Skip to contentSkip to conversation

cdx: Fix driver_managed_dma check

PATCH In review Robin Murphy 3 participants 3 messages Apr 25, 2025 – May 5, 2025 AMD cdx bus driver

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 cdx: Fix driver_managed_dma check

Since it's not currently safe to take device_lock() in the IOMMU probe path, that can race against really_probe() setting dev->driver before attempting to bind. The race itself isn't so bad, since we're only concerned with dereferencing dev->driver itself anyway, but sadly my attempt to implement the check with minimal churn leads to a kind of TOCTOU issue, where dev->driver becomes valid after to_cdx_driver(NULL) is already computed, and thus the check fails to work as intended.

Will and I both hit this with the platform bus, but the pattern here is the same, so fix it for correctness too.

Fixes bcb81ac6ae3c ("iommu: Get DT/ACPI parsing into the proper probe path")
---
 drivers/cdx/cdx.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/cdx/cdx.c +3 -3
index 092306ca2541..8aa92a198b71 100644
@@ -348,7 +348,7 @@ static void cdx_shutdown(struct device *dev)
 
 static int cdx_dma_configure(struct device *dev)
 {
-	struct cdx_driver *cdx_drv = to_cdx_driver(dev->driver);
+	const struct device_driver *drv = READ_ONCE(dev->driver);
 	struct cdx_device *cdx_dev = to_cdx_device(dev);
 	struct cdx_controller *cdx = cdx_dev->cdx;
 	u32 input_id = cdx_dev->req_id;
@@ -360,8 +360,8 @@ static int cdx_dma_configure(struct device *dev)
 		return ret;
 	}
 
-	/* @cdx_drv may not be valid when we're called from the IOMMU layer */
-	if (!ret && dev->driver && !cdx_drv->driver_managed_dma) {
+	/* @drv may not be valid when we're called from the IOMMU layer */
+	if (!ret && drv && !to_cdx_driver(drv)->driver_managed_dma) {
 		ret = iommu_device_use_default_domain(dev);
 		if (ret)
 			arch_teardown_dma_ops(dev);
Signature · 1 line
2.39.2.101.g768bb238c484.dirty
Reviewed-by on v1 1/1
#1 lore raw
Quoted 10 lines · Robin Murphy

Since it's not currently safe to take device_lock() in the IOMMU probe path, that can race against really_probe() setting dev->driver before attempting to bind. The race itself isn't so bad, since we're only concerned with dereferencing dev->driver itself anyway, but sadly my attempt to implement the check with minimal churn leads to a kind of TOCTOU issue, where dev->driver becomes valid after to_cdx_driver(NULL) is already computed, and thus the check fails to work as intended.

Will and I both hit this with the platform bus, but the pattern here is the same, so fix it for correctness too.

Thanks!

Quoted 35 lines
Fixes bcb81ac6ae3c ("iommu: Get DT/ACPI parsing into the proper probe path")
---
 drivers/cdx/cdx.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/cdx/cdx.c +3 -3
index 092306ca2541..8aa92a198b71 100644
@@ -348,7 +348,7 @@ static void cdx_shutdown(struct device *dev)
 
 static int cdx_dma_configure(struct device *dev)
 {
-	struct cdx_driver *cdx_drv = to_cdx_driver(dev->driver);
+	const struct device_driver *drv = READ_ONCE(dev->driver);
 	struct cdx_device *cdx_dev = to_cdx_device(dev);
 	struct cdx_controller *cdx = cdx_dev->cdx;
 	u32 input_id = cdx_dev->req_id;
@@ -360,8 +360,8 @@ static int cdx_dma_configure(struct device *dev)
 		return ret;
 	}
 
-	/* @cdx_drv may not be valid when we're called from the IOMMU layer */
-	if (!ret && dev->driver && !cdx_drv->driver_managed_dma) {
+	/* @drv may not be valid when we're called from the IOMMU layer */
+	if (!ret && drv && !to_cdx_driver(drv)->driver_managed_dma) {
 		ret = iommu_device_use_default_domain(dev);
 		if (ret)
 			arch_teardown_dma_ops(dev);
Signature · 1 line
2.39.2.101.g768bb238c484.dirty
maintainer Acked-by on v1 1/1
#2 lore raw
In reply to Robin Murphy
Quoted 18 lines

-----Original Message-----
From: Robin Murphy <[email protected]>
Sent: Friday, April 25, 2025 7:09 PM
To: [email protected]; [email protected]; Gupta, Nipun <[email protected]>; Agarwal, Nikhil <[email protected]>
Cc: [email protected]; [email protected]; Will McVicker <[email protected]>
Subject: [PATCH] cdx: Fix driver_managed_dma check

Since it's not currently safe to take device_lock() in the IOMMU probe path, that can race against really_probe() setting dev->driver before attempting to bind. The race itself isn't so bad, since we're only concerned with dereferencing dev->driver itself anyway, but sadly my attempt to implement the check with minimal churn leads to a kind of TOCTOU issue, where dev->driver becomes valid after to_cdx_driver(NULL) is already computed, and thus the check fails to work as intended.

Quoted 6 lines

Will and I both hit this with the platform bus, but the pattern here is the same, so fix it for correctness too.

Fixes bcb81ac6ae3c ("iommu: Get DT/ACPI parsing into the proper probe path")