Skip to contentSkip to conversation

PM: runtime: fix recursive locking warning of lockdep from rpm_resume

PATCH Applied [email protected] 3 participants 3 messages Oct 22, 2010

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 PM: runtime: fix recursive locking warning of lockdep from rpm_resume

From: Ming Lei <[email protected]>

For device with no_callbacks flag set, its power lock and its parent's power lock may be held nestedly in rpm_resume, so we should take spin_lock_nested(lock, SINGLE_DEPTH_NESTING) to acquire parent power lock to avoid lockdep warning.

---
 drivers/base/power/runtime.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/base/power/runtime.c +1 -1
index 1dd8676..126ca49 100644
@@ -503,7 +503,7 @@ static int rpm_resume(struct device *dev, int rpmflags)
 	 * the resume will actually succeed.
 	 */
 	if (dev->power.no_callbacks && !parent && dev->parent) {
-		spin_lock(&dev->parent->power.lock);
+		spin_lock_nested(&dev->parent->power.lock, SINGLE_DEPTH_NESTING);
 		if (dev->parent->power.disable_depth > 0
 		    || dev->parent->power.ignore_children
 		    || dev->parent->power.runtime_status == RPM_ACTIVE) {
Signature · 1 line
1.7.3
Acked-by on v1 1/1
#1 lore raw
Quoted 17 lines · On Fri, 22 Oct 2010 [email protected] wrote:

From: Ming Lei <[email protected]>

For device with no_callbacks flag set, its power lock and its parent's power lock may be held nestedly in rpm_resume, so we should take spin_lock_nested(lock, SINGLE_DEPTH_NESTING) to acquire parent power lock to avoid lockdep warning.

---
 drivers/base/power/runtime.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/base/power/runtime.c +0 -0
index 1dd8676..126ca49 100644

drivers/base/power/runtime.c static int rpm_resume(struct device *dev, int rpmflags) · line 509

Quoted 8 lines
drivers/base/power/runtime.c +1 -1
@@ -503,7 +503,7 @@ static int rpm_resume(struct device *dev, int rpmflags)
 	 * the resume will actually succeed.
 	 */
 	if (dev->power.no_callbacks && !parent && dev->parent) {
-		spin_lock(&dev->parent->power.lock);
+		spin_lock_nested(&dev->parent->power.lock, SINGLE_DEPTH_NESTING);
 		if (dev->parent->power.disable_depth > 0
 		    || dev->parent->power.ignore_children
 		    || dev->parent->power.runtime_status == RPM_ACTIVE) {

Quite correct; I don't know why this didn't show up during my testing.

#2 lore raw
Quoted 30 lines · Alan Stern
Quoted 17 lines · On Fri, 22 Oct 2010 [email protected] wrote:

From: Ming Lei <[email protected]>

For device with no_callbacks flag set, its power lock and its parent's power lock may be held nestedly in rpm_resume, so we should take spin_lock_nested(lock, SINGLE_DEPTH_NESTING) to acquire parent power lock to avoid lockdep warning.

---
 drivers/base/power/runtime.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/base/power/runtime.c +0 -0
index 1dd8676..126ca49 100644

drivers/base/power/runtime.c static int rpm_resume(struct device *dev, int rpmflags) · line 509

Quoted 8 lines
drivers/base/power/runtime.c +1 -1
@@ -503,7 +503,7 @@ static int rpm_resume(struct device *dev, int rpmflags)
 	 * the resume will actually succeed.
 	 */
 	if (dev->power.no_callbacks && !parent && dev->parent) {
-		spin_lock(&dev->parent->power.lock);
+		spin_lock_nested(&dev->parent->power.lock, SINGLE_DEPTH_NESTING);
 		if (dev->parent->power.disable_depth > 0
 		    || dev->parent->power.ignore_children
 		    || dev->parent->power.runtime_status == RPM_ACTIVE) {

Quite correct; I don't know why this didn't show up during my testing.

Quoted 1 line

Applied to suspend-2.6/pm-fixes, will push to Linus early next week.

Thanks,
Rafael