diff mbox

[3/4] cpufreq: Fix governor module removal race

Message ID 2007526.uAUpsoPmOG@aspire.rjw.lan (mailing list archive)
State Superseded, archived
Headers show

Commit Message

Rafael J. Wysocki Nov. 23, 2017, 12:29 a.m. UTC
From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

It is possible to remove a cpufreq governor module after
cpufreq_parse_governor() has returned success in
store_scaling_governor() and before cpufreq_set_policy()
acquires a reference to it, because the governor list is
not protected during that period and nothing prevents the
governor from being unregistered then.  The pointer to the
governor structure coming from cpufreq_parse_governor() may
become stale as a result of that.

Prevent that from happening by acquiring an extra reference
to the governor module temporarily in cpufreq_parse_governor(),
under cpufreq_governor_mutex, and dropping it in
store_scaling_governor(), when cpufreq_set_policy() returns.

Note that the second cpufreq_parse_governor() call site is fine,
because it only cares about the policy member of new_policy.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/cpufreq/cpufreq.c |    8 ++++++++
 1 file changed, 8 insertions(+)

Comments

Viresh Kumar Nov. 23, 2017, 4:01 a.m. UTC | #1
On 23-11-17, 01:29, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> 
> It is possible to remove a cpufreq governor module after
> cpufreq_parse_governor() has returned success in
> store_scaling_governor() and before cpufreq_set_policy()
> acquires a reference to it, because the governor list is
> not protected during that period and nothing prevents the
> governor from being unregistered then.  The pointer to the
> governor structure coming from cpufreq_parse_governor() may
> become stale as a result of that.
> 
> Prevent that from happening by acquiring an extra reference
> to the governor module temporarily in cpufreq_parse_governor(),
> under cpufreq_governor_mutex, and dropping it in
> store_scaling_governor(), when cpufreq_set_policy() returns.
> 
> Note that the second cpufreq_parse_governor() call site is fine,
> because it only cares about the policy member of new_policy.
> 
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> ---
>  drivers/cpufreq/cpufreq.c |    8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> Index: linux-pm/drivers/cpufreq/cpufreq.c
> ===================================================================
> --- linux-pm.orig/drivers/cpufreq/cpufreq.c
> +++ linux-pm/drivers/cpufreq/cpufreq.c
> @@ -607,11 +607,13 @@ static int cpufreq_parse_governor(char *
>  	if (cpufreq_driver->setpolicy) {
>  		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
>  			policy->policy = CPUFREQ_POLICY_PERFORMANCE;
> +			policy->governor = NULL;
>  			return 0;
>  		}
>  
>  		if (!strncasecmp(str_governor, "powersave", CPUFREQ_NAME_LEN)) {
>  			policy->policy = CPUFREQ_POLICY_POWERSAVE;
> +			policy->governor = NULL;

Why are the above two changes required? policy->governor should always be NULL
for setpolicy drivers anyway.
Rafael J. Wysocki Nov. 23, 2017, 1:16 p.m. UTC | #2
On Thursday, November 23, 2017 5:01:17 AM CET Viresh Kumar wrote:
> On 23-11-17, 01:29, Rafael J. Wysocki wrote:
> > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > 
> > It is possible to remove a cpufreq governor module after
> > cpufreq_parse_governor() has returned success in
> > store_scaling_governor() and before cpufreq_set_policy()
> > acquires a reference to it, because the governor list is
> > not protected during that period and nothing prevents the
> > governor from being unregistered then.  The pointer to the
> > governor structure coming from cpufreq_parse_governor() may
> > become stale as a result of that.
> > 
> > Prevent that from happening by acquiring an extra reference
> > to the governor module temporarily in cpufreq_parse_governor(),
> > under cpufreq_governor_mutex, and dropping it in
> > store_scaling_governor(), when cpufreq_set_policy() returns.
> > 
> > Note that the second cpufreq_parse_governor() call site is fine,
> > because it only cares about the policy member of new_policy.
> > 
> > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > ---
> >  drivers/cpufreq/cpufreq.c |    8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > Index: linux-pm/drivers/cpufreq/cpufreq.c
> > ===================================================================
> > --- linux-pm.orig/drivers/cpufreq/cpufreq.c
> > +++ linux-pm/drivers/cpufreq/cpufreq.c
> > @@ -607,11 +607,13 @@ static int cpufreq_parse_governor(char *
> >  	if (cpufreq_driver->setpolicy) {
> >  		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
> >  			policy->policy = CPUFREQ_POLICY_PERFORMANCE;
> > +			policy->governor = NULL;
> >  			return 0;
> >  		}
> >  
> >  		if (!strncasecmp(str_governor, "powersave", CPUFREQ_NAME_LEN)) {
> >  			policy->policy = CPUFREQ_POLICY_POWERSAVE;
> > +			policy->governor = NULL;
> 
> Why are the above two changes required? policy->governor should always be NULL
> for setpolicy drivers anyway.

OK, I'll drop them.

Thanks,
Rafael
diff mbox

Patch

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================
--- linux-pm.orig/drivers/cpufreq/cpufreq.c
+++ linux-pm/drivers/cpufreq/cpufreq.c
@@ -607,11 +607,13 @@  static int cpufreq_parse_governor(char *
 	if (cpufreq_driver->setpolicy) {
 		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
 			policy->policy = CPUFREQ_POLICY_PERFORMANCE;
+			policy->governor = NULL;
 			return 0;
 		}
 
 		if (!strncasecmp(str_governor, "powersave", CPUFREQ_NAME_LEN)) {
 			policy->policy = CPUFREQ_POLICY_POWERSAVE;
+			policy->governor = NULL;
 			return 0;
 		}
 	} else {
@@ -633,6 +635,8 @@  static int cpufreq_parse_governor(char *
 
 			t = find_governor(str_governor);
 		}
+		if (t && !try_module_get(t->owner))
+			t = NULL;
 
 		mutex_unlock(&cpufreq_governor_mutex);
 
@@ -766,6 +770,10 @@  static ssize_t store_scaling_governor(st
 		return -EINVAL;
 
 	ret = cpufreq_set_policy(policy, &new_policy);
+
+	if (new_policy.governor)
+		module_put(new_policy.governor->owner);
+
 	return ret ? ret : count;
 }