Skip to contentSkip to conversation

Generic line based metadata support, internal pads

PATCH Superseded Sakari Ailus 6 participants 149 messages Mar 13, 2024 – Apr 20, 2024 Page 1 of 2

Subsystems (5)

v8 of 9 Superseded Latest is v9 →

lore
Share & tools
lore

Sakari Ailus

author maintainer
#0 lore raw

v8 cover Generic line based metadata support, internal pads

Hi folks,

Here are a few patches to add support generic, line based metadata as well as internal pads. While the amount of code is not very large, to the contrary it is quite small actually IMO, I presume what this is about and why it is being proposed requires some explaining.

Metadata mbus codes and formats have existed for some time in V4L2. They however have been only used by drivers that produce the data itself and effectively this metadata has always been statistics of some sort (at least when it comes to ISPs). What is different here is that we intend to add support for metadata originating from camera sensors.

Camera sensors produce different kinds of metadata, embedded data (usually register address--value pairs used to capture the frame, in a more or less sensor specific format), histograms (in a very sensor specific format), dark pixels etc. The number of these formats is probably going to be about as large as image data formats if not larger, as the image data formats are much better standardised but a smaller subset of them will be supported by V4L2, at least initially but possibly much more in the long run.

Having this many device specific formats would be a major problem for all the other drivers along that pipeline (not to mention the users of those drivers), including bridge (e.g. CSI-2 to parallel) but especially CSI-2 receiver drivers that have DMA: the poor driver developer would not only need to know camera sensor specific formats but to choose the specific packing of that format suitable for the DMA used by the hardware. It is unlikely many of these would ever get tested while being present on the driver API. Also adding new sensors with new embedded data formats would involve updating all bridge and CSI-2 receiver drivers. I don't expect this to be a workable approach.

Instead what I'm proposing is to use specific metadata formats on the sensor devices only, on internal pads (more about those soon) of the sensors, only visible in the UAPI, and then generic mbus formats along the pipeline and finally generic V4L2 metadata formats on the DMAs (specific to bit depth and packing). This would unsnarl the two, defining what data there is (specific mbus code) and how that is transported and packed (generic mbus codes and V4L2 formats).

The user space would be required to "know" the path of that data from the sensor's internal pad to the V4L2 video node. I do not see this as these devices require at least some knowledge of the pipeline, i.e. hardware at hand. Separating what the data means and how it is packed may even be
beneficial: it allows separating code that interprets the data (sensor internal mbus code) from the code that accesses it (packing).

These formats are in practice line based, meaning that there may be padding at the end of the line, depending on the bus as well as the DMA. If non-line based formats are needed, it is always possible to set the "height" field to 1.

The internal (source) pads are an alternative to source routes [1]. The source routes were not universally liked and I do have to say I like re-using existing interface concepts (pads and everything you can do with pads, including access format, selections etc.) wherever it makes sense, instead of duplicating functionality.

Effectively internal source pads behave mostly just like sink pads, but they describe a flow of data that originates from a sub-device instead of arriving to a sub-device. The SUBDEV_S_ROUTING IOCTLs are used to enable and disable routes from internal source pads to sub-device's source pads. The subdev format IOCTLs are usable, too, so one can find which subdev format is available on given internal source pad.

I've also pushed these here and I'll keep updating the branch, I've also included untested OV2740 patches:

<URL:https://git.linuxtv.org/sailus/media_tree.git/log/?h=metadata>

Questions and comments are most welcome.

Preliminary media-ctl and yavta patches can be found here:

<URL:https://git.retiisi.eu/?p=~sailus/yavta.git;a=shortlog;h=refs/heads/metadata> <URL:https://git.retiisi.eu/?p=~sailus/v4l-utils.git;a=shortlog;h=refs/heads/metadata>

I have used IMX219 as an example on routing in a sensor driver in this version.

The patches are on my master branch <URL:https://git.linuxtv.org/sailus/media_tree.git/log/>.

[1] <URL:https://lore.kernel.org/linux-media/[email protected]/>

since v7:

- Add embedded data support for the ov2740 driver.

- Add three patches on top, to add an IMMUTABLE flag to source streams when they cannot be disabled.

- Improved documentation of len_routes and num_routes arguments of [GS]_ROUTING.

- Remove one inclusion of twice-included media/v4l2-fwnode.h in drivers/media/i2c/ccs/ccs-core.c .

- Add missing forward declaration of ccs_internal_ops in drivers/media/i2c/ccs/ccs-core.c .

since v6:

- Improve embedded data UAPI documentation on camera sensors.

- Improve wording of stream glossary entry.

- Improve internal pad flag documentation.

- Fix definition of "data unit" and remove an extra "only" in INTERNAL pad flag description (1st patch).

- Use IMX219 driver in examples consistently.

- Remove the CSI-2 to parallel bridge from the example to simplify the example.

- Minor rewording of some parts of the routing examples.

- Rebase on unified sub-device state information access functions:
<URL:https://lore.kernel.org/linux-media/[email protected]/T/#t>

- In CCS driver, do not maintain current active configuration in driver's device context struct (apart from mbus codes). Rely on sub-device state locking and clean up the code. (Multiple patches towards the end of the set.)

- Arrange the CCS patches early in the set towards the end of the set.

- Move the patch enabling streams API to the end of the set.

- Rework IOCTL argument copying condition for [GS]_ROUTING).

- Handle copying back routes in S_ROUTING, do not rely on G_ROUTING IOCTL implementation.

- Rebase on metadata preparation patchset v6:
<URL:https://lore.kernel.org/linux-media/[email protected]/T/#t>.

since v5:

- Rebase on new set of preparation patches.

- Switch CCS driver from s_stream to enable_streams/disable_streams. Keep streaming state information --- the sensor remains in streaming state if any of the streams is enabled.

- Fix setting mbus code on embedded data in get_frame_desc() op in the CCS driver.

since v4:

- Add a patch to acquire two sub-device states that may use the same lock.

- Add a patch for CCS driver to remove ccs_get_crop_compose() helper.

- Add a patch for CCS driver moving acquiring and releasing the mutex to the s_stream callback.

- Add a patch for CCS driver to rely on sub-device state locking using a single driver-provided lock.

- Fixed calculating minimum number of routes in copying the routes (thanks, Laurent).

- Moved a label in S_ROUTING handling to make Clang happy (hopefully).

- Fixed setting emb_data_ctrl register for CCS embedded data support.

- Rebase on Laurent's cleanup patches.

- Wrap a few long lines.

- Write in embedded data documentation sensor drivers generally don't allow configuring it.

since v3:

- Separate preparation patches from this set.

- Add a definition for "Data unit", a pixel that is not image data and use it instead in format documentation.

- Fix more numbered lists in dev-subdev.rst.

- Remove a redundant definition for V4L2_META_FMT_GENERIC_CSI2_2_24 --- V4L2_META_FMT_GENERIC_CSI2_12 can be used instead.

- Use "X" instead of "p" to denote padding in format documentation.

- Use IMX219 in examples instead of CCS.

- Document that the generic V4L2 CSI-2 metadata formats use padding defined in CSI-2 spec and packing defined in CCS spec.

- Add patches to align [GS]_ROUTING behaviour with V4L2. This means mainly returning configured routes as part of S_ROUTING as well. "len_routes" field is added to denote the length of the array and having more routes than fits in the array is no longer an error. Also added more reserved fields.

- Added trivial support for S_ROUTING (via G_ROUTING implementation) for use in drivers with static-only routes.

- Added helper functions to obtain mbus format as well as crop and compose rectangles that are streams-independent.

- Added a patch to define generic CSI-2 long packet types.

- Removed MEDIA_BUS_FMT_IS_META() macro. It didn't seem useful in the end.

- Use a single CCS embedded data format. The bit depth can be selected using the meta stream on the source pad.

- Fix mbus code numbers (there were holes due to removed redundant formats).

- Fix generic mbus code documentation (byte was being used instead of bit).

- Fix spelling of "length".

- Added a patch to remove v4l2_subdev_enable_streams_api that disables streams API. This should be merged once libcamera support for streams works nicely.

- Don't use strings in printing frame descriptor flags.

- Warn on string truncation in printing frame descriptor.

since v2:

- Add a better example, with formats.

- Add CCS static data media bus codes.

- Added an example demonstrating the use of internal pads. --- Is the level of detail enough for the purpose?

- Improved documentation.

- Added a macro to tell whether a format is a metadata format. (Documentation could be added.)

- A small ReST syntax fix in the same section.

- Drop leftovers of a patch checking for the INTERNAL_SOURCE flag.

since v1:

- Make the new pad flag just "INTERNAL", requiring either SINK or SOURCE pad flag to accompany it. Removed the union in struct v4l2_subdev_route.

- Add the term "stream" to MC glossary.

- Improved and fixed documentation (according to comments).

- Note these formats are little endian.

- Remove 1X8 from the names of the mbus codes. These formats have generally 8 bits per pixel.

- Fix mbus code numbering (had holes in RFC).

- Add new metadata fields to debug prints.

- Fix a minor documentation build issue.

Sakari Ailus (38):
media: mc: Add INTERNAL pad flag
media: Documentation: Add "stream" into glossary
media: uapi: Add generic serial metadata mbus formats
media: uapi: Document which mbus format fields are valid for metadata
media: uapi: Add generic 8-bit metadata format definitions
media: v4l: Support line-based metadata capture
media: Documentation: Additional streams generally don't harm capture
media: Documentation: Document embedded data guidelines for camera
sensors
media: Documentation: v4l: Document internal source pads
media: Documentation: Document S_ROUTING behaviour
media: v4l: subdev: Add a function to lock two sub-device states, use
it
media: v4l: subdev: Move G_ROUTING handling below S_ROUTING
media: v4l: subdev: Copy argument back to user also for S_ROUTING
media: v4l: subdev: Add len_routes field to struct v4l2_subdev_routing
media: v4l: subdev: Return routes set using S_ROUTING
media: v4l: subdev: Allow a larger number of routes than there's room
for
media: v4l: subdev: Add trivial set_routing support
media: ccs: No need to set streaming to false in power off
media: ccs: Use {enable,disable}_streams operations
media: ccs: Track streaming state
media: ccs: Move ccs_validate_csi_data_format up
media: ccs: Support frame descriptors
media: uapi: ccs: Add media bus code for MIPI CCS embedded data
media: ccs: Add support for embedded data stream
media: ccs: Remove ccs_get_crop_compose helper
media: ccs: Rely on sub-device state locking
media: ccs: Compute binning configuration from sub-device state
media: ccs: Compute scaling configuration from sub-device state
media: ccs: Remove which parameter from ccs_propagate
media: Documentation: ccs: Document routing
media: uapi: v4l: subdev: Enable streams API
media: uapi: Add media bus code for ov2740 embedded data
media: ov2740: Switch to {enable,disable}_streams
media: ov2740: Track streaming state
media: ov2740: Add support for embedded data
media: v4l: Add V4L2_SUBDEV_ROUTE_FL_IMMUTABLE sub-device routing flag
media: ccs: Add IMMUTABLE route flag
media: ov2740: Add IMMUTABLE route flag

 .../media/drivers/camera-sensor.rst           |   29 +
 .../userspace-api/media/drivers/ccs.rst       |   34 +-
 .../userspace-api/media/glossary.rst          |   15 +
 .../media/mediactl/media-types.rst            |    8 +
 .../userspace-api/media/v4l/dev-meta.rst      |   15 +
 .../userspace-api/media/v4l/dev-subdev.rst    |  178 ++-
 .../userspace-api/media/v4l/meta-formats.rst  |    1 +
 .../media/v4l/metafmt-generic.rst             |  304 +++++
 .../media/v4l/subdev-formats.rst              |  354 ++++++
 .../media/v4l/vidioc-enum-fmt.rst             |    7 +
 .../media/v4l/vidioc-subdev-g-routing.rst     |   47 +-
 .../media/videodev2.h.rst.exceptions          |    1 +
 drivers/media/i2c/ccs/ccs-core.c              | 1033 +++++++++++------
 drivers/media/i2c/ccs/ccs-quirk.h             |    7 +
 drivers/media/i2c/ccs/ccs.h                   |   27 +-
 drivers/media/i2c/ov2740.c                    |  234 +++-
 drivers/media/mc/mc-entity.c                  |   10 +-
 drivers/media/v4l2-core/v4l2-ioctl.c          |   25 +-
 drivers/media/v4l2-core/v4l2-subdev.c         |  116 +-
 include/media/v4l2-subdev.h                   |   42 +
 include/uapi/linux/media-bus-format.h         |   13 +
 include/uapi/linux/media.h                    |    1 +
 include/uapi/linux/v4l2-mediabus.h            |   18 +-
 include/uapi/linux/v4l2-subdev.h              |   14 +-
 include/uapi/linux/videodev2.h                |   18 +
 25 files changed, 2062 insertions(+), 489 deletions(-)
 create mode 100644 Documentation/userspace-api/media/v4l/metafmt-generic.rst

base-commit: 10ca30599cad9d577bb3f4003c74ed2000a7693e

Signature · 1 line
2.39.2
author maintainer
#1 lore raw

PATCH v8 1/38 media: mc: Add INTERNAL pad flag

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
 .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
 drivers/media/mc/mc-entity.c                           | 10 ++++++++--
 include/uapi/linux/media.h                             |  1 +
 3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +8 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
 .. _MEDIA-PAD-FL-SINK:
 .. _MEDIA-PAD-FL-SOURCE:
 .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
 .. flat-table:: Media pad flags
     :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
 	  enabled links even when this flag isn't set; the absence of the flag
 	  doesn't imply there is none.
 
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.
+
+	  The internal flag may currently be present only in a source pad where
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
 One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
 must be set for every pad.
drivers/media/mc/mc-entity.c +8 -2
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
 		iter->index = i++;
 
 		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
 			ret = -EINVAL;
 			break;
 		}
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
 	for (i = 0; i < entity->num_pads; i++) {
 		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
 			continue;
 
 		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
 		return -EINVAL;
 	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
 		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
 	link = media_add_link(&source->links);
 	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
 #define MEDIA_PAD_FL_SINK			(1U << 0)
 #define MEDIA_PAD_FL_SOURCE			(1U << 1)
 #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
 struct media_pad_desc {
 	__u32 entity;		/* entity ID */
Signature · 1 line
2.39.2
Reviewed-by on v8 1/38
#2 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Quoted 27 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 8 lines
Documentation/userspace-api/media/mediactl/media-types.rst +5 -0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Tomi

Quoted 52 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +8 -2
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
author maintainer
#3 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Quoted 11 lines · Tomi Valkeinen

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 388

Quoted 8 lines
Documentation/userspace-api/media/mediactl/media-types.rst +5 -0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 1 line

Thanks, I'll fix that for v9.

Signature · 1 line
Sakari Ailus
maintainer Reviewed-by
#4 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

In reply to Tomi Valkeinen
Quoted 31 lines · Tomi Valkeinen
Quoted 30 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
Quoted 3 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 387

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +3 -0
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.

I would expand this slightly, as it's the only source of documentation regarding internal pads.

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

Quoted 18 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 2 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 12 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +0 -0
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
Quoted 8 lines

drivers/media/mc/mc-entity.c int media_entity_pads_init(struct media_entity *entity, u16 num_pads, · line 221

Quoted 8 lines
drivers/media/mc/mc-entity.c +3 -1
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}

This may become more readable written as a switch statement:

		const u32 pad_flags_mask = MEDIA_PAD_FL_SINK |
					   MEDIA_PAD_FL_SOURCE |
					   MEDIA_PAD_FL_INTERNAL;

		switch (iter->flags & pad_flags_mask) {
		case MEDIA_PAD_FL_SINK:
		case MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL:
		case MEDIA_PAD_FL_SOURCE:
			break;

		default:
			ret = -EINVAL;
			break;
		}

		if (ret)
			break;

And now that I've written this, I'm not too sure anymore :-) Another option would be

	const u32 pad_flags = iter->flags & (MEDIA_PAD_FL_SINK |
					     MEDIA_PAD_FL_SOURCE |
					     MEDIA_PAD_FL_INTERNAL);

	if (pad_flags != MEDIA_PAD_FL_SINK &&
	    pad_flags != MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL
	    pad_flags != MEDIA_PAD_FL_SOURCE) {
		ret = -EINVAL;
		break;
	}

Up to you.

Quoted 32 lines
Quoted 32 lines
drivers/media/mc/mc-entity.c +5 -1
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#5 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Hi Laurent,

Quoted 32 lines · Laurent Pinchart
Quoted 31 lines · Tomi Valkeinen
Quoted 30 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
Quoted 6 lines
Quoted 3 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 387

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +3 -0
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.

I would expand this slightly, as it's the only source of documentation regarding internal pads.

Patch 9 adds more documentation, this patch is for MC only.

Quoted 5 lines

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

I'd drop the sentence related to sensors.

Quoted 63 lines
Quoted 19 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 12 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +0 -0
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
Quoted 8 lines

drivers/media/mc/mc-entity.c int media_entity_pads_init(struct media_entity *entity, u16 num_pads, · line 221

Quoted 8 lines
drivers/media/mc/mc-entity.c +3 -1
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}

This may become more readable written as a switch statement:

		const u32 pad_flags_mask = MEDIA_PAD_FL_SINK |
					   MEDIA_PAD_FL_SOURCE |
					   MEDIA_PAD_FL_INTERNAL;

		switch (iter->flags & pad_flags_mask) {
		case MEDIA_PAD_FL_SINK:
		case MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL:
		case MEDIA_PAD_FL_SOURCE:
			break;

		default:
			ret = -EINVAL;
			break;
		}

		if (ret)
			break;

And now that I've written this, I'm not too sure anymore :-) Another option would be

	const u32 pad_flags = iter->flags & (MEDIA_PAD_FL_SINK |
					     MEDIA_PAD_FL_SOURCE |
					     MEDIA_PAD_FL_INTERNAL);

	if (pad_flags != MEDIA_PAD_FL_SINK &&
	    pad_flags != MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL
	    pad_flags != MEDIA_PAD_FL_SOURCE) {
		ret = -EINVAL;
		break;
	}
Quoted 1 line

Up to you.

I'd prefer to keep it as-is since the original check is testing two independent things instead of merging them: that only either SINK or SOURCE is set, and then separately that if INTERNAL is set, then SINK is set, too.

Of the two options you suggested, I prefer the latter.

Quoted 2 lines

Thank you!

Quoted 34 lines
Quoted 33 lines
Quoted 33 lines
drivers/media/mc/mc-entity.c +5 -1
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
Signature · 3 lines
Regards,

Sakari Ailus
maintainer
#6 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Quoted 33 lines · Sakari Ailus
Quoted 32 lines · Laurent Pinchart
Quoted 31 lines · Tomi Valkeinen
Quoted 30 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 387

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +3 -0
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.

I would expand this slightly, as it's the only source of documentation regarding internal pads.

Patch 9 adds more documentation, this patch is for MC only.

That's my point, it's the only source of documentation for internal pads from an MC point of view, so expanding the documentation would be good :-)

Quoted 7 lines
Quoted 5 lines

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

I'd drop the sentence related to sensors.

I'm fine with another example, or a more generic explanation, but with that sentence dropped, I think this will leave the reader wondering what an internal pad is and what it's used for.

Quoted 62 lines
Quoted 62 lines
Quoted 18 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 2 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 12 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +0 -0
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
Quoted 8 lines

drivers/media/mc/mc-entity.c int media_entity_pads_init(struct media_entity *entity, u16 num_pads, · line 221

Quoted 8 lines
drivers/media/mc/mc-entity.c +3 -1
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}

This may become more readable written as a switch statement:

		const u32 pad_flags_mask = MEDIA_PAD_FL_SINK |
					   MEDIA_PAD_FL_SOURCE |
					   MEDIA_PAD_FL_INTERNAL;

		switch (iter->flags & pad_flags_mask) {
		case MEDIA_PAD_FL_SINK:
		case MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL:
		case MEDIA_PAD_FL_SOURCE:
			break;

		default:
			ret = -EINVAL;
			break;
		}

		if (ret)
			break;

And now that I've written this, I'm not too sure anymore :-) Another option would be

	const u32 pad_flags = iter->flags & (MEDIA_PAD_FL_SINK |
					     MEDIA_PAD_FL_SOURCE |
					     MEDIA_PAD_FL_INTERNAL);

	if (pad_flags != MEDIA_PAD_FL_SINK &&
	    pad_flags != MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL
	    pad_flags != MEDIA_PAD_FL_SOURCE) {
		ret = -EINVAL;
		break;
	}
Quoted 7 lines
Quoted 1 line

Up to you.

I'd prefer to keep it as-is since the original check is testing two independent things instead of merging them: that only either SINK or SOURCE is set, and then separately that if INTERNAL is set, then SINK is set, too.

Of the two options you suggested, I prefer the latter.

I prefer the latter too, and I think it's more readable than the current code. If we later end up having to test for more rules, we can separate the checks.

Quoted 37 lines
Quoted 2 lines

Thank you!

Quoted 32 lines
Quoted 32 lines
Quoted 32 lines
drivers/media/mc/mc-entity.c +5 -1
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#7 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Hi Laurent,

Quoted 41 lines · Laurent Pinchart
Quoted 40 lines · Sakari Ailus
Quoted 38 lines · Laurent Pinchart
Quoted 31 lines · Tomi Valkeinen
Quoted 30 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
Quoted 3 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 387

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +3 -0
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.

I would expand this slightly, as it's the only source of documentation regarding internal pads.

Quoted 5 lines
Quoted 1 line

Patch 9 adds more documentation, this patch is for MC only.

That's my point, it's the only source of documentation for internal pads from an MC point of view, so expanding the documentation would be good :-)

This is MC documentation, camera sensors aren't relevant in this context.

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

Quoted 5 lines
Quoted 1 line

I'd drop the sentence related to sensors.

I'm fine with another example, or a more generic explanation, but with that sentence dropped, I think this will leave the reader wondering what an internal pad is and what it's used for.

What we could and probably have here is that the internal sink pad indicates a source of data. That's what it really is, whether that data is image data or something else.

So I'd change this to:

The internal flag indicates an internal pad that has no external connections. Such a pad shall not be connected with a link. The internal pad flag is allowed only in conjunction with the sink pad flag. Together the two flags indicate the pad is a source of data inside the entity.

Quoted 69 lines
Quoted 69 lines
Quoted 63 lines
Quoted 19 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 12 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +0 -0
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
Quoted 8 lines

drivers/media/mc/mc-entity.c int media_entity_pads_init(struct media_entity *entity, u16 num_pads, · line 221

Quoted 8 lines
drivers/media/mc/mc-entity.c +3 -1
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}

This may become more readable written as a switch statement:

		const u32 pad_flags_mask = MEDIA_PAD_FL_SINK |
					   MEDIA_PAD_FL_SOURCE |
					   MEDIA_PAD_FL_INTERNAL;

		switch (iter->flags & pad_flags_mask) {
		case MEDIA_PAD_FL_SINK:
		case MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL:
		case MEDIA_PAD_FL_SOURCE:
			break;

		default:
			ret = -EINVAL;
			break;
		}

		if (ret)
			break;

And now that I've written this, I'm not too sure anymore :-) Another option would be

	const u32 pad_flags = iter->flags & (MEDIA_PAD_FL_SINK |
					     MEDIA_PAD_FL_SOURCE |
					     MEDIA_PAD_FL_INTERNAL);

	if (pad_flags != MEDIA_PAD_FL_SINK &&
	    pad_flags != MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL
	    pad_flags != MEDIA_PAD_FL_SOURCE) {
		ret = -EINVAL;
		break;
	}
Quoted 1 line

Up to you.

I'd prefer to keep it as-is since the original check is testing two independent things instead of merging them: that only either SINK or SOURCE is set, and then separately that if INTERNAL is set, then SINK is set, too.

Quoted 5 lines
Quoted 1 line

Of the two options you suggested, I prefer the latter.

I prefer the latter too, and I think it's more readable than the current code. If we later end up having to test for more rules, we can separate the checks.

I can switch to the latter.

Quoted 38 lines
Quoted 38 lines
Quoted 3 lines

Thank you!

Quoted 32 lines
Quoted 32 lines
Quoted 32 lines
drivers/media/mc/mc-entity.c +5 -1
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#8 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Hi Laurent,

Quoted 15 lines · Sakari Ailus
Quoted 10 lines
Quoted 6 lines
Quoted 4 lines

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

I'd drop the sentence related to sensors.

I'm fine with another example, or a more generic explanation, but with that sentence dropped, I think this will leave the reader wondering what an internal pad is and what it's used for.

What we could and probably have here is that the internal sink pad indicates a source of data. That's what it really is, whether that data is image data or something else.

Quoted 6 lines

So I'd change this to:

The internal flag indicates an internal pad that has no external connections. Such a pad shall not be connected with a link. The internal pad flag is allowed only in conjunction with the sink pad flag. Together the two flags indicate the pad is a source of data inside the entity.

A similar text already exists in the following paragraph already so I don't think additions should be needed.

Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 5 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 5 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
Signature · 3 lines
Regards,

Sakari Ailus
maintainer
#9 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

In reply to Sakari Ailus

Hi Sakari,

Quoted 42 lines · Sakari Ailus
Quoted 41 lines · Laurent Pinchart
Quoted 40 lines · Sakari Ailus
Quoted 38 lines · Laurent Pinchart
Quoted 31 lines · Tomi Valkeinen
Quoted 30 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
Quoted 3 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 387

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +3 -0
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.

I would expand this slightly, as it's the only source of documentation regarding internal pads.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Patch 9 adds more documentation, this patch is for MC only.

That's my point, it's the only source of documentation for internal pads from an MC point of view, so expanding the documentation would be good :-)

This is MC documentation, camera sensors aren't relevant in this context.

They are only relevant in the same that they are one of the many possible users of the MC API.

Quoted 16 lines
Quoted 11 lines
Quoted 7 lines
Quoted 5 lines

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

I'd drop the sentence related to sensors.

I'm fine with another example, or a more generic explanation, but with that sentence dropped, I think this will leave the reader wondering what an internal pad is and what it's used for.

What we could and probably have here is that the internal sink pad indicates a source of data. That's what it really is, whether that data is image data or something else.

Quoted 6 lines

So I'd change this to:

The internal flag indicates an internal pad that has no external connections. Such a pad shall not be connected with a link. The internal pad flag is allowed only in conjunction with the sink pad flag. Together the two flags indicate the pad is a source of data inside the entity.

I think that's still difficult to understand. What's the issue with mentioning a particular example to make the concept easier to understand ?

Quoted 113 lines
Quoted 68 lines
Quoted 68 lines
Quoted 62 lines
Quoted 18 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 2 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 12 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +0 -0
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
Quoted 8 lines

drivers/media/mc/mc-entity.c int media_entity_pads_init(struct media_entity *entity, u16 num_pads, · line 221

Quoted 8 lines
drivers/media/mc/mc-entity.c +3 -1
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}

This may become more readable written as a switch statement:

		const u32 pad_flags_mask = MEDIA_PAD_FL_SINK |
					   MEDIA_PAD_FL_SOURCE |
					   MEDIA_PAD_FL_INTERNAL;

		switch (iter->flags & pad_flags_mask) {
		case MEDIA_PAD_FL_SINK:
		case MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL:
		case MEDIA_PAD_FL_SOURCE:
			break;

		default:
			ret = -EINVAL;
			break;
		}

		if (ret)
			break;

And now that I've written this, I'm not too sure anymore :-) Another option would be

	const u32 pad_flags = iter->flags & (MEDIA_PAD_FL_SINK |
					     MEDIA_PAD_FL_SOURCE |
					     MEDIA_PAD_FL_INTERNAL);

	if (pad_flags != MEDIA_PAD_FL_SINK &&
	    pad_flags != MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL
	    pad_flags != MEDIA_PAD_FL_SOURCE) {
		ret = -EINVAL;
		break;
	}
Quoted 1 line

Up to you.

I'd prefer to keep it as-is since the original check is testing two independent things instead of merging them: that only either SINK or SOURCE is set, and then separately that if INTERNAL is set, then SINK is set, too.

Quoted 5 lines
Quoted 1 line

Of the two options you suggested, I prefer the latter.

I prefer the latter too, and I think it's more readable than the current code. If we later end up having to test for more rules, we can separate the checks.

I can switch to the latter.

Quoted 37 lines
Quoted 37 lines
Quoted 2 lines

Thank you!

Quoted 32 lines
Quoted 32 lines
Quoted 32 lines
drivers/media/mc/mc-entity.c +5 -1
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#10 lore raw

Re: [PATCH v8 01/38] media: mc: Add INTERNAL pad flag

Hi Laurent,

Quoted 47 lines · Laurent Pinchart

Hi Sakari,

Quoted 44 lines · Sakari Ailus
Quoted 43 lines · Laurent Pinchart
Quoted 41 lines · Sakari Ailus
Quoted 32 lines · Laurent Pinchart
Quoted 31 lines · Tomi Valkeinen
Quoted 30 lines · Sakari Ailus

Internal source pads will be used as routing endpoints in V4L2 [GS]_ROUTING IOCTLs, to indicate that the stream begins in the entity. Internal source pads are pads that have both SINK and INTERNAL flags set.

Also prevent creating links to pads that have been flagged as internal and initialising SOURCE pads with INTERNAL flag set.

---
  .../userspace-api/media/mediactl/media-types.rst       |  8 ++++++++
  drivers/media/mc/mc-entity.c                           | 10 ++++++++--
  include/uapi/linux/media.h                             |  1 +
  3 files changed, 17 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/mediactl/media-types.rst +1 -0
index 6332e8395263..f55ef055bcf8 100644
@@ -361,6 +361,7 @@ Types and flags used to represent the media graph elements
  .. _MEDIA-PAD-FL-SINK:
  .. _MEDIA-PAD-FL-SOURCE:
  .. _MEDIA-PAD-FL-MUST-CONNECT:
+.. _MEDIA-PAD-FL-INTERNAL:
 
  .. flat-table:: Media pad flags
      :header-rows:  0
@@ -381,6 +382,13 @@ Types and flags used to represent the media graph elements
  	  enabled links even when this flag isn't set; the absence of the flag
  	  doesn't imply there is none.
 
Quoted 6 lines
Quoted 3 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 387

Quoted 3 lines
Documentation/userspace-api/media/mediactl/media-types.rst +3 -0
+    *  -  ``MEDIA_PAD_FL_INTERNAL``
+       -  The internal flag indicates an internal pad that has no external
+	  connections. Such a pad shall not be connected with a link.

I would expand this slightly, as it's the only source of documentation regarding internal pads.

Patch 9 adds more documentation, this patch is for MC only.

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

That's my point, it's the only source of documentation for internal pads from an MC point of view, so expanding the documentation would be good :-)

This is MC documentation, camera sensors aren't relevant in this context.

They are only relevant in the same that they are one of the many possible users of the MC API.

Cameras are mentioned in MC documentation only in generic sense, outside the context of V4L2. The patch is currently in line with existing practices with it comes to MC documentation.

If some V4L2 documentation would be added under MC documentation, it should be discussed (and implemented if so decided) separately from this patch.

There is more documentation later in the series, also relating to internal pads.

Quoted 142 lines
Quoted 17 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

- The internal flag indicates an internal pad that has no external connections. This can be used to model, for instance, the pixel array internal to an image sensor. As they are internal to entities, internal pads shall not be connected with links.

Quoted 5 lines
Quoted 1 line

I'd drop the sentence related to sensors.

I'm fine with another example, or a more generic explanation, but with that sentence dropped, I think this will leave the reader wondering what an internal pad is and what it's used for.

What we could and probably have here is that the internal sink pad indicates a source of data. That's what it really is, whether that data is image data or something else.

Quoted 6 lines

So I'd change this to:

The internal flag indicates an internal pad that has no external connections. Such a pad shall not be connected with a link. The internal pad flag is allowed only in conjunction with the sink pad flag. Together the two flags indicate the pad is a source of data inside the entity.

I think that's still difficult to understand. What's the issue with mentioning a particular example to make the concept easier to understand ?

Quoted 113 lines
Quoted 68 lines
Quoted 68 lines
Quoted 62 lines
Quoted 18 lines

Documentation/userspace-api/media/mediactl/media-types.rst Types and flags used to represent the media graph elements · line 389

Quoted 2 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+
+	  The internal flag may currently be present only in a source pad where

s/source/sink/

Quoted 12 lines
Documentation/userspace-api/media/mediactl/media-types.rst +2 -0
+	  it indicates that the :ref:``stream <media-glossary-stream>``
+	  originates from within the entity.
 
  One and only one of ``MEDIA_PAD_FL_SINK`` and ``MEDIA_PAD_FL_SOURCE``
  must be set for every pad.
drivers/media/mc/mc-entity.c +0 -0
index 0e28b9a7936e..1973e9e1013e 100644
@@ -213,7 +213,9 @@ int media_entity_pads_init(struct media_entity *entity, u16 num_pads,
  		iter->index = i++;
 
Quoted 8 lines

drivers/media/mc/mc-entity.c int media_entity_pads_init(struct media_entity *entity, u16 num_pads, · line 221

Quoted 8 lines
drivers/media/mc/mc-entity.c +3 -1
  		if (hweight32(iter->flags & (MEDIA_PAD_FL_SINK |
-					     MEDIA_PAD_FL_SOURCE)) != 1) {
+					     MEDIA_PAD_FL_SOURCE)) != 1 ||
+		    (iter->flags & MEDIA_PAD_FL_INTERNAL &&
+		     !(iter->flags & MEDIA_PAD_FL_SINK))) {
  			ret = -EINVAL;
  			break;
  		}

This may become more readable written as a switch statement:

		const u32 pad_flags_mask = MEDIA_PAD_FL_SINK |
					   MEDIA_PAD_FL_SOURCE |
					   MEDIA_PAD_FL_INTERNAL;

		switch (iter->flags & pad_flags_mask) {
		case MEDIA_PAD_FL_SINK:
		case MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL:
		case MEDIA_PAD_FL_SOURCE:
			break;

		default:
			ret = -EINVAL;
			break;
		}

		if (ret)
			break;

And now that I've written this, I'm not too sure anymore :-) Another option would be

	const u32 pad_flags = iter->flags & (MEDIA_PAD_FL_SINK |
					     MEDIA_PAD_FL_SOURCE |
					     MEDIA_PAD_FL_INTERNAL);

	if (pad_flags != MEDIA_PAD_FL_SINK &&
	    pad_flags != MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL
	    pad_flags != MEDIA_PAD_FL_SOURCE) {
		ret = -EINVAL;
		break;
	}
Quoted 1 line

Up to you.

I'd prefer to keep it as-is since the original check is testing two independent things instead of merging them: that only either SINK or SOURCE is set, and then separately that if INTERNAL is set, then SINK is set, too.

Quoted 5 lines
Quoted 1 line

Of the two options you suggested, I prefer the latter.

I prefer the latter too, and I think it's more readable than the current code. If we later end up having to test for more rules, we can separate the checks.

I can switch to the latter.

Quoted 37 lines
Quoted 37 lines
Quoted 2 lines

Thank you!

Quoted 32 lines
Quoted 32 lines
Quoted 32 lines
drivers/media/mc/mc-entity.c +5 -1
@@ -1112,7 +1114,8 @@ int media_get_pad_index(struct media_entity *entity, u32 pad_type,
 
  	for (i = 0; i < entity->num_pads; i++) {
  		if ((entity->pads[i].flags &
-		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE)) != pad_type)
+		     (MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_SOURCE |
+		      MEDIA_PAD_FL_INTERNAL)) != pad_type)
  			continue;
 
  		if (entity->pads[i].sig_type == sig_type)
@@ -1142,6 +1145,9 @@ media_create_pad_link(struct media_entity *source, u16 source_pad,
  		return -EINVAL;
  	if (WARN_ON(!(sink->pads[sink_pad].flags & MEDIA_PAD_FL_SINK)))
  		return -EINVAL;
+	if (WARN_ON(source->pads[source_pad].flags & MEDIA_PAD_FL_INTERNAL) ||
+	    WARN_ON(sink->pads[sink_pad].flags & MEDIA_PAD_FL_INTERNAL))
+		return -EINVAL;
 
  	link = media_add_link(&source->links);
  	if (link == NULL)
include/uapi/linux/media.h +1 -0
index 1c80b1d6bbaf..80cfd12a43fc 100644
@@ -208,6 +208,7 @@ struct media_entity_desc {
  #define MEDIA_PAD_FL_SINK			(1U << 0)
  #define MEDIA_PAD_FL_SOURCE			(1U << 1)
  #define MEDIA_PAD_FL_MUST_CONNECT		(1U << 2)
+#define MEDIA_PAD_FL_INTERNAL			(1U << 3)
 
  struct media_pad_desc {
  	__u32 entity;		/* entity ID */
Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#11 lore raw

PATCH v8 2/38 media: Documentation: Add "stream" into glossary

In reply to Sakari Ailus

Add term "stream" to the glossary of the Media subsystem documentation.

---
 Documentation/userspace-api/media/glossary.rst | 6 ++++++
 1 file changed, 6 insertions(+)
Documentation/userspace-api/media/glossary.rst +6 -0
index 96a360edbf3b..ef0ab601b5bf 100644
@@ -173,6 +173,12 @@ Glossary
 	An integrated circuit that integrates all components of a computer
 	or other electronic systems.
 
+_media-glossary-stream:
+    Stream
+	A distinct flow of data (image data or metadata) from an initial source
+	to a final sink. The initial source may be e.g. an image sensor and the
+	final sink e.g. a memory buffer.
+
     V4L2 API
 	**V4L2 userspace API**
 
Signature · 1 line
2.39.2
Reviewed-by on v8 2/38
#12 lore raw

Re: [PATCH v8 02/38] media: Documentation: Add "stream" into glossary

Quoted 14 lines · Sakari Ailus

Add term "stream" to the glossary of the Media subsystem documentation.

---
  Documentation/userspace-api/media/glossary.rst | 6 ++++++
  1 file changed, 6 insertions(+)
Documentation/userspace-api/media/glossary.rst +0 -0
index 96a360edbf3b..ef0ab601b5bf 100644
@@ -173,6 +173,12 @@ Glossary
  	An integrated circuit that integrates all components of a computer
  	or other electronic systems.

Documentation/userspace-api/media/glossary.rst Glossary · line 184

Quoted 10 lines
Documentation/userspace-api/media/glossary.rst +6 -0
 
+_media-glossary-stream:
+    Stream
+	A distinct flow of data (image data or metadata) from an initial source
+	to a final sink. The initial source may be e.g. an image sensor and the
+	final sink e.g. a memory buffer.
+
      V4L2 API
  	**V4L2 userspace API**
 

Tomi

maintainer Reviewed-by on v8 2/38
#13 lore raw

Re: [PATCH v8 02/38] media: Documentation: Add "stream" into glossary

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 12 lines · Sakari Ailus

Add term "stream" to the glossary of the Media subsystem documentation.

---
 Documentation/userspace-api/media/glossary.rst | 6 ++++++
 1 file changed, 6 insertions(+)
Documentation/userspace-api/media/glossary.rst +0 -0
index 96a360edbf3b..ef0ab601b5bf 100644

Documentation/userspace-api/media/glossary.rst Glossary · line 180

Quoted 8 lines
Documentation/userspace-api/media/glossary.rst +5 -0
@@ -173,6 +173,12 @@ Glossary
 	An integrated circuit that integrates all components of a computer
 	or other electronic systems.
 
+_media-glossary-stream:
+    Stream
+	A distinct flow of data (image data or metadata) from an initial source
+	to a final sink. The initial source may be e.g. an image sensor and the
+	final sink e.g. a memory buffer.

I would add

A link may carry multiple streams.

or something similar.

Quoted 4 lines
Documentation/userspace-api/media/glossary.rst +1 -0
+
     V4L2 API
 	**V4L2 userspace API**
 
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#14 lore raw

PATCH v8 3/38 media: uapi: Add generic serial metadata mbus formats

In reply to Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

---
 .../userspace-api/media/glossary.rst          |   9 +
 .../media/v4l/subdev-formats.rst              | 258 ++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |   9 +
 3 files changed, 276 insertions(+)
Documentation/userspace-api/media/glossary.rst +9 -0
index ef0ab601b5bf..7078141894c5 100644
@@ -25,6 +25,15 @@ Glossary
 
 	See :ref:`cec`.
 
+.. _media-glossary-data-unit:
+
+    Data unit
+
+	Unit of data transported by a bus. On parallel buses, the data unit
+	consists of one or more related samples while on serial buses the data
+	unit is logical. If the data unit is image data, it may also be called a
+	pixel.
+
     Device Driver
 	Part of the Linux Kernel that implements support for a hardware
 	component.
Documentation/userspace-api/media/v4l/subdev-formats.rst +258 -0

261 lines not on this page · raw message

include/uapi/linux/media-bus-format.h +9 -0
index f05f747e444d..d4c1d991014b 100644
@@ -174,4 +174,13 @@
  */
 #define MEDIA_BUS_FMT_METADATA_FIXED		0x7001
 
+/* Generic line based metadata formats for serial buses. Next is 0x8008. */
+#define MEDIA_BUS_FMT_META_8			0x8001
+#define MEDIA_BUS_FMT_META_10			0x8002
+#define MEDIA_BUS_FMT_META_12			0x8003
+#define MEDIA_BUS_FMT_META_14			0x8004
+#define MEDIA_BUS_FMT_META_16			0x8005
+#define MEDIA_BUS_FMT_META_20			0x8006
+#define MEDIA_BUS_FMT_META_24			0x8007
+
 #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 1 line
2.39.2
#15 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

Tomi

author maintainer
#16 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Moi,

Quoted 9 lines · Tomi Valkeinen
Quoted 8 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Quoted 5 lines
Quoted 2 lines

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Signature · 3 lines
Terveisin,

Sakari Ailus
#17 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Quoted 15 lines · Sakari Ailus

Moi,

Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

Tomi

maintainer
#18 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Quoted 32 lines · Tomi Valkeinen
Quoted 13 lines · Sakari Ailus
Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

Quoted 5 lines

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

I don't have a strong objection against multiplying the width, but we need to figure out the impact on other kernel space components, as well as on userspace. I suppose the media bus code for the embedded data stream on the sensor source pad when using optimized packing and capturing RAW20 images would be MEDIA_BUS_FMT_META_10 ? In that case I think the sensor driver should be able to handle the width calculations on its own, and the value would just be propagated by userspace.

Signature · 3 lines
Regards,

Laurent Pinchart
maintainer
#19 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Quoted 39 lines · Laurent Pinchart
Quoted 32 lines · Tomi Valkeinen
Quoted 13 lines · Sakari Ailus
Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

Quoted 5 lines

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

Quoted 7 lines

I don't have a strong objection against multiplying the width, but we need to figure out the impact on other kernel space components, as well as on userspace. I suppose the media bus code for the embedded data stream on the sensor source pad when using optimized packing and capturing RAW20 images would be MEDIA_BUS_FMT_META_10 ? In that case I think the sensor driver should be able to handle the width calculations on its own, and the value would just be propagated by userspace.

This should be documented somewhere in this series by the way (not in this patch).

Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#20 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Hi Laurent,

Quoted 48 lines · Laurent Pinchart
Quoted 39 lines · Laurent Pinchart
Quoted 32 lines · Tomi Valkeinen
Quoted 13 lines · Sakari Ailus
Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

Quoted 5 lines

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

Quoted 7 lines

I don't have a strong objection against multiplying the width, but we need to figure out the impact on other kernel space components, as well as on userspace. I suppose the media bus code for the embedded data stream on the sensor source pad when using optimized packing and capturing RAW20 images would be MEDIA_BUS_FMT_META_10 ? In that case I think the sensor driver should be able to handle the width calculations on its own, and the value would just be propagated by userspace.

Quoted 2 lines

This should be documented somewhere in this series by the way (not in this patch).

This could go to the CCS driver documentation. I modified the last paragraph and added a new one:

------8<-----------
Devices supporting embedded data for bit depths greater than or equal to 16 may support more dense packing or legacy single metadata byte per data unit, or both of these. The supported embedded data formats can be enumerated and configured on stream 1 of the source pad (1) of the CCS source sub-device.

The use of the denser packing results in embedded data lines being longer than the pixel data in data units since the data units are smaller. In bytes the embedded data lines are still not longer than the image data lines. ------8<-----------

I believe the reason for the specs requiring embedded data lines not being longer (in bytes) is most likely that some hardware may have issues reciving the data otherwise for various reasons.

Signature · 3 lines
Regards,

Sakari Ailus
maintainer
#21 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Quoted 61 lines · Sakari Ailus
Quoted 48 lines · Laurent Pinchart
Quoted 39 lines · Laurent Pinchart
Quoted 32 lines · Tomi Valkeinen
Quoted 13 lines · Sakari Ailus
Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

Quoted 5 lines

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

Quoted 7 lines

I don't have a strong objection against multiplying the width, but we need to figure out the impact on other kernel space components, as well as on userspace. I suppose the media bus code for the embedded data stream on the sensor source pad when using optimized packing and capturing RAW20 images would be MEDIA_BUS_FMT_META_10 ? In that case I think the sensor driver should be able to handle the width calculations on its own, and the value would just be propagated by userspace.

Quoted 2 lines

This should be documented somewhere in this series by the way (not in this patch).

This could go to the CCS driver documentation. I modified the last paragraph and added a new one:

------8<-----------
Devices supporting embedded data for bit depths greater than or equal to 16 may support more dense packing or legacy single metadata byte per data unit, or both of these. The supported embedded data formats can be enumerated and configured on stream 1 of the source pad (1) of the CCS source sub-device.

Quoted 3 lines

The use of the denser packing results in embedded data lines being longer than the pixel data in data units since the data units are smaller. In bytes the embedded data lines are still not longer than the image data lines.

Please document explicitly that e.g. V4L2_META_FMT_GENERIC_CSI2_10 is used for the RAW20 denser packing (in a general way that covers the other formats too). You should also explain more explicitly that the width is doubled in the relevant uAPI data structures.

This is not limited to CCS but is applicable to other sensors too, so I'd like that documentation to be in a more generic place.

Quoted 5 lines

------8<-----------

I believe the reason for the specs requiring embedded data lines not being longer (in bytes) is most likely that some hardware may have issues reciving the data otherwise for various reasons.

Signature · 3 lines
Regards,

Laurent Pinchart
#22 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

In reply to Laurent Pinchart
Quoted 39 lines · Laurent Pinchart
Quoted 32 lines · Tomi Valkeinen
Quoted 13 lines · Sakari Ailus
Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

Quoted 5 lines

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

Quoted 7 lines

I don't have a strong objection against multiplying the width, but we need to figure out the impact on other kernel space components, as well as on userspace. I suppose the media bus code for the embedded data stream on the sensor source pad when using optimized packing and capturing RAW20 images would be MEDIA_BUS_FMT_META_10 ? In that case I think the sensor driver should be able to handle the width calculations on its own, and the value would just be propagated by userspace.

Yes, I think it works. I just find it more logical if the widths of both the video and embedded streams are the same (which is the case for all other embedded formats).

However, even the CCS spec says "for RAW16, RAW20, and/or RAW24 Visible pixels, top-embedded data may instead be more optimally packed using RAW8, RAW10, and/or RAW12 pixels", so that's in line with what Sakari suggests. Although the spec specifically says "top-embedded", so does it mean that the optimized data is not allowed for bottom data?

Tomi

author maintainer
#23 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Moi,

Quoted 52 lines · Tomi Valkeinen
Quoted 39 lines · Laurent Pinchart
Quoted 32 lines · Tomi Valkeinen
Quoted 13 lines · Sakari Ailus
Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

Quoted 5 lines

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

Quoted 7 lines

I don't have a strong objection against multiplying the width, but we need to figure out the impact on other kernel space components, as well as on userspace. I suppose the media bus code for the embedded data stream on the sensor source pad when using optimized packing and capturing RAW20 images would be MEDIA_BUS_FMT_META_10 ? In that case I think the sensor driver should be able to handle the width calculations on its own, and the value would just be propagated by userspace.

Yes, I think it works. I just find it more logical if the widths of both the video and embedded streams are the same (which is the case for all other embedded formats).

Quoted 5 lines

However, even the CCS spec says "for RAW16, RAW20, and/or RAW24 Visible pixels, top-embedded data may instead be more optimally packed using RAW8, RAW10, and/or RAW12 pixels", so that's in line with what Sakari suggests. Although the spec specifically says "top-embedded", so does it mean that the optimized data is not allowed for bottom data?

I haven't read the spec regarding this but I don't think there should be anything that prevents it: it's the embedded data format. The CCS driver patches only adds support for the top embedded data anyway.

Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#24 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

In reply to Tomi Valkeinen

Moi,

Quoted 29 lines · Tomi Valkeinen
Quoted 15 lines · Sakari Ailus

Moi,

Quoted 12 lines · Tomi Valkeinen
Quoted 10 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

Quoted 7 lines
Quoted 2 lines

The CCS spec talks about legacy packing and optimized packing for 16+ bit formats. You cover only the "legacy" ones here. Did you look at those?

The reason is that the bus data layout of the new packing at higher bit depth matches with packing at lower bit depths (half to be precise). That's why there's no need to define formats for the new packing methods at higher bit depths (the driver simply uses the packing at half of the bit depth).

Hmm. If we're capturing 10-bit raw format, say, with the width of 640 pixels, we'll configure the video stream format according to those. For the embedded stream, we'll set it to V4L2_META_FMT_GENERIC_CSI2_10 and 640 width, right?

Quoted 7 lines

If we're capturing 20-bit raw, we'll configure the video stream format again accordingly, width to 640, and 20 bit fourcc/mbus code. If the embedded stream uses the "legacy" packing, we'll set the format to V4L2_META_FMT_GENERIC_CSI2_20 with width of 640, right?

But if it's using packed format for the embedded stream, we set the format to V4L2_META_FMT_GENERIC_CSI2_10 and width to 1280?

The driver sets the embedded data sub-device format apart from the mbus code. The width would be 1280 in this case, yes, and fewer embedded data lines could be needed.

Quoted 3 lines

Considering that the video and (line-based) embedded data come from the same source, I'd expect the widths to be the same.

The width in bytes (with the packing) of embedded data is still the same as for the image data. Pixels don't matter in this case.

Signature · 3 lines
Terveisin,

Sakari Ailus
maintainer Reviewed-by on v8 3/38
#25 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 24 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

---
 .../userspace-api/media/glossary.rst          |   9 +
 .../media/v4l/subdev-formats.rst              | 258 ++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |   9 +
 3 files changed, 276 insertions(+)
Documentation/userspace-api/media/glossary.rst +0 -0
index ef0ab601b5bf..7078141894c5 100644
@@ -25,6 +25,15 @@ Glossary
 

Documentation/userspace-api/media/glossary.rst Glossary · line 35

Quoted 10 lines
Documentation/userspace-api/media/glossary.rst +8 -0
 	See :ref:`cec`.
 
+.. _media-glossary-data-unit:
+
+    Data unit
+
+	Unit of data transported by a bus. On parallel buses, the data unit
+	consists of one or more related samples while on serial buses the data
+	unit is logical. If the data unit is image data, it may also be called a
+	pixel.

I'm pretty sure nobody will be able to understand what this means, but I don't have a better proposal at the moment.

Quoted 13 lines
Documentation/userspace-api/media/glossary.rst +1 -0
+
     Device Driver
 	Part of the Linux Kernel that implements support for a hardware
 	component.
Documentation/userspace-api/media/v4l/subdev-formats.rst +1 -0
index eb3cd20b0cf2..cbd475f7cae9 100644
@@ -8306,3 +8306,261 @@ The following table lists the existing metadata formats.
 	both sides of the link and the bus format is a fixed
 	metadata format that is not configurable from userspace.
 	Width and height will be set to 0 for this format.
+

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8319

Quoted 10 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +10 -0
+Generic Serial Metadata Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Generic serial metadata formats are used on serial buses where the actual data
+content is more or less device specific but the data is transmitted and received
+by multiple devices that do not process the data in any way, simply writing
+it to system memory for processing in software at the end of the pipeline.
+
+The more specific variant describing the actual data is used on the internal
+source pad of the originating sub-device.

Maybe this paragraph would be best added in the patch that adds the specific metadata formats, you could then mention one of them as an example:

The exact format of the data generated by the device is reported on the internal source pad of the originating sub-device, using one of the more specific metadata formats such as MEDIA_BUS_FMT_CCS_EMBEDDED.

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8321

Quoted 2 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +2 -0
+
+"b" in an array cell signifies a byte of data, followed by the number of the bit

s/bit$/byte/

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8322

Quoted 1 line
Documentation/userspace-api/media/v4l/subdev-formats.rst +1 -0
+and finally the bit number in subscript. "X" indicates a padding bit.

We use a lower-case x in pixfmt-rgb.rst, I would do the same here. We also use single quotes there, turning "b" and "x" into 'b' and 'x'.

Quoted 262 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +244 -0
+
+.. _media-bus-format-generic-meta:
+
+.. cssclass: longtable
+
+.. flat-table:: Generic Serial Metadata Formats
+    :header-rows:  2
+    :stub-columns: 0
+
+    * - Identifier
+      - Code
+      -
+      - :cspan:`23` Data organization within bus ``Data unit
+	<media-glossary-data-unit>``
+    * -
+      -
+      - Bit
+      - 23
+      - 22
+      - 21
+      - 20
+      - 19
+      - 18
+      - 17
+      - 16
+      - 15
+      - 14
+      - 13
+      - 12
+      - 11
+      - 10
+      - 9
+      - 8
+      - 7
+      - 6
+      - 5
+      - 4
+      - 3
+      - 2
+      - 1
+      - 0
+    * .. _MEDIA-BUS-FMT-META-8:
+
+      - MEDIA_BUS_FMT_META_8
+      - 0x8001
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+    * .. _MEDIA-BUS-FMT-META-10:
+
+      - MEDIA_BUS_FMT_META_10
+      - 0x8002
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+      - X
+      - X
+    * .. _MEDIA-BUS-FMT-META-12:
+
+      - MEDIA_BUS_FMT_META_12
+      - 0x8003
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+      - X
+      - X
+      - X
+      - X
+    * .. _MEDIA-BUS-FMT-META-14:
+
+      - MEDIA_BUS_FMT_META_14
+      - 0x8004
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+    * .. _MEDIA-BUS-FMT-META-16:
+
+      - MEDIA_BUS_FMT_META_16
+      - 0x8005
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+    * .. _MEDIA-BUS-FMT-META-20:
+
+      - MEDIA_BUS_FMT_META_20
+      - 0x8006
+      -
+      -
+      -
+      -
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+    * .. _MEDIA-BUS-FMT-META-24:
+
+      - MEDIA_BUS_FMT_META_24
+      - 0x8007
+      -
+      - b0\ :sub:`7`
+      - b0\ :sub:`6`
+      - b0\ :sub:`5`
+      - b0\ :sub:`4`
+      - b0\ :sub:`3`
+      - b0\ :sub:`2`
+      - b0\ :sub:`1`
+      - b0\ :sub:`0`
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
+      - X
include/uapi/linux/media-bus-format.h +9 -0
index f05f747e444d..d4c1d991014b 100644
@@ -174,4 +174,13 @@
  */
 #define MEDIA_BUS_FMT_METADATA_FIXED		0x7001
 
+/* Generic line based metadata formats for serial buses. Next is 0x8008. */
+#define MEDIA_BUS_FMT_META_8			0x8001
+#define MEDIA_BUS_FMT_META_10			0x8002
+#define MEDIA_BUS_FMT_META_12			0x8003
+#define MEDIA_BUS_FMT_META_14			0x8004
+#define MEDIA_BUS_FMT_META_16			0x8005
+#define MEDIA_BUS_FMT_META_20			0x8006
+#define MEDIA_BUS_FMT_META_24			0x8007
+
 #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#26 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Hi Laurent,

Quoted 71 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 24 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

---
 .../userspace-api/media/glossary.rst          |   9 +
 .../media/v4l/subdev-formats.rst              | 258 ++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |   9 +
 3 files changed, 276 insertions(+)
Documentation/userspace-api/media/glossary.rst +0 -0
index ef0ab601b5bf..7078141894c5 100644
@@ -25,6 +25,15 @@ Glossary
 

Documentation/userspace-api/media/glossary.rst Glossary · line 35

Quoted 10 lines
Documentation/userspace-api/media/glossary.rst +8 -0
 	See :ref:`cec`.
 
+.. _media-glossary-data-unit:
+
+    Data unit
+
+	Unit of data transported by a bus. On parallel buses, the data unit
+	consists of one or more related samples while on serial buses the data
+	unit is logical. If the data unit is image data, it may also be called a
+	pixel.

I'm pretty sure nobody will be able to understand what this means, but I don't have a better proposal at the moment.

Quoted 13 lines
Documentation/userspace-api/media/glossary.rst +1 -0
+
     Device Driver
 	Part of the Linux Kernel that implements support for a hardware
 	component.
Documentation/userspace-api/media/v4l/subdev-formats.rst +1 -0
index eb3cd20b0cf2..cbd475f7cae9 100644
@@ -8306,3 +8306,261 @@ The following table lists the existing metadata formats.
 	both sides of the link and the bus format is a fixed
 	metadata format that is not configurable from userspace.
 	Width and height will be set to 0 for this format.
+

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8319

Quoted 10 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +10 -0
+Generic Serial Metadata Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Generic serial metadata formats are used on serial buses where the actual data
+content is more or less device specific but the data is transmitted and received
+by multiple devices that do not process the data in any way, simply writing
+it to system memory for processing in software at the end of the pipeline.
+
+The more specific variant describing the actual data is used on the internal
+source pad of the originating sub-device.

Maybe this paragraph would be best added in the patch that adds the specific metadata formats, you could then mention one of them as an example:

Quoted 3 lines

The exact format of the data generated by the device is reported on the internal source pad of the originating sub-device, using one of the more specific metadata formats such as MEDIA_BUS_FMT_CCS_EMBEDDED.

I'd really like to get rid of the "internal source" pads as the naming is really confusing (it's present still in this version but not in many locations). They're sink pads after all, so I'd call them such. In a few locations there's text that explains they do represent sources of data within the sub-device itself.

Quoted 5 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8321

Quoted 3 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +2 -0
+
+"b" in an array cell signifies a byte of data, followed by the number of the bit

s/bit$/byte/

Uh, yes.

Quoted 5 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8322

Quoted 2 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +1 -0
+and finally the bit number in subscript. "X" indicates a padding bit.

We use a lower-case x in pixfmt-rgb.rst, I would do the same here. We also use single quotes there, turning "b" and "x" into 'b' and 'x'.

This is documentation, not C source code where we'd want to denote a single character. Double quotes should thus be used instead.

I'll switch to lower case X.

Quoted 2 lines

Thank you!

Signature · 3 lines
Regards,

Sakari Ailus
maintainer
#27 lore raw

Re: [PATCH v8 03/38] media: uapi: Add generic serial metadata mbus formats

Hi Sakari,

Quoted 72 lines · Sakari Ailus
Quoted 67 lines · Laurent Pinchart
Quoted 24 lines · Sakari Ailus

Add generic serial metadata mbus formats. These formats describe data width and packing but not the content itself. The reason for specifying such formats is that the formats as such are fairly device specific but they are still handled by CSI-2 receiver drivers that should not be aware of device specific formats. What makes generic metadata formats possible is that these formats are parsed by software only, after capturing the data to system memory.

Also add a definition for "Data unit" to cover what is essentially a pixel but is not image data.

---
 .../userspace-api/media/glossary.rst          |   9 +
 .../media/v4l/subdev-formats.rst              | 258 ++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |   9 +
 3 files changed, 276 insertions(+)
Documentation/userspace-api/media/glossary.rst +0 -0
index ef0ab601b5bf..7078141894c5 100644
@@ -25,6 +25,15 @@ Glossary
 

Documentation/userspace-api/media/glossary.rst Glossary · line 35

Quoted 10 lines
Documentation/userspace-api/media/glossary.rst +8 -0
 	See :ref:`cec`.
 
+.. _media-glossary-data-unit:
+
+    Data unit
+
+	Unit of data transported by a bus. On parallel buses, the data unit
+	consists of one or more related samples while on serial buses the data
+	unit is logical. If the data unit is image data, it may also be called a
+	pixel.

I'm pretty sure nobody will be able to understand what this means, but I don't have a better proposal at the moment.

Quoted 13 lines
Documentation/userspace-api/media/glossary.rst +1 -0
+
     Device Driver
 	Part of the Linux Kernel that implements support for a hardware
 	component.
Documentation/userspace-api/media/v4l/subdev-formats.rst +1 -0
index eb3cd20b0cf2..cbd475f7cae9 100644
@@ -8306,3 +8306,261 @@ The following table lists the existing metadata formats.
 	both sides of the link and the bus format is a fixed
 	metadata format that is not configurable from userspace.
 	Width and height will be set to 0 for this format.
+

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8319

Quoted 10 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +10 -0
+Generic Serial Metadata Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Generic serial metadata formats are used on serial buses where the actual data
+content is more or less device specific but the data is transmitted and received
+by multiple devices that do not process the data in any way, simply writing
+it to system memory for processing in software at the end of the pipeline.
+
+The more specific variant describing the actual data is used on the internal
+source pad of the originating sub-device.

Maybe this paragraph would be best added in the patch that adds the specific metadata formats, you could then mention one of them as an example:

Quoted 3 lines

The exact format of the data generated by the device is reported on the internal source pad of the originating sub-device, using one of the more specific metadata formats such as MEDIA_BUS_FMT_CCS_EMBEDDED.

Quoted 5 lines

I'd really like to get rid of the "internal source" pads as the naming is really confusing (it's present still in this version but not in many locations). They're sink pads after all, so I'd call them such. In a few locations there's text that explains they do represent sources of data within the sub-device itself.

I thought we had agreed to use the term "internal sink pad", but I saw a patch in this series that consistently makes use of "internal source pad". Both are confusing for different reasons, but regardless of which one we pick (that discussion belongs to the other patch that documents the internal source pads), please make sure to use the same terms consistently through the series in v9.

Quoted 7 lines
Quoted 4 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8321

Quoted 2 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +2 -0
+
+"b" in an array cell signifies a byte of data, followed by the number of the bit

s/bit$/byte/

Uh, yes.

Quoted 7 lines
Quoted 4 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst The following table lists the existing metadata formats. · line 8322

Quoted 1 line
Documentation/userspace-api/media/v4l/subdev-formats.rst +1 -0
+and finally the bit number in subscript. "X" indicates a padding bit.

We use a lower-case x in pixfmt-rgb.rst, I would do the same here. We also use single quotes there, turning "b" and "x" into 'b' and 'x'.

This is documentation, not C source code where we'd want to denote a single character. Double quotes should thus be used instead.

We seem to use single quotes in format documentations, I'd continue doing so for consistency.

Quoted 5 lines

I'll switch to lower case X.

Quoted 1 line

Thank you!

Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#28 lore raw

PATCH v8 4/38 media: uapi: Document which mbus format fields are valid for metadata

In reply to Sakari Ailus

Now that metadata mbus formats have been added, it is necessary to define which fields in struct v4l2_mbus_format are applicable to them (not many).

---
 include/uapi/linux/v4l2-mediabus.h | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
include/uapi/linux/v4l2-mediabus.h +12 -6
index 6b07b73473b5..de1d6161bf62 100644
@@ -19,12 +19,18 @@
  * @width:	image width
  * @height:	image height
  * @code:	data format code (from enum v4l2_mbus_pixelcode)
- * @field:	used interlacing type (from enum v4l2_field)
- * @colorspace:	colorspace of the data (from enum v4l2_colorspace)
- * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding)
- * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding)
- * @quantization: quantization of the data (from enum v4l2_quantization)
- * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func)
+ * @field:	used interlacing type (from enum v4l2_field), zero on metadata
+ *		mbus codes
+ * @colorspace:	colorspace of the data (from enum v4l2_colorspace), zero on
+ *		metadata mbus codes
+ * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding), zero
+ *		on metadata mbus codes
+ * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding), zero on
+ *		metadata mbus codes
+ * @quantization: quantization of the data (from enum v4l2_quantization), zero
+ *		on metadata mbus codes
+ * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func), zero
+ *		on metadata mbus codes
  * @flags:	flags (V4L2_MBUS_FRAMEFMT_*)
  * @reserved:  reserved bytes that can be later used
  */
Signature · 1 line
2.39.2
Reviewed-by on v8 4/38
#29 lore raw

Re: [PATCH v8 04/38] media: uapi: Document which mbus format fields are valid for metadata

Quoted 27 lines · Sakari Ailus

Now that metadata mbus formats have been added, it is necessary to define which fields in struct v4l2_mbus_format are applicable to them (not many).

---
  include/uapi/linux/v4l2-mediabus.h | 18 ++++++++++++------
  1 file changed, 12 insertions(+), 6 deletions(-)
include/uapi/linux/v4l2-mediabus.h +5 -6
index 6b07b73473b5..de1d6161bf62 100644
@@ -19,12 +19,18 @@
   * @width:	image width
   * @height:	image height
   * @code:	data format code (from enum v4l2_mbus_pixelcode)
- * @field:	used interlacing type (from enum v4l2_field)
- * @colorspace:	colorspace of the data (from enum v4l2_colorspace)
- * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding)
- * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding)
- * @quantization: quantization of the data (from enum v4l2_quantization)
- * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func)
+ * @field:	used interlacing type (from enum v4l2_field), zero on metadata
+ *		mbus codes
+ * @colorspace:	colorspace of the data (from enum v4l2_colorspace), zero on
+ *		metadata mbus codes
+ * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding), zero

include/uapi/linux/v4l2-mediabus.h line 36

Quoted 10 lines
include/uapi/linux/v4l2-mediabus.h +7 -0
+ *		on metadata mbus codes
+ * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding), zero on
+ *		metadata mbus codes
+ * @quantization: quantization of the data (from enum v4l2_quantization), zero
+ *		on metadata mbus codes
+ * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func), zero
+ *		on metadata mbus codes
   * @flags:	flags (V4L2_MBUS_FRAMEFMT_*)
   * @reserved:  reserved bytes that can be later used
   */

Tomi

maintainer
#30 lore raw

Re: [PATCH v8 04/38] media: uapi: Document which mbus format fields are valid for metadata

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 14 lines · Sakari Ailus

Now that metadata mbus formats have been added, it is necessary to define which fields in struct v4l2_mbus_format are applicable to them (not many).

---
 include/uapi/linux/v4l2-mediabus.h | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
include/uapi/linux/v4l2-mediabus.h +0 -0
index 6b07b73473b5..de1d6161bf62 100644
@@ -19,12 +19,18 @@
  * @width:	image width

include/uapi/linux/v4l2-mediabus.h line 23

Quoted 10 lines
include/uapi/linux/v4l2-mediabus.h +2 -6
  * @height:	image height
  * @code:	data format code (from enum v4l2_mbus_pixelcode)
- * @field:	used interlacing type (from enum v4l2_field)
- * @colorspace:	colorspace of the data (from enum v4l2_colorspace)
- * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding)
- * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding)
- * @quantization: quantization of the data (from enum v4l2_quantization)
- * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func)
+ * @field:	used interlacing type (from enum v4l2_field), zero on metadata
+ *		mbus codes

I would write "zero for metadata formats". Up to you.

Shouldn't you also update Documentation/userspace-api/media/v4l/subdev-formats.rst ?

Quoted 13 lines
include/uapi/linux/v4l2-mediabus.h +10 -0
+ * @colorspace:	colorspace of the data (from enum v4l2_colorspace), zero on
+ *		metadata mbus codes
+ * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding), zero
+ *		on metadata mbus codes
+ * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding), zero on
+ *		metadata mbus codes
+ * @quantization: quantization of the data (from enum v4l2_quantization), zero
+ *		on metadata mbus codes
+ * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func), zero
+ *		on metadata mbus codes
  * @flags:	flags (V4L2_MBUS_FRAMEFMT_*)
  * @reserved:  reserved bytes that can be later used
  */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#31 lore raw

Re: [PATCH v8 04/38] media: uapi: Document which mbus format fields are valid for metadata

Hi Laurent,

Quoted 23 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 18 lines · Sakari Ailus

Now that metadata mbus formats have been added, it is necessary to define which fields in struct v4l2_mbus_format are applicable to them (not many).

---
 include/uapi/linux/v4l2-mediabus.h | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
include/uapi/linux/v4l2-mediabus.h +0 -2
index 6b07b73473b5..de1d6161bf62 100644
@@ -19,12 +19,18 @@
  * @width:	image width
  * @height:	image height
  * @code:	data format code (from enum v4l2_mbus_pixelcode)
- * @field:	used interlacing type (from enum v4l2_field)
- * @colorspace:	colorspace of the data (from enum v4l2_colorspace)
Quoted 8 lines

include/uapi/linux/v4l2-mediabus.h line 23

Quoted 6 lines
include/uapi/linux/v4l2-mediabus.h +2 -4
- * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding)
- * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding)
- * @quantization: quantization of the data (from enum v4l2_quantization)
- * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func)
+ * @field:	used interlacing type (from enum v4l2_field), zero on metadata
+ *		mbus codes

I would write "zero for metadata formats". Up to you.

The code field determines this, and elsewhere it's called mbus code. I prefer thus "mbus code".

Quoted 3 lines

Shouldn't you also update Documentation/userspace-api/media/v4l/subdev-formats.rst ?

Good point. I'll do that for v9, with similar additions than here.

Quoted 14 lines
Quoted 14 lines
include/uapi/linux/v4l2-mediabus.h +10 -0
+ * @colorspace:	colorspace of the data (from enum v4l2_colorspace), zero on
+ *		metadata mbus codes
+ * @ycbcr_enc:	YCbCr encoding of the data (from enum v4l2_ycbcr_encoding), zero
+ *		on metadata mbus codes
+ * @hsv_enc:	HSV encoding of the data (from enum v4l2_hsv_encoding), zero on
+ *		metadata mbus codes
+ * @quantization: quantization of the data (from enum v4l2_quantization), zero
+ *		on metadata mbus codes
+ * @xfer_func:  transfer function of the data (from enum v4l2_xfer_func), zero
+ *		on metadata mbus codes
  * @flags:	flags (V4L2_MBUS_FRAMEFMT_*)
  * @reserved:  reserved bytes that can be later used
  */
Signature · 3 lines
Kind regards,

Sakari Ailus
author maintainer
#32 lore raw

PATCH v8 5/38 media: uapi: Add generic 8-bit metadata format definitions

In reply to Sakari Ailus

Generic 8-bit metadata formats define the in-memory data layout but not the format of the data itself. The reasoning for having such formats is to allow CSI-2 receiver drivers to receive and DMA drivers to write the data to memory without knowing a large number of device specific formats.

These formats may be used only in conjunction of a Media controller pipeline where the internal pad of the source sub-device defines the specific format of the data (using an mbus code).

---
 .../userspace-api/media/v4l/dev-subdev.rst    |   2 +
 .../userspace-api/media/v4l/meta-formats.rst  |   1 +
 .../media/v4l/metafmt-generic.rst             | 304 ++++++++++++++++++
 drivers/media/v4l2-core/v4l2-ioctl.c          |   7 +
 include/uapi/linux/videodev2.h                |   8 +
 5 files changed, 322 insertions(+)
 create mode 100644 Documentation/userspace-api/media/v4l/metafmt-generic.rst
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
index 43988516acdd..f375b820ab68 100644
@@ -506,6 +506,8 @@ source pads.
 
     subdev-formats
 
+.. _subdev-routing:
+
 Streams, multiplexed media pads and internal routing
 ----------------------------------------------------
 
Documentation/userspace-api/media/v4l/meta-formats.rst +1 -0
index 0bb61fc5bc00..919f595576b9 100644
@@ -19,3 +19,4 @@ These formats are used for the :ref:`metadata` interface only.
     metafmt-vsp1-hgo
     metafmt-vsp1-hgt
     metafmt-vivid
+    metafmt-generic
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +304 -0

304 lines not on this page · raw message

drivers/media/v4l2-core/v4l2-ioctl.c +7 -0
index 33076af4dfdb..7cb6063f7056 100644
@@ -1452,6 +1452,13 @@ static void v4l_fill_fmtdesc(struct v4l2_fmtdesc *fmt)
 	case V4L2_PIX_FMT_Y210:		descr = "10-bit YUYV Packed"; break;
 	case V4L2_PIX_FMT_Y212:		descr = "12-bit YUYV Packed"; break;
 	case V4L2_PIX_FMT_Y216:		descr = "16-bit YUYV Packed"; break;
+	case V4L2_META_FMT_GENERIC_8:	descr = "8-bit Generic Metadata"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_10:	descr = "8b Generic Meta, 10b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_12:	descr = "8b Generic Meta, 12b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_14:	descr = "8b Generic Meta, 14b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_16:	descr = "8b Generic Meta, 16b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_20:	descr = "8b Generic Meta, 20b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_24:	descr = "8b Generic Meta, 24b CSI-2"; break;
 
 	default:
 		/* Compressed formats */
include/uapi/linux/videodev2.h +8 -0
index a8015e5e7fa4..6a4f8ae30186 100644
@@ -839,6 +839,14 @@ struct v4l2_pix_format {
 #define V4L2_META_FMT_RK_ISP1_PARAMS	v4l2_fourcc('R', 'K', '1', 'P') /* Rockchip ISP1 3A Parameters */
 #define V4L2_META_FMT_RK_ISP1_STAT_3A	v4l2_fourcc('R', 'K', '1', 'S') /* Rockchip ISP1 3A Statistics */
 
+#define V4L2_META_FMT_GENERIC_8		v4l2_fourcc('M', 'E', 'T', '8') /* Generic 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_10	v4l2_fourcc('M', 'C', '1', 'A') /* 10-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_12	v4l2_fourcc('M', 'C', '1', 'C') /* 12-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_14	v4l2_fourcc('M', 'C', '1', 'E') /* 14-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_16	v4l2_fourcc('M', 'C', '1', 'G') /* 16-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_20	v4l2_fourcc('M', 'C', '1', 'K') /* 20-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_24	v4l2_fourcc('M', 'C', '1', 'O') /* 24-bit CSI-2 packed 8-bit metadata */
+
 /* priv field value to indicates that subsequent fields are valid. */
 #define V4L2_PIX_FMT_PRIV_MAGIC		0xfeedcafe
 
Signature · 1 line
2.39.2
maintainer
#33 lore raw

Re: [PATCH v8 05/38] media: uapi: Add generic 8-bit metadata format definitions

Hi Sakari,

Thank you for the patch.

Mentioning V4L2 on the subject line would help differentiating this from the media bus codes.

Quoted 4 lines · Sakari Ailus

Generic 8-bit metadata formats define the in-memory data layout but not the format of the data itself. The reasoning for having such formats is to allow CSI-2 receiver drivers to receive and DMA drivers to write the data to memory without knowing a large number of device specific formats.

s/device specific/device-specific/

Quoted 1 line

These formats may be used only in conjunction of a Media controller

s/of a/with a/

Quoted 31 lines

pipeline where the internal pad of the source sub-device defines the specific format of the data (using an mbus code).

---
 .../userspace-api/media/v4l/dev-subdev.rst    |   2 +
 .../userspace-api/media/v4l/meta-formats.rst  |   1 +
 .../media/v4l/metafmt-generic.rst             | 304 ++++++++++++++++++
 drivers/media/v4l2-core/v4l2-ioctl.c          |   7 +
 include/uapi/linux/videodev2.h                |   8 +
 5 files changed, 322 insertions(+)
 create mode 100644 Documentation/userspace-api/media/v4l/metafmt-generic.rst
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
index 43988516acdd..f375b820ab68 100644
@@ -506,6 +506,8 @@ source pads.
 
     subdev-formats
 
+.. _subdev-routing:
+
 Streams, multiplexed media pads and internal routing
 ----------------------------------------------------
 
Documentation/userspace-api/media/v4l/meta-formats.rst +0 -0
index 0bb61fc5bc00..919f595576b9 100644

Documentation/userspace-api/media/v4l/meta-formats.rst These formats are used for the :ref:`metadata` interface only. · line 22

Quoted 4 lines
Documentation/userspace-api/media/v4l/meta-formats.rst +1 -0
@@ -19,3 +19,4 @@ These formats are used for the :ref:`metadata` interface only.
     metafmt-vsp1-hgo
     metafmt-vsp1-hgt
     metafmt-vivid
+    metafmt-generic

I think the intention was to keep those alphabetically sorted, even if vivid is out of order already.

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 5

Quoted 11 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +5 -0
new file mode 100644
index 000000000000..2ebab1f895e0
@@ -0,0 +1,304 @@
+.. SPDX-License-Identifier: GPL-2.0 OR GFDL-1.1-no-invariants-or-later
+
+**************************************************************************************************************************************************************************************************************************************************************************************************************************
+V4L2_META_FMT_GENERIC_8 ('MET8'), V4L2_META_FMT_GENERIC_CSI2_10 ('MC1A'), V4L2_META_FMT_GENERIC_CSI2_12 ('MC1C'), V4L2_META_FMT_GENERIC_CSI2_14 ('MC1E'), V4L2_META_FMT_GENERIC_CSI2_16 ('MC1G'), V4L2_META_FMT_GENERIC_CSI2_20 ('MC1K'), V4L2_META_FMT_GENERIC_CSI2_24 ('MC1O')
+**************************************************************************************************************************************************************************************************************************************************************************************************************************

Aren't you supposed to match the line lengths ?

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 16

Quoted 11 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +11 -0
+
+
+Generic line-based metadata formats
+
+
+Description
+===========
+
+These generic line-based metadata formats define the memory layout of the data
+without defining the format or meaning of the metadata itself. These formats may
+only be used with a Media controller pipeline where the more specific format is

"Media Controller" or "media controller", but not a hybrid mix.

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 17

Quoted 1 line
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +1 -0
+defined in an :ref:`internal source pad <MEDIA-PAD-FL-INTERNAL>` of the source

s/defined in/reported by/

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 27

Quoted 10 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +10 -0
+sub-device. See also :ref:`source routes <subdev-routing>`.
+
+.. _v4l2-meta-fmt-generic-8:
+
+V4L2_META_FMT_GENERIC_8
+-----------------------
+
+The V4L2_META_FMT_GENERIC_8 format is a plain 8-bit metadata format.
+
+This format is also used on CSI-2 for both 8 bits per ``Data unit

s/Data unit/Data Unit/ (through the whole patch series)

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 36

Quoted 9 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +9 -0
+<media-glossary-data-unit>`` as well as for 16 bits per Data unit when two bytes
+of metadata are packed into one 16-bit Data unit.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_8.**
+Each cell is one byte. "M" denotes a byte of metadata.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|
+
+.. flat-table::

Let's add a title here

.. flat-table:: Sample 4x2 Metadata Frame

(feel free to adjust). Same below.

Quoted 14 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +14 -0
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+    * - start + 4:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - M\ :sub:`21`
+      - M\ :sub:`31`

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 60

Quoted 10 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +10 -0
+
+.. _v4l2-meta-fmt-generic-csi2-10:
+
+V4L2_META_FMT_GENERIC_CSI2_10
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_10 contains packed 8-bit generic metadata, 10 bits
+for each 8 bits of data. Every four bytes of metadata is followed by a single
+byte of padding. The way the data is packed follows the MIPI CSI-2 specification
+and the padding is defined in the MIPI CCS specification.

Doesn't really sound packed, when you have one byte of padding after four bytes of data :-) I understand where this "packing" is from, but the documentation seems confusing for people not familiar with the topic. How about simplifying it as follows ?

V4L2_META_FMT_GENERIC_CSI2_10 contains 8-bit generic metadata, stored with one padding byte after every four bytes of metadata. This is usually produced by CSI-2 receivers with a source that transmits MEDIA_BUS_FMT_META_10, when the CSI-2 receiver packs the received data as defined in the MIPI CSI-2 specification.

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 66

Quoted 6 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +6 -0
+
+This format is also used in conjunction with 20 bits per ``Data unit
+<media-glossary-data-unit>`` formats that pack two bytes of metadata into one
+Data unit.
+
+This format is little endian.

How is it little endian when it's byte-oriented ?

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 69

Quoted 3 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +3 -0
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_10.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.

'M' and 'x'.

These comments apply to the formats below too.

Quoted 240 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +235 -0
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+      - X
+    * - start + 5:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - M\ :sub:`21`
+      - M\ :sub:`31`
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-12:
+
+V4L2_META_FMT_GENERIC_CSI2_12
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_12 contains packed 8-bit generic metadata, 12 bits
+for each 8 bits of data. Every four bytes of metadata is followed by a single
+byte of padding. The way the data is packed follows the MIPI CSI-2 specification
+and the padding is defined in the MIPI CCS specification.
+
+This format is also used in conjunction with 24 bits per ``Data unit
+<media-glossary-data-unit>`` formats that pack two bytes of metadata into one
+Data unit.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_12.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - X
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+      - X
+    * - start + 6:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - X
+      - M\ :sub:`21`
+      - M\ :sub:`31`
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-14:
+
+V4L2_META_FMT_GENERIC_CSI2_14
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_14 contains packed 8-bit generic metadata, 14 bits
+for each 8 bits of data. Every four bytes of metadata is followed by three bytes
+of padding. The way the data is packed follows the MIPI CSI-2 specification and
+the padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_14.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+      - X
+      - X
+      - X
+    * - start + 7:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - M\ :sub:`21`
+      - M\ :sub:`31`
+      - X
+      - X
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-16:
+
+V4L2_META_FMT_GENERIC_CSI2_16
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_16 contains packed 8-bit generic metadata, 16 bits
+for each 8 bits of data. Every byte of metadata is followed by one byte of
+padding. The way the data is packed follows the MIPI CSI-2 specification and the
+padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_16.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - X
+      - M\ :sub:`10`
+      - X
+      - M\ :sub:`20`
+      - X
+      - M\ :sub:`30`
+      - X
+    * - start + 8:
+      - M\ :sub:`01`
+      - X
+      - M\ :sub:`11`
+      - X
+      - M\ :sub:`21`
+      - X
+      - M\ :sub:`31`
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-20:
+
+V4L2_META_FMT_GENERIC_CSI2_20
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_20 contains packed 8-bit generic metadata, 20 bits
+for each 8 bits of data. Every byte of metadata is followed by alternating one
+and two bytes of padding. The way the data is packed follows the MIPI CSI-2
+specification and the padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_20.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - X
+      - M\ :sub:`10`
+      - X
+      - X
+      - M\ :sub:`20`
+      - X
+      - M\ :sub:`30`
+      - X
+      - X
+    * - start + 10:
+      - M\ :sub:`01`
+      - X
+      - M\ :sub:`11`
+      - X
+      - X
+      - M\ :sub:`21`
+      - X
+      - M\ :sub:`31`
+      - X
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-24:
+
+V4L2_META_FMT_GENERIC_CSI2_24
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_24 contains packed 8-bit generic metadata, 24 bits
+for each 8 bits of data. Every byte of metadata is followed by two bytes of
+padding. The way the data is packed follows the MIPI CSI-2 specification and the
+padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_24.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - X
+      - X
+      - M\ :sub:`10`
+      - X
+      - X
+      - M\ :sub:`20`
+      - X
+      - X
+      - M\ :sub:`30`
+      - X
+      - X
+    * - start + 12:
+      - M\ :sub:`01`
+      - X
+      - X
+      - M\ :sub:`11`
+      - X
+      - X
+      - M\ :sub:`21`
+      - X
+      - X
+      - M\ :sub:`31`
+      - X
+      - X
drivers/media/v4l2-core/v4l2-ioctl.c +0 -0
index 33076af4dfdb..7cb6063f7056 100644

drivers/media/v4l2-core/v4l2-ioctl.c static void v4l_fill_fmtdesc(struct v4l2_fmtdesc *fmt) · line 1456

Quoted 5 lines
drivers/media/v4l2-core/v4l2-ioctl.c +2 -0
@@ -1452,6 +1452,13 @@ static void v4l_fill_fmtdesc(struct v4l2_fmtdesc *fmt)
 	case V4L2_PIX_FMT_Y210:		descr = "10-bit YUYV Packed"; break;
 	case V4L2_PIX_FMT_Y212:		descr = "12-bit YUYV Packed"; break;
 	case V4L2_PIX_FMT_Y216:		descr = "16-bit YUYV Packed"; break;
+	case V4L2_META_FMT_GENERIC_8:	descr = "8-bit Generic Metadata"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_10:	descr = "8b Generic Meta, 10b CSI-2"; break;

You could write "8-bit Generic Meta, 10b CSI-2" and it would still fit. Up to you.

Quoted 27 lines
drivers/media/v4l2-core/v4l2-ioctl.c +5 -0
+	case V4L2_META_FMT_GENERIC_CSI2_12:	descr = "8b Generic Meta, 12b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_14:	descr = "8b Generic Meta, 14b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_16:	descr = "8b Generic Meta, 16b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_20:	descr = "8b Generic Meta, 20b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_24:	descr = "8b Generic Meta, 24b CSI-2"; break;
 
 	default:
 		/* Compressed formats */
include/uapi/linux/videodev2.h +8 -0
index a8015e5e7fa4..6a4f8ae30186 100644
@@ -839,6 +839,14 @@ struct v4l2_pix_format {
 #define V4L2_META_FMT_RK_ISP1_PARAMS	v4l2_fourcc('R', 'K', '1', 'P') /* Rockchip ISP1 3A Parameters */
 #define V4L2_META_FMT_RK_ISP1_STAT_3A	v4l2_fourcc('R', 'K', '1', 'S') /* Rockchip ISP1 3A Statistics */
 
+#define V4L2_META_FMT_GENERIC_8		v4l2_fourcc('M', 'E', 'T', '8') /* Generic 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_10	v4l2_fourcc('M', 'C', '1', 'A') /* 10-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_12	v4l2_fourcc('M', 'C', '1', 'C') /* 12-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_14	v4l2_fourcc('M', 'C', '1', 'E') /* 14-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_16	v4l2_fourcc('M', 'C', '1', 'G') /* 16-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_20	v4l2_fourcc('M', 'C', '1', 'K') /* 20-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_24	v4l2_fourcc('M', 'C', '1', 'O') /* 24-bit CSI-2 packed 8-bit metadata */
+
 /* priv field value to indicates that subsequent fields are valid. */
 #define V4L2_PIX_FMT_PRIV_MAGIC		0xfeedcafe
 
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#34 lore raw

Re: [PATCH v8 05/38] media: uapi: Add generic 8-bit metadata format definitions

Quoted 6 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Mentioning V4L2 on the subject line would help differentiating this from the media bus codes.

Yes, I'll add that.

Quoted 11 lines · Sakari Ailus
Quoted 4 lines

Generic 8-bit metadata formats define the in-memory data layout but not the format of the data itself. The reasoning for having such formats is to allow CSI-2 receiver drivers to receive and DMA drivers to write the data to memory without knowing a large number of device specific formats.

s/device specific/device-specific/

Quoted 1 line

These formats may be used only in conjunction of a Media controller

s/of a/with a/

Yes...

Quoted 27 lines
Quoted 27 lines

pipeline where the internal pad of the source sub-device defines the specific format of the data (using an mbus code).

---
 .../userspace-api/media/v4l/dev-subdev.rst    |   2 +
 .../userspace-api/media/v4l/meta-formats.rst  |   1 +
 .../media/v4l/metafmt-generic.rst             | 304 ++++++++++++++++++
 drivers/media/v4l2-core/v4l2-ioctl.c          |   7 +
 include/uapi/linux/videodev2.h                |   8 +
 5 files changed, 322 insertions(+)
 create mode 100644 Documentation/userspace-api/media/v4l/metafmt-generic.rst
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
index 43988516acdd..f375b820ab68 100644
@@ -506,6 +506,8 @@ source pads.
 
     subdev-formats
 
+.. _subdev-routing:
+
 Streams, multiplexed media pads and internal routing
 ----------------------------------------------------
 
Quoted 12 lines

Documentation/userspace-api/media/v4l/meta-formats.rst These formats are used for the :ref:`metadata` interface only. · line 22

Quoted 9 lines
Documentation/userspace-api/media/v4l/meta-formats.rst +1 -0
index 0bb61fc5bc00..919f595576b9 100644
@@ -19,3 +19,4 @@ These formats are used for the :ref:`metadata` interface only.
     metafmt-vsp1-hgo
     metafmt-vsp1-hgt
     metafmt-vivid
+    metafmt-generic

I think the intention was to keep those alphabetically sorted, even if vivid is out of order already.

I'll sort these while at it.

Quoted 14 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 5

Quoted 12 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +5 -0
new file mode 100644
index 000000000000..2ebab1f895e0
@@ -0,0 +1,304 @@
+.. SPDX-License-Identifier: GPL-2.0 OR GFDL-1.1-no-invariants-or-later
+
+**************************************************************************************************************************************************************************************************************************************************************************************************************************
+V4L2_META_FMT_GENERIC_8 ('MET8'), V4L2_META_FMT_GENERIC_CSI2_10 ('MC1A'), V4L2_META_FMT_GENERIC_CSI2_12 ('MC1C'), V4L2_META_FMT_GENERIC_CSI2_14 ('MC1E'), V4L2_META_FMT_GENERIC_CSI2_16 ('MC1G'), V4L2_META_FMT_GENERIC_CSI2_20 ('MC1K'), V4L2_META_FMT_GENERIC_CSI2_24 ('MC1O')
+**************************************************************************************************************************************************************************************************************************************************************************************************************************

Aren't you supposed to match the line lengths ?

Fixed.

Quoted 14 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 16

Quoted 12 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +11 -0
+
+
+Generic line-based metadata formats
+
+
+Description
+===========
+
+These generic line-based metadata formats define the memory layout of the data
+without defining the format or meaning of the metadata itself. These formats may
+only be used with a Media controller pipeline where the more specific format is

"Media Controller" or "media controller", but not a hybrid mix.

I'll use Media Controller.

Quoted 4 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 17

Quoted 2 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +1 -0
+defined in an :ref:`internal source pad <MEDIA-PAD-FL-INTERNAL>` of the source

s/defined in/reported by/

Yes...

Quoted 13 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 27

Quoted 11 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +10 -0
+sub-device. See also :ref:`source routes <subdev-routing>`.
+
+.. _v4l2-meta-fmt-generic-8:
+
+V4L2_META_FMT_GENERIC_8
+-----------------------
+
+The V4L2_META_FMT_GENERIC_8 format is a plain 8-bit metadata format.
+
+This format is also used on CSI-2 for both 8 bits per ``Data unit

s/Data unit/Data Unit/ (through the whole patch series)

Yes.

Quoted 13 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 36

Quoted 10 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +9 -0
+<media-glossary-data-unit>`` as well as for 16 bits per Data unit when two bytes
+of metadata are packed into one 16-bit Data unit.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_8.**
+Each cell is one byte. "M" denotes a byte of metadata.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|
+
+.. flat-table::

Let's add a title here

Quoted 3 lines

.. flat-table:: Sample 4x2 Metadata Frame

(feel free to adjust). Same below.

Sounds good.

Quoted 31 lines
Quoted 15 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +14 -0
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+    * - start + 4:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - M\ :sub:`21`
+      - M\ :sub:`31`

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 60

Quoted 10 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +10 -0
+
+.. _v4l2-meta-fmt-generic-csi2-10:
+
+V4L2_META_FMT_GENERIC_CSI2_10
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_10 contains packed 8-bit generic metadata, 10 bits
+for each 8 bits of data. Every four bytes of metadata is followed by a single
+byte of padding. The way the data is packed follows the MIPI CSI-2 specification
+and the padding is defined in the MIPI CCS specification.

Doesn't really sound packed, when you have one byte of padding after four bytes of data :-) I understand where this "packing" is from, but the documentation seems confusing for people not familiar with the topic. How about simplifying it as follows ?

Quoted 5 lines

V4L2_META_FMT_GENERIC_CSI2_10 contains 8-bit generic metadata, stored with one padding byte after every four bytes of metadata. This is usually produced by CSI-2 receivers with a source that transmits MEDIA_BUS_FMT_META_10, when the CSI-2 receiver packs the received data as defined in the MIPI CSI-2 specification.

The receiver doesn't pack the data, it already arrives packed on the bus.

"Packing" is also generally used term to describe such arrangements (see e.g. the CCS spec). This also includes packed V4L2 pixel formats.

How about:

V4L2_META_FMT_GENERIC_CSI2_10 contains 8-bit generic metadata packed in 10-bit data units, with one padding byte after every four bytes of metadata, compliant with the MIPI CSI-2 specification. This format is typically used by CSI-2 receivers with a source that transmits MEDIA_BUS_FMT_META_10 and the CSI-2 receiver writes the received data to memory as-is.

Quoted 9 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 66

Quoted 7 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +6 -0
+
+This format is also used in conjunction with 20 bits per ``Data unit
+<media-glossary-data-unit>`` formats that pack two bytes of metadata into one
+Data unit.
+
+This format is little endian.

How is it little endian when it's byte-oriented ?

Some bits are still split across bytes.

Quoted 8 lines

Documentation/userspace-api/media/v4l/metafmt-generic.rst line 69

Quoted 4 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +3 -0
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_10.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.

'M' and 'x'.

These comments apply to the formats below too.

Sounds good.

Quoted 241 lines
Quoted 241 lines
Documentation/userspace-api/media/v4l/metafmt-generic.rst new file +235 -0
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+      - X
+    * - start + 5:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - M\ :sub:`21`
+      - M\ :sub:`31`
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-12:
+
+V4L2_META_FMT_GENERIC_CSI2_12
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_12 contains packed 8-bit generic metadata, 12 bits
+for each 8 bits of data. Every four bytes of metadata is followed by a single
+byte of padding. The way the data is packed follows the MIPI CSI-2 specification
+and the padding is defined in the MIPI CCS specification.
+
+This format is also used in conjunction with 24 bits per ``Data unit
+<media-glossary-data-unit>`` formats that pack two bytes of metadata into one
+Data unit.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_12.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - X
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+      - X
+    * - start + 6:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - X
+      - M\ :sub:`21`
+      - M\ :sub:`31`
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-14:
+
+V4L2_META_FMT_GENERIC_CSI2_14
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_14 contains packed 8-bit generic metadata, 14 bits
+for each 8 bits of data. Every four bytes of metadata is followed by three bytes
+of padding. The way the data is packed follows the MIPI CSI-2 specification and
+the padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_14.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - M\ :sub:`10`
+      - M\ :sub:`20`
+      - M\ :sub:`30`
+      - X
+      - X
+      - X
+    * - start + 7:
+      - M\ :sub:`01`
+      - M\ :sub:`11`
+      - M\ :sub:`21`
+      - M\ :sub:`31`
+      - X
+      - X
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-16:
+
+V4L2_META_FMT_GENERIC_CSI2_16
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_16 contains packed 8-bit generic metadata, 16 bits
+for each 8 bits of data. Every byte of metadata is followed by one byte of
+padding. The way the data is packed follows the MIPI CSI-2 specification and the
+padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_16.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - X
+      - M\ :sub:`10`
+      - X
+      - M\ :sub:`20`
+      - X
+      - M\ :sub:`30`
+      - X
+    * - start + 8:
+      - M\ :sub:`01`
+      - X
+      - M\ :sub:`11`
+      - X
+      - M\ :sub:`21`
+      - X
+      - M\ :sub:`31`
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-20:
+
+V4L2_META_FMT_GENERIC_CSI2_20
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_20 contains packed 8-bit generic metadata, 20 bits
+for each 8 bits of data. Every byte of metadata is followed by alternating one
+and two bytes of padding. The way the data is packed follows the MIPI CSI-2
+specification and the padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_20.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - X
+      - M\ :sub:`10`
+      - X
+      - X
+      - M\ :sub:`20`
+      - X
+      - M\ :sub:`30`
+      - X
+      - X
+    * - start + 10:
+      - M\ :sub:`01`
+      - X
+      - M\ :sub:`11`
+      - X
+      - X
+      - M\ :sub:`21`
+      - X
+      - M\ :sub:`31`
+      - X
+      - X
+
+.. _v4l2-meta-fmt-generic-csi2-24:
+
+V4L2_META_FMT_GENERIC_CSI2_24
+-----------------------------
+
+V4L2_META_FMT_GENERIC_CSI2_24 contains packed 8-bit generic metadata, 24 bits
+for each 8 bits of data. Every byte of metadata is followed by two bytes of
+padding. The way the data is packed follows the MIPI CSI-2 specification and the
+padding is defined in the MIPI CCS specification.
+
+This format is little endian.
+
+**Byte Order Of V4L2_META_FMT_GENERIC_CSI2_24.**
+Each cell is one byte. "M" denotes a byte of metadata and "X" a byte of padding.
+
+.. tabularcolumns:: |p{2.4cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|p{1.2cm}|p{.8cm}|p{.8cm}|
+
+.. flat-table::
+    :header-rows:  0
+    :stub-columns: 0
+    :widths: 12 8 8 8 8 8 8 8 8 8 8 8 8
+
+    * - start + 0:
+      - M\ :sub:`00`
+      - X
+      - X
+      - M\ :sub:`10`
+      - X
+      - X
+      - M\ :sub:`20`
+      - X
+      - X
+      - M\ :sub:`30`
+      - X
+      - X
+    * - start + 12:
+      - M\ :sub:`01`
+      - X
+      - X
+      - M\ :sub:`11`
+      - X
+      - X
+      - M\ :sub:`21`
+      - X
+      - X
+      - M\ :sub:`31`
+      - X
+      - X
drivers/media/v4l2-core/v4l2-ioctl.c +0 -0
index 33076af4dfdb..7cb6063f7056 100644
@@ -1452,6 +1452,13 @@ static void v4l_fill_fmtdesc(struct v4l2_fmtdesc *fmt)
Quoted 8 lines

drivers/media/v4l2-core/v4l2-ioctl.c static void v4l_fill_fmtdesc(struct v4l2_fmtdesc *fmt) · line 1456

Quoted 5 lines
drivers/media/v4l2-core/v4l2-ioctl.c +2 -0
 	case V4L2_PIX_FMT_Y210:		descr = "10-bit YUYV Packed"; break;
 	case V4L2_PIX_FMT_Y212:		descr = "12-bit YUYV Packed"; break;
 	case V4L2_PIX_FMT_Y216:		descr = "16-bit YUYV Packed"; break;
+	case V4L2_META_FMT_GENERIC_8:	descr = "8-bit Generic Metadata"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_10:	descr = "8b Generic Meta, 10b CSI-2"; break;

You could write "8-bit Generic Meta, 10b CSI-2" and it would still fit. Up to you.

I'll change it. The rest are using "x-bit" already.

Quoted 29 lines
Quoted 28 lines
drivers/media/v4l2-core/v4l2-ioctl.c +5 -0
+	case V4L2_META_FMT_GENERIC_CSI2_12:	descr = "8b Generic Meta, 12b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_14:	descr = "8b Generic Meta, 14b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_16:	descr = "8b Generic Meta, 16b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_20:	descr = "8b Generic Meta, 20b CSI-2"; break;
+	case V4L2_META_FMT_GENERIC_CSI2_24:	descr = "8b Generic Meta, 24b CSI-2"; break;
 
 	default:
 		/* Compressed formats */
include/uapi/linux/videodev2.h +8 -0
index a8015e5e7fa4..6a4f8ae30186 100644
@@ -839,6 +839,14 @@ struct v4l2_pix_format {
 #define V4L2_META_FMT_RK_ISP1_PARAMS	v4l2_fourcc('R', 'K', '1', 'P') /* Rockchip ISP1 3A Parameters */
 #define V4L2_META_FMT_RK_ISP1_STAT_3A	v4l2_fourcc('R', 'K', '1', 'S') /* Rockchip ISP1 3A Statistics */
 
+#define V4L2_META_FMT_GENERIC_8		v4l2_fourcc('M', 'E', 'T', '8') /* Generic 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_10	v4l2_fourcc('M', 'C', '1', 'A') /* 10-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_12	v4l2_fourcc('M', 'C', '1', 'C') /* 12-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_14	v4l2_fourcc('M', 'C', '1', 'E') /* 14-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_16	v4l2_fourcc('M', 'C', '1', 'G') /* 16-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_20	v4l2_fourcc('M', 'C', '1', 'K') /* 20-bit CSI-2 packed 8-bit metadata */
+#define V4L2_META_FMT_GENERIC_CSI2_24	v4l2_fourcc('M', 'C', '1', 'O') /* 24-bit CSI-2 packed 8-bit metadata */
+
 /* priv field value to indicates that subsequent fields are valid. */
 #define V4L2_PIX_FMT_PRIV_MAGIC		0xfeedcafe
 
Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#35 lore raw

PATCH v8 6/38 media: v4l: Support line-based metadata capture

In reply to Sakari Ailus

many camera sensors, among other devices, transmit embedded data and image data for each CSI-2 frame. This embedded data typically contains register configuration of the sensor that has been used to capture the image data of the same frame.

The embedded data is received by the CSI-2 receiver and has the same properties as the image data, including that it is line based: it has width, height and bytesperline (stride).

Add these fields to struct v4l2_meta_format and document them.

Also add V4L2_FMT_FLAG_META_LINE_BASED to tell a given format is line-based i.e. these fields of struct v4l2_meta_format are valid for it.

---
 .../userspace-api/media/v4l/dev-meta.rst          | 15 +++++++++++++++
 .../userspace-api/media/v4l/vidioc-enum-fmt.rst   |  7 +++++++
 .../media/videodev2.h.rst.exceptions              |  1 +
 drivers/media/v4l2-core/v4l2-ioctl.c              |  5 +++--
 include/uapi/linux/videodev2.h                    | 10 ++++++++++
 5 files changed, 36 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/v4l/dev-meta.rst +15 -0
index 0e7e1ee1471a..4dfd79e0a705 100644
@@ -65,3 +65,18 @@ to 0.
       - ``buffersize``
       - Maximum buffer size in bytes required for data. The value is set by the
         driver.
+    * - __u32
+      - ``width``
+      - Width of a line of metadata in Data units. Valid when
+	:c:type`v4l2_fmtdesc` flag ``V4L2_FMT_FLAG_META_LINE_BASED`` is set,
+	otherwise zero. See :c:func:`VIDIOC_ENUM_FMT`.
+    * - __u32
+      - ``height``
+      - Number of rows of metadata. Valid when :c:type`v4l2_fmtdesc` flag
+	``V4L2_FMT_FLAG_META_LINE_BASED`` is set, otherwise zero. See
+	:c:func:`VIDIOC_ENUM_FMT`.
+    * - __u32
+      - ``bytesperline``
+      - Offset in bytes between the beginning of two consecutive lines. Valid
+	when :c:type`v4l2_fmtdesc` flag ``V4L2_FMT_FLAG_META_LINE_BASED`` is
+	set, otherwise zero. See :c:func:`VIDIOC_ENUM_FMT`.
Documentation/userspace-api/media/v4l/vidioc-enum-fmt.rst +7 -0
index 000c154b0f98..a79abf4428c8 100644
@@ -227,6 +227,13 @@ the ``mbus_code`` field is handled differently:
 	The application can ask to configure the quantization of the capture
 	device when calling the :ref:`VIDIOC_S_FMT <VIDIOC_G_FMT>` ioctl with
 	:ref:`V4L2_PIX_FMT_FLAG_SET_CSC <v4l2-pix-fmt-flag-set-csc>` set.
+    * - ``V4L2_FMT_FLAG_META_LINE_BASED``
+      - 0x0200
+      - The metadata format is line-based. In this case the ``width``,
+	``height`` and ``bytesperline`` fields of :c:type:`v4l2_meta_format` are
+	valid. The buffer consists of ``height`` lines, each having ``width``
+	Data units of data and offset (in bytes) between the beginning of each
+	two consecutive lines is ``bytesperline``.
 
 Return Value
 ============
Documentation/userspace-api/media/videodev2.h.rst.exceptions +1 -0
index 3e58aac4ef0b..bdc628e8c1d6 100644
@@ -215,6 +215,7 @@ replace define V4L2_FMT_FLAG_CSC_XFER_FUNC fmtdesc-flags
 replace define V4L2_FMT_FLAG_CSC_YCBCR_ENC fmtdesc-flags
 replace define V4L2_FMT_FLAG_CSC_HSV_ENC fmtdesc-flags
 replace define V4L2_FMT_FLAG_CSC_QUANTIZATION fmtdesc-flags
+replace define V4L2_FMT_FLAG_META_LINE_BASED fmtdesc-flags
 
 # V4L2 timecode types
 replace define V4L2_TC_TYPE_24FPS timecode-type
drivers/media/v4l2-core/v4l2-ioctl.c +3 -2
index 7cb6063f7056..d125d23e4e61 100644
@@ -343,8 +343,9 @@ static void v4l_print_format(const void *arg, bool write_only)
 	case V4L2_BUF_TYPE_META_OUTPUT:
 		meta = &p->fmt.meta;
 		pixelformat = meta->dataformat;
-		pr_cont(", dataformat=%p4cc, buffersize=%u\n",
-			&pixelformat, meta->buffersize);
+		pr_cont(", dataformat=%p4cc, buffersize=%u, width=%u, height=%u, bytesperline=%u\n",
+			&pixelformat, meta->buffersize, meta->width,
+			meta->height, meta->bytesperline);
 		break;
 	}
 }
include/uapi/linux/videodev2.h +10 -0
index 6a4f8ae30186..88fdf190a437 100644
@@ -877,6 +877,7 @@ struct v4l2_fmtdesc {
 #define V4L2_FMT_FLAG_CSC_YCBCR_ENC		0x0080
 #define V4L2_FMT_FLAG_CSC_HSV_ENC		V4L2_FMT_FLAG_CSC_YCBCR_ENC
 #define V4L2_FMT_FLAG_CSC_QUANTIZATION		0x0100
+#define V4L2_FMT_FLAG_META_LINE_BASED		0x0200
 
 	/* Frame Size and frame rate enumeration */
 /*
@@ -2423,10 +2424,19 @@ struct v4l2_sdr_format {
  * struct v4l2_meta_format - metadata format definition
  * @dataformat:		little endian four character code (fourcc)
  * @buffersize:		maximum size in bytes required for data
+ * @width:		number of data units of data per line (valid for line
+ *			based formats only, see format documentation)
+ * @height:		number of lines of data per buffer (valid for line based
+ *			formats only)
+ * @bytesperline:	offset between the beginnings of two adjacent lines in
+ *			bytes (valid for line based formats only)
  */
 struct v4l2_meta_format {
 	__u32				dataformat;
 	__u32				buffersize;
+	__u32				width;
+	__u32				height;
+	__u32				bytesperline;
 } __attribute__ ((packed));
 
 /**
Signature · 1 line
2.39.2
maintainer
#36 lore raw

Re: [PATCH v8 06/38] media: v4l: Support line-based metadata capture

Hi Sakari,

Thank you for the patch.

Quoted 1 line · Sakari Ailus

many camera sensors, among other devices, transmit embedded data and image

s/many/Many/

Quoted 50 lines

data for each CSI-2 frame. This embedded data typically contains register configuration of the sensor that has been used to capture the image data of the same frame.

The embedded data is received by the CSI-2 receiver and has the same properties as the image data, including that it is line based: it has width, height and bytesperline (stride).

Add these fields to struct v4l2_meta_format and document them.

Also add V4L2_FMT_FLAG_META_LINE_BASED to tell a given format is line-based i.e. these fields of struct v4l2_meta_format are valid for it.

---
 .../userspace-api/media/v4l/dev-meta.rst          | 15 +++++++++++++++
 .../userspace-api/media/v4l/vidioc-enum-fmt.rst   |  7 +++++++
 .../media/videodev2.h.rst.exceptions              |  1 +
 drivers/media/v4l2-core/v4l2-ioctl.c              |  5 +++--
 include/uapi/linux/videodev2.h                    | 10 ++++++++++
 5 files changed, 36 insertions(+), 2 deletions(-)
Documentation/userspace-api/media/v4l/dev-meta.rst +15 -0
index 0e7e1ee1471a..4dfd79e0a705 100644
@@ -65,3 +65,18 @@ to 0.
       - ``buffersize``
       - Maximum buffer size in bytes required for data. The value is set by the
         driver.
+    * - __u32
+      - ``width``
+      - Width of a line of metadata in Data units. Valid when
+	:c:type`v4l2_fmtdesc` flag ``V4L2_FMT_FLAG_META_LINE_BASED`` is set,
+	otherwise zero. See :c:func:`VIDIOC_ENUM_FMT`.
+    * - __u32
+      - ``height``
+      - Number of rows of metadata. Valid when :c:type`v4l2_fmtdesc` flag
+	``V4L2_FMT_FLAG_META_LINE_BASED`` is set, otherwise zero. See
+	:c:func:`VIDIOC_ENUM_FMT`.
+    * - __u32
+      - ``bytesperline``
+      - Offset in bytes between the beginning of two consecutive lines. Valid
+	when :c:type`v4l2_fmtdesc` flag ``V4L2_FMT_FLAG_META_LINE_BASED`` is
+	set, otherwise zero. See :c:func:`VIDIOC_ENUM_FMT`.
Documentation/userspace-api/media/v4l/vidioc-enum-fmt.rst +0 -0
index 000c154b0f98..a79abf4428c8 100644

Documentation/userspace-api/media/v4l/vidioc-enum-fmt.rst the ``mbus_code`` field is handled differently: · line 235

Quoted 9 lines
Documentation/userspace-api/media/v4l/vidioc-enum-fmt.rst +6 -0
@@ -227,6 +227,13 @@ the ``mbus_code`` field is handled differently:
 	The application can ask to configure the quantization of the capture
 	device when calling the :ref:`VIDIOC_S_FMT <VIDIOC_G_FMT>` ioctl with
 	:ref:`V4L2_PIX_FMT_FLAG_SET_CSC <v4l2-pix-fmt-flag-set-csc>` set.
+    * - ``V4L2_FMT_FLAG_META_LINE_BASED``
+      - 0x0200
+      - The metadata format is line-based. In this case the ``width``,
+	``height`` and ``bytesperline`` fields of :c:type:`v4l2_meta_format` are
+	valid. The buffer consists of ``height`` lines, each having ``width``
+	Data units of data and offset (in bytes) between the beginning of each

s/and offset/, and the offset/

Quoted 37 lines
Documentation/userspace-api/media/v4l/vidioc-enum-fmt.rst +1 -0
+	two consecutive lines is ``bytesperline``.
 
 Return Value
 ============
Documentation/userspace-api/media/videodev2.h.rst.exceptions +1 -0
index 3e58aac4ef0b..bdc628e8c1d6 100644
@@ -215,6 +215,7 @@ replace define V4L2_FMT_FLAG_CSC_XFER_FUNC fmtdesc-flags
 replace define V4L2_FMT_FLAG_CSC_YCBCR_ENC fmtdesc-flags
 replace define V4L2_FMT_FLAG_CSC_HSV_ENC fmtdesc-flags
 replace define V4L2_FMT_FLAG_CSC_QUANTIZATION fmtdesc-flags
+replace define V4L2_FMT_FLAG_META_LINE_BASED fmtdesc-flags
 
 # V4L2 timecode types
 replace define V4L2_TC_TYPE_24FPS timecode-type
drivers/media/v4l2-core/v4l2-ioctl.c +3 -2
index 7cb6063f7056..d125d23e4e61 100644
@@ -343,8 +343,9 @@ static void v4l_print_format(const void *arg, bool write_only)
 	case V4L2_BUF_TYPE_META_OUTPUT:
 		meta = &p->fmt.meta;
 		pixelformat = meta->dataformat;
-		pr_cont(", dataformat=%p4cc, buffersize=%u\n",
-			&pixelformat, meta->buffersize);
+		pr_cont(", dataformat=%p4cc, buffersize=%u, width=%u, height=%u, bytesperline=%u\n",
+			&pixelformat, meta->buffersize, meta->width,
+			meta->height, meta->bytesperline);
 		break;
 	}
 }
include/uapi/linux/videodev2.h +0 -0
index 6a4f8ae30186..88fdf190a437 100644

include/uapi/linux/videodev2.h struct v4l2_fmtdesc { · line 880

Quoted 4 lines
include/uapi/linux/videodev2.h +1 -0
@@ -877,6 +877,7 @@ struct v4l2_fmtdesc {
 #define V4L2_FMT_FLAG_CSC_YCBCR_ENC		0x0080
 #define V4L2_FMT_FLAG_CSC_HSV_ENC		V4L2_FMT_FLAG_CSC_YCBCR_ENC
 #define V4L2_FMT_FLAG_CSC_QUANTIZATION		0x0100
+#define V4L2_FMT_FLAG_META_LINE_BASED		0x0200

Could the V4L2_FMT_FLAG_META_LINE_BASED flag be set by the V4L2 core for line-based metadata formats, instead of requiring drivers to set it ? It would ensure consistency.

Quoted 23 lines
include/uapi/linux/videodev2.h +9 -0
 	/* Frame Size and frame rate enumeration */
 /*
@@ -2423,10 +2424,19 @@ struct v4l2_sdr_format {
  * struct v4l2_meta_format - metadata format definition
  * @dataformat:		little endian four character code (fourcc)
  * @buffersize:		maximum size in bytes required for data
+ * @width:		number of data units of data per line (valid for line
+ *			based formats only, see format documentation)
+ * @height:		number of lines of data per buffer (valid for line based
+ *			formats only)
+ * @bytesperline:	offset between the beginnings of two adjacent lines in
+ *			bytes (valid for line based formats only)
  */
 struct v4l2_meta_format {
 	__u32				dataformat;
 	__u32				buffersize;
+	__u32				width;
+	__u32				height;
+	__u32				bytesperline;
 } __attribute__ ((packed));
 
 /**
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#37 lore raw

PATCH v8 7/38 media: Documentation: Additional streams generally don't harm capture

In reply to Sakari Ailus

Having extra streams on the source end of the link that cannot be captured by the sink sub-device generally are not an issue, at least not on CSI-2 bus. Still document that there may be hardware specific limitations. For example on parallel bus this might not work on all cases.

---
 Documentation/userspace-api/media/v4l/dev-subdev.rst | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -3
index f375b820ab68..a387e8a15b8d 100644
@@ -529,9 +529,9 @@ the its sink pad and allows to route them individually to one of its source
 pads.
 
 Subdevice drivers that support multiplexed streams are compatible with
-non-multiplexed subdev drivers, but, of course, require a routing configuration
-where the link between those two types of drivers contains only a single
-stream.
+non-multiplexed subdev drivers. However, if the driver at the sink end of a link
+does not support streams, then only the stream 0 on source end may be
+captured. There may be additional hardware specific limitations.
 
 Understanding streams
 ^^^^^^^^^^^^^^^^^^^^^
Signature · 1 line
2.39.2
maintainer Reviewed-by on v8 7/38
#38 lore raw

Re: [PATCH v8 07/38] media: Documentation: Additional streams generally don't harm capture

Hi Sakari,

Thank you for the patch.

Quoted 3 lines · Sakari Ailus

Having extra streams on the source end of the link that cannot be captured by the sink sub-device generally are not an issue, at least not on CSI-2 bus. Still document that there may be hardware specific limitations. For

s/hardware specific/hardware-specific/

Quoted 1 line

example on parallel bus this might not work on all cases.

s/bus/buses/

Quoted 12 lines
---
 Documentation/userspace-api/media/v4l/dev-subdev.rst | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Documentation/userspace-api/media/v4l/dev-subdev.rst +0 -0
index f375b820ab68..a387e8a15b8d 100644

Documentation/userspace-api/media/v4l/dev-subdev.rst the its sink pad and allows to route them individually to one of its source · line 533

Quoted 8 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -3
@@ -529,9 +529,9 @@ the its sink pad and allows to route them individually to one of its source
 pads.
 
 Subdevice drivers that support multiplexed streams are compatible with
-non-multiplexed subdev drivers, but, of course, require a routing configuration
-where the link between those two types of drivers contains only a single
-stream.
+non-multiplexed subdev drivers. However, if the driver at the sink end of a link
+does not support streams, then only the stream 0 on source end may be

s/the stream 0 on source end/stream 0 of the source/

Documentation/userspace-api/media/v4l/dev-subdev.rst the its sink pad and allows to route them individually to one of its source · line 534

Quoted 1 line
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+captured. There may be additional hardware specific limitations.

s/hardware specific/hardware-specific/

Or maybe

There may be additional limitations specific to the sink device.

Quoted 3 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +0 -0
 Understanding streams
 ^^^^^^^^^^^^^^^^^^^^^
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#39 lore raw

Re: [PATCH v8 07/38] media: Documentation: Additional streams generally don't harm capture

Hi Laurent,

Quoted 10 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 3 lines · Sakari Ailus

Having extra streams on the source end of the link that cannot be captured by the sink sub-device generally are not an issue, at least not on CSI-2 bus. Still document that there may be hardware specific limitations. For

s/hardware specific/hardware-specific/

Yes.

