Skip to contentSkip to conversation

i2c: core: fix lockdep warning for sparsely nested adapter chain

PATCH Applied Daniel Mack 2 participants 3 messages Oct 18, 2023 – Oct 28, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 i2c: core: fix lockdep warning for sparsely nested adapter chain

When adapters are chained in a sparse manner (with intermediate MFD devices, for instance) the code currently fails to use the correct subclass for the adapter's bus_lock which leads to false-positive lockdep warnings.

Fix this by walking the entire pedigree of the device and count all adapters along the way instead of just checking the immediate parent.

---
This hit me when during the development of a driver stack that isn't submitted mainline yet. This patch could however be discussed independently I think.

 drivers/i2c/i2c-core-base.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/i2c/i2c-core-base.c +4 -2
index 60746652fd52..4692a1e5ea0a 100644
@@ -1189,9 +1189,11 @@ static void i2c_adapter_dev_release(struct device *dev)
 unsigned int i2c_adapter_depth(struct i2c_adapter *adapter)
 {
 	unsigned int depth = 0;
+	struct device *parent;
 
-	while ((adapter = i2c_parent_is_i2c_adapter(adapter)))
-		depth++;
+	for (parent = adapter->dev.parent; parent; parent = parent->parent)
+		if (parent->type == &i2c_adapter_type)
+			depth++;
 
 	WARN_ONCE(depth >= MAX_LOCKDEP_SUBCLASSES,
 		  "adapter depth exceeds lockdep subclass limit\n");
Signature · 1 line
2.41.0
#1 lore raw

Hi Daniel,

Quoted 1 line · Daniel Mack

When adapters are chained in a sparse manner (with intermediate MFD devices,

So, you have an MFD including an i2c-mux or something?

Quoted 5 lines

for instance) the code currently fails to use the correct subclass for the adapter's bus_lock which leads to false-positive lockdep warnings.

Fix this by walking the entire pedigree of the device and count all adapters along the way instead of just checking the immediate parent.

Sounds reasonable to me.

Quoted 6 lines

---
This hit me when during the development of a driver stack that isn't submitted mainline yet. This patch could however be discussed independently I think.

Yes, it can :)

Quoted 9 lines
 drivers/i2c/i2c-core-base.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/i2c/i2c-core-base.c +0 -0
index 60746652fd52..4692a1e5ea0a 100644

drivers/i2c/i2c-core-base.c static void i2c_adapter_dev_release(struct device *dev) · line 1193

Quoted 6 lines
drivers/i2c/i2c-core-base.c +1 -1
@@ -1189,9 +1189,11 @@ static void i2c_adapter_dev_release(struct device *dev)
 unsigned int i2c_adapter_depth(struct i2c_adapter *adapter)
 {
 	unsigned int depth = 0;
+	struct device *parent;
 
-	while ((adapter = i2c_parent_is_i2c_adapter(adapter)))

I never noticed we overwrite the 'adapter' function argument. Much better with your version and the local variable.

drivers/i2c/i2c-core-base.c static void i2c_adapter_dev_release(struct device *dev) · line 1196

Quoted 4 lines
drivers/i2c/i2c-core-base.c +3 -1
-		depth++;
+	for (parent = adapter->dev.parent; parent; parent = parent->parent)
+		if (parent->type == &i2c_adapter_type)
+			depth++;

I am not sure myself. Is the code explaining itself or should we add a short comment why we use a for-loop? I tend to leave it as is.

Thanks for this patch!

Wolfram

Applied
#2 lore raw
In reply to Daniel Mack
Quoted 8 lines · Daniel Mack

When adapters are chained in a sparse manner (with intermediate MFD devices, for instance) the code currently fails to use the correct subclass for the adapter's bus_lock which leads to false-positive lockdep warnings.

Fix this by walking the entire pedigree of the device and count all adapters along the way instead of just checking the immediate parent.

Applied to for-next, thanks!