Skip to contentSkip to conversation

media: v4l2-subdev: Fail {enable,disable}_streams and s_streaming nicely

PATCH Superseded Sakari Ailus 3 participants 5 messages Mar 23 – Mar 25 Media (V4L/DVB)

v2 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 media: v4l2-subdev: Fail {enable,disable}_streams and s_streaming nicely

If a sub-device does not set enable_streams() and disable_streams() pad ops while it sets the s_stream() video op to v4l2_subdev_s_stream_helper(), enabling or disabling streaming either way on the sub-device will result calling v4l2_subdev_s_stream_helper() and v4l2_subdev_{enable,disable}_streams() recursively, exhausting the stack. Return -ENOIOCTLCMD in this case to handle the situation gracefully.

Fixes b62949ddaa52 ("media: subdev: Support single-stream case in v4l2_subdev_enable/disable_streams()")

---
since v1:

- Better patch description.

 drivers/media/v4l2-core/v4l2-subdev.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +10 -6
index 831c69c958b8..f8ea4afc6cbb 100644
@@ -2334,11 +2334,13 @@ int v4l2_subdev_enable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .enable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, enable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
@@ -2435,11 +2437,13 @@ int v4l2_subdev_disable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .disable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, disable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
Signature · 1 line
2.47.3
#1 lore raw
Quoted 6 lines · Sakari Ailus

If a sub-device does not set enable_streams() and disable_streams() pad ops while it sets the s_stream() video op to v4l2_subdev_s_stream_helper(), enabling or disabling streaming either way on the sub-device will result calling v4l2_subdev_s_stream_helper() and v4l2_subdev_{enable,disable}_streams() recursively, exhausting the stack. Return -ENOIOCTLCMD in this case to handle the situation gracefully.

This is not a valid use case, right ? Can we WARN() ?

Quoted 50 lines
Fixes b62949ddaa52 ("media: subdev: Support single-stream case in v4l2_subdev_enable/disable_streams()")

---
since v1:

- Better patch description.

 drivers/media/v4l2-core/v4l2-subdev.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +10 -6
index 831c69c958b8..f8ea4afc6cbb 100644
@@ -2334,11 +2334,13 @@ int v4l2_subdev_enable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .enable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, enable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
@@ -2435,11 +2437,13 @@ int v4l2_subdev_disable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .disable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, disable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
Signature · 3 lines
Regards,

Laurent Pinchart
#2 lore raw

Hi Laurent,

Quoted 9 lines · Laurent Pinchart
Quoted 6 lines · Sakari Ailus

If a sub-device does not set enable_streams() and disable_streams() pad ops while it sets the s_stream() video op to v4l2_subdev_s_stream_helper(), enabling or disabling streaming either way on the sub-device will result calling v4l2_subdev_s_stream_helper() and v4l2_subdev_{enable,disable}_streams() recursively, exhausting the stack. Return -ENOIOCTLCMD in this case to handle the situation gracefully.

This is not a valid use case, right ? Can we WARN() ?

Partly yes, and partly it requires a driver bug.

The CCS driver soon will differentiate its ops for source (scaler/binner and CSI-2 receiver) sub-device, but only pad ops, not video ops. This means that on non-source sub-devices the s_stream pad op is set to v4l2_subdev_s_stream_helper() while {enable,disable}_streams() pad ops are NULL.

It also takes a bug in the camss driver to trigger this -- it enables streaming on all sub-devices of the pipeline instead of the next upstream sub-device. That bug should be fixed but it's a separate issue.

Quoted 52 lines
Quoted 51 lines
Fixes b62949ddaa52 ("media: subdev: Support single-stream case in v4l2_subdev_enable/disable_streams()")

---
since v1:

- Better patch description.

 drivers/media/v4l2-core/v4l2-subdev.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +10 -6
index 831c69c958b8..f8ea4afc6cbb 100644
@@ -2334,11 +2334,13 @@ int v4l2_subdev_enable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .enable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, enable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
@@ -2435,11 +2437,13 @@ int v4l2_subdev_disable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .disable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, disable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
Signature · 3 lines
Regards,

Sakari Ailus
#3 lore raw
Quoted 19 lines · Sakari Ailus
Quoted 9 lines · Laurent Pinchart
Quoted 6 lines · Sakari Ailus

If a sub-device does not set enable_streams() and disable_streams() pad ops while it sets the s_stream() video op to v4l2_subdev_s_stream_helper(), enabling or disabling streaming either way on the sub-device will result calling v4l2_subdev_s_stream_helper() and v4l2_subdev_{enable,disable}_streams() recursively, exhausting the stack. Return -ENOIOCTLCMD in this case to handle the situation gracefully.

