[PATCH 0/5] cpufreq: Fixes for 3.12

[PATCH 0/5] cpufreq: Fixes for 3.12

[Viresh Kumar]

Hi Rafael,

You recently did this:

commit 878f6e074e9a7784a6e351512eace4ccb3542eef
Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Date:   Sun Aug 18 15:35:59 2013 +0200

    Revert "cpufreq: Use cpufreq_policy_list for iterating over policies"
    
    Revert commit eb60852 (cpufreq: Use cpufreq_policy_list for iterating
    over policies), because it breaks system suspend/resume on multiple
    machines.
    
    It either causes resume to block indefinitely or causes the BUG_ON()
    in lock_policy_rwsem_##mode() to trigger on sysfs accesses to cpufreq
    attributes.
    
------x------------x---------------

This patchset gets the reverted patch back along with few supporting patches.
Cause of the initial problem you observed was this:

- At suspend all CPUs are removed leaving boot cpu. At this time policies aren't
  freed and also aren't removed from cpufreq_policy_list. And per-cpu variable
  cpufreq_cpu_data is marked as NULL.
- At resume CPUs other than boot cpu called __cpufreq_add_dev(). The tricky
  change that was introduced by my patch was: We iterate over list of policies
  instead of CPUs, where we used to get policy structure associated with
  CPUs using per-cpu variable. Which used to be NULL for first CPU of a policy
  that turned up. For the first cpu we don't want to call
  cpufreq_add_policy_cpu() but want __cpufreq_add_add() to continue.

  When we called cpufreq_add_policy_cpu() it tried to stop the governor (which
  was already stopped) and hence errors leading into unstable state.

This patchset fixes these issues and is tested with suspend-resume over my
thinkpad with ubuntu. Apart from minor cleanups it removes policy from
cpufreq_policy_list in case of suspend/resume as well and hence we will never
call cpufreq_add_policy_cpu() for first cpu of a policy.

--
viresh

Viresh Kumar (5):
  cpufreq: align closing brace '}' of an if block
  cpufreq: remove policy from cpufreq_policy_list in system suspend
  cpufreq: remove unnecessary check in __cpufreq_governor()
  cpufreq: remove cpufreq_policy_cpu per-cpu variable
  cpufreq: Use cpufreq_policy_list for iterating over policies

 drivers/cpufreq/cpufreq.c | 77 +++++++++++++++--------------------------------
 1 file changed, 24 insertions(+), 53 deletions(-)

-- 
1.7.12.rc2.18.g61b472e

[PATCH 1/5] cpufreq: align closing brace '}' of an if block

[Viresh Kumar]

Following patch caused this:

commit 3de9bdeb28638e164d1f0eb38dd68e3f5d2ac95c
Author: Viresh Kumar <viresh.kumar@linaro.org>
Date:   Tue Aug 6 22:53:13 2013 +0530

    cpufreq: improve error checking on return values of __cpufreq_governor()

Lets fix it.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/cpufreq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index c0ef84d..fedc842 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -1253,7 +1253,7 @@ static int __cpufreq_remove_dev(struct device *dev,
 						__func__);
 				return ret;
 			}
-	}
+		}
 
 		if (!frozen) {
 			lock_policy_rwsem_read(cpu);
-- 
1.7.12.rc2.18.g61b472e

[PATCH 2/5] cpufreq: remove policy from cpufreq_policy_list in system suspend

[Viresh Kumar]

cpufreq_policy_list is a list of active policies. We do remove policies from
this list when all CPUs belonging to that policy are removed. But during suspend
we don't really free a policy struct as it will be used again during resume. And
so we didn't remove it from cpufreq_policy_list as well..

This is incorrect. We are saying this policy isn't valid anymore and must not be
referenced (though we haven't freed it), but it can still be used by code that
iterates over cpufreq_policy_list.

Lets remove policy from this list even in case of suspend as well.. Similarly we
must add it back whenever the first cpu belonging to that policy turns up.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/cpufreq.c | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index fedc842..0302121 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -950,12 +950,6 @@ err_free_policy:
 
 static void cpufreq_policy_free(struct cpufreq_policy *policy)
 {
-	unsigned long flags;
-
-	write_lock_irqsave(&cpufreq_driver_lock, flags);
-	list_del(&policy->policy_list);
-	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
-
 	free_cpumask_var(policy->related_cpus);
 	free_cpumask_var(policy->cpus);
 	kfree(policy);
@@ -1069,12 +1063,12 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 		ret = cpufreq_add_dev_interface(policy, dev);
 		if (ret)
 			goto err_out_unregister;
-
-		write_lock_irqsave(&cpufreq_driver_lock, flags);
-		list_add(&policy->policy_list, &cpufreq_policy_list);
-		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 	}
 