Quoted 4 lines
Quoted 2 lines

example on parallel bus this might not work on all cases.

s/bus/buses/

Works for me. We should probably consider renaming "the parallel bus" as DVP (a spec exists) or something alike, outside this context naturally.

Quoted 24 lines
Quoted 13 lines
---
 Documentation/userspace-api/media/v4l/dev-subdev.rst | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Documentation/userspace-api/media/v4l/dev-subdev.rst +0 -0
index f375b820ab68..a387e8a15b8d 100644

Documentation/userspace-api/media/v4l/dev-subdev.rst the its sink pad and allows to route them individually to one of its source · line 533

Quoted 8 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -3
@@ -529,9 +529,9 @@ the its sink pad and allows to route them individually to one of its source
 pads.
 
 Subdevice drivers that support multiplexed streams are compatible with
-non-multiplexed subdev drivers, but, of course, require a routing configuration
-where the link between those two types of drivers contains only a single
-stream.
+non-multiplexed subdev drivers. However, if the driver at the sink end of a link
+does not support streams, then only the stream 0 on source end may be

s/the stream 0 on source end/stream 0 of the source/

Quoted 7 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst the its sink pad and allows to route them individually to one of its source · line 534

Quoted 1 line
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+captured. There may be additional hardware specific limitations.

s/hardware specific/hardware-specific/

Or maybe

There may be additional limitations specific to the sink device.

Sounds good.

Quoted 2 lines

Thanks!

Quoted 5 lines
Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +0 -0
 Understanding streams
 ^^^^^^^^^^^^^^^^^^^^^
Signature · 3 lines
Kind regards,

Sakari Ailus
author maintainer
#40 lore raw

PATCH v8 8/38 media: Documentation: Document embedded data guidelines for camera sensors

In reply to Sakari Ailus

Document how embedded data support should be implemented for camera sensors, and when and how CCS embedded data format should be referenced.

---
 .../media/drivers/camera-sensor.rst           | 29 +++++++++++++++++++
 1 file changed, 29 insertions(+)
Documentation/userspace-api/media/drivers/camera-sensor.rst +29 -0
index 919a50e8b9d9..92545538b855 100644
@@ -102,3 +102,32 @@ register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP
 values programmed by the register sequences. The default values of these
 controls shall be 0 (disabled). Especially these controls shall not be inverted,
 independently of the sensor's mounting rotation.
+
+Embedded data
+-------------
+
+Many sensors, mostly raw sensors, support embedded data which is used to convey
+the sensor configuration for the captured frame back to the host. While CSI-2 is
+the most common bus used by such sensors, embedded data can be available on
+other bus types as well.
+
+Embedded data support shall use an internal source pad (a pad that has
+``MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK>`` and ``MEDIA_PAD_FL_INTERNAL
+<MEDIA-PAD-FL-INTERNAL>`` flags set) and route to the external pad. If embedded
+data output can be disabled in hardware, it should be possible to disable the
+embedded data route via ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL.
+
+In general, changing the embedded data format from the driver-configured values
+is not supported. The height of the metadata is hardware specific and the width
+is that (or less of that) of the image width, as configured on the pixel data
+stream.
+
+CCS and non-CCS embedded data
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Embedded data which is compliant with CCS definitions shall use ``CCS embedded
+data format <MEDIA-BUS-FMT-CCS-EMBEDDED>``. Device specific embedded data which
+is compliant up to MIPI CCS embedded data levels 1 or 2 only shall refer to CCS
+embedded data formats and document the level of conformance. The rest of the
+device specific embedded data format shall be documented in the context of the
+data format itself.
Signature · 1 line
2.39.2
Reviewed-by on v8 8/38
#41 lore raw

Re: [PATCH v8 08/38] media: Documentation: Document embedded data guidelines for camera sensors

Quoted 35 lines · Sakari Ailus

Document how embedded data support should be implemented for camera sensors, and when and how CCS embedded data format should be referenced.

---
  .../media/drivers/camera-sensor.rst           | 29 +++++++++++++++++++
  1 file changed, 29 insertions(+)
Documentation/userspace-api/media/drivers/camera-sensor.rst +19 -0
index 919a50e8b9d9..92545538b855 100644
@@ -102,3 +102,32 @@ register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP
  values programmed by the register sequences. The default values of these
  controls shall be 0 (disabled). Especially these controls shall not be inverted,
  independently of the sensor's mounting rotation.
+
+Embedded data
+-------------
+
+Many sensors, mostly raw sensors, support embedded data which is used to convey
+the sensor configuration for the captured frame back to the host. While CSI-2 is
+the most common bus used by such sensors, embedded data can be available on
+other bus types as well.
+
+Embedded data support shall use an internal source pad (a pad that has
+``MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK>`` and ``MEDIA_PAD_FL_INTERNAL
+<MEDIA-PAD-FL-INTERNAL>`` flags set) and route to the external pad. If embedded
+data output can be disabled in hardware, it should be possible to disable the
+embedded data route via ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL.
+
+In general, changing the embedded data format from the driver-configured values
+is not supported. The height of the metadata is hardware specific and the width
+is that (or less of that) of the image width, as configured on the pixel data
+stream.

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 133

Quoted 10 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +10 -0
+
+CCS and non-CCS embedded data
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Embedded data which is compliant with CCS definitions shall use ``CCS embedded
+data format <MEDIA-BUS-FMT-CCS-EMBEDDED>``. Device specific embedded data which
+is compliant up to MIPI CCS embedded data levels 1 or 2 only shall refer to CCS
+embedded data formats and document the level of conformance. The rest of the
+device specific embedded data format shall be documented in the context of the
+data format itself.

Julien

maintainer
#42 lore raw

Re: [PATCH v8 08/38] media: Documentation: Document embedded data guidelines for camera sensors

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 16 lines · Sakari Ailus

Document how embedded data support should be implemented for camera sensors, and when and how CCS embedded data format should be referenced.

---
 .../media/drivers/camera-sensor.rst           | 29 +++++++++++++++++++
 1 file changed, 29 insertions(+)
Documentation/userspace-api/media/drivers/camera-sensor.rst +0 -0
index 919a50e8b9d9..92545538b855 100644
@@ -102,3 +102,32 @@ register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP
 values programmed by the register sequences. The default values of these
 controls shall be 0 (disabled). Especially these controls shall not be inverted,
 independently of the sensor's mounting rotation.

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 114

Quoted 10 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +10 -0
+
+Embedded data
+-------------
+
+Many sensors, mostly raw sensors, support embedded data which is used to convey
+the sensor configuration for the captured frame back to the host. While CSI-2 is
+the most common bus used by such sensors, embedded data can be available on
+other bus types as well.
+
+Embedded data support shall use an internal source pad (a pad that has

s/source pad/sink pad/

Or do we call those "internal source pad" ?

As this is uAPI documentation, it is mainly targetting (in my opinion) userspace developers. I would write "Embedded data support uses ..." to describe the behaviour seen from userspace, instead of using "shall" to describe the requirements towards drivers.

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 118

Quoted 4 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +4 -0
+``MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK>`` and ``MEDIA_PAD_FL_INTERNAL
+<MEDIA-PAD-FL-INTERNAL>`` flags set) and route to the external pad. If embedded
+data output can be disabled in hardware, it should be possible to disable the
+embedded data route via ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL.

You should mention the image pad here too:

Such sensors expose two internal sink pads (pads that have both the ``MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK>`` and ``MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`` flags set) to model the source of the image and embedded data streams. Each of those pads produces a single stream, and the sub-device routes those streams to the external (source) pad. If embedded data output can be disabled in hardware, the sub-device allows disabling the embedded data route via the ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL.

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 121

Quoted 3 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +3 -0
+
+In general, changing the embedded data format from the driver-configured values
+is not supported. The height of the metadata is hardware specific and the width

s/hardware specific/device-specific/

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 129

Quoted 8 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +8 -0
+is that (or less of that) of the image width, as configured on the pixel data
+stream.
+
+CCS and non-CCS embedded data
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Embedded data which is compliant with CCS definitions shall use ``CCS embedded
+data format <MEDIA-BUS-FMT-CCS-EMBEDDED>``. Device specific embedded data which

You should clarify here that you're talking about the internal sink pad.

s/Device specific/Only device-specific/

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 130

Quoted 1 line
Documentation/userspace-api/media/drivers/camera-sensor.rst +1 -0
+is compliant up to MIPI CCS embedded data levels 1 or 2 only shall refer to CCS

s/up to/with the/
s/only shall/may/

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 132

Quoted 2 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +2 -0
+embedded data formats and document the level of conformance. The rest of the
+device specific embedded data format shall be documented in the context of the

s/device specific/device-specific/

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 133

Quoted 1 line
Documentation/userspace-api/media/drivers/camera-sensor.rst +1 -0
+data format itself.

Not sure what you mean by the context in the last sentence.

Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#43 lore raw

Re: [PATCH v8 08/38] media: Documentation: Document embedded data guidelines for camera sensors

Hi Laurent,

Quoted 34 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 16 lines · Sakari Ailus

Document how embedded data support should be implemented for camera sensors, and when and how CCS embedded data format should be referenced.

---
 .../media/drivers/camera-sensor.rst           | 29 +++++++++++++++++++
 1 file changed, 29 insertions(+)
Documentation/userspace-api/media/drivers/camera-sensor.rst +0 -0
index 919a50e8b9d9..92545538b855 100644
@@ -102,3 +102,32 @@ register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP
 values programmed by the register sequences. The default values of these
 controls shall be 0 (disabled). Especially these controls shall not be inverted,
 independently of the sensor's mounting rotation.

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 114

Quoted 10 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +10 -0
+
+Embedded data
+-------------
+
+Many sensors, mostly raw sensors, support embedded data which is used to convey
+the sensor configuration for the captured frame back to the host. While CSI-2 is
+the most common bus used by such sensors, embedded data can be available on
+other bus types as well.
+
+Embedded data support shall use an internal source pad (a pad that has

s/source pad/sink pad/

Quoted 1 line

Or do we call those "internal source pad" ?

As I wrote in the earlier e-mail, I'm quite unhappy with the "internal source pad" term, mainly because of the conflict with the term itself and the flags that indicate it. I'd just call them "internal sink pads" and then explain they're actually a source of data.

Quoted 5 lines

As this is uAPI documentation, it is mainly targetting (in my opinion) userspace developers. I would write "Embedded data support uses ..." to describe the behaviour seen from userspace, instead of using "shall" to describe the requirements towards drivers.

Sounds good.

Quoted 8 lines

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 118

Quoted 5 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +4 -0
+``MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK>`` and ``MEDIA_PAD_FL_INTERNAL
+<MEDIA-PAD-FL-INTERNAL>`` flags set) and route to the external pad. If embedded
+data output can be disabled in hardware, it should be possible to disable the
+embedded data route via ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL.

You should mention the image pad here too:

Quoted 4 lines

Such sensors expose two internal sink pads (pads that have both the ``MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK>`` and ``MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`` flags set) to model the source of the image and embedded data streams. Each of those pads produces a single stream, and the

s/Each of those/Both of these/

Quoted 3 lines

sub-device routes those streams to the external (source) pad. If embedded data output can be disabled in hardware, the sub-device allows disabling the embedded data route via the ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL.

In the last sentence, we need to refer to the driver.

But generally I agree. I'll use:

"If the sub-device driver supports disabling embedded data, this can be done by disabling the embedded data route via the ``VIDIOC_SUBDEV_S_ROUTING`` IOCTL."

Quoted 6 lines

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 121

Quoted 4 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +3 -0
+
+In general, changing the embedded data format from the driver-configured values
+is not supported. The height of the metadata is hardware specific and the width

s/hardware specific/device-specific/

Yes.

Quoted 19 lines

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 129

Quoted 9 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +8 -0
+is that (or less of that) of the image width, as configured on the pixel data
+stream.
+
+CCS and non-CCS embedded data
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Embedded data which is compliant with CCS definitions shall use ``CCS embedded
+data format <MEDIA-BUS-FMT-CCS-EMBEDDED>``. Device specific embedded data which

You should clarify here that you're talking about the internal sink pad.

s/Device specific/Only device-specific/

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 130

Quoted 1 line
Documentation/userspace-api/media/drivers/camera-sensor.rst +1 -0
+is compliant up to MIPI CCS embedded data levels 1 or 2 only shall refer to CCS

s/up to/with the/
s/only shall/may/

Quoted 8 lines

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 132

Quoted 2 lines
Documentation/userspace-api/media/drivers/camera-sensor.rst +2 -0
+embedded data formats and document the level of conformance. The rest of the
+device specific embedded data format shall be documented in the context of the

s/device specific/device-specific/

Documentation/userspace-api/media/drivers/camera-sensor.rst register programming sequences shall initialize the :ref:`V4L2_CID_HFLIP · line 133

Quoted 1 line
Documentation/userspace-api/media/drivers/camera-sensor.rst +1 -0
+data format itself.

Not sure what you mean by the context in the last sentence.

This refers to CCS embedded data support levels which appears to be documented in a later patch in the series ("media: uapi: ccs: Add media bus code for MIPI CCS embedded data"). I'll add this paragraph after that patch.

The paragraph became finally:

Embedded data which is fully compliant with CCS definitions uses ``CCS embedded data format <MEDIA-BUS-FMT-CCS-EMBEDDED>`` media bus code (level 3). Device-specific embedded data compliant with either MIPI CCS embedded data levels 1 or 2 only shall not use CCS embedded data mbus code, but may refer to CCS embedded data documentation with the level of conformance specified and omit documenting these aspects of the format. The rest of the device-specific embedded data format is documented in the context of the data format itself.

Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#44 lore raw

PATCH v8 9/38 media: Documentation: v4l: Document internal source pads

In reply to Sakari Ailus

Document internal source pads, pads that have both SINK and INTERNAL flags set. Use the IMX219 camera sensor as an example.

---
 .../userspace-api/media/v4l/dev-subdev.rst    | 145 ++++++++++++++++++
 1 file changed, 145 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +145 -0
index a387e8a15b8d..1808f40f63e3 100644
@@ -553,6 +553,27 @@ A stream at a specific point in the media pipeline is identified by the
 sub-device and a (pad, stream) pair. For sub-devices that do not support
 multiplexed streams the 'stream' field is always 0.
 
+.. _v4l2-subdev-internal-source-pads:
+
+Internal source pads and routing
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Cases where a single sub-device source pad is traversed by multiple streams, one
+or more of which originate from within the sub-device itself, are special as
+there is no external sink pad for such routes. In those cases, the sources of
+the internally generated streams are represented by internal source pads, which
+are sink pads that have the :ref:`MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`
+pad flag set.
+
+Internal pads have all the properties of an external pad, including formats and
+selections. The format in this case is the source format of the stream. An
+internal pad always has a single stream only (0).
+
+Routes from an internal source pad to an external source pad are typically not
+modifiable but they can be activated and deactivated using the
+:ref:`V4L2_SUBDEV_ROUTE_FL_ACTIVE <v4l2-subdev-routing-flags>` flag, depending
+on driver capabilities.
+
 Interaction between routes, streams, formats and selections
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
@@ -668,3 +689,127 @@ To configure this pipeline, the userspace must take the following steps:
    the configurations along the stream towards the receiver, using
    :ref:`VIDIOC_SUBDEV_S_FMT <VIDIOC_SUBDEV_G_FMT>` ioctls to configure each
    stream endpoint in each sub-device.
+
+Internal pads setup example
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+A simple example of a multiplexed stream setup might be as follows:
+
+- An IMX219 camera sensor source sub-device, with one sink pad (0), one source pad
+  (1), an internal sink pad (2) that represents the source of embedded
+  data. There are two routes, one from the sink pad to the source, and another
+  from the internal sink pad to the source pad. Both streams are always active,
+  i.e. there is no need to separately enable the embedded data stream. The
+  sensor uses the CSI-2 bus.
+
+- A CSI-2 receiver in the SoC (Receiver). The receiver has a single sink pad
+  (pad 0), connected to the bridge, and two source pads (pads 1-2), going to the
+  DMA engine. The receiver demultiplexes the incoming streams to the source
+  pads.
+
+- DMA Engines in the SoC (DMA Engine), one for each stream. Each DMA engine is
+  connected to a single source pad in the receiver.
+
+The sensor, the bridge and the receiver are modeled as V4L2 sub-devices,
+exposed to userspace via /dev/v4l-subdevX device nodes. The DMA engines are
+modeled as V4L2 devices, exposed to userspace via /dev/videoX nodes.
+
+To configure this pipeline, the userspace must take the following steps:
+
+1) Set up media links between entities: connect the sensors to the bridge,
+   bridge to the receiver, and the receiver to the DMA engines. This step does
+   not differ from normal non-multiplexed media controller setup.
+
+2) Configure routing
+
+.. flat-table:: Camera sensor. There are no configurable routes.
+    :header-rows: 1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from the sink pad
+    * - 2/0
+      - 1/1
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from the internal sink pad
+
+.. flat-table:: Receiver routing table. Typically both routes need to be
+		explicitly set.
+    :header-rows:  1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from camera sensor
+    * - 0/1
+      - 2/0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from camera sensor
+
+The options available in sensor's routing configuration are dictated by hardware
+capabilities: typically camera sensors always produce an image data stream while
+it may be possible to enable and disable the embedded data stream.
+
+3) Configure formats and selections
+
+   This example assumes that the formats are propagated from sink pad to the
+   source pad as-is. The tables contain fields of both struct v4l2_subdev_format
+   and struct v4l2_mbus_framefmt.
+
+.. flat-table:: Formats set on the sub-devices. Bold values are set, others are
+                static or propagated. The order is aligned with configured
+                routes.
+    :header-rows: 1
+    :fill-cells:
+
+    * - Sub-device
+      - Pad/Stream
+      - Width
+      - Height
+      - Code
+    * - :rspan:`3` Camera sensor sub-device (IMX219)
+      - 1/0
+      - 3296
+      - 2480
+      - MEDIA_BUS_FMT_SRGGB10
+    * - 0/0
+      - **3296**
+      - **2480**
+      - **MEDIA_BUS_FMT_SRGGB10**
+    * - 2/0
+      - 3296
+      - 2
+      - MEDIA_BUS_FMT_IMX219_EMBEDDED
+    * - 1/1
+      - 3296
+      - 2
+      - MEDIA_BUS_FMT_META_10
+    * - :rspan:`3` Receiver
+      - 0/0
+      - **3296**
+      - **2480**
+      - **MEDIA_BUS_FMT_SRGGB10**
+    * - 1/0
+      - 3296
+      - 2480
+      - MEDIA_BUS_FMT_SRGGB10
+    * - 0/1
+      - **3296**
+      - **2**
+      - **MEDIA_BUS_FMT_META_10**
+    * - 2/0
+      - 3296
+      - 2
+      - MEDIA_BUS_FMT_META_10
+
+The embedded data format does not need to be configured as the format is
+dictated by the pixel data format in this case.
Signature · 1 line
2.39.2
#45 lore raw

Re: [PATCH v8 09/38] media: Documentation: v4l: Document internal source pads

Quoted 42 lines · Sakari Ailus

Document internal source pads, pads that have both SINK and INTERNAL flags set. Use the IMX219 camera sensor as an example.

---
  .../userspace-api/media/v4l/dev-subdev.rst    | 145 ++++++++++++++++++
  1 file changed, 145 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +21 -0
index a387e8a15b8d..1808f40f63e3 100644
@@ -553,6 +553,27 @@ A stream at a specific point in the media pipeline is identified by the
  sub-device and a (pad, stream) pair. For sub-devices that do not support
  multiplexed streams the 'stream' field is always 0.
 
+.. _v4l2-subdev-internal-source-pads:
+
+Internal source pads and routing
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Cases where a single sub-device source pad is traversed by multiple streams, one
+or more of which originate from within the sub-device itself, are special as
+there is no external sink pad for such routes. In those cases, the sources of
+the internally generated streams are represented by internal source pads, which
+are sink pads that have the :ref:`MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`
+pad flag set.
+
+Internal pads have all the properties of an external pad, including formats and
+selections. The format in this case is the source format of the stream. An
+internal pad always has a single stream only (0).
+
+Routes from an internal source pad to an external source pad are typically not
+modifiable but they can be activated and deactivated using the
+:ref:`V4L2_SUBDEV_ROUTE_FL_ACTIVE <v4l2-subdev-routing-flags>` flag, depending
+on driver capabilities.
+
  Interaction between routes, streams, formats and selections
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
@@ -668,3 +689,127 @@ To configure this pipeline, the userspace must take the following steps:
     the configurations along the stream towards the receiver, using

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 699

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +8 -0
     :ref:`VIDIOC_SUBDEV_S_FMT <VIDIOC_SUBDEV_G_FMT>` ioctls to configure each
     stream endpoint in each sub-device.
+
+Internal pads setup example
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+A simple example of a multiplexed stream setup might be as follows:
+
+- An IMX219 camera sensor source sub-device, with one sink pad (0), one source pad
+  (1), an internal sink pad (2) that represents the source of embedded

The pixel pad is an internal pad as well ?

Quoted 24 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +24 -0
+  data. There are two routes, one from the sink pad to the source, and another
+  from the internal sink pad to the source pad. Both streams are always active,
+  i.e. there is no need to separately enable the embedded data stream. The
+  sensor uses the CSI-2 bus.
+
+- A CSI-2 receiver in the SoC (Receiver). The receiver has a single sink pad
+  (pad 0), connected to the bridge, and two source pads (pads 1-2), going to the
+  DMA engine. The receiver demultiplexes the incoming streams to the source
+  pads.
+
+- DMA Engines in the SoC (DMA Engine), one for each stream. Each DMA engine is
+  connected to a single source pad in the receiver.
+
+The sensor, the bridge and the receiver are modeled as V4L2 sub-devices,
+exposed to userspace via /dev/v4l-subdevX device nodes. The DMA engines are
+modeled as V4L2 devices, exposed to userspace via /dev/videoX nodes.
+
+To configure this pipeline, the userspace must take the following steps:
+
+1) Set up media links between entities: connect the sensors to the bridge,
+   bridge to the receiver, and the receiver to the DMA engines. This step does
+   not differ from normal non-multiplexed media controller setup.
+
+2) Configure routing

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 733

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+
+.. flat-table:: Camera sensor. There are no configurable routes.
+    :header-rows: 1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0

- 1/0
- 0/0

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 737

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +4 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from the sink pad
+    * - 2/0
+      - 1/1

- 0/1

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 739

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from the internal sink pad

In latest patch "[PATCH v6 05/15] media: i2c: imx219: Add embedded data support" we have:
- 0 -> source pad
- 1 -> pixel/image
- 2 -> EDATA

This is also what you did for ov2740.

With that fixed:
Reviewed-by Julien Massot <[email protected]> Julien

author maintainer
#46 lore raw

Re: [PATCH v8 09/38] media: Documentation: v4l: Document internal source pads

Hi Julien,

Quoted 46 lines · Julien Massot
Quoted 43 lines · Sakari Ailus

Document internal source pads, pads that have both SINK and INTERNAL flags set. Use the IMX219 camera sensor as an example.

---
  .../userspace-api/media/v4l/dev-subdev.rst    | 145 ++++++++++++++++++
  1 file changed, 145 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +22 -0
index a387e8a15b8d..1808f40f63e3 100644
@@ -553,6 +553,27 @@ A stream at a specific point in the media pipeline is identified by the
  sub-device and a (pad, stream) pair. For sub-devices that do not support
  multiplexed streams the 'stream' field is always 0.
+.. _v4l2-subdev-internal-source-pads:
+
+Internal source pads and routing
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Cases where a single sub-device source pad is traversed by multiple streams, one
+or more of which originate from within the sub-device itself, are special as
+there is no external sink pad for such routes. In those cases, the sources of
+the internally generated streams are represented by internal source pads, which
+are sink pads that have the :ref:`MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`
+pad flag set.
+
+Internal pads have all the properties of an external pad, including formats and
+selections. The format in this case is the source format of the stream. An
+internal pad always has a single stream only (0).
+
+Routes from an internal source pad to an external source pad are typically not
+modifiable but they can be activated and deactivated using the
+:ref:`V4L2_SUBDEV_ROUTE_FL_ACTIVE <v4l2-subdev-routing-flags>` flag, depending
+on driver capabilities.
+
  Interaction between routes, streams, formats and selections
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -668,3 +689,127 @@ To configure this pipeline, the userspace must take the following steps:
     the configurations along the stream towards the receiver, using
     :ref:`VIDIOC_SUBDEV_S_FMT <VIDIOC_SUBDEV_G_FMT>` ioctls to configure each
     stream endpoint in each sub-device.
+
Quoted 8 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 699

Quoted 7 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +7 -0
+Internal pads setup example
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+A simple example of a multiplexed stream setup might be as follows:
+
+- An IMX219 camera sensor source sub-device, with one sink pad (0), one source pad
+  (1), an internal sink pad (2) that represents the source of embedded

The pixel pad is an internal pad as well ?

How about:

- An IMX219 camera sensor source sub-device, with one internal sink pad
(0) for image data, one source pad (1), an internal sink pad (2) that represents the source of embedded

Quoted 45 lines
Quoted 25 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +24 -0
+  data. There are two routes, one from the sink pad to the source, and another
+  from the internal sink pad to the source pad. Both streams are always active,
+  i.e. there is no need to separately enable the embedded data stream. The
+  sensor uses the CSI-2 bus.
+
+- A CSI-2 receiver in the SoC (Receiver). The receiver has a single sink pad
+  (pad 0), connected to the bridge, and two source pads (pads 1-2), going to the
+  DMA engine. The receiver demultiplexes the incoming streams to the source
+  pads.
+
+- DMA Engines in the SoC (DMA Engine), one for each stream. Each DMA engine is
+  connected to a single source pad in the receiver.
+
+The sensor, the bridge and the receiver are modeled as V4L2 sub-devices,
+exposed to userspace via /dev/v4l-subdevX device nodes. The DMA engines are
+modeled as V4L2 devices, exposed to userspace via /dev/videoX nodes.
+
+To configure this pipeline, the userspace must take the following steps:
+
+1) Set up media links between entities: connect the sensors to the bridge,
+   bridge to the receiver, and the receiver to the DMA engines. This step does
+   not differ from normal non-multiplexed media controller setup.
+
+2) Configure routing

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 733

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+
+.. flat-table:: Camera sensor. There are no configurable routes.
+    :header-rows: 1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0

- 1/0
- 0/0

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 737

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +4 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from the sink pad
+    * - 2/0
+      - 1/1

- 0/1

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 739

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from the internal sink pad
Quoted 8 lines

In latest patch "[PATCH v6 05/15] media: i2c: imx219: Add embedded data support"
we have:
- 0 -> source pad
- 1 -> pixel/image
- 2 -> EDATA

This is also what you did for ov2740.

Yes, these were leftovers from the CCS example. I've fixed them.

Quoted 3 lines

With that fixed:
Reviewed-by Julien Massot <[email protected]>

Thank you!

Signature · 3 lines
Regards,

Sakari Ailus
#47 lore raw

Re: [PATCH v8 09/38] media: Documentation: v4l: Document internal source pads

Hi Sakari,

Quoted 58 lines · Sakari Ailus

Hi Julien,

Quoted 46 lines · Julien Massot
Quoted 43 lines · Sakari Ailus

Document internal source pads, pads that have both SINK and INTERNAL flags set. Use the IMX219 camera sensor as an example.

---
   .../userspace-api/media/v4l/dev-subdev.rst    | 145 ++++++++++++++++++
   1 file changed, 145 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +22 -0
index a387e8a15b8d..1808f40f63e3 100644
@@ -553,6 +553,27 @@ A stream at a specific point in the media pipeline is identified by the
   sub-device and a (pad, stream) pair. For sub-devices that do not support
   multiplexed streams the 'stream' field is always 0.
+.. _v4l2-subdev-internal-source-pads:
+
+Internal source pads and routing
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Cases where a single sub-device source pad is traversed by multiple streams, one
+or more of which originate from within the sub-device itself, are special as
+there is no external sink pad for such routes. In those cases, the sources of
+the internally generated streams are represented by internal source pads, which
+are sink pads that have the :ref:`MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`
+pad flag set.
+
+Internal pads have all the properties of an external pad, including formats and
+selections. The format in this case is the source format of the stream. An
+internal pad always has a single stream only (0).
+
+Routes from an internal source pad to an external source pad are typically not
+modifiable but they can be activated and deactivated using the
+:ref:`V4L2_SUBDEV_ROUTE_FL_ACTIVE <v4l2-subdev-routing-flags>` flag, depending
+on driver capabilities.
+
   Interaction between routes, streams, formats and selections
   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -668,3 +689,127 @@ To configure this pipeline, the userspace must take the following steps:
      the configurations along the stream towards the receiver, using
      :ref:`VIDIOC_SUBDEV_S_FMT <VIDIOC_SUBDEV_G_FMT>` ioctls to configure each
      stream endpoint in each sub-device.
+
Quoted 8 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 699

Quoted 7 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +7 -0
+Internal pads setup example
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+A simple example of a multiplexed stream setup might be as follows:
+
+- An IMX219 camera sensor source sub-device, with one sink pad (0), one source pad
+  (1), an internal sink pad (2) that represents the source of embedded

The pixel pad is an internal pad as well ?

Quoted 5 lines

How about:

- An IMX219 camera sensor source sub-device, with one internal sink pad
(0) for image data, one source pad (1), an internal sink pad (2) that represents the source of embedded

sure looks good to me!

Quoted 63 lines
Quoted 46 lines
Quoted 26 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +24 -0
+  data. There are two routes, one from the sink pad to the source, and another
+  from the internal sink pad to the source pad. Both streams are always active,
+  i.e. there is no need to separately enable the embedded data stream. The
+  sensor uses the CSI-2 bus.
+
+- A CSI-2 receiver in the SoC (Receiver). The receiver has a single sink pad
+  (pad 0), connected to the bridge, and two source pads (pads 1-2), going to the
+  DMA engine. The receiver demultiplexes the incoming streams to the source
+  pads.
+
+- DMA Engines in the SoC (DMA Engine), one for each stream. Each DMA engine is
+  connected to a single source pad in the receiver.
+
+The sensor, the bridge and the receiver are modeled as V4L2 sub-devices,
+exposed to userspace via /dev/v4l-subdevX device nodes. The DMA engines are
+modeled as V4L2 devices, exposed to userspace via /dev/videoX nodes.
+
+To configure this pipeline, the userspace must take the following steps:
+
+1) Set up media links between entities: connect the sensors to the bridge,
+   bridge to the receiver, and the receiver to the DMA engines. This step does
+   not differ from normal non-multiplexed media controller setup.
+
+2) Configure routing

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 733

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+
+.. flat-table:: Camera sensor. There are no configurable routes.
+    :header-rows: 1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0

- 1/0
- 0/0

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 737

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +4 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from the sink pad
+    * - 2/0
+      - 1/1

- 0/1

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 739

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from the internal sink pad
Quoted 8 lines

In latest patch "[PATCH v6 05/15] media: i2c: imx219: Add embedded data support"
we have:
- 0 -> source pad
- 1 -> pixel/image
- 2 -> EDATA

This is also what you did for ov2740.

Yes, these were leftovers from the CCS example. I've fixed them.

Quoted 3 lines

With that fixed:
Reviewed-by Julien Massot <[email protected]>

Thank you!

maintainer
#48 lore raw

Re: [PATCH v8 09/38] media: Documentation: v4l: Document internal source pads

In reply to Sakari Ailus
Quoted 20 lines · Sakari Ailus
Quoted 19 lines · Julien Massot
Quoted 18 lines · Sakari Ailus

Document internal source pads, pads that have both SINK and INTERNAL flags set. Use the IMX219 camera sensor as an example.

---
  .../userspace-api/media/v4l/dev-subdev.rst    | 145 ++++++++++++++++++
  1 file changed, 145 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -0
index a387e8a15b8d..1808f40f63e3 100644
@@ -553,6 +553,27 @@ A stream at a specific point in the media pipeline is identified by the
  sub-device and a (pad, stream) pair. For sub-devices that do not support
  multiplexed streams the 'stream' field is always 0.
+.. _v4l2-subdev-internal-source-pads:
+
+Internal source pads and routing
Quoted 8 lines
Quoted 8 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst A stream at a specific point in the media pipeline is identified by the · line 566

Quoted 8 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +8 -0
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Cases where a single sub-device source pad is traversed by multiple streams, one
+or more of which originate from within the sub-device itself, are special as
+there is no external sink pad for such routes. In those cases, the sources of
+the internally generated streams are represented by internal source pads, which
+are sink pads that have the :ref:`MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`
+pad flag set.

This seems to answer a question in a previous patch, where I wondered if you meant "internal sink pad" instead of "internal source pad".

Given that this is will most likely be a source of confusion, I think you should explain this very clearly.

.. note::

Internal source pads model the source of data *within* a sub-device, and are therefore conceptually located on the "sink" side of the sub-device from the point of view of sub-device internal routes. For this reason, and despite their name, they use the :ref:MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK> flag. When used in the routing API, they are addressed through the ``sink_pad`` and ``sink_stream`` fields of the :c:type:v4l2_subdev_route structure.

I hope this will be enough... We'll need to be extra careful during review in the beginning, making sure to always use the right terms.

Quoted 27 lines
Quoted 18 lines
Quoted 18 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +12 -0
+
+Internal pads have all the properties of an external pad, including formats and
+selections. The format in this case is the source format of the stream. An
+internal pad always has a single stream only (0).
+
+Routes from an internal source pad to an external source pad are typically not
+modifiable but they can be activated and deactivated using the
+:ref:`V4L2_SUBDEV_ROUTE_FL_ACTIVE <v4l2-subdev-routing-flags>` flag, depending
+on driver capabilities.
+
  Interaction between routes, streams, formats and selections
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -668,3 +689,127 @@ To configure this pipeline, the userspace must take the following steps:
     the configurations along the stream towards the receiver, using
     :ref:`VIDIOC_SUBDEV_S_FMT <VIDIOC_SUBDEV_G_FMT>` ioctls to configure each
     stream endpoint in each sub-device.
+
+Internal pads setup example
Quoted 8 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 699

Quoted 6 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +6 -0
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+A simple example of a multiplexed stream setup might be as follows:
+
+- An IMX219 camera sensor source sub-device, with one sink pad (0), one source pad
+  (1), an internal sink pad (2) that represents the source of embedded

The pixel pad is an internal pad as well ?

Quoted 5 lines

How about:

- An IMX219 camera sensor source sub-device, with one internal sink pad
(0) for image data, one source pad (1), an internal sink pad (2) that represents the source of embedded

Sounds better, except you're now talking about internal sink pads, not internal source pads :-) We have to pick one option and stick to it. Same below.

Quoted 1 line
Quoted 1 line

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 700

Quoted 1 line
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+  data. There are two routes, one from the sink pad to the source, and another

"from the internal image source pad to the source" sounds weird, so you may need to write "from the internal image source pad to the external source pad". If we instead talked about "internal sink pads" to model the source of the streams, you could write "from the internal image sink pad to the source pad".

Quoted 6 lines
Quoted 6 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 706

Quoted 6 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +6 -0
+  from the internal sink pad to the source pad. Both streams are always active,
+  i.e. there is no need to separately enable the embedded data stream. The
+  sensor uses the CSI-2 bus.
+
+- A CSI-2 receiver in the SoC (Receiver). The receiver has a single sink pad
+  (pad 0), connected to the bridge, and two source pads (pads 1-2), going to the

I think you meant s/bridge/sensor/

Quoted 1 line
Quoted 1 line

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 707

Quoted 1 line
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+  DMA engine. The receiver demultiplexes the incoming streams to the source

And "DMA engines" here ?

Quoted 3 lines
Quoted 3 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 710

Quoted 3 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -0
+  pads.
+
+- DMA Engines in the SoC (DMA Engine), one for each stream. Each DMA engine is

s/Engines/engine/

Quoted 1 line
Quoted 1 line

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 711

Quoted 1 line
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+  connected to a single source pad in the receiver.

s/in the/of the/

Quoted 2 lines
Quoted 2 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 713

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+
+The sensor, the bridge and the receiver are modeled as V4L2 sub-devices,

What bridge ? Same below.

Quoted 49 lines
Quoted 34 lines
Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+exposed to userspace via /dev/v4l-subdevX device nodes. The DMA engines are
+modeled as V4L2 devices, exposed to userspace via /dev/videoX nodes.
+
+To configure this pipeline, the userspace must take the following steps:
+
+1) Set up media links between entities: connect the sensors to the bridge,
+   bridge to the receiver, and the receiver to the DMA engines. This step does
+   not differ from normal non-multiplexed media controller setup.
+
+2) Configure routing

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 733

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+
+.. flat-table:: Camera sensor. There are no configurable routes.
+    :header-rows: 1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0

- 1/0
- 0/0

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 737

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +4 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from the sink pad
+    * - 2/0
+      - 1/1

- 0/1

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 739

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from the internal sink pad
Quoted 8 lines

In latest patch "[PATCH v6 05/15] media: i2c: imx219: Add embedded data support"
we have:
- 0 -> source pad
- 1 -> pixel/image
- 2 -> EDATA

This is also what you did for ov2740.

Yes, these were leftovers from the CCS example. I've fixed them.

Quoted 2 lines

With that fixed:
Reviewed-by Julien Massot <[email protected]>

Thank you!

Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#49 lore raw

Re: [PATCH v8 09/38] media: Documentation: v4l: Document internal source pads

Hi Laurent,

Quoted 45 lines · Laurent Pinchart
Quoted 20 lines · Sakari Ailus
Quoted 19 lines · Julien Massot
Quoted 18 lines · Sakari Ailus

Document internal source pads, pads that have both SINK and INTERNAL flags set. Use the IMX219 camera sensor as an example.

---
  .../userspace-api/media/v4l/dev-subdev.rst    | 145 ++++++++++++++++++
  1 file changed, 145 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -0
index a387e8a15b8d..1808f40f63e3 100644
@@ -553,6 +553,27 @@ A stream at a specific point in the media pipeline is identified by the
  sub-device and a (pad, stream) pair. For sub-devices that do not support
  multiplexed streams the 'stream' field is always 0.
+.. _v4l2-subdev-internal-source-pads:
+
+Internal source pads and routing
Quoted 8 lines
Quoted 8 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst A stream at a specific point in the media pipeline is identified by the · line 566

Quoted 8 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +8 -0
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Cases where a single sub-device source pad is traversed by multiple streams, one
+or more of which originate from within the sub-device itself, are special as
+there is no external sink pad for such routes. In those cases, the sources of
+the internally generated streams are represented by internal source pads, which
+are sink pads that have the :ref:`MEDIA_PAD_FL_INTERNAL <MEDIA-PAD-FL-INTERNAL>`
+pad flag set.

This seems to answer a question in a previous patch, where I wondered if you meant "internal sink pad" instead of "internal source pad".

Given that this is will most likely be a source of confusion, I think you should explain this very clearly.

.. note::

Internal source pads model the source of data *within* a sub-device, and are therefore conceptually located on the "sink" side of the sub-device from the point of view of sub-device internal routes. For this reason, and despite their name, they use the :ref:MEDIA_PAD_FL_SINK <MEDIA-PAD-FL-SINK> flag. When used in the routing API, they are addressed through the ``sink_pad`` and ``sink_stream`` fields of the :c:type:v4l2_subdev_route structure.

Quoted 2 lines

I hope this will be enough... We'll need to be extra careful during review in the beginning, making sure to always use the right terms.

Let's see if we need something like this with the "internal source pad" concept removed. But let's revisit the topic in v9.

Quoted 30 lines
Quoted 30 lines
Quoted 19 lines
Quoted 19 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +12 -0
+
+Internal pads have all the properties of an external pad, including formats and
+selections. The format in this case is the source format of the stream. An
+internal pad always has a single stream only (0).
+
+Routes from an internal source pad to an external source pad are typically not
+modifiable but they can be activated and deactivated using the
+:ref:`V4L2_SUBDEV_ROUTE_FL_ACTIVE <v4l2-subdev-routing-flags>` flag, depending
+on driver capabilities.
+
  Interaction between routes, streams, formats and selections
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -668,3 +689,127 @@ To configure this pipeline, the userspace must take the following steps:
     the configurations along the stream towards the receiver, using
     :ref:`VIDIOC_SUBDEV_S_FMT <VIDIOC_SUBDEV_G_FMT>` ioctls to configure each
     stream endpoint in each sub-device.
+
+Internal pads setup example
Quoted 8 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 699

Quoted 6 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +6 -0
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+A simple example of a multiplexed stream setup might be as follows:
+
+- An IMX219 camera sensor source sub-device, with one sink pad (0), one source pad
+  (1), an internal sink pad (2) that represents the source of embedded

The pixel pad is an internal pad as well ?

How about:

Quoted 7 lines
Quoted 3 lines

- An IMX219 camera sensor source sub-device, with one internal sink pad
(0) for image data, one source pad (1), an internal sink pad (2) that represents the source of embedded

Sounds better, except you're now talking about internal sink pads, not internal source pads :-) We have to pick one option and stick to it. Same below.

Let's discuss this in the context of patch 8.

Quoted 8 lines
Quoted 2 lines
Quoted 2 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 700

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+  data. There are two routes, one from the sink pad to the source, and another

"from the internal image source pad to the source" sounds weird, so you may need to write "from the internal image source pad to the external source pad". If we instead talked about "internal sink pads" to model the source of the streams, you could write "from the internal image sink pad to the source pad".

I think this text was from the original CCS example. I'll update it for IMX290.

Quoted 9 lines
Quoted 7 lines
Quoted 7 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 706

Quoted 7 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +6 -0
+  from the internal sink pad to the source pad. Both streams are always active,
+  i.e. there is no need to separately enable the embedded data stream. The
+  sensor uses the CSI-2 bus.
+
+- A CSI-2 receiver in the SoC (Receiver). The receiver has a single sink pad
+  (pad 0), connected to the bridge, and two source pads (pads 1-2), going to the

I think you meant s/bridge/sensor/

Thanks. In an earlier version there was a bridge, this was a leftover from that it seems. I'll check if there are other remaining references, too.

Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 707

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+  DMA engine. The receiver demultiplexes the incoming streams to the source

And "DMA engines" here ?

Correct.

Quoted 6 lines
Quoted 4 lines
Quoted 4 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 710

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -0
+  pads.
+
+- DMA Engines in the SoC (DMA Engine), one for each stream. Each DMA engine is

s/Engines/engine/

Sounds good.

Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 711

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +1 -0
+  connected to a single source pad in the receiver.

s/in the/of the/

Works for me.

Quoted 56 lines
Quoted 3 lines
Quoted 3 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 713

Quoted 3 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+
+The sensor, the bridge and the receiver are modeled as V4L2 sub-devices,

What bridge ? Same below.

Quoted 49 lines
Quoted 34 lines
Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+exposed to userspace via /dev/v4l-subdevX device nodes. The DMA engines are
+modeled as V4L2 devices, exposed to userspace via /dev/videoX nodes.
+
+To configure this pipeline, the userspace must take the following steps:
+
+1) Set up media links between entities: connect the sensors to the bridge,
+   bridge to the receiver, and the receiver to the DMA engines. This step does
+   not differ from normal non-multiplexed media controller setup.
+
+2) Configure routing

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 733

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+
+.. flat-table:: Camera sensor. There are no configurable routes.
+    :header-rows: 1
+
+    * - Sink Pad/Stream
+      - Source Pad/Stream
+      - Routing Flags
+      - Comments
+    * - 0/0
+      - 1/0

- 1/0
- 0/0

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 737

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +4 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Pixel data stream from the sink pad
+    * - 2/0
+      - 1/1

- 0/1

Documentation/userspace-api/media/v4l/dev-subdev.rst To configure this pipeline, the userspace must take the following steps: · line 739

Quoted 2 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
+      - V4L2_SUBDEV_ROUTE_FL_ACTIVE
+      - Metadata stream from the internal sink pad
Quoted 8 lines

In latest patch "[PATCH v6 05/15] media: i2c: imx219: Add embedded data support"
we have:
- 0 -> source pad
- 1 -> pixel/image
- 2 -> EDATA

This is also what you did for ov2740.

Yes, these were leftovers from the CCS example. I've fixed them.

Quoted 2 lines

With that fixed:
Reviewed-by Julien Massot <[email protected]>

Thank you!

Signature · 3 lines
Kind regards,

Sakari Ailus
author maintainer
#50 lore raw

PATCH v8 10/38 media: Documentation: Document S_ROUTING behaviour

In reply to Sakari Ailus

Document S_ROUTING behaviour for different devices.

Generally in devices that produce streams the streams are static and some can be enabled and disabled, whereas in devices that just transport them or write them to memory, more configurability is allowed. Document this.

---
 .../userspace-api/media/v4l/dev-subdev.rst    | 24 +++++++++++++++++++
 1 file changed, 24 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +24 -0
index 1808f40f63e3..08495cc6f4a6 100644
@@ -593,6 +593,30 @@ Any configurations of a stream within a pad, such as format or selections,
 are independent of similar configurations on other streams. This is
 subject to change in the future.
 
+Device types and routing setup
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Different kinds of sub-devices have differing behaviour for route activation,
+depending on the hardware. In all cases, however, only routes that have the
+``V4L2_SUBDEV_STREAM_FL_ACTIVE`` flag set are active.
+
+Devices generating the streams may allow enabling and disabling some of the
+routes or the configuration is fixed. If the routes can be disabled, not
+declaring the routes (or declaring them without
+``VIDIOC_SUBDEV_STREAM_FL_ACTIVE`` flag set) in ``VIDIOC_SUBDEV_S_ROUTING`` will
+disable the routes while the sub-device driver retains the streams and their
+configuration. The ``VIDIOC_SUBDEV_S_ROUTING`` will still return such routes
+back to the user in the routes array, with the ``V4L2_SUBDEV_STREAM_FL_ACTIVE``
+flag unset.
+
+Devices transporting the streams almost always have more configurability with
+respect to routing. Typically any route between the sub-device's sink and source
+pads is possible, and multiple routes (usually up to certain limited number) may
+be active simultaneously. For such devices, no routes are created by the driver
+and user-created routes are fully replaced when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called on the sub-device. Such newly created routes have the device's default
+configuration for format and selection rectangles.
+
 Configuring streams
 ^^^^^^^^^^^^^^^^^^^
 
Signature · 1 line
2.39.2
Reviewed-by on v8 10/38
#51 lore raw

Re: [PATCH v8 10/38] media: Documentation: Document S_ROUTING behaviour

Quoted 36 lines · Sakari Ailus

Document S_ROUTING behaviour for different devices.

Generally in devices that produce streams the streams are static and some can be enabled and disabled, whereas in devices that just transport them or write them to memory, more configurability is allowed. Document this.

---
  .../userspace-api/media/v4l/dev-subdev.rst    | 24 +++++++++++++++++++
  1 file changed, 24 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +17 -0
index 1808f40f63e3..08495cc6f4a6 100644
@@ -593,6 +593,30 @@ Any configurations of a stream within a pad, such as format or selections,
  are independent of similar configurations on other streams. This is
  subject to change in the future.
 
+Device types and routing setup
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Different kinds of sub-devices have differing behaviour for route activation,
+depending on the hardware. In all cases, however, only routes that have the
+``V4L2_SUBDEV_STREAM_FL_ACTIVE`` flag set are active.
+
+Devices generating the streams may allow enabling and disabling some of the
+routes or the configuration is fixed. If the routes can be disabled, not
+declaring the routes (or declaring them without
+``VIDIOC_SUBDEV_STREAM_FL_ACTIVE`` flag set) in ``VIDIOC_SUBDEV_S_ROUTING`` will
+disable the routes while the sub-device driver retains the streams and their
+configuration. The ``VIDIOC_SUBDEV_S_ROUTING`` will still return such routes
+back to the user in the routes array, with the ``V4L2_SUBDEV_STREAM_FL_ACTIVE``
+flag unset.
+
+Devices transporting the streams almost always have more configurability with

Documentation/userspace-api/media/v4l/dev-subdev.rst Any configurations of a stream within a pad, such as format or selections, · line 622

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +7 -0
+respect to routing. Typically any route between the sub-device's sink and source
+pads is possible, and multiple routes (usually up to certain limited number) may
+be active simultaneously. For such devices, no routes are created by the driver
+and user-created routes are fully replaced when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called on the sub-device. Such newly created routes have the device's default
+configuration for format and selection rectangles.
+
  Configuring streams
  ^^^^^^^^^^^^^^^^^^^
 

Julien

maintainer
#52 lore raw

Re: [PATCH v8 10/38] media: Documentation: Document S_ROUTING behaviour

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 21 lines · Sakari Ailus

Document S_ROUTING behaviour for different devices.

Generally in devices that produce streams the streams are static and some can be enabled and disabled, whereas in devices that just transport them or write them to memory, more configurability is allowed. Document this.

---
 .../userspace-api/media/v4l/dev-subdev.rst    | 24 +++++++++++++++++++
 1 file changed, 24 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +2 -0
index 1808f40f63e3..08495cc6f4a6 100644
@@ -593,6 +593,30 @@ Any configurations of a stream within a pad, such as format or selections,
 are independent of similar configurations on other streams. This is
 subject to change in the future.
 
+Device types and routing setup
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Documentation/userspace-api/media/v4l/dev-subdev.rst Any configurations of a stream within a pad, such as format or selections, · line 607

Quoted 10 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +10 -0
+
+Different kinds of sub-devices have differing behaviour for route activation,
+depending on the hardware. In all cases, however, only routes that have the
+``V4L2_SUBDEV_STREAM_FL_ACTIVE`` flag set are active.
+
+Devices generating the streams may allow enabling and disabling some of the
+routes or the configuration is fixed. If the routes can be disabled, not
+declaring the routes (or declaring them without
+``VIDIOC_SUBDEV_STREAM_FL_ACTIVE`` flag set) in ``VIDIOC_SUBDEV_S_ROUTING`` will
+disable the routes while the sub-device driver retains the streams and their

No idea what you mean by "retains the streams and their configuration".

Documentation/userspace-api/media/v4l/dev-subdev.rst Any configurations of a stream within a pad, such as format or selections, · line 610

Quoted 3 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -0
+configuration. The ``VIDIOC_SUBDEV_S_ROUTING`` will still return such routes
+back to the user in the routes array, with the ``V4L2_SUBDEV_STREAM_FL_ACTIVE``
+flag unset.

I find this quite confusing. There's no explanation as to why the API behaves this way. I think you should rewrite this documentation, trying to evaluate it from the point of view of someone who doesn't know about the routing API.

Quoted 12 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +9 -0
+
+Devices transporting the streams almost always have more configurability with
+respect to routing. Typically any route between the sub-device's sink and source
+pads is possible, and multiple routes (usually up to certain limited number) may
+be active simultaneously. For such devices, no routes are created by the driver
+and user-created routes are fully replaced when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called on the sub-device. Such newly created routes have the device's default
+configuration for format and selection rectangles.
+
 Configuring streams
 ^^^^^^^^^^^^^^^^^^^
 
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#53 lore raw

Re: [PATCH v8 10/38] media: Documentation: Document S_ROUTING behaviour

Hi Laurent,

Quoted 24 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 19 lines · Sakari Ailus

Document S_ROUTING behaviour for different devices.

Generally in devices that produce streams the streams are static and some can be enabled and disabled, whereas in devices that just transport them or write them to memory, more configurability is allowed. Document this.

---
 .../userspace-api/media/v4l/dev-subdev.rst    | 24 +++++++++++++++++++
 1 file changed, 24 insertions(+)
Documentation/userspace-api/media/v4l/dev-subdev.rst +0 -0
index 1808f40f63e3..08495cc6f4a6 100644
@@ -593,6 +593,30 @@ Any configurations of a stream within a pad, such as format or selections,
 are independent of similar configurations on other streams. This is
 subject to change in the future.
 
Quoted 14 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst Any configurations of a stream within a pad, such as format or selections, · line 607

Quoted 12 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +12 -0
+Device types and routing setup
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Different kinds of sub-devices have differing behaviour for route activation,
+depending on the hardware. In all cases, however, only routes that have the
+``V4L2_SUBDEV_STREAM_FL_ACTIVE`` flag set are active.
+
+Devices generating the streams may allow enabling and disabling some of the
+routes or the configuration is fixed. If the routes can be disabled, not
+declaring the routes (or declaring them without
+``VIDIOC_SUBDEV_STREAM_FL_ACTIVE`` flag set) in ``VIDIOC_SUBDEV_S_ROUTING`` will
+disable the routes while the sub-device driver retains the streams and their

No idea what you mean by "retains the streams and their configuration".

I'll add that this means format and selection configuration.

Quoted 9 lines

Documentation/userspace-api/media/v4l/dev-subdev.rst Any configurations of a stream within a pad, such as format or selections, · line 610

Quoted 4 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +3 -0
+configuration. The ``VIDIOC_SUBDEV_S_ROUTING`` will still return such routes
+back to the user in the routes array, with the ``V4L2_SUBDEV_STREAM_FL_ACTIVE``
+flag unset.

I find this quite confusing. There's no explanation as to why the API behaves this way. I think you should rewrite this documentation, trying to evaluate it from the point of view of someone who doesn't know about the routing API.

I wonder if this would be easier for the user if such routes had a flag that would tell they're always there. It could be called e.g. STATIC. Then the user would be able to expect certain behaviour before disabling a link.

The purpose is really to avoid removing streams that the user would not know how to re-create in order to make the device function again.

Quoted 14 lines
Quoted 13 lines
Documentation/userspace-api/media/v4l/dev-subdev.rst +9 -0
+
+Devices transporting the streams almost always have more configurability with
+respect to routing. Typically any route between the sub-device's sink and source
+pads is possible, and multiple routes (usually up to certain limited number) may
+be active simultaneously. For such devices, no routes are created by the driver
+and user-created routes are fully replaced when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called on the sub-device. Such newly created routes have the device's default
+configuration for format and selection rectangles.
+
 Configuring streams
 ^^^^^^^^^^^^^^^^^^^
 
Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#54 lore raw

PATCH v8 11/38 media: v4l: subdev: Add a function to lock two sub-device states, use it

In reply to Sakari Ailus

Add two new functions, v4l2_subdev_lock_states() and v4l2_subdev_unclock_states(), to acquire and release the state of two sub-devices. They differ from calling v4l2_subdev_{un,}lock_state() so that if the two states share the same lock, the lock is acquired only once.

Also use the new functions in v4l2_subdev_link_validate().

---
 drivers/media/v4l2-core/v4l2-subdev.c | 12 +++-----
 include/media/v4l2-subdev.h           | 40 +++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 8 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +4 -8
index 4c6198c48dd6..6c34df19ea2f 100644
@@ -1400,17 +1400,13 @@ int v4l2_subdev_link_validate(struct media_link *link)
 
 	states_locked = sink_state && source_state;
 
-	if (states_locked) {
-		v4l2_subdev_lock_state(sink_state);
-		v4l2_subdev_lock_state(source_state);
-	}
+	if (states_locked)
+		v4l2_subdev_lock_states(sink_state, source_state);
 
 	ret = v4l2_subdev_link_validate_locked(link, states_locked);
 
-	if (states_locked) {
-		v4l2_subdev_unlock_state(sink_state);
-		v4l2_subdev_unlock_state(source_state);
-	}
+	if (states_locked)
+		v4l2_subdev_unlock_states(sink_state, source_state);
 
 	return ret;
 }
include/media/v4l2-subdev.h +40 -0
index a9e6b8146279..9cce48365975 100644
@@ -1724,6 +1724,46 @@ static inline void v4l2_subdev_unlock_state(struct v4l2_subdev_state *state)
 	mutex_unlock(state->lock);
 }
 
+/**
+ * v4l2_subdev_lock_states - Lock two sub-device states
+ * @state1: One subdevice state
+ * @state2: The other subdevice state
+ *
+ * Locks the state of two sub-devices.
+ *
+ * The states must be unlocked with v4l2_subdev_unlock_states() after use.
+ *
+ * This differs from calling v4l2_subdev_lock_state() on both states so that if
+ * the states share the same lock, the lock is acquired only once (so no
+ * deadlock occurs). The caller is responsible for ensuring the locks will
+ * always be acquired in the same order.
+ */
+static inline void v4l2_subdev_lock_states(struct v4l2_subdev_state *state1,
+					   struct v4l2_subdev_state *state2)
+{
+	mutex_lock(state1->lock);
+	if (state1->lock != state2->lock)
+		mutex_lock(state2->lock);
+}
+
+/**
+ * v4l2_subdev_unlock_states() - Unlock two sub-device states
+ * @state1: One subdevice state
+ * @state2: The other subdevice state
+ *
+ * Unlocks the state of two sub-devices.
+ *
+ * This differs from calling v4l2_subdev_unlock_state() on both states so that
+ * if the states share the same lock, the lock is released only once.
+ */
+static inline void v4l2_subdev_unlock_states(struct v4l2_subdev_state *state1,
+					     struct v4l2_subdev_state *state2)
+{
+	mutex_unlock(state1->lock);
+	if (state1->lock != state2->lock)
+		mutex_unlock(state2->lock);
+}
+
 /**
  * v4l2_subdev_get_unlocked_active_state() - Checks that the active subdev state
  *					     is unlocked and returns it
Signature · 1 line
2.39.2
Reviewed-by on v8 11/38
#55 lore raw

Re: [PATCH v8 11/38] media: v4l: subdev: Add a function to lock two sub-device states, use it

Quoted 9 lines · Sakari Ailus

Add two new functions, v4l2_subdev_lock_states() and v4l2_subdev_unclock_states(), to acquire and release the state of two sub-devices. They differ from calling v4l2_subdev_{un,}lock_state() so that if the two states share the same lock, the lock is acquired only once.

Also use the new functions in v4l2_subdev_link_validate().

