Message-ID: <20251024105023.GA13023@pendragon.ideasonboard.com>
Date: Fri, 24 Oct 2025 13:50:23 +0300
From: Laurent Pinchart <laurent.pinchart@...asonboard.com>
To: Ricardo Ribalda <ribalda@...omium.org>
Cc: Hans de Goede <hansg@...nel.org>,
	Mauro Carvalho Chehab <mchehab@...nel.org>,
	linux-media@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] media: uvcvideo: Create a specific id namespace for
 output entities

On Wed, Oct 22, 2025 at 01:12:09PM +0000, Ricardo Ribalda wrote:
> Nothing can be connected to an output terminal. Which means that no
> other entity can reference an output terminal as baSourceId.
> 
> Use this fact to move all the output streaming entities, which have no
> controls, to a different namespace.
> 
> The output entities are usually named after the dev_name() of the usb
> device, so there should not be any uAPI change from this change.
> 
> Although with this change we can handle some id collisions
> automagically, change the logic of uvc_alloc_new_entity() to keep
> showing a warning when a camera has invalid descriptors. Hopefully this
> message will help vendors fix their invalid descriptors.

This is one of the commits that I expect we'll go back to in the future
when trying to remember why we namespace the streeaming terminal IDs.
Let's make the rationale very clear, and also mention the name of the
devices we know to be affected.

----
Some devices, such as the Grandstream GUV3100 and the LSK Meeting Eye
for Business & Home, exhibit entity ID collisions between units and
streaming output terminals.

The UVC specification requires unit and terminal IDs to be unique, and
uses the ID to reference entities:

- In control requests, to identify the target entity
- In the UVC units and terminals descriptors' bSourceID field, to
  identify source entities
- In the UVC input header descriptor's bTerminalLink, to identify the
  terminal associated with a streaming interface

Entity ID collisions break accessing controls and make the graph
description in the UVC descriptors ambiguous. However, collisions where
one of the entities is a streaming output terminal and the other entity
is not a streaming terminal are less severe. Streaming output terminals
have no controls, and, as they are the final entity in pipelines, they
are never referenced in descriptors as source entities. They are
referenced by ID only from innput header descriptors, which by
definition only reference streaming terminals.

For these reasons, we can work around the collision by giving streaming
output terminals their own ID namespace. Do so by setting bit
UVC_TERM_OUTPUT (15) in the uvc_entity.id field, which is normally never
set as the ID is a 8-bit value.

This ID change doesn't affect the entity name in the media controller
graph as the name isn't constructed from the ID, so there should not be
any impact on the uAPI.

Although this change handles some ID collisions automagically, keep
printing an error in uvc_alloc_new_entity() when a camera has invalid
descriptors. Hopefully this message will help vendors fix their invalid
descriptors.
----

> 
> Suggested-by: Laurent Pinchart <laurent.pinchart@...asonboard.com>
> Signed-off-by: Ricardo Ribalda <ribalda@...omium.org>
> ---
> Hi, this patch fixes support for some devices with invalid USB
> descriptor.
> 
> It is orthogonal to:
> https://lore.kernel.org/linux-media/20251021184213.GC19043@pendragon.ideasonboard.com/T/#t
> 
> Some devices will be fixed by the other patch, other devices will be
> fixed by this. In my opinion is worth to land both patches.

I would prefer merging this fix only (or rather, if we merge the other
one as a quick fix, reverting it once we merge this one). The rationale
is that enabling non-compliant behaviour should be limited to issues we
know about, otherwise it could encourage making more non-compliant
devices.

The UVC specification states

  Each Unit and Terminal within the video function is assigned a unique
  identification number, the Unit ID (UID) or Terminal ID (TID),
  contained in the bUnitID or bTerminalID field of the descriptor. The
  value 0x00 is reserved for undefined ID, effectively restricting the
  total number of addressable entities in the video function (both Units
  and Terminals) to 255.

To me it's clear that the unit and terminal IDs share the same
namespace, but I suppose some vendors can overlook that. When combined
with the fact that the streaming output terminals do not have controls,
and therefore do not need to be addressable and lookup up by ID for
anything else than matching with a streaming interface, this makes me
think that ID collision between a streaming output terminal and another
unit will likely be the most common case.

> Tested with GRANDSTREAM GUV3100 in a 6.6 kernel.
> ---
> Changes in v2:
> - Change Macro name
> - Apply quirk only to TT_STEAMING
> - Add missing suggested by
> - uvc_stream_for_terminal
> - Note, v2 has not been tested yet in real hardware, only v1.
> - Link to v1: https://lore.kernel.org/r/20251022-uvc-grandstream-laurent-v1-1-0925738a3484@chromium.org
> ---
>  drivers/media/usb/uvc/uvc_driver.c | 31 ++++++++++++++++++++++++-------
>  drivers/media/usb/uvc/uvcvideo.h   |  3 ++-
>  2 files changed, 26 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/media/usb/uvc/uvc_driver.c b/drivers/media/usb/uvc/uvc_driver.c
> index fb6afb8e84f00961f86fd8f840fba48d706d7a9a..c0a2c05b0f13a8c3b14018c47dfb0be2614340ce 100644
> --- a/drivers/media/usb/uvc/uvc_driver.c
> +++ b/drivers/media/usb/uvc/uvc_driver.c
> @@ -165,8 +165,10 @@ static struct uvc_entity *uvc_entity_by_reference(struct uvc_device *dev,
>  	return NULL;
>  }
>  
> -static struct uvc_streaming *uvc_stream_by_id(struct uvc_device *dev, int id)
> +static struct uvc_streaming *uvc_stream_for_terminal(struct uvc_device *dev,
> +						     struct uvc_entity *term)
>  {
> +	u16 id = UVC_HARDWARE_ENTITY_ID(term->id);
>  	struct uvc_streaming *stream;
>  
>  	list_for_each_entry(stream, &dev->streams, list) {
> @@ -810,10 +812,12 @@ static struct uvc_entity *uvc_alloc_new_entity(struct uvc_device *dev, u16 type,
>  	}
>  
>  	/* Per UVC 1.1+ spec 3.7.2, the ID is unique. */
> -	if (uvc_entity_by_id(dev, id)) {
> -		dev_err(&dev->intf->dev, "Found multiple Units with ID %u\n", id);
> +	if (uvc_entity_by_id(dev, UVC_HARDWARE_ENTITY_ID(id)))
> +		dev_err(&dev->intf->dev, "Found multiple Units with ID %u\n",
> +			UVC_HARDWARE_ENTITY_ID(id));
> +
> +	if (uvc_entity_by_id(dev, id))
>  		id = UVC_INVALID_ENTITY_ID;
> -	}
>  
>  	extra_size = roundup(extra_size, sizeof(*entity->pads));
>  	if (num_pads)
> @@ -969,6 +973,7 @@ static int uvc_parse_standard_control(struct uvc_device *dev,
>  	struct usb_host_interface *alts = dev->intf->cur_altsetting;
>  	unsigned int i, n, p, len;
>  	const char *type_name;
> +	unsigned int id;
>  	u16 type;
>  
>  	switch (buffer[2]) {
> @@ -1107,8 +1112,20 @@ static int uvc_parse_standard_control(struct uvc_device *dev,
>  			return 0;
>  		}
>  
> +		id = buffer[3];
> +
> +		/*
> +		 * Nothing can be connected to an output terminal. To avoid
> +		 * entity-id's collisions in devices with invalid USB
> +		 * descriptors, move the output terminal id to its own
> +		 * namespace. Do this only for UVC_TT_STREAMING entities, to
> +		 * avoid changing the id of terminals with controls.
> +		 */

How about expanding this too, similarly to the commit message to avoid
digging all the information from the git history ? I've found that
explaining hacks in more details helped me before when I had to rework
the code years later.

		/*
		 * Some devices, such as the Grandstream GUV3100 and the LSK
		 * Meeting Eye for Business & Home, exhibit entity ID collisions
		 * between units and streaming output terminals. Move streaming
		 * output terminals to their own ID namespace by setting bit
		 * UVC_TT_STREAMING (15), above the ID's 8-bit value. The bit is
		 * ignored in uvc_stream_for_terminal() when looking up the
		 * streaming interface for the terminal.
		 *
		 * This hack is safe to enable unconditionally, asthe ID is not
		 * used for any other purpose (streaming output terminals have
		 * no controls and are never referenced as sources in UVC
		 * descriptors). Other types output terminals can have controls,
		 * so limit usage of this separate namespace to streaming output
		 * terminals.
		 */

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

> +		if (type & UVC_TT_STREAMING)
> +			id |= UVC_TERM_OUTPUT;
> +
>  		term = uvc_alloc_new_entity(dev, type | UVC_TERM_OUTPUT,
> -					    buffer[3], 1, 0);
> +					    id, 1, 0);
>  		if (IS_ERR(term))
>  			return PTR_ERR(term);
>  
> @@ -2105,8 +2122,8 @@ static int uvc_register_terms(struct uvc_device *dev,
>  		if (UVC_ENTITY_TYPE(term) != UVC_TT_STREAMING)
>  			continue;
>  
> -		stream = uvc_stream_by_id(dev, term->id);
> -		if (stream == NULL) {
> +		stream = uvc_stream_for_terminal(dev, term);
> +		if (!stream) {
>  			dev_info(&dev->intf->dev,
>  				 "No streaming interface found for terminal %u.",
>  				 term->id);
> diff --git a/drivers/media/usb/uvc/uvcvideo.h b/drivers/media/usb/uvc/uvcvideo.h
> index ed7bad31f75ca474c1037d666d5310c78dd764df..3f2e832025e712585edc324afa6cad760d4edafc 100644
> --- a/drivers/media/usb/uvc/uvcvideo.h
> +++ b/drivers/media/usb/uvc/uvcvideo.h
> @@ -41,7 +41,8 @@
>  #define UVC_EXT_GPIO_UNIT		0x7ffe
>  #define UVC_EXT_GPIO_UNIT_ID		0x100
>  
> -#define UVC_INVALID_ENTITY_ID          0xffff
> +#define UVC_HARDWARE_ENTITY_ID(id)	((id) & 0xff)
> +#define UVC_INVALID_ENTITY_ID		0xffff
>  
>  /* ------------------------------------------------------------------------
>   * Driver specific constants.
> 
> ---
> base-commit: ea299a2164262ff787c9d33f46049acccd120672
> change-id: 20251022-uvc-grandstream-laurent-3f9abb8a0d5b

-- 
Regards,

Laurent Pinchart

Powered by blists - more mailing lists