+	write_lock_irqsave(&cpufreq_driver_lock, flags);
+	list_add(&policy->policy_list, &cpufreq_policy_list);
+	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
+
 	cpufreq_init_policy(policy);
 
 	kobject_uevent(&policy->kobj, KOBJ_ADD);
@@ -1280,6 +1274,11 @@ static int __cpufreq_remove_dev(struct device *dev,
 		if (cpufreq_driver->exit)
 			cpufreq_driver->exit(policy);
 
+		/* Remove policy from list of active policies */
+		write_lock_irqsave(&cpufreq_driver_lock, flags);
+		list_del(&policy->policy_list);
+		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
+
 		if (!frozen)
 			cpufreq_policy_free(policy);
 	} else {
-- 
1.7.12.rc2.18.g61b472e

[PATCH 3/5] cpufreq: remove unnecessary check in __cpufreq_governor()

[Viresh Kumar]

We don't need to check if event is CPUFREQ_GOV_POLICY_INIT and put governor
module as we are sure event can only be START/STOP here.

And so remove this useless check.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/cpufreq.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 0302121..b03a851 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -1719,8 +1719,6 @@ static int __cpufreq_governor(struct cpufreq_policy *policy,
 	if ((!policy->governor_enabled && (event == CPUFREQ_GOV_STOP)) ||
 	    (policy->governor_enabled && (event == CPUFREQ_GOV_START))) {
 		mutex_unlock(&cpufreq_governor_lock);
-		if (event == CPUFREQ_GOV_POLICY_INIT)
-			module_put(policy->governor->owner);
 		return -EBUSY;
 	}
 
-- 
1.7.12.rc2.18.g61b472e

[PATCH 4/5] cpufreq: remove cpufreq_policy_cpu per-cpu variable

[Viresh Kumar]

cpufreq_policy_cpu per-cpu variables are used for storing cpu that manages a cpu
or its policy->cpu. We are also storing policy for each cpu in cpufreq_cpu_data
and so this information is just redundant.

Better use cpufreq_cpu_data to retrieve policy and get policy->cpu from there.
And so this patch removes cpufreq_policy_cpu per-cpu variable.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/cpufreq.c | 45 ++++++++++-----------------------------------
 1 file changed, 10 insertions(+), 35 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index b03a851..0586bd2 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -64,15 +64,14 @@ static DEFINE_PER_CPU(char[CPUFREQ_NAME_LEN], cpufreq_cpu_governor);
  * - Lock should not be held across
  *     __cpufreq_governor(data, CPUFREQ_GOV_STOP);
  */
-static DEFINE_PER_CPU(int, cpufreq_policy_cpu);
 static DEFINE_PER_CPU(struct rw_semaphore, cpu_policy_rwsem);
 
 #define lock_policy_rwsem(mode, cpu)					\
 static int lock_policy_rwsem_##mode(int cpu)				\
 {									\
-	int policy_cpu = per_cpu(cpufreq_policy_cpu, cpu);		\
-	BUG_ON(policy_cpu == -1);					\
-	down_##mode(&per_cpu(cpu_policy_rwsem, policy_cpu));		\
+	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);	\
+	BUG_ON(!policy);						\
+	down_##mode(&per_cpu(cpu_policy_rwsem, policy->cpu));		\
 									\
 	return 0;							\
 }
@@ -83,9 +82,9 @@ lock_policy_rwsem(write, cpu);
 #define unlock_policy_rwsem(mode, cpu)					\
 static void unlock_policy_rwsem_##mode(int cpu)				\
 {									\
-	int policy_cpu = per_cpu(cpufreq_policy_cpu, cpu);		\
-	BUG_ON(policy_cpu == -1);					\
-	up_##mode(&per_cpu(cpu_policy_rwsem, policy_cpu));		\
+	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);	\
+	BUG_ON(!policy);						\
+	up_##mode(&per_cpu(cpu_policy_rwsem, policy->cpu));		\
 }
 
 unlock_policy_rwsem(read, cpu);
@@ -887,7 +886,6 @@ static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
 	write_lock_irqsave(&cpufreq_driver_lock, flags);
 
 	cpumask_set_cpu(cpu, policy->cpus);
-	per_cpu(cpufreq_policy_cpu, cpu) = policy->cpu;
 	per_cpu(cpufreq_cpu_data, cpu) = policy;
 	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
@@ -1013,9 +1011,6 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 	policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
 	cpumask_copy(policy->cpus, cpumask_of(cpu));
 
-	/* Initially set CPU itself as the policy_cpu */
-	per_cpu(cpufreq_policy_cpu, cpu) = cpu;
-
 	init_completion(&policy->kobj_unregister);
 	INIT_WORK(&policy->update, handle_update);
 
@@ -1053,10 +1048,8 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 #endif
 
 	write_lock_irqsave(&cpufreq_driver_lock, flags);
-	for_each_cpu(j, policy->cpus) {
+	for_each_cpu(j, policy->cpus)
 		per_cpu(cpufreq_cpu_data, j) = policy;
-		per_cpu(cpufreq_policy_cpu, j) = policy->cpu;
-	}
 	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
 	if (!frozen) {
@@ -1080,15 +1073,11 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 
 err_out_unregister:
 	write_lock_irqsave(&cpufreq_driver_lock, flags);
-	for_each_cpu(j, policy->cpus) {
+	for_each_cpu(j, policy->cpus)
 		per_cpu(cpufreq_cpu_data, j) = NULL;
-		if (j != cpu)
-			per_cpu(cpufreq_policy_cpu, j) = -1;
-	}
 	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
 err_set_policy_cpu:
-	per_cpu(cpufreq_policy_cpu, cpu) = -1;
 	cpufreq_policy_free(policy);
 nomem_out:
 	up_read(&cpufreq_rwsem);
@@ -1112,14 +1101,9 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 
 static void update_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu)
 {
-	int j;
-
 	policy->last_cpu = policy->cpu;
 	policy->cpu = cpu;
 
-	for_each_cpu(j, policy->cpus)
-		per_cpu(cpufreq_policy_cpu, j) = cpu;
-
 #ifdef CONFIG_CPU_FREQ_TABLE
 	cpufreq_frequency_table_update_policy_cpu(policy);
 #endif
@@ -1131,7 +1115,6 @@ static int cpufreq_nominate_new_policy_cpu(struct cpufreq_policy *policy,
 					   unsigned int old_cpu, bool frozen)
 {
 	struct device *cpu_dev;
-	unsigned long flags;
 	int ret;
 
 	/* first sibling now owns the new sysfs dir */
@@ -1148,11 +1131,6 @@ static int cpufreq_nominate_new_policy_cpu(struct cpufreq_policy *policy,
 
 		WARN_ON(lock_policy_rwsem_write(old_cpu));
 		cpumask_set_cpu(old_cpu, policy->cpus);
-
-		write_lock_irqsave(&cpufreq_driver_lock, flags);
-		per_cpu(cpufreq_cpu_data, old_cpu) = policy;
-		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
-
 		unlock_policy_rwsem_write(old_cpu);
 
 		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
@@ -1186,7 +1164,6 @@ static int __cpufreq_remove_dev(struct device *dev,
 	write_lock_irqsave(&cpufreq_driver_lock, flags);
 
 	policy = per_cpu(cpufreq_cpu_data, cpu);
-	per_cpu(cpufreq_cpu_data, cpu) = NULL;
 
 	/* Save the policy somewhere when doing a light-weight tear-down */
 	if (frozen)
@@ -1292,7 +1269,7 @@ static int __cpufreq_remove_dev(struct device *dev,
 		}
 	}
 
-	per_cpu(cpufreq_policy_cpu, cpu) = -1;
+	per_cpu(cpufreq_cpu_data, cpu) = NULL;
 	return 0;
 }
 
@@ -2148,10 +2125,8 @@ static int __init cpufreq_core_init(void)
 	if (cpufreq_disabled())
 		return -ENODEV;
 
-	for_each_possible_cpu(cpu) {
-		per_cpu(cpufreq_policy_cpu, cpu) = -1;
+	for_each_possible_cpu(cpu)
 		init_rwsem(&per_cpu(cpu_policy_rwsem, cpu));
-	}
 
 	cpufreq_global_kobject = kobject_create();
 	BUG_ON(!cpufreq_global_kobject);
-- 
1.7.12.rc2.18.g61b472e

[PATCH 5/5] cpufreq: Use cpufreq_policy_list for iterating over policies

[Viresh Kumar]

To iterate over all policies we currently iterate over all CPUs and
then get the policy for each of them.  Let's use the newly created
cpufreq_policy_list for this purpose.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/cpufreq.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 0586bd2..81ceea6 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -961,8 +961,8 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 	struct cpufreq_policy *policy;
 	unsigned long flags;
 #ifdef CONFIG_HOTPLUG_CPU
+	struct cpufreq_policy *tpolicy;
 	struct cpufreq_governor *gov;
-	int sibling;
 #endif
 
 	if (cpu_is_offline(cpu))
@@ -985,11 +985,10 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 #ifdef CONFIG_HOTPLUG_CPU
 	/* Check if this cpu was hot-unplugged earlier and has siblings */
 	read_lock_irqsave(&cpufreq_driver_lock, flags);
-	for_each_online_cpu(sibling) {
-		struct cpufreq_policy *cp = per_cpu(cpufreq_cpu_data, sibling);
-		if (cp && cpumask_test_cpu(cpu, cp->related_cpus)) {
+	list_for_each_entry(tpolicy, &cpufreq_policy_list, policy_list) {
+		if (cpumask_test_cpu(cpu, tpolicy->related_cpus)) {
 			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
-			ret = cpufreq_add_policy_cpu(cp, cpu, dev, frozen);
+			ret = cpufreq_add_policy_cpu(tpolicy, cpu, dev, frozen);
 			up_read(&cpufreq_rwsem);
 			return ret;
 		}
-- 
1.7.12.rc2.18.g61b472e

