Re: [PATCH v3] media: uvcvideo: Recover stalled ElGato devices

[Laurent Pinchart]

Hi Ricardo,

Thank you for the patch.

On Mon, Dec 05, 2022 at 01:53:01PM +0100, Ricardo Ribalda wrote:
> Elgato Cam Link 4k can be in a stalled state if the resolution of
> the external source has changed while the firmware initializes.
> Once in this state, the device is useless until it receives a
> USB reset. It has even been observed that the stalled state will
> continue even after unplugging the device.

Why, ô, why :'-(

> lsusb -v
> 
> Bus 002 Device 002: ID 0fd9:0066 Elgato Systems GmbH Cam Link 4K
> Device Descriptor:
>   bLength                18
>   bDescriptorType         1
>   bcdUSB               3.00
>   bDeviceClass          239 Miscellaneous Device
>   bDeviceSubClass         2
>   bDeviceProtocol         1 Interface Association
>   bMaxPacketSize0         9
>   idVendor           0x0fd9 Elgato Systems GmbH
>   idProduct          0x0066
>   bcdDevice            0.00
>   iManufacturer           1 Elgato
>   iProduct                2 Cam Link 4K
>   iSerial                 4 0005AC52FE000
>   bNumConfigurations      1
> 
> Reviewed-by: Sergey Senozhatsky <senozhatsky@xxxxxxxxxxxx>
> Signed-off-by: Ricardo Ribalda <ribalda@xxxxxxxxxxxx>
> ---
> Recover stalled ElGato devices
> 
> Just a resend of this hw fix.
> 
> To: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
> To: Mauro Carvalho Chehab <mchehab@xxxxxxxxxx>
> Cc: linux-media@xxxxxxxxxxxxxxx
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Cc: Sergey Senozhatsky <senozhatsky@xxxxxxxxxxxx>
> Cc: Yunke Cao <yunkec@xxxxxxxxxxxx>
> ---
> Changes in v3:
> - Add Reviewed-by: Sergey
> - Improve identation (Thanks Sergey!)
> - Link to v2: https://lore.kernel.org/r/20220920-resend-elgato-v2-0-06b48b3b486a@xxxxxxxxxxxx
> 
> Changes in v2:
> - Remove info from lsusb 
> - Link to v1: https://lore.kernel.org/r/20220920-resend-elgato-v1-0-8672a2380e3d@xxxxxxxxxxxx
> ---
>  drivers/media/usb/uvc/uvc_video.c | 23 +++++++++++++++++++----
>  1 file changed, 19 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c
> index 170a008f4006..a8323cf0e9f9 100644
> --- a/drivers/media/usb/uvc/uvc_video.c
> +++ b/drivers/media/usb/uvc/uvc_video.c
> @@ -129,12 +129,13 @@ int uvc_query_ctrl(struct uvc_device *dev, u8 query, u8 unit,
>  	return -EPIPE;
>  }
>  
> +static const struct usb_device_id elgato_cam_link_4k = {
> +	USB_DEVICE(0x0fd9, 0x0066)
> +};
> +
>  static void uvc_fixup_video_ctrl(struct uvc_streaming *stream,
>  	struct uvc_streaming_control *ctrl)
>  {
> -	static const struct usb_device_id elgato_cam_link_4k = {
> -		USB_DEVICE(0x0fd9, 0x0066)
> -	};
>  	struct uvc_format *format = NULL;
>  	struct uvc_frame *frame = NULL;
>  	unsigned int i;
> @@ -297,7 +298,7 @@ static int uvc_get_video_ctrl(struct uvc_streaming *stream,
>  		dev_err(&stream->intf->dev,
>  			"Failed to query (%u) UVC %s control : %d (exp. %u).\n",
>  			query, probe ? "probe" : "commit", ret, size);
> -		ret = -EIO;
> +		ret = (ret == -EPROTO) ? -EPROTO : -EIO;

This will change the error returned by VIDIOC_S_FMT and VIDIOC_TRY_FMT,
but I don't think that will cause issues.

>  		goto out;
>  	}
>  
> @@ -2121,6 +2122,20 @@ int uvc_video_init(struct uvc_streaming *stream)
>  	 * request on the probe control, as required by the UVC specification.
>  	 */
>  	ret = uvc_get_video_ctrl(stream, probe, 1, UVC_GET_CUR);
> +
> +	/*
> +	 * Elgato Cam Link 4k can be in a stalled state if the resolution of
> +	 * the external source has changed while the firmware initializes.
> +	 * Once in this state, the device is useless until it receives a
> +	 * USB reset. It has even been observed that the stalled state will
> +	 * continue even after unplugging the device.
> +	 */
> +	if (usb_match_one_id(stream->dev->intf, &elgato_cam_link_4k) &&
> +	    (ret == -EPROTO)) {

I'd swap the tests, as the second one is cheaper. You also don't need
the inner parentheses.

> +		usb_reset_device(stream->dev->udev);
> +		dev_err(&stream->intf->dev, "Restarting Elgato Cam Link 4K\n");

I would also swap those two lines, if an error occurs when resetting the
device, it's good to have the "restarting" message printed first.

If I understand this correctly, probe will still fail, and the user will
have to unplug and replug the device to fix this, right ? If so I'd
expand the error message, maybe something along those lines:

		dev_err(&stream->intf->dev, "Elgato Cam Link 4K firmware crash detected\n");
		dev_err(&stream->intf->dev, "Resetting the device, unplug and replug to recover\n");

Reviewed-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>

> +	}
> +
>  	if (ret < 0)
>  		return ret;
>  
> 
> ---
> base-commit: 521a547ced6477c54b4b0cc206000406c221b4d6
> change-id: 20220920-resend-elgato-a845482bdd02

-- 
Regards,

Laurent Pinchart