Skip to contentSkip to conversation

media: qcom: iris: fix runtime PM error handling

PATCH New Hungyu Lin 2 participants 4 messages Jun 8

Subsystems (4)
lore
Share & tools
lore

Hungyu Lin

author
#0 lore raw

v5 cover media: qcom: iris: fix runtime PM error handling

This series fixes two issues in the Iris power management path.

Patch 1 updates the runtime PM handling to use pm_runtime_resume_and_get() and ensures that runtime PM references remain balanced during power-down error handling.

Patch 2 rolls back the maximum OPP vote when pm_runtime_resume_and_get() fails during power-up, preventing the driver from leaving an unnecessary OPP vote active after a failed resume.

Changes in v5:
- Handle pm_runtime_put_sync() return value in iris_disable_power_domains().
- Add a blank line before return statement in rollback path.

Changes in v4:
- Added Dmitry Baryshkov's Reviewed-by tag to patch 1.
- Reworked rollback handling in patch 2 to use a standard goto-based error path.
- Roll back the OPP vote from a dedicated error label when pm_runtime_resume_and_get() fails.

Hungyu Lin (2):
media: qcom: iris: fix runtime PM reference leaks
media: qcom: iris: rollback OPP vote on PM resume failure

 .../media/platform/qcom/iris/iris_resources.c | 20 ++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)
Signature · 1 line
2.34.1
author
#1 lore raw

PATCH v5 1/2 media: qcom: iris: fix runtime PM reference leaks

Use pm_runtime_resume_and_get() in iris_enable_power_domains() to avoid leaking a runtime PM usage count on failure.

Also ensure pm_runtime_put_sync() is always called in iris_disable_power_domains(), even when iris_opp_set_rate() fails, so runtime PM references remain balanced.

Fixes bb8a95aa038e ("media: iris: implement power management")
---
 drivers/media/platform/qcom/iris/iris_resources.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
drivers/media/platform/qcom/iris/iris_resources.c +6 -9
index 773f6548370a..6d3339423eec 100644
@@ -78,24 +78,21 @@ int iris_enable_power_domains(struct iris_core *core, struct device *pd_dev)
 	if (ret)
 		return ret;
 
-	ret = pm_runtime_get_sync(pd_dev);
-	if (ret < 0)
-		return ret;
-
-	return ret;
+	return pm_runtime_resume_and_get(pd_dev);
 }
 
 int iris_disable_power_domains(struct iris_core *core, struct device *pd_dev)
 {
 	int ret;
+	int pm_ret;
 
 	ret = iris_opp_set_rate(core->dev, 0);
-	if (ret)
-		return ret;
 
-	pm_runtime_put_sync(pd_dev);
+	pm_ret = pm_runtime_put_sync(pd_dev);
+	if (!ret)
+		ret = pm_ret;
 
-	return 0;
+	return ret;
 }
 
 static struct clk *iris_get_clk_by_type(struct iris_core *core, enum platform_clk_type clk_type)
Signature · 1 line
2.34.1
#2 lore raw

Re: [PATCH v5 1/2] media: qcom: iris: fix runtime PM reference leaks

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] `iris_disable_power_domains()` incorrectly propagates `1` on success when the device is already…

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] iris_disable_power_domains() incorrectly propagates 1 on success when the device is already suspended. --

commit 7b8fb44887231870efc4a92ebbb558afa99135b1
Author: Hungyu Lin <[email protected]>

media: qcom: iris: fix runtime PM reference leaks

This commit uses pm_runtime_resume_and_get() in iris_enable_power_domains() to prevent reference leaks on failure. It also updates iris_disable_power_domains() to always call pm_runtime_put_sync() and capture its result.

Quoted 3 lines
drivers/media/platform/qcom/iris/iris_resources.c +0 -0

[ ... ]

Quoted 7 lines

int iris_disable_power_domains(struct iris_core *core, struct device *pd_dev) {

drivers/media/platform/qcom/iris/iris_resources.c +1 -1
 	int ret;
+	int pm_ret;
 
 	ret = iris_opp_set_rate(core->dev, 0);
-	if (ret)

drivers/media/platform/qcom/iris/iris_resources.c int iris_enable_power_domains(struct iris_core *core, struct device *pd_dev) · line 96

Quoted 10 lines
drivers/media/platform/qcom/iris/iris_resources.c +4 -3
-		return ret;
 
-	pm_runtime_put_sync(pd_dev);
+	pm_ret = pm_runtime_put_sync(pd_dev);
+	if (!ret)
+		ret = pm_ret;
 
-	return 0;
+	return ret;
 }

[Severity: Low]
Is it intentional to propagate a positive return value here?

According to the runtime PM API, pm_runtime_put_sync() returns 1 as a success case when the device usage count drops to zero but it is already in the RPM_SUSPENDED state.

By capturing and returning this value directly from iris_disable_power_domains(), the function will return 1 instead of 0 on a successful suspend. While current callers may ignore the return value, future callers checking if (ret) could incorrectly treat this normal success case as a failure.

Would it be safer to normalize positive return values to 0?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
author
#3 lore raw

PATCH v5 2/2 media: qcom: iris: rollback OPP vote on PM resume failure

In reply to Hungyu Lin

Rollback the maximum OPP vote when
pm_runtime_resume_and_get() fails in
iris_enable_power_domains().

Use a standard rollback path to release the OPP vote when power domain enablement does not complete successfully.

Fixes bb8a95aa038e ("media: iris: implement power management")
---
 drivers/media/platform/qcom/iris/iris_resources.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/media/platform/qcom/iris/iris_resources.c +10 -1
index 6d3339423eec..3e18c6998f92 100644
@@ -78,7 +78,16 @@ int iris_enable_power_domains(struct iris_core *core, struct device *pd_dev)
 	if (ret)
 		return ret;
 
-	return pm_runtime_resume_and_get(pd_dev);
+	ret = pm_runtime_resume_and_get(pd_dev);
+	if (ret)
+		goto err_opp;
+
+	return 0;
+
+err_opp:
+	iris_opp_set_rate(core->dev, 0);
+
+	return ret;
 }
 
 int iris_disable_power_domains(struct iris_core *core, struct device *pd_dev)
Signature · 1 line
2.34.1