++ linux-pm/drivers/cpufreq/cpufreq.c @@ -908,8 +908,10 @@ static int cpufreq_add_policy_cpu(unsign unsigned long flags; policy = cpufreq_cpu_get(sibling); - WARN_ON(!policy); + if (WARN_ON_ONCE(!policy)) + return -ENODATA; + kobject_get(&policy->kobj); if (has_target) __cpufreq_governor(policy, CPUFREQ_GOV_STOP); @@ -932,14 +934,14 @@ static int cpufreq_add_policy_cpu(unsign /* Don't touch sysfs links during light-weight init */ if (frozen) { /* Drop the extra refcount that we took above */ - cpufreq_cpu_put(policy); - return 0; + kobject_put(&policy->kobj); + } else { + ret = sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq"); + if (ret) + kobject_put(&policy->kobj); } - ret = sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq"); - if (ret) - cpufreq_cpu_put(policy); - + cpufreq_cpu_put(policy); return ret; } #endif @@ -1298,10 +1300,14 @@ static int __cpufreq_remove_dev(struct d if (!frozen) cpufreq_policy_free(data); } else { - + /* + * There are more CPUs using the same policy, so only drop the + * reference taken by cpufreq_add_policy_cpu() (unless the + * system is suspending). + */ if (!frozen) { pr_debug("%s: removing link, cpu: %d\n", __func__, cpu); - cpufreq_cpu_put(data); + kobject_put(&data->kobj); } if (cpufreq_driver->target) {