From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: "Pol Fernández Fernández" <fernandezfernandezpol@gmail.com>
Cc: linux-media@vger.kernel.org, linux-usb@vger.kernel.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/2] media: uvcvideo: add quirks for per-device stream error handling
Date: Mon, 29 Jun 2026 21:23:47 +0300 [thread overview]
Message-ID: <20260629182347.GI3054459@killaraus.ideasonboard.com> (raw)
In-Reply-To: <20260629163805.37879-1-fernandezfernandezpol@gmail.com>
On Mon, Jun 29, 2026 at 06:38:02PM +0200, Pol Fernández Fernández wrote:
> The UVC_STREAM_ERR bit in the payload header signals that the device
> could not deliver a frame correctly. By default, uvcvideo delivers
> these frames to userspace with V4L2_BUF_FLAG_ERROR when uvc_no_drop_param
> is set (the default). Applications that rely on strict JPEG decoders
> (e.g. libjpeg) will display gray bands for the incomplete portion of
> the image, since libjpeg fills undecodable MCU rows with a neutral gray
> when the EOI marker is missing.
>
> Add UVC_QUIRK_DROP_STREAM_ERR to force-drop frames flagged with
> UVC_STREAM_ERR for specific devices, regardless of the uvc_no_drop_param
> module parameter. This ensures applications never receive truncated frames
> from devices known to set UVC_STREAM_ERR on genuine encoding errors.
That completely defeats the point of uvc_no_drop_param. The issue should
be fixed in userspace, where you should ignore frames with the error
flag set if desired.
> The existing uvc_queue_to_stream() helper is used to reach the device
> quirks from within uvc_queue_buffer_complete().
>
> Additionally, uvcvideo unconditionally overrides the camera's probed
> wCompQuality with the maximum value returned by GET_MAX. For devices
> whose firmware treats wCompQuality=0 as an adaptive encoding mode (where
> the encoder adjusts quality dynamically to fit within the available USB
> bandwidth),
The UVC 1.1 specification documents the range wCompQuality values as
"Values for this property range from 0 to 10000 (0 indicates the lowest
quality, 10000 the highest)."
While the UVC 1.5 specification documents it as
"Values for this property range from 1 to 10000 (1 indicates the lowest
quality, 10000 the highest)."
There is no mention of 0 being allowed as a magic value. Where did you
find information about the adaptive encoding mode ?
> this override increases encoding pressure and can contribute
> to UVC_STREAM_ERR events. Add UVC_QUIRK_NO_FORCE_QUALITY to preserve
> the camera's default quality value during probe negotiation.
>
> Signed-off-by: Pol Fernández Fernández <fernandezfernandezpol@gmail.com>
Please split this patch in two as it addresses two different issues.
> ---
> drivers/media/usb/uvc/uvc_queue.c | 3 ++-
> drivers/media/usb/uvc/uvc_video.c | 3 ++-
> drivers/media/usb/uvc/uvcvideo.h | 2 ++
> 3 files changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/media/usb/uvc/uvcvideo.h b/drivers/media/usb/uvc/uvcvideo.h
> index XXXXXXX..XXXXXXX 100644
> --- a/drivers/media/usb/uvc/uvcvideo.h
> +++ b/drivers/media/usb/uvc/uvcvideo.h
> @@ -81,6 +81,8 @@
> #define UVC_QUIRK_INVALID_DEVICE_SOF 0x00010000
> #define UVC_QUIRK_MJPEG_NO_EOF 0x00020000
> #define UVC_QUIRK_MSXU_META 0x00040000
> +#define UVC_QUIRK_DROP_STREAM_ERR 0x00080000
> +#define UVC_QUIRK_NO_FORCE_QUALITY 0x00100000
>
> /* Format flags */
> #define UVC_FMT_FLAG_COMPRESSED 0x00000001
> diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c
> index XXXXXXX..XXXXXXX 100644
> --- a/drivers/media/usb/uvc/uvc_video.c
> +++ b/drivers/media/usb/uvc/uvc_video.c
> @@ -451,7 +451,8 @@ int uvc_probe_video(struct uvc_streaming *stream,
> if (ret < 0)
> goto done;
>
> - probe->wCompQuality = probe_max.wCompQuality;
> + if (!(stream->dev->quirks & UVC_QUIRK_NO_FORCE_QUALITY))
> + probe->wCompQuality = probe_max.wCompQuality;
> }
>
> for (i = 0; i < 2; ++i) {
> diff --git a/drivers/media/usb/uvc/uvc_queue.c b/drivers/media/usb/uvc/uvc_queue.c
> index XXXXXXX..XXXXXXX 100644
> --- a/drivers/media/usb/uvc/uvc_queue.c
> +++ b/drivers/media/usb/uvc/uvc_queue.c
> @@ -357,7 +357,8 @@ static void uvc_queue_buffer_complete(struct kref *ref)
> struct vb2_buffer *vb = &buf->buf.vb2_buf;
> struct uvc_video_queue *queue = vb2_get_drv_priv(vb->vb2_queue);
>
> - if (buf->error && !uvc_no_drop_param) {
> + if (buf->error && (!uvc_no_drop_param ||
> + (uvc_queue_to_stream(queue)->dev->quirks & UVC_QUIRK_DROP_STREAM_ERR))) {
> uvc_queue_buffer_requeue(queue, buf);
> return;
> }
--
Regards,
Laurent Pinchart
prev parent reply other threads:[~2026-06-29 18:23 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-29 16:38 [PATCH 1/2] media: uvcvideo: add quirks for per-device stream error handling Pol Fernández Fernández
2026-06-29 16:38 ` [PATCH 2/2] media: uvcvideo: add device entry for Logitech C920 (046d:08e5) Pol Fernández Fernández
2026-06-29 16:55 ` [PATCH 1/2] media: uvcvideo: add quirks for per-device stream error handling Ricardo Ribalda
2026-06-29 18:23 ` Laurent Pinchart [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260629182347.GI3054459@killaraus.ideasonboard.com \
--to=laurent.pinchart@ideasonboard.com \
--cc=fernandezfernandezpol@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox