Neither lock in struct most_video_dev carries a comment describing what
it protects.

list_lock protects the pending_mbos list. Buffers are added by the
component's rx_completion callback, comp_rx_data(), which is invoked
from the interface driver's completion path (a URB completion handler
in the case of the USB HDM), and are removed from process context by
comp_vdev_read() and comp_vdev_close(). Note that in the comment, since
it is what makes the irq-safe spinlock variants necessary here.

lock is never acquired by this driver directly. comp_register_videodev()
installs it as struct video_device::lock, where v4l2_ioctl_get_lock()
returns it and video_ioctl2() holds it across ioctl dispatch.

No functional change.

Signed-off-by: Godana Emiru <godanaemiru@gmail.com>
---
 drivers/staging/most/video/video.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/staging/most/video/video.c b/drivers/staging/most/video/video.c
index 88892b5db..c354d1820 100644
--- a/drivers/staging/most/video/video.c
+++ b/drivers/staging/most/video/video.c
@@ -33,6 +33,7 @@ struct most_video_dev {
 	bool mute;
 
 	struct list_head pending_mbos;
+	/* protects pending_mbos; taken from rx completion (softirq) */
 	spinlock_t list_lock;
 
 	struct v4l2_device v4l2_dev;
@@ -40,6 +41,7 @@ struct most_video_dev {
 	struct video_device *vdev;
 	unsigned int ctrl_input;
 
+	/* serializes V4L2 ioctls; used as struct video_device::lock */
 	struct mutex lock;
 
 	wait_queue_head_t wait_data;
-- 
2.53.0

On Tue, Sep 01, 2026 at 08:20:55PM +0300, Godana Emiru wrote:
> Neither lock in struct most_video_dev carries a comment describing what
> it protects.
> 
> list_lock protects the pending_mbos list. Buffers are added by the
> component's rx_completion callback, comp_rx_data(), which is invoked
> from the interface driver's completion path (a URB completion handler
> in the case of the USB HDM), and are removed from process context by
> comp_vdev_read() and comp_vdev_close(). Note that in the comment, since
> it is what makes the irq-safe spinlock variants necessary here.
> 
> lock is never acquired by this driver directly. comp_register_videodev()
> installs it as struct video_device::lock, where v4l2_ioctl_get_lock()
> returns it and video_ioctl2() holds it across ioctl dispatch.

What tool did you use to generate this changelog?

thanks,

greg k-h

Hi Greg,
I used an LLM (Claude) to help trace the locking and draft the changelog
for this patch. I reviewed it and sent it under my own Signed-off-by, but
I failed to add the "Assisted-by: LLM" tag required by
Documentation/process/coding-assistants.rst. The same applies to my
earlier rtl8723bs patch. I apologize for the omission.

I'm happy to resend with the correct tags if the patches are otherwise
useful, or drop them if you prefer.

thanks,
Godana

On Thu, Sep 03, 2026 at 04:29:46PM +0300, Godana Emiru wrote:
> Hi Greg,
> I used an LLM (Claude) to help trace the locking and draft the changelog
> for this patch. I reviewed it and sent it under my own Signed-off-by, but
> I failed to add the "Assisted-by: LLM" tag required by
> Documentation/process/coding-assistants.rst. The same applies to my
> earlier rtl8723bs patch. I apologize for the omission.
> 
> I'm happy to resend with the correct tags if the patches are otherwise
> useful, or drop them if you prefer.

Please read this:
	https://lore.kernel.org/r/2026080354-skater-urgent-31b2@gregkh