This is not a valid use case, right ? Can we WARN() ?

Partly yes, and partly it requires a driver bug.

The CCS driver soon will differentiate its ops for source (scaler/binner and CSI-2 receiver) sub-device, but only pad ops, not video ops. This means that on non-source sub-devices the s_stream pad op is set to v4l2_subdev_s_stream_helper() while {enable,disable}_streams() pad ops are NULL.

Quoted 3 lines

It also takes a bug in the camss driver to trigger this -- it enables streaming on all sub-devices of the pipeline instead of the next upstream sub-device. That bug should be fixed but it's a separate issue.

Sounds like a candidate for a WARN() then, doesn't it ?

Quoted 50 lines
Quoted 50 lines
Quoted 50 lines
Fixes b62949ddaa52 ("media: subdev: Support single-stream case in v4l2_subdev_enable/disable_streams()")

---
since v1:

- Better patch description.

 drivers/media/v4l2-core/v4l2-subdev.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +10 -6
index 831c69c958b8..f8ea4afc6cbb 100644
@@ -2334,11 +2334,13 @@ int v4l2_subdev_enable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .enable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, enable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
@@ -2435,11 +2437,13 @@ int v4l2_subdev_disable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .disable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, disable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
Signature · 3 lines
Regards,

Laurent Pinchart
#4 lore raw

Hi Laurent,

Quoted 20 lines · Laurent Pinchart
Quoted 19 lines · Sakari Ailus
Quoted 9 lines · Laurent Pinchart
Quoted 6 lines · Sakari Ailus

If a sub-device does not set enable_streams() and disable_streams() pad ops while it sets the s_stream() video op to v4l2_subdev_s_stream_helper(), enabling or disabling streaming either way on the sub-device will result calling v4l2_subdev_s_stream_helper() and v4l2_subdev_{enable,disable}_streams() recursively, exhausting the stack. Return -ENOIOCTLCMD in this case to handle the situation gracefully.

This is not a valid use case, right ? Can we WARN() ?

Partly yes, and partly it requires a driver bug.

The CCS driver soon will differentiate its ops for source (scaler/binner and CSI-2 receiver) sub-device, but only pad ops, not video ops. This means that on non-source sub-devices the s_stream pad op is set to v4l2_subdev_s_stream_helper() while {enable,disable}_streams() pad ops are NULL.

Quoted 5 lines
Quoted 3 lines

It also takes a bug in the camss driver to trigger this -- it enables streaming on all sub-devices of the pipeline instead of the next upstream sub-device. That bug should be fixed but it's a separate issue.

Sounds like a candidate for a WARN() then, doesn't it ?

I'll add WARN_ON_ONCE() for using v4l2_subdev_s_straem_helper() without both {enable,disable}_streams()_ in v3. I came to think also that this check can be moved to the helper where I think it'll look nicer.

Quoted 52 lines
Quoted 51 lines
Quoted 51 lines
Quoted 51 lines
Fixes b62949ddaa52 ("media: subdev: Support single-stream case in v4l2_subdev_enable/disable_streams()")

---
since v1:

- Better patch description.

 drivers/media/v4l2-core/v4l2-subdev.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +10 -6
index 831c69c958b8..f8ea4afc6cbb 100644
@@ -2334,11 +2334,13 @@ int v4l2_subdev_enable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .enable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, enable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
@@ -2435,11 +2437,13 @@ int v4l2_subdev_disable_streams(struct v4l2_subdev *sd, u32 pad,
 
 	/* Fallback on .s_stream() if .disable_streams() isn't available. */
 	use_s_stream = !v4l2_subdev_has_op(sd, pad, disable_streams);
-
-	if (!use_s_stream)
+	if (!use_s_stream) {
 		state = v4l2_subdev_lock_and_get_active_state(sd);
-	else
+	} else {
+		if (!v4l2_subdev_has_op(sd, video, s_stream))
+			return -ENOIOCTLCMD;
 		state = NULL;
+	}
 
 	/*
 	 * Verify that the requested streams exist and that they are not
Signature · 3 lines
Regards,

Sakari Ailus