Quoted 82 lines
---
  drivers/media/v4l2-core/v4l2-subdev.c | 12 +++-----
  include/media/v4l2-subdev.h           | 40 +++++++++++++++++++++++++++
  2 files changed, 44 insertions(+), 8 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +4 -8
index 4c6198c48dd6..6c34df19ea2f 100644
@@ -1400,17 +1400,13 @@ int v4l2_subdev_link_validate(struct media_link *link)
 
  	states_locked = sink_state && source_state;
 
-	if (states_locked) {
-		v4l2_subdev_lock_state(sink_state);
-		v4l2_subdev_lock_state(source_state);
-	}
+	if (states_locked)
+		v4l2_subdev_lock_states(sink_state, source_state);
 
  	ret = v4l2_subdev_link_validate_locked(link, states_locked);
 
-	if (states_locked) {
-		v4l2_subdev_unlock_state(sink_state);
-		v4l2_subdev_unlock_state(source_state);
-	}
+	if (states_locked)
+		v4l2_subdev_unlock_states(sink_state, source_state);
 
  	return ret;
  }
include/media/v4l2-subdev.h +40 -0
index a9e6b8146279..9cce48365975 100644
@@ -1724,6 +1724,46 @@ static inline void v4l2_subdev_unlock_state(struct v4l2_subdev_state *state)
  	mutex_unlock(state->lock);
  }
 
+/**
+ * v4l2_subdev_lock_states - Lock two sub-device states
+ * @state1: One subdevice state
+ * @state2: The other subdevice state
+ *
+ * Locks the state of two sub-devices.
+ *
+ * The states must be unlocked with v4l2_subdev_unlock_states() after use.
+ *
+ * This differs from calling v4l2_subdev_lock_state() on both states so that if
+ * the states share the same lock, the lock is acquired only once (so no
+ * deadlock occurs). The caller is responsible for ensuring the locks will
+ * always be acquired in the same order.
+ */
+static inline void v4l2_subdev_lock_states(struct v4l2_subdev_state *state1,
+					   struct v4l2_subdev_state *state2)
+{
+	mutex_lock(state1->lock);
+	if (state1->lock != state2->lock)
+		mutex_lock(state2->lock);
+}
+
+/**
+ * v4l2_subdev_unlock_states() - Unlock two sub-device states
+ * @state1: One subdevice state
+ * @state2: The other subdevice state
+ *
+ * Unlocks the state of two sub-devices.
+ *
+ * This differs from calling v4l2_subdev_unlock_state() on both states so that
+ * if the states share the same lock, the lock is released only once.
+ */
+static inline void v4l2_subdev_unlock_states(struct v4l2_subdev_state *state1,
+					     struct v4l2_subdev_state *state2)
+{
+	mutex_unlock(state1->lock);
+	if (state1->lock != state2->lock)
+		mutex_unlock(state2->lock);
+}
+
  /**
   * v4l2_subdev_get_unlocked_active_state() - Checks that the active subdev state
   *					     is unlocked and returns it
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer Reviewed-by on v8 11/38
#56 lore raw

Re: [PATCH v8 11/38] media: v4l: subdev: Add a function to lock two sub-device states, use it

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 5 lines · Sakari Ailus

Add two new functions, v4l2_subdev_lock_states() and v4l2_subdev_unclock_states(), to acquire and release the state of two sub-devices. They differ from calling v4l2_subdev_{un,}lock_state() so that if the two states share the same lock, the lock is acquired only once.

A bit of a hack, but I think we can live with it for now. A better long term implementation will be to acquire all state locks from media_pipeline_start() before doing any validation, and always using different locks for different subdevs.

Quoted 85 lines

Also use the new functions in v4l2_subdev_link_validate().

---
 drivers/media/v4l2-core/v4l2-subdev.c | 12 +++-----
 include/media/v4l2-subdev.h           | 40 +++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 8 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +4 -8
index 4c6198c48dd6..6c34df19ea2f 100644
@@ -1400,17 +1400,13 @@ int v4l2_subdev_link_validate(struct media_link *link)
 
 	states_locked = sink_state && source_state;
 
-	if (states_locked) {
-		v4l2_subdev_lock_state(sink_state);
-		v4l2_subdev_lock_state(source_state);
-	}
+	if (states_locked)
+		v4l2_subdev_lock_states(sink_state, source_state);
 
 	ret = v4l2_subdev_link_validate_locked(link, states_locked);
 
-	if (states_locked) {
-		v4l2_subdev_unlock_state(sink_state);
-		v4l2_subdev_unlock_state(source_state);
-	}
+	if (states_locked)
+		v4l2_subdev_unlock_states(sink_state, source_state);
 
 	return ret;
 }
include/media/v4l2-subdev.h +40 -0
index a9e6b8146279..9cce48365975 100644
@@ -1724,6 +1724,46 @@ static inline void v4l2_subdev_unlock_state(struct v4l2_subdev_state *state)
 	mutex_unlock(state->lock);
 }
 
+/**
+ * v4l2_subdev_lock_states - Lock two sub-device states
+ * @state1: One subdevice state
+ * @state2: The other subdevice state
+ *
+ * Locks the state of two sub-devices.
+ *
+ * The states must be unlocked with v4l2_subdev_unlock_states() after use.
+ *
+ * This differs from calling v4l2_subdev_lock_state() on both states so that if
+ * the states share the same lock, the lock is acquired only once (so no
+ * deadlock occurs). The caller is responsible for ensuring the locks will
+ * always be acquired in the same order.
+ */
+static inline void v4l2_subdev_lock_states(struct v4l2_subdev_state *state1,
+					   struct v4l2_subdev_state *state2)
+{
+	mutex_lock(state1->lock);
+	if (state1->lock != state2->lock)
+		mutex_lock(state2->lock);
+}
+
+/**
+ * v4l2_subdev_unlock_states() - Unlock two sub-device states
+ * @state1: One subdevice state
+ * @state2: The other subdevice state
+ *
+ * Unlocks the state of two sub-devices.
+ *
+ * This differs from calling v4l2_subdev_unlock_state() on both states so that
+ * if the states share the same lock, the lock is released only once.
+ */
+static inline void v4l2_subdev_unlock_states(struct v4l2_subdev_state *state1,
+					     struct v4l2_subdev_state *state2)
+{
+	mutex_unlock(state1->lock);
+	if (state1->lock != state2->lock)
+		mutex_unlock(state2->lock);
+}
+
 /**
  * v4l2_subdev_get_unlocked_active_state() - Checks that the active subdev state
  *					     is unlocked and returns it
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#57 lore raw

PATCH v8 12/38 media: v4l: subdev: Move G_ROUTING handling below S_ROUTING

In reply to Sakari Ailus

Move G_ROUTING IOCTL handling below that of S_ROUTING. G_ROUTING implementation will soon needed in handling S_ROUTING as well.

---
 drivers/media/v4l2-core/v4l2-subdev.c | 54 +++++++++++++--------------
 1 file changed, 27 insertions(+), 27 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +27 -27
index 6c34df19ea2f..ec8de3f872b9 100644
@@ -909,33 +909,6 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 	case VIDIOC_SUBDEV_QUERYSTD:
 		return v4l2_subdev_call(sd, video, querystd, arg);
 
-	case VIDIOC_SUBDEV_G_ROUTING: {
-		struct v4l2_subdev_routing *routing = arg;
-		struct v4l2_subdev_krouting *krouting;
-
-		if (!v4l2_subdev_enable_streams_api)
-			return -ENOIOCTLCMD;
-
-		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
-			return -ENOIOCTLCMD;
-
-		memset(routing->reserved, 0, sizeof(routing->reserved));
-
-		krouting = &state->routing;
-
-		if (routing->num_routes < krouting->num_routes) {
-			routing->num_routes = krouting->num_routes;
-			return -ENOSPC;
-		}
-
-		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
-		       krouting->routes,
-		       krouting->num_routes * sizeof(*krouting->routes));
-		routing->num_routes = krouting->num_routes;
-
-		return 0;
-	}
-
 	case VIDIOC_SUBDEV_S_ROUTING: {
 		struct v4l2_subdev_routing *routing = arg;
 		struct v4l2_subdev_route *routes =
@@ -984,6 +957,33 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 					routing->which, &krouting);
 	}
 
+	case VIDIOC_SUBDEV_G_ROUTING: {
+		struct v4l2_subdev_routing *routing = arg;
+		struct v4l2_subdev_krouting *krouting;
+
+		if (!v4l2_subdev_enable_streams_api)
+			return -ENOIOCTLCMD;
+
+		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
+			return -ENOIOCTLCMD;
+
+		memset(routing->reserved, 0, sizeof(routing->reserved));
+
+		krouting = &state->routing;
+
+		if (routing->num_routes < krouting->num_routes) {
+			routing->num_routes = krouting->num_routes;
+			return -ENOSPC;
+		}
+
+		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+		       krouting->routes,
+		       krouting->num_routes * sizeof(*krouting->routes));
+		routing->num_routes = krouting->num_routes;
+
+		return 0;
+	}
+
 	case VIDIOC_SUBDEV_G_CLIENT_CAP: {
 		struct v4l2_subdev_client_capability *client_cap = arg;
 
Signature · 1 line
2.39.2
Reviewed-by on v8 12/38
#58 lore raw

Re: [PATCH v8 12/38] media: v4l: subdev: Move G_ROUTING handling below S_ROUTING

Quoted 4 lines · Sakari Ailus

Move G_ROUTING IOCTL handling below that of S_ROUTING. G_ROUTING implementation will soon needed in handling S_ROUTING as well.

Quoted 76 lines
---
  drivers/media/v4l2-core/v4l2-subdev.c | 54 +++++++++++++--------------
  1 file changed, 27 insertions(+), 27 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +27 -27
index 6c34df19ea2f..ec8de3f872b9 100644
@@ -909,33 +909,6 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
  	case VIDIOC_SUBDEV_QUERYSTD:
  		return v4l2_subdev_call(sd, video, querystd, arg);
 
-	case VIDIOC_SUBDEV_G_ROUTING: {
-		struct v4l2_subdev_routing *routing = arg;
-		struct v4l2_subdev_krouting *krouting;
-
-		if (!v4l2_subdev_enable_streams_api)
-			return -ENOIOCTLCMD;
-
-		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
-			return -ENOIOCTLCMD;
-
-		memset(routing->reserved, 0, sizeof(routing->reserved));
-
-		krouting = &state->routing;
-
-		if (routing->num_routes < krouting->num_routes) {
-			routing->num_routes = krouting->num_routes;
-			return -ENOSPC;
-		}
-
-		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
-		       krouting->routes,
-		       krouting->num_routes * sizeof(*krouting->routes));
-		routing->num_routes = krouting->num_routes;
-
-		return 0;
-	}
-
  	case VIDIOC_SUBDEV_S_ROUTING: {
  		struct v4l2_subdev_routing *routing = arg;
  		struct v4l2_subdev_route *routes =
@@ -984,6 +957,33 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
  					routing->which, &krouting);
  	}
 
+	case VIDIOC_SUBDEV_G_ROUTING: {
+		struct v4l2_subdev_routing *routing = arg;
+		struct v4l2_subdev_krouting *krouting;
+
+		if (!v4l2_subdev_enable_streams_api)
+			return -ENOIOCTLCMD;
+
+		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
+			return -ENOIOCTLCMD;
+
+		memset(routing->reserved, 0, sizeof(routing->reserved));
+
+		krouting = &state->routing;
+
+		if (routing->num_routes < krouting->num_routes) {
+			routing->num_routes = krouting->num_routes;
+			return -ENOSPC;
+		}
+
+		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+		       krouting->routes,
+		       krouting->num_routes * sizeof(*krouting->routes));
+		routing->num_routes = krouting->num_routes;
+
+		return 0;
+	}
+
  	case VIDIOC_SUBDEV_G_CLIENT_CAP: {
  		struct v4l2_subdev_client_capability *client_cap = arg;
 
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer Reviewed-by on v8 12/38
#59 lore raw

Re: [PATCH v8 12/38] media: v4l: subdev: Move G_ROUTING handling below S_ROUTING

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 4 lines · Sakari Ailus

Move G_ROUTING IOCTL handling below that of S_ROUTING. G_ROUTING implementation will soon needed in handling S_ROUTING as well.

Quoted 76 lines
---
 drivers/media/v4l2-core/v4l2-subdev.c | 54 +++++++++++++--------------
 1 file changed, 27 insertions(+), 27 deletions(-)
drivers/media/v4l2-core/v4l2-subdev.c +27 -27
index 6c34df19ea2f..ec8de3f872b9 100644
@@ -909,33 +909,6 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 	case VIDIOC_SUBDEV_QUERYSTD:
 		return v4l2_subdev_call(sd, video, querystd, arg);
 
-	case VIDIOC_SUBDEV_G_ROUTING: {
-		struct v4l2_subdev_routing *routing = arg;
-		struct v4l2_subdev_krouting *krouting;
-
-		if (!v4l2_subdev_enable_streams_api)
-			return -ENOIOCTLCMD;
-
-		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
-			return -ENOIOCTLCMD;
-
-		memset(routing->reserved, 0, sizeof(routing->reserved));
-
-		krouting = &state->routing;
-
-		if (routing->num_routes < krouting->num_routes) {
-			routing->num_routes = krouting->num_routes;
-			return -ENOSPC;
-		}
-
-		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
-		       krouting->routes,
-		       krouting->num_routes * sizeof(*krouting->routes));
-		routing->num_routes = krouting->num_routes;
-
-		return 0;
-	}
-
 	case VIDIOC_SUBDEV_S_ROUTING: {
 		struct v4l2_subdev_routing *routing = arg;
 		struct v4l2_subdev_route *routes =
@@ -984,6 +957,33 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 					routing->which, &krouting);
 	}
 
+	case VIDIOC_SUBDEV_G_ROUTING: {
+		struct v4l2_subdev_routing *routing = arg;
+		struct v4l2_subdev_krouting *krouting;
+
+		if (!v4l2_subdev_enable_streams_api)
+			return -ENOIOCTLCMD;
+
+		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
+			return -ENOIOCTLCMD;
+
+		memset(routing->reserved, 0, sizeof(routing->reserved));
+
+		krouting = &state->routing;
+
+		if (routing->num_routes < krouting->num_routes) {
+			routing->num_routes = krouting->num_routes;
+			return -ENOSPC;
+		}
+
+		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+		       krouting->routes,
+		       krouting->num_routes * sizeof(*krouting->routes));
+		routing->num_routes = krouting->num_routes;
+
+		return 0;
+	}
+
 	case VIDIOC_SUBDEV_G_CLIENT_CAP: {
 		struct v4l2_subdev_client_capability *client_cap = arg;
 
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#60 lore raw

PATCH v8 13/38 media: v4l: subdev: Copy argument back to user also for S_ROUTING

In reply to Sakari Ailus

As the user needs to know what went wrong for S_ROUTING, copy array arguments back to the user.

---
 drivers/media/v4l2-core/v4l2-ioctl.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/media/v4l2-core/v4l2-ioctl.c +6 -3
index d125d23e4e61..95bd56145d38 100644
@@ -3415,11 +3415,14 @@ video_usercopy(struct file *file, unsigned int orig_cmd, unsigned long arg,
 	 * FIXME: subdev IOCTLS are partially handled here and partially in
 	 * v4l2-subdev.c and the 'always_copy' flag can only be set for IOCTLS
 	 * defined here as part of the 'v4l2_ioctls' array. As
-	 * VIDIOC_SUBDEV_G_ROUTING needs to return results to applications even
-	 * in case of failure, but it is not defined here as part of the
+	 * VIDIOC_SUBDEV_[GS]_ROUTING needs to return results to applications
+	 * even in case of failure, but it is not defined here as part of the
 	 * 'v4l2_ioctls' array, insert an ad-hoc check to address that.
 	 */
-	if (err < 0 && !always_copy && cmd != VIDIOC_SUBDEV_G_ROUTING)
+	if (cmd == VIDIOC_SUBDEV_G_ROUTING || cmd == VIDIOC_SUBDEV_S_ROUTING)
+		always_copy = true;
+
+	if (err < 0 && !always_copy)
 		goto out;
 
 	if (has_array_args) {
Signature · 1 line
2.39.2
Reviewed-by on v8 13/38
#61 lore raw

Re: [PATCH v8 13/38] media: v4l: subdev: Copy argument back to user also for S_ROUTING

Quoted 4 lines · Sakari Ailus

As the user needs to know what went wrong for S_ROUTING, copy array arguments back to the user.

Quoted 26 lines
---
  drivers/media/v4l2-core/v4l2-ioctl.c | 9 ++++++---
  1 file changed, 6 insertions(+), 3 deletions(-)
drivers/media/v4l2-core/v4l2-ioctl.c +6 -3
index d125d23e4e61..95bd56145d38 100644
@@ -3415,11 +3415,14 @@ video_usercopy(struct file *file, unsigned int orig_cmd, unsigned long arg,
  	 * FIXME: subdev IOCTLS are partially handled here and partially in
  	 * v4l2-subdev.c and the 'always_copy' flag can only be set for IOCTLS
  	 * defined here as part of the 'v4l2_ioctls' array. As
-	 * VIDIOC_SUBDEV_G_ROUTING needs to return results to applications even
-	 * in case of failure, but it is not defined here as part of the
+	 * VIDIOC_SUBDEV_[GS]_ROUTING needs to return results to applications
+	 * even in case of failure, but it is not defined here as part of the
  	 * 'v4l2_ioctls' array, insert an ad-hoc check to address that.
  	 */
-	if (err < 0 && !always_copy && cmd != VIDIOC_SUBDEV_G_ROUTING)
+	if (cmd == VIDIOC_SUBDEV_G_ROUTING || cmd == VIDIOC_SUBDEV_S_ROUTING)
+		always_copy = true;
+
+	if (err < 0 && !always_copy)
  		goto out;
 
  	if (has_array_args) {
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer Reviewed-by on v8 13/38
#62 lore raw

Re: [PATCH v8 13/38] media: v4l: subdev: Copy argument back to user also for S_ROUTING

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 13 lines · Sakari Ailus

As the user needs to know what went wrong for S_ROUTING, copy array arguments back to the user.

---
 drivers/media/v4l2-core/v4l2-ioctl.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/media/v4l2-core/v4l2-ioctl.c +0 -0
index d125d23e4e61..95bd56145d38 100644

drivers/media/v4l2-core/v4l2-ioctl.c video_usercopy(struct file *file, unsigned int orig_cmd, unsigned long arg, · line 3418

Quoted 6 lines
drivers/media/v4l2-core/v4l2-ioctl.c +1 -2
@@ -3415,11 +3415,14 @@ video_usercopy(struct file *file, unsigned int orig_cmd, unsigned long arg,
 	 * FIXME: subdev IOCTLS are partially handled here and partially in
 	 * v4l2-subdev.c and the 'always_copy' flag can only be set for IOCTLS
 	 * defined here as part of the 'v4l2_ioctls' array. As
-	 * VIDIOC_SUBDEV_G_ROUTING needs to return results to applications even
-	 * in case of failure, but it is not defined here as part of the
+	 * VIDIOC_SUBDEV_[GS]_ROUTING needs to return results to applications

s/needs/need/

Quoted 11 lines
drivers/media/v4l2-core/v4l2-ioctl.c +5 -1
+	 * even in case of failure, but it is not defined here as part of the
 	 * 'v4l2_ioctls' array, insert an ad-hoc check to address that.
 	 */
-	if (err < 0 && !always_copy && cmd != VIDIOC_SUBDEV_G_ROUTING)
+	if (cmd == VIDIOC_SUBDEV_G_ROUTING || cmd == VIDIOC_SUBDEV_S_ROUTING)
+		always_copy = true;
+
+	if (err < 0 && !always_copy)
 		goto out;
 
 	if (has_array_args) {
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#63 lore raw

Re: [PATCH v8 13/38] media: v4l: subdev: Copy argument back to user also for S_ROUTING

Hi Laurent,

Thanks for the review.

Quoted 13 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 8 lines · Sakari Ailus

As the user needs to know what went wrong for S_ROUTING, copy array arguments back to the user.

---
 drivers/media/v4l2-core/v4l2-ioctl.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
Quoted 13 lines

drivers/media/v4l2-core/v4l2-ioctl.c video_usercopy(struct file *file, unsigned int orig_cmd, unsigned long arg, · line 3418

Quoted 11 lines
drivers/media/v4l2-core/v4l2-ioctl.c +1 -2
index d125d23e4e61..95bd56145d38 100644
@@ -3415,11 +3415,14 @@ video_usercopy(struct file *file, unsigned int orig_cmd, unsigned long arg,
 	 * FIXME: subdev IOCTLS are partially handled here and partially in
 	 * v4l2-subdev.c and the 'always_copy' flag can only be set for IOCTLS
 	 * defined here as part of the 'v4l2_ioctls' array. As
-	 * VIDIOC_SUBDEV_G_ROUTING needs to return results to applications even
-	 * in case of failure, but it is not defined here as part of the
+	 * VIDIOC_SUBDEV_[GS]_ROUTING needs to return results to applications

s/needs/need/

I'd say "needs" is correct.

Quoted 15 lines
Quoted 11 lines
drivers/media/v4l2-core/v4l2-ioctl.c +5 -1
+	 * even in case of failure, but it is not defined here as part of the
 	 * 'v4l2_ioctls' array, insert an ad-hoc check to address that.
 	 */
-	if (err < 0 && !always_copy && cmd != VIDIOC_SUBDEV_G_ROUTING)
+	if (cmd == VIDIOC_SUBDEV_G_ROUTING || cmd == VIDIOC_SUBDEV_S_ROUTING)
+		always_copy = true;
+
+	if (err < 0 && !always_copy)
 		goto out;
 
 	if (has_array_args) {
Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#64 lore raw

PATCH v8 14/38 media: v4l: subdev: Add len_routes field to struct v4l2_subdev_routing

In reply to Sakari Ailus

The len_routes field is used to tell the size of the routes array in struct v4l2_subdev_routing. This way the number of routes returned from S_ROUTING IOCTL may be larger than the number of routes provided, in case there are more routes returned by the driver.

Note that this uAPI is still disabled in the code, so this change can safely be done. Anyone who manually patched the code to enable this uAPI must update their code.

---
 .../media/v4l/vidioc-subdev-g-routing.rst     | 33 ++++++++++++-------
 drivers/media/v4l2-core/v4l2-ioctl.c          |  4 +--
 drivers/media/v4l2-core/v4l2-subdev.c         |  6 +++-
 include/media/v4l2-subdev.h                   |  2 ++
 include/uapi/linux/v4l2-subdev.h              |  9 +++--
 5 files changed, 37 insertions(+), 17 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +22 -11
index 26b5004bfe6d..8f9aa83275c5 100644
@@ -46,20 +46,28 @@ with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and
 setting or clearing flags of the  ``flags`` field of a
 struct :c:type:`v4l2_subdev_route`.
 
-All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is called. This
-means that the userspace must reconfigure all streams after calling the ioctl
-with e.g. ``VIDIOC_SUBDEV_S_FMT``.
+All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called. This means that the userspace must reconfigure all streams after calling
+the ioctl with e.g. ``VIDIOC_SUBDEV_S_FMT``.
 
 Only subdevices which have both sink and source pads can support routing.
 
-When inspecting routes through ``VIDIOC_SUBDEV_G_ROUTING`` and the application
-provided ``num_routes`` is not big enough to contain all the available routes
-the subdevice exposes, drivers return the ENOSPC error code and adjust the
-value of the ``num_routes`` field. Application should then reserve enough memory
-for all the route entries and call ``VIDIOC_SUBDEV_G_ROUTING`` again.
+The ``num_routes`` field is used to denote the number of routes set (set by user
+space on ``VIDIOC_SUBDEV_S_ROUTING`` argument) on the routing table as well as
+the number of routes returned back from both IOCTLs. ``len_routes`` signifies
+the number of routes that can fit into the ``routes`` array. The userspace shall
+set ``len_routes`` for both IOCTLs and ``num_routes`` for
+``VIDIOC_SUBDEV_S_ROUTING``.
 
-On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
-``num_routes`` field to reflect the actual number of routes returned.
+On a ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the ``num_routes``
+field to reflect the actual number of routes known by the driver.
+``num_routes`` larger than ``len_routes`` may be returned by both IOCTLs to
+indicate there are more routes than fits to the ``routes`` array. In this
+case first ``len_routes`` were returned back to the userspace in the
+``routes`` array. This is not considered as an error.
+
+Also ``VIDIOC_SUBDEV_S_ROUTING`` may return more route than the user provided in
+``num_routes`` field due to e.g. hardware properties.
 
 .. tabularcolumns:: |p{4.4cm}|p{4.4cm}|p{8.7cm}|
 
@@ -74,6 +82,9 @@ On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
       - ``which``
       - Routing table to be accessed, from enum
         :ref:`v4l2_subdev_format_whence <v4l2-subdev-format-whence>`.
+    * - __u32
+      - ``len_routes``
+      - The length of the array (as in memory reserved for the array)
     * - struct :c:type:`v4l2_subdev_route`
       - ``routes[]``
       - Array of struct :c:type:`v4l2_subdev_route` entries
@@ -81,7 +92,7 @@ On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
       - ``num_routes``
       - Number of entries of the routes array
     * - __u32
-      - ``reserved``\ [5]
+      - ``reserved``\ [11]
       - Reserved for future extensions. Applications and drivers must set
 	the array to zero.
 
drivers/media/v4l2-core/v4l2-ioctl.c +2 -2
index 95bd56145d38..018e7efc21ca 100644
@@ -3155,13 +3155,13 @@ static int check_array_args(unsigned int cmd, void *parg, size_t *array_size,
 	case VIDIOC_SUBDEV_S_ROUTING: {
 		struct v4l2_subdev_routing *routing = parg;
 
-		if (routing->num_routes > 256)
+		if (routing->len_routes > 256)
 			return -E2BIG;
 
 		*user_ptr = u64_to_user_ptr(routing->routes);
 		*kernel_ptr = (void **)&routing->routes;
 		*array_size = sizeof(struct v4l2_subdev_route)
-			    * routing->num_routes;
+			    * routing->len_routes;
 		ret = 1;
 		break;
 	}
drivers/media/v4l2-core/v4l2-subdev.c +5 -1
index ec8de3f872b9..486a5c3016de 100644
@@ -925,6 +925,9 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		if (routing->which != V4L2_SUBDEV_FORMAT_TRY && ro_subdev)
 			return -EPERM;
 
+		if (routing->num_routes > routing->len_routes)
+			return -EINVAL;
+
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
 		for (i = 0; i < routing->num_routes; ++i) {
@@ -951,6 +954,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		}
 
 		krouting.num_routes = routing->num_routes;
+		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
 
 		return v4l2_subdev_call(sd, pad, set_routing, state,
@@ -971,7 +975,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		krouting = &state->routing;
 
-		if (routing->num_routes < krouting->num_routes) {
+		if (routing->len_routes < krouting->num_routes) {
 			routing->num_routes = krouting->num_routes;
 			return -ENOSPC;
 		}
include/media/v4l2-subdev.h +2 -0
index 9cce48365975..1df6b963a1c9 100644
@@ -728,12 +728,14 @@ struct v4l2_subdev_stream_configs {
 /**
  * struct v4l2_subdev_krouting - subdev routing table
  *
+ * @len_routes: length of routes array, in routes
  * @num_routes: number of routes
  * @routes: &struct v4l2_subdev_route
  *
  * This structure contains the routing table for a subdev.
  */
 struct v4l2_subdev_krouting {
+	unsigned int len_routes;
 	unsigned int num_routes;
 	struct v4l2_subdev_route *routes;
 };
include/uapi/linux/v4l2-subdev.h +6 -3
index 7048c51581c6..ca543982460c 100644
@@ -224,15 +224,18 @@ struct v4l2_subdev_route {
  * struct v4l2_subdev_routing - Subdev routing information
  *
  * @which: configuration type (from enum v4l2_subdev_format_whence)
- * @num_routes: the total number of routes in the routes array
+ * @len_routes: the length of the routes array, in routes
  * @routes: pointer to the routes array
+ * @num_routes: the total number of routes, possibly more than fits in the
+ *		routes array
  * @reserved: drivers and applications must zero this array
  */
 struct v4l2_subdev_routing {
 	__u32 which;
-	__u32 num_routes;
+	__u32 len_routes;
 	__u64 routes;
-	__u32 reserved[6];
+	__u32 num_routes;
+	__u32 reserved[11];
 };
 
 /*
Signature · 1 line
2.39.2
maintainer
#65 lore raw

Re: [PATCH v8 14/38] media: v4l: subdev: Add len_routes field to struct v4l2_subdev_routing

Hi Sakari,

Thank you for the patch.

Quoted 23 lines · Sakari Ailus

The len_routes field is used to tell the size of the routes array in struct v4l2_subdev_routing. This way the number of routes returned from S_ROUTING IOCTL may be larger than the number of routes provided, in case there are more routes returned by the driver.

Note that this uAPI is still disabled in the code, so this change can safely be done. Anyone who manually patched the code to enable this uAPI must update their code.

---
 .../media/v4l/vidioc-subdev-g-routing.rst     | 33 ++++++++++++-------
 drivers/media/v4l2-core/v4l2-ioctl.c          |  4 +--
 drivers/media/v4l2-core/v4l2-subdev.c         |  6 +++-
 include/media/v4l2-subdev.h                   |  2 ++
 include/uapi/linux/v4l2-subdev.h              |  9 +++--
 5 files changed, 37 insertions(+), 17 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -0
index 26b5004bfe6d..8f9aa83275c5 100644

Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and · line 51

Quoted 9 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +3 -3
@@ -46,20 +46,28 @@ with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and
 setting or clearing flags of the  ``flags`` field of a
 struct :c:type:`v4l2_subdev_route`.
 
-All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is called. This
-means that the userspace must reconfigure all streams after calling the ioctl
-with e.g. ``VIDIOC_SUBDEV_S_FMT``.
+All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called. This means that the userspace must reconfigure all streams after calling
+the ioctl with e.g. ``VIDIOC_SUBDEV_S_FMT``.

Unrelated. If you really want to reflow, the first line is too short, 'called.' should be moved there.

Quoted 17 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +6 -7
 Only subdevices which have both sink and source pads can support routing.
 
