[v1] usb: gadget: uvc_video: unlock before submitting a request to ep

[PATCH] usb: gadget: uvc_video: unlock before submitting a request to ep

Posted by Piyush Mehta 2 years, 1 month ago

There could be chances where the usb_ep_queue() could fail and trigger
complete() handler with error status. In this case, if usb_ep_queue()
is called with lock held and the triggered complete() handler is waiting
for the same lock to be cleared could result in a deadlock situation and
could result in system hang. To aviod this scenerio, call usb_ep_queue()
with lock removed. This patch does the same.

Signed-off-by: Piyush Mehta <piyush.mehta@amd.com>
---
 drivers/usb/gadget/function/uvc_video.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/usb/gadget/function/uvc_video.c b/drivers/usb/gadget/function/uvc_video.c
index 91af3b1ef0d4..0a5d9ac145e7 100644
--- a/drivers/usb/gadget/function/uvc_video.c
+++ b/drivers/usb/gadget/function/uvc_video.c
@@ -460,11 +460,12 @@ static void uvcg_video_pump(struct work_struct *work)
 			req->no_interrupt = 1;
 		}
 
-		/* Queue the USB request */
-		ret = uvcg_video_ep_queue(video, req);
 		spin_unlock_irqrestore(&queue->irqlock, flags);
 
+		/* Queue the USB request */
+		ret = uvcg_video_ep_queue(video, req);
 		if (ret < 0) {
+			usb_ep_set_halt(video->ep);
 			uvcg_queue_cancel(queue, 0);
 			break;
 		}
-- 
2.25.1

Re: [PATCH] usb: gadget: uvc_video: unlock before submitting a request to ep

Posted by Dan Scally 2 years, 1 month ago

Hi Piyush - thanks for the patch

On 02/11/2023 07:11, Piyush Mehta wrote:
> There could be chances where the usb_ep_queue() could fail and trigger
> complete() handler with error status. In this case, if usb_ep_queue()
> is called with lock held and the triggered complete() handler is waiting
> for the same lock to be cleared could result in a deadlock situation and
> could result in system hang. To aviod this scenerio, call usb_ep_queue()
> with lock removed. This patch does the same.


s/aviod/avoid. s/scenerio/scenario/

>
> Signed-off-by: Piyush Mehta <piyush.mehta@amd.com>
> ---
>   drivers/usb/gadget/function/uvc_video.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/usb/gadget/function/uvc_video.c b/drivers/usb/gadget/function/uvc_video.c
> index 91af3b1ef0d4..0a5d9ac145e7 100644
> --- a/drivers/usb/gadget/function/uvc_video.c
> +++ b/drivers/usb/gadget/function/uvc_video.c
> @@ -460,11 +460,12 @@ static void uvcg_video_pump(struct work_struct *work)
>   			req->no_interrupt = 1;
>   		}
>   
> -		/* Queue the USB request */
> -		ret = uvcg_video_ep_queue(video, req);
>   		spin_unlock_irqrestore(&queue->irqlock, flags);
>   
> +		/* Queue the USB request */
> +		ret = uvcg_video_ep_queue(video, req);
>   		if (ret < 0) {
> +			usb_ep_set_halt(video->ep);


This change isn't mentioned, and shouldn't be necessary - uvcg_video_ep_queue() will already call 
usb_ep_set_halt() if it's in the error path.

>   			uvcg_queue_cancel(queue, 0);
>   			break;
>   		}

Re: [PATCH] usb: gadget: uvc_video: unlock before submitting a request to ep

Posted by Sergey Shtylyov 2 years, 1 month ago

Hello!

On 11/2/23 10:11 AM, Piyush Mehta wrote:

> There could be chances where the usb_ep_queue() could fail and trigger
> complete() handler with error status. In this case, if usb_ep_queue()
> is called with lock held and the triggered complete() handler is waiting
> for the same lock to be cleared could result in a deadlock situation and
> could result in system hang. To aviod this scenerio, call usb_ep_queue()

   Scenario. :-)

> with lock removed. This patch does the same.

   The last sentence is hardly needed.

> Signed-off-by: Piyush Mehta <piyush.mehta@amd.com>
> ---
>  drivers/usb/gadget/function/uvc_video.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/usb/gadget/function/uvc_video.c b/drivers/usb/gadget/function/uvc_video.c
> index 91af3b1ef0d4..0a5d9ac145e7 100644
> --- a/drivers/usb/gadget/function/uvc_video.c
> +++ b/drivers/usb/gadget/function/uvc_video.c
> @@ -460,11 +460,12 @@ static void uvcg_video_pump(struct work_struct *work)
>  			req->no_interrupt = 1;
>  		}
>  
> -		/* Queue the USB request */
> -		ret = uvcg_video_ep_queue(video, req);
>  		spin_unlock_irqrestore(&queue->irqlock, flags);
>  
> +		/* Queue the USB request */
> +		ret = uvcg_video_ep_queue(video, req);
>  		if (ret < 0) {
> +			usb_ep_set_halt(video->ep);

   Hm, you don't say anything about this change in the patch
description...

[...]

MBR, Sergey