Re: [PATCH 0/5] cpufreq: Fixes for 3.12

[Rafael J. Wysocki]

On Tuesday, August 20, 2013 12:08:21 PM Viresh Kumar wrote:
> Hi Rafael,
> 
> You recently did this:
> 
> commit 878f6e074e9a7784a6e351512eace4ccb3542eef
> Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> Date:   Sun Aug 18 15:35:59 2013 +0200
> 
>     Revert "cpufreq: Use cpufreq_policy_list for iterating over policies"
>     
>     Revert commit eb60852 (cpufreq: Use cpufreq_policy_list for iterating
>     over policies), because it breaks system suspend/resume on multiple
>     machines.
>     
>     It either causes resume to block indefinitely or causes the BUG_ON()
>     in lock_policy_rwsem_##mode() to trigger on sysfs accesses to cpufreq
>     attributes.
>     
> ------x------------x---------------
> 
> This patchset gets the reverted patch back along with few supporting patches.
> Cause of the initial problem you observed was this:
> 
> - At suspend all CPUs are removed leaving boot cpu. At this time policies aren't
>   freed and also aren't removed from cpufreq_policy_list. And per-cpu variable
>   cpufreq_cpu_data is marked as NULL.
> - At resume CPUs other than boot cpu called __cpufreq_add_dev(). The tricky
>   change that was introduced by my patch was: We iterate over list of policies
>   instead of CPUs, where we used to get policy structure associated with
>   CPUs using per-cpu variable. Which used to be NULL for first CPU of a policy
>   that turned up. For the first cpu we don't want to call
>   cpufreq_add_policy_cpu() but want __cpufreq_add_add() to continue.
> 
>   When we called cpufreq_add_policy_cpu() it tried to stop the governor (which
>   was already stopped) and hence errors leading into unstable state.
> 
> This patchset fixes these issues and is tested with suspend-resume over my
> thinkpad with ubuntu. Apart from minor cleanups it removes policy from
> cpufreq_policy_list in case of suspend/resume as well and hence we will never
> call cpufreq_add_policy_cpu() for first cpu of a policy.

Well, this looks good, but do we really need it in 3.12?  It doesn't look
like 3.12 will be missing these changes a lot?

Rafael

Re: [PATCH 0/5] cpufreq: Fixes for 3.12

[Viresh Kumar]

On 20 August 2013 18:13, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> Well, this looks good, but do we really need it in 3.12?  It doesn't look
> like 3.12 will be missing these changes a lot?

Hmm.. If its too late for 3.12 we can skip them, but first two are bug
fixes really (though nobody is affected as cpufreq_policy_list isn't used
yet :) ) and other three are cleanups..

So, if there is any chance that we can queue them up for 3.12, it would
be good.. but definitely not a must to have stuff. :)

--
viresh

Re: [PATCH 0/5] cpufreq: Fixes for 3.12

[Rafael J. Wysocki]

On Tuesday, August 20, 2013 06:07:14 PM Viresh Kumar wrote:
> On 20 August 2013 18:13, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > Well, this looks good, but do we really need it in 3.12?  It doesn't look
> > like 3.12 will be missing these changes a lot?
> 
> Hmm.. If its too late for 3.12 we can skip them, but first two are bug
> fixes really (though nobody is affected as cpufreq_policy_list isn't used
> yet :) ) and other three are cleanups..
> 
> So, if there is any chance that we can queue them up for 3.12, it would
> be good.. but definitely not a must to have stuff. :)

OK, queued up for 3.12.  Hopefully, there won't be any fallout from [3/5].

Thanks,
Rafael

Re: [PATCH 0/5] cpufreq: Fixes for 3.12

[Viresh Kumar]

On 20 August 2013 19:33, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> OK, queued up for 3.12.  Hopefully, there won't be any fallout from [3/5].

Hopefully :)

Thanks..

Re: [PATCH 0/5] cpufreq: Fixes for 3.12

[Rafael J. Wysocki]

On Tuesday, August 20, 2013 07:24:41 PM Viresh Kumar wrote:
> On 20 August 2013 19:33, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > OK, queued up for 3.12.  Hopefully, there won't be any fallout from [3/5].

Argh, I meant [4/5], typed a wrong number.  Whatever.