-When inspecting routes through ``VIDIOC_SUBDEV_G_ROUTING`` and the application
-provided ``num_routes`` is not big enough to contain all the available routes
-the subdevice exposes, drivers return the ENOSPC error code and adjust the
-value of the ``num_routes`` field. Application should then reserve enough memory
-for all the route entries and call ``VIDIOC_SUBDEV_G_ROUTING`` again.
+The ``num_routes`` field is used to denote the number of routes set (set by user
+space on ``VIDIOC_SUBDEV_S_ROUTING`` argument) on the routing table as well as
+the number of routes returned back from both IOCTLs. ``len_routes`` signifies
+the number of routes that can fit into the ``routes`` array. The userspace shall
+set ``len_routes`` for both IOCTLs and ``num_routes`` for
+``VIDIOC_SUBDEV_S_ROUTING``.
 
-On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
-``num_routes`` field to reflect the actual number of routes returned.

Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and · line 71

Quoted 10 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +9 -0
+On a ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the ``num_routes``
+field to reflect the actual number of routes known by the driver.
+``num_routes`` larger than ``len_routes`` may be returned by both IOCTLs to
+indicate there are more routes than fits to the ``routes`` array. In this
+case first ``len_routes`` were returned back to the userspace in the
+``routes`` array. This is not considered as an error.
+
+Also ``VIDIOC_SUBDEV_S_ROUTING`` may return more route than the user provided in
+``num_routes`` field due to e.g. hardware properties.
 

An attempt at making this clearer (in my opinion, feel free to pick the pieces you like, or reject it outright).

The ``len_routes`` field indicates the number of routes that can fit in the ``routes`` array allocated by userspace. It is set by applications for both ioctls to indicate how many routes the kernel can return, and is never modified by the kernel.

The ``num_routes`` field indicates the number of routes in the subdevice routing table. For ``VIDIOC_SUBDEV_S_ROUTING``, it is set by userspace to the number of routes that the application stored in the ``routes`` array. For both ioctls, it is returned by the kernel and indicates how many routes are stored in the subdevice routing table. This may be smaller or larger than the value of ``num_routes`` set by the application for ``VIDIOC_SUBDEV_S_ROUTING``, as drivers may adjust the requested routing table.

The kernel can return a ``num_routes`` value larger than ``len_routes`` from both ioctls. This indicates thare are more routes in the routing table than fits the ``routes`` array. In this case, the ``routes`` array is filled by the kernel with the first ``len_routes`` entries of the subdevice routing table. This is not considered an array, and the ioctl call succeeds. If the applications wants to retrieve the missing routes, it can issue a new ``VIDIOC_SUBDEV_G_ROUTING`` call with a large enough ``routes`` array.

Quoted 64 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +4 -1
 .. tabularcolumns:: |p{4.4cm}|p{4.4cm}|p{8.7cm}|
 
@@ -74,6 +82,9 @@ On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
       - ``which``
       - Routing table to be accessed, from enum
         :ref:`v4l2_subdev_format_whence <v4l2-subdev-format-whence>`.
+    * - __u32
+      - ``len_routes``
+      - The length of the array (as in memory reserved for the array)
     * - struct :c:type:`v4l2_subdev_route`
       - ``routes[]``
       - Array of struct :c:type:`v4l2_subdev_route` entries
@@ -81,7 +92,7 @@ On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
       - ``num_routes``
       - Number of entries of the routes array
     * - __u32
-      - ``reserved``\ [5]
+      - ``reserved``\ [11]
       - Reserved for future extensions. Applications and drivers must set
 	the array to zero.
 
drivers/media/v4l2-core/v4l2-ioctl.c +2 -2
index 95bd56145d38..018e7efc21ca 100644
@@ -3155,13 +3155,13 @@ static int check_array_args(unsigned int cmd, void *parg, size_t *array_size,
 	case VIDIOC_SUBDEV_S_ROUTING: {
 		struct v4l2_subdev_routing *routing = parg;
 
-		if (routing->num_routes > 256)
+		if (routing->len_routes > 256)
 			return -E2BIG;
 
 		*user_ptr = u64_to_user_ptr(routing->routes);
 		*kernel_ptr = (void **)&routing->routes;
 		*array_size = sizeof(struct v4l2_subdev_route)
-			    * routing->num_routes;
+			    * routing->len_routes;
 		ret = 1;
 		break;
 	}
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
index ec8de3f872b9..486a5c3016de 100644
@@ -925,6 +925,9 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		if (routing->which != V4L2_SUBDEV_FORMAT_TRY && ro_subdev)
 			return -EPERM;
 
+		if (routing->num_routes > routing->len_routes)
+			return -EINVAL;
+
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
 		for (i = 0; i < routing->num_routes; ++i) {
@@ -951,6 +954,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		}
 
 		krouting.num_routes = routing->num_routes;
+		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
 
 		return v4l2_subdev_call(sd, pad, set_routing, state,

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 980

Quoted 7 lines
drivers/media/v4l2-core/v4l2-subdev.c +1 -1
@@ -971,7 +975,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		krouting = &state->routing;
 
-		if (routing->num_routes < krouting->num_routes) {
+		if (routing->len_routes < krouting->num_routes) {
 			routing->num_routes = krouting->num_routes;
 			return -ENOSPC;

Do we still want to return -ENOSPC here ?

Quoted 46 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
 		}
include/media/v4l2-subdev.h +2 -0
index 9cce48365975..1df6b963a1c9 100644
@@ -728,12 +728,14 @@ struct v4l2_subdev_stream_configs {
 /**
  * struct v4l2_subdev_krouting - subdev routing table
  *
+ * @len_routes: length of routes array, in routes
  * @num_routes: number of routes
  * @routes: &struct v4l2_subdev_route
  *
  * This structure contains the routing table for a subdev.
  */
 struct v4l2_subdev_krouting {
+	unsigned int len_routes;
 	unsigned int num_routes;
 	struct v4l2_subdev_route *routes;
 };
include/uapi/linux/v4l2-subdev.h +6 -3
index 7048c51581c6..ca543982460c 100644
@@ -224,15 +224,18 @@ struct v4l2_subdev_route {
  * struct v4l2_subdev_routing - Subdev routing information
  *
  * @which: configuration type (from enum v4l2_subdev_format_whence)
- * @num_routes: the total number of routes in the routes array
+ * @len_routes: the length of the routes array, in routes
  * @routes: pointer to the routes array
+ * @num_routes: the total number of routes, possibly more than fits in the
+ *		routes array
  * @reserved: drivers and applications must zero this array
  */
 struct v4l2_subdev_routing {
 	__u32 which;
-	__u32 num_routes;
+	__u32 len_routes;
 	__u64 routes;
-	__u32 reserved[6];
+	__u32 num_routes;
+	__u32 reserved[11];
 };
 
 /*
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#66 lore raw

Re: [PATCH v8 14/38] media: v4l: subdev: Add len_routes field to struct v4l2_subdev_routing

Hi Laurent,

Quoted 31 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 26 lines · Sakari Ailus

The len_routes field is used to tell the size of the routes array in struct v4l2_subdev_routing. This way the number of routes returned from S_ROUTING IOCTL may be larger than the number of routes provided, in case there are more routes returned by the driver.

Note that this uAPI is still disabled in the code, so this change can safely be done. Anyone who manually patched the code to enable this uAPI must update their code.

---
 .../media/v4l/vidioc-subdev-g-routing.rst     | 33 ++++++++++++-------
 drivers/media/v4l2-core/v4l2-ioctl.c          |  4 +--
 drivers/media/v4l2-core/v4l2-subdev.c         |  6 +++-
 include/media/v4l2-subdev.h                   |  2 ++
 include/uapi/linux/v4l2-subdev.h              |  9 +++--
 5 files changed, 37 insertions(+), 17 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -0
index 26b5004bfe6d..8f9aa83275c5 100644
@@ -46,20 +46,28 @@ with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and
 setting or clearing flags of the  ``flags`` field of a
 struct :c:type:`v4l2_subdev_route`.
 
Quoted 9 lines

Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and · line 51

Quoted 6 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +3 -3
-All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is called. This
-means that the userspace must reconfigure all streams after calling the ioctl
-with e.g. ``VIDIOC_SUBDEV_S_FMT``.
+All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is
+called. This means that the userspace must reconfigure all streams after calling
+the ioctl with e.g. ``VIDIOC_SUBDEV_S_FMT``.

Unrelated. If you really want to reflow, the first line is too short, 'called.' should be moved there.

Seems to be, indeed. This would look like a bug in an Emacs major mode. :-I

Quoted 33 lines
Quoted 18 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +6 -7
 Only subdevices which have both sink and source pads can support routing.
 
-When inspecting routes through ``VIDIOC_SUBDEV_G_ROUTING`` and the application
-provided ``num_routes`` is not big enough to contain all the available routes
-the subdevice exposes, drivers return the ENOSPC error code and adjust the
-value of the ``num_routes`` field. Application should then reserve enough memory
-for all the route entries and call ``VIDIOC_SUBDEV_G_ROUTING`` again.
+The ``num_routes`` field is used to denote the number of routes set (set by user
+space on ``VIDIOC_SUBDEV_S_ROUTING`` argument) on the routing table as well as
+the number of routes returned back from both IOCTLs. ``len_routes`` signifies
+the number of routes that can fit into the ``routes`` array. The userspace shall
+set ``len_routes`` for both IOCTLs and ``num_routes`` for
+``VIDIOC_SUBDEV_S_ROUTING``.
 
-On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
-``num_routes`` field to reflect the actual number of routes returned.

Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and · line 71

Quoted 10 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +9 -0
+On a ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the ``num_routes``
+field to reflect the actual number of routes known by the driver.
+``num_routes`` larger than ``len_routes`` may be returned by both IOCTLs to
+indicate there are more routes than fits to the ``routes`` array. In this
+case first ``len_routes`` were returned back to the userspace in the
+``routes`` array. This is not considered as an error.
+
+Also ``VIDIOC_SUBDEV_S_ROUTING`` may return more route than the user provided in
+``num_routes`` field due to e.g. hardware properties.
 

An attempt at making this clearer (in my opinion, feel free to pick the pieces you like, or reject it outright).

Quoted 7 lines

The ``len_routes`` field indicates the number of routes that can fit in the ``routes`` array allocated by userspace. It is set by applications for both ioctls to indicate how many routes the kernel can return, and is never modified by the kernel.

The ``num_routes`` field indicates the number of routes in the subdevice routing table. For ``VIDIOC_SUBDEV_S_ROUTING``, it is set by userspace to the number of

For S_ROUTING this is the number of routes in the IOCTL argument. The routing table may contain more (static routes).

Quoted 5 lines

routes that the application stored in the ``routes`` array. For both ioctls, it is returned by the kernel and indicates how many routes are stored in the subdevice routing table. This may be smaller or larger than the value of ``num_routes`` set by the application for ``VIDIOC_SUBDEV_S_ROUTING``, as drivers may adjust the requested routing table.

How about this (first sentence modified):

The ``num_routes`` field, when returned from the kernel on both IOCTLs, indicates the number of routes in the subdevice routing table and when calling ``VIDIOC_SUBDEV_S_ROUTING``, it is set by userspace to the number of routes that the application stored in the ``routes`` array. The value returned by the kernel may be smaller or larger than the value of ``num_routes`` set by the application for ``VIDIOC_SUBDEV_S_ROUTING``, as drivers may adjust the requested routing table.

Quoted 6 lines

The kernel can return a ``num_routes`` value larger than ``len_routes`` from both ioctls. This indicates thare are more routes in the routing table than fits the ``routes`` array. In this case, the ``routes`` array is filled by the kernel with the first ``len_routes`` entries of the subdevice routing table. This is not considered an array, and the ioctl call succeeds. If the applications wants

Too many arrays? :-)

s/not considered \Kan array/to be an error/

Quoted 68 lines

to retrieve the missing routes, it can issue a new ``VIDIOC_SUBDEV_G_ROUTING`` call with a large enough ``routes`` array.

Quoted 65 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +4 -1
 .. tabularcolumns:: |p{4.4cm}|p{4.4cm}|p{8.7cm}|
 
@@ -74,6 +82,9 @@ On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
       - ``which``
       - Routing table to be accessed, from enum
         :ref:`v4l2_subdev_format_whence <v4l2-subdev-format-whence>`.
+    * - __u32
+      - ``len_routes``
+      - The length of the array (as in memory reserved for the array)
     * - struct :c:type:`v4l2_subdev_route`
       - ``routes[]``
       - Array of struct :c:type:`v4l2_subdev_route` entries
@@ -81,7 +92,7 @@ On a successful ``VIDIOC_SUBDEV_G_ROUTING`` call the driver updates the
       - ``num_routes``
       - Number of entries of the routes array
     * - __u32
-      - ``reserved``\ [5]
+      - ``reserved``\ [11]
       - Reserved for future extensions. Applications and drivers must set
 	the array to zero.
 
drivers/media/v4l2-core/v4l2-ioctl.c +2 -2
index 95bd56145d38..018e7efc21ca 100644
@@ -3155,13 +3155,13 @@ static int check_array_args(unsigned int cmd, void *parg, size_t *array_size,
 	case VIDIOC_SUBDEV_S_ROUTING: {
 		struct v4l2_subdev_routing *routing = parg;
 
-		if (routing->num_routes > 256)
+		if (routing->len_routes > 256)
 			return -E2BIG;
 
 		*user_ptr = u64_to_user_ptr(routing->routes);
 		*kernel_ptr = (void **)&routing->routes;
 		*array_size = sizeof(struct v4l2_subdev_route)
-			    * routing->num_routes;
+			    * routing->len_routes;
 		ret = 1;
 		break;
 	}
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
index ec8de3f872b9..486a5c3016de 100644
@@ -925,6 +925,9 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		if (routing->which != V4L2_SUBDEV_FORMAT_TRY && ro_subdev)
 			return -EPERM;
 
+		if (routing->num_routes > routing->len_routes)
+			return -EINVAL;
+
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
 		for (i = 0; i < routing->num_routes; ++i) {
@@ -951,6 +954,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		}
 
 		krouting.num_routes = routing->num_routes;
+		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
 
 		return v4l2_subdev_call(sd, pad, set_routing, state,
@@ -971,7 +975,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
Quoted 8 lines

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 980

Quoted 6 lines
drivers/media/v4l2-core/v4l2-subdev.c +1 -1
 		krouting = &state->routing;
 
-		if (routing->num_routes < krouting->num_routes) {
+		if (routing->len_routes < krouting->num_routes) {
 			routing->num_routes = krouting->num_routes;
 			return -ENOSPC;

Do we still want to return -ENOSPC here ?

Indeed, this part should have been changed, too.

The code in the case would then become:

		struct v4l2_subdev_routing *routing = arg;
		struct v4l2_subdev_krouting *krouting;

		if (!v4l2_subdev_enable_streams_api)
			return -ENOIOCTLCMD;

		if (!(sd->flags & V4L2_SUBDEV_FL_STREAMS))
			return -ENOIOCTLCMD;

		memset(routing->reserved, 0, sizeof(routing->reserved));

		krouting = &state->routing;

		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
		       krouting->routes,
		       min(krouting->num_routes, routing->len_routes) *
		       sizeof(*krouting->routes));
		routing->num_routes = krouting->num_routes;

		return 0;
Quoted 48 lines
Quoted 47 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
 		}
include/media/v4l2-subdev.h +2 -0
index 9cce48365975..1df6b963a1c9 100644
@@ -728,12 +728,14 @@ struct v4l2_subdev_stream_configs {
 /**
  * struct v4l2_subdev_krouting - subdev routing table
  *
+ * @len_routes: length of routes array, in routes
  * @num_routes: number of routes
  * @routes: &struct v4l2_subdev_route
  *
  * This structure contains the routing table for a subdev.
  */
 struct v4l2_subdev_krouting {
+	unsigned int len_routes;
 	unsigned int num_routes;
 	struct v4l2_subdev_route *routes;
 };
include/uapi/linux/v4l2-subdev.h +6 -3
index 7048c51581c6..ca543982460c 100644
@@ -224,15 +224,18 @@ struct v4l2_subdev_route {
  * struct v4l2_subdev_routing - Subdev routing information
  *
  * @which: configuration type (from enum v4l2_subdev_format_whence)
- * @num_routes: the total number of routes in the routes array
+ * @len_routes: the length of the routes array, in routes
  * @routes: pointer to the routes array
+ * @num_routes: the total number of routes, possibly more than fits in the
+ *		routes array
  * @reserved: drivers and applications must zero this array
  */
 struct v4l2_subdev_routing {
 	__u32 which;
-	__u32 num_routes;
+	__u32 len_routes;
 	__u64 routes;
-	__u32 reserved[6];
+	__u32 num_routes;
+	__u32 reserved[11];
 };
 
 /*
Signature · 3 lines
Kind regards,

Sakari Ailus
author maintainer
#67 lore raw

PATCH v8 15/38 media: v4l: subdev: Return routes set using S_ROUTING

In reply to Sakari Ailus

Return the routes set using S_ROUTING back to the user. Also reflect this in documentation.

---
 .../media/v4l/vidioc-subdev-g-routing.rst            |  5 +++--
 drivers/media/v4l2-core/v4l2-subdev.c                | 12 +++++++++++-
 2 files changed, 14 insertions(+), 3 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +3 -2
index 8f9aa83275c5..6eb6a59570dc 100644
@@ -43,8 +43,9 @@ The routing configuration determines the flows of data inside an entity.
 Drivers report their current routing tables using the
 ``VIDIOC_SUBDEV_G_ROUTING`` ioctl and application may enable or disable routes
 with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and
-setting or clearing flags of the  ``flags`` field of a
-struct :c:type:`v4l2_subdev_route`.
+setting or clearing flags of the ``flags`` field of a struct
+:c:type:`v4l2_subdev_route`. Similarly to ``VIDIOC_SUBDEV_G_ROUTING``, also
+``VIDIOC_SUBDEV_S_ROUTING`` returns the routes back to the user.
 
 All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is
 called. This means that the userspace must reconfigure all streams after calling
drivers/media/v4l2-core/v4l2-subdev.c +11 -1
index 486a5c3016de..a357ce318192 100644
@@ -957,8 +957,18 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
 
-		return v4l2_subdev_call(sd, pad, set_routing, state,
+		rval = v4l2_subdev_call(sd, pad, set_routing, state,
 					routing->which, &krouting);
+		if (rval < 0)
+			return rval;
+
+		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+		       state->routing.routes,
+		       state->routing.num_routes *
+		       sizeof(*state->routing.routes));
+		routing->num_routes = state->routing.num_routes;
+
+		return 0;
 	}
 
 	case VIDIOC_SUBDEV_G_ROUTING: {
Signature · 1 line
2.39.2
maintainer
#68 lore raw

Re: [PATCH v8 15/38] media: v4l: subdev: Return routes set using S_ROUTING

Hi Sakari,

Thank you for the patch.

Quoted 32 lines · Sakari Ailus

Return the routes set using S_ROUTING back to the user. Also reflect this in documentation.

---
 .../media/v4l/vidioc-subdev-g-routing.rst            |  5 +++--
 drivers/media/v4l2-core/v4l2-subdev.c                | 12 +++++++++++-
 2 files changed, 14 insertions(+), 3 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +3 -2
index 8f9aa83275c5..6eb6a59570dc 100644
@@ -43,8 +43,9 @@ The routing configuration determines the flows of data inside an entity.
 Drivers report their current routing tables using the
 ``VIDIOC_SUBDEV_G_ROUTING`` ioctl and application may enable or disable routes
 with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and
-setting or clearing flags of the  ``flags`` field of a
-struct :c:type:`v4l2_subdev_route`.
+setting or clearing flags of the ``flags`` field of a struct
+:c:type:`v4l2_subdev_route`. Similarly to ``VIDIOC_SUBDEV_G_ROUTING``, also
+``VIDIOC_SUBDEV_S_ROUTING`` returns the routes back to the user.
 
 All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is
 called. This means that the userspace must reconfigure all streams after calling
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index 486a5c3016de..a357ce318192 100644
@@ -957,8 +957,18 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 967

Quoted 10 lines
drivers/media/v4l2-core/v4l2-subdev.c +7 -1
 
-		return v4l2_subdev_call(sd, pad, set_routing, state,
+		rval = v4l2_subdev_call(sd, pad, set_routing, state,
 					routing->which, &krouting);
+		if (rval < 0)
+			return rval;
+
+		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+		       state->routing.routes,
+		       state->routing.num_routes *

I think this should be capped at len_routes.

Quoted 10 lines
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
+		       sizeof(*state->routing.routes));
+		routing->num_routes = state->routing.num_routes;
+
+		return 0;
 	}
 
 	case VIDIOC_SUBDEV_G_ROUTING: {
Signature · 1 line
2.39.2
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#69 lore raw

Re: [PATCH v8 15/38] media: v4l: subdev: Return routes set using S_ROUTING

Quoted 38 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 33 lines · Sakari Ailus

Return the routes set using S_ROUTING back to the user. Also reflect this in documentation.

---
 .../media/v4l/vidioc-subdev-g-routing.rst            |  5 +++--
 drivers/media/v4l2-core/v4l2-subdev.c                | 12 +++++++++++-
 2 files changed, 14 insertions(+), 3 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +3 -2
index 8f9aa83275c5..6eb6a59570dc 100644
@@ -43,8 +43,9 @@ The routing configuration determines the flows of data inside an entity.
 Drivers report their current routing tables using the
 ``VIDIOC_SUBDEV_G_ROUTING`` ioctl and application may enable or disable routes
 with the ``VIDIOC_SUBDEV_S_ROUTING`` ioctl, by adding or removing routes and
-setting or clearing flags of the  ``flags`` field of a
-struct :c:type:`v4l2_subdev_route`.
+setting or clearing flags of the ``flags`` field of a struct
+:c:type:`v4l2_subdev_route`. Similarly to ``VIDIOC_SUBDEV_G_ROUTING``, also
+``VIDIOC_SUBDEV_S_ROUTING`` returns the routes back to the user.
 
 All stream configurations are reset when ``VIDIOC_SUBDEV_S_ROUTING`` is
 called. This means that the userspace must reconfigure all streams after calling
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index 486a5c3016de..a357ce318192 100644
@@ -957,8 +957,18 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
 
Quoted 11 lines

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 967

Quoted 9 lines
drivers/media/v4l2-core/v4l2-subdev.c +7 -1
-		return v4l2_subdev_call(sd, pad, set_routing, state,
+		rval = v4l2_subdev_call(sd, pad, set_routing, state,
 					routing->which, &krouting);
+		if (rval < 0)
+			return rval;
+
+		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+		       state->routing.routes,
+		       state->routing.num_routes *

I think this should be capped at len_routes.

Good catch!

Quoted 16 lines
Quoted 11 lines
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
+		       sizeof(*state->routing.routes));
+		routing->num_routes = state->routing.num_routes;
+
+		return 0;
 	}
 
 	case VIDIOC_SUBDEV_G_ROUTING: {
Signature · 1 line
2.39.2
Signature · 3 lines
Regards,

Laurent Pinchart
Signature · 1 line
Sakari Ailus
author maintainer
#70 lore raw

PATCH v8 16/38 media: v4l: subdev: Allow a larger number of routes than there's room for

In reply to Sakari Ailus

On VIDIOC_SUBDEV_[GS]_ROUTING, only return as many routes back to the user as there's room. Do not consider it an error if more routes existed. Simply inform the user there are more routes.

---
 .../media/v4l/vidioc-subdev-g-routing.rst              |  4 ----
 drivers/media/v4l2-core/v4l2-subdev.c                  | 10 +++-------
 2 files changed, 3 insertions(+), 11 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -4
index 6eb6a59570dc..08b8d17cef3f 100644
@@ -147,10 +147,6 @@ On success 0 is returned, on error -1 and the ``errno`` variable is set
 appropriately. The generic error codes are described at the
 :ref:`Generic Error Codes <gen-errors>` chapter.
 
-ENOSPC
-   The application provided ``num_routes`` is not big enough to contain
-   all the available routes the subdevice exposes.
-
 EINVAL
    The sink or source pad identifiers reference a non-existing pad or reference
    pads of different types (ie. the sink_pad identifiers refers to a source
drivers/media/v4l2-core/v4l2-subdev.c +3 -7
index a357ce318192..a6107e440ef0 100644
@@ -964,7 +964,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
 		       state->routing.routes,
-		       state->routing.num_routes *
+		       min(state->routing.num_routes, routing->len_routes)  *
 		       sizeof(*state->routing.routes));
 		routing->num_routes = state->routing.num_routes;
 
@@ -985,14 +985,10 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		krouting = &state->routing;
 
-		if (routing->len_routes < krouting->num_routes) {
-			routing->num_routes = krouting->num_routes;
-			return -ENOSPC;
-		}
-
 		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
 		       krouting->routes,
-		       krouting->num_routes * sizeof(*krouting->routes));
+		       min(krouting->num_routes, routing->len_routes) *
+		       sizeof(*krouting->routes));
 		routing->num_routes = krouting->num_routes;
 
 		return 0;
Signature · 1 line
2.39.2
maintainer
#71 lore raw

Re: [PATCH v8 16/38] media: v4l: subdev: Allow a larger number of routes than there's room for

Hi Sakari,

Thank you for the patch.

Quoted 15 lines · Sakari Ailus

On VIDIOC_SUBDEV_[GS]_ROUTING, only return as many routes back to the user as there's room. Do not consider it an error if more routes existed. Simply inform the user there are more routes.

---
 .../media/v4l/vidioc-subdev-g-routing.rst              |  4 ----
 drivers/media/v4l2-core/v4l2-subdev.c                  | 10 +++-------
 2 files changed, 3 insertions(+), 11 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -0
index 6eb6a59570dc..08b8d17cef3f 100644

Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst On success 0 is returned, on error -1 and the ``errno`` variable is set · line 153

Quoted 7 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -4
@@ -147,10 +147,6 @@ On success 0 is returned, on error -1 and the ``errno`` variable is set
 appropriately. The generic error codes are described at the
 :ref:`Generic Error Codes <gen-errors>` chapter.
 
-ENOSPC
-   The application provided ``num_routes`` is not big enough to contain
-   all the available routes the subdevice exposes.
-

I wonder if this patch should be squashed with the previous two. You update the documentation in 14/38 already.

Quoted 33 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -0
 EINVAL
    The sink or source pad identifiers reference a non-existing pad or reference
    pads of different types (ie. the sink_pad identifiers refers to a source
drivers/media/v4l2-core/v4l2-subdev.c +3 -7
index a357ce318192..a6107e440ef0 100644
@@ -964,7 +964,7 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
 		       state->routing.routes,
-		       state->routing.num_routes *
+		       min(state->routing.num_routes, routing->len_routes)  *
 		       sizeof(*state->routing.routes));
 		routing->num_routes = state->routing.num_routes;
 
@@ -985,14 +985,10 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		krouting = &state->routing;
 
-		if (routing->len_routes < krouting->num_routes) {
-			routing->num_routes = krouting->num_routes;
-			return -ENOSPC;
-		}
-
 		memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
 		       krouting->routes,
-		       krouting->num_routes * sizeof(*krouting->routes));
+		       min(krouting->num_routes, routing->len_routes) *
+		       sizeof(*krouting->routes));
 		routing->num_routes = krouting->num_routes;
 
 		return 0;
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#72 lore raw

Re: [PATCH v8 16/38] media: v4l: subdev: Allow a larger number of routes than there's room for

Quoted 23 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 18 lines · Sakari Ailus

On VIDIOC_SUBDEV_[GS]_ROUTING, only return as many routes back to the user as there's room. Do not consider it an error if more routes existed. Simply inform the user there are more routes.

---
 .../media/v4l/vidioc-subdev-g-routing.rst              |  4 ----
 drivers/media/v4l2-core/v4l2-subdev.c                  | 10 +++-------
 2 files changed, 3 insertions(+), 11 deletions(-)
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -0
index 6eb6a59570dc..08b8d17cef3f 100644
@@ -147,10 +147,6 @@ On success 0 is returned, on error -1 and the ``errno`` variable is set
 appropriately. The generic error codes are described at the
 :ref:`Generic Error Codes <gen-errors>` chapter.
 
Quoted 7 lines

Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst On success 0 is returned, on error -1 and the ``errno`` variable is set · line 153

Quoted 4 lines
Documentation/userspace-api/media/v4l/vidioc-subdev-g-routing.rst +0 -4
-ENOSPC
-   The application provided ``num_routes`` is not big enough to contain
-   all the available routes the subdevice exposes.
-

I wonder if this patch should be squashed with the previous two. You update the documentation in 14/38 already.

I'll do that.

Signature · 1 line
Sakari Ailus
author maintainer
#73 lore raw

PATCH v8 17/38 media: v4l: subdev: Add trivial set_routing support

In reply to Sakari Ailus

Add trivial S_ROUTING IOCTL support for drivers where routing is static. Essentially this means returning the same information G_ROUTING call would have done.

---
 drivers/media/v4l2-core/v4l2-subdev.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/media/v4l2-core/v4l2-subdev.c +14 -0
index a6107e440ef0..c8c435df92c8 100644
@@ -930,6 +930,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
Signature · 1 line
2.39.2
Reviewed-by on v8 17/38
#74 lore raw

Re: [PATCH v8 17/38] media: v4l: subdev: Add trivial set_routing support

Quoted 5 lines · Sakari Ailus

Add trivial S_ROUTING IOCTL support for drivers where routing is static. Essentially this means returning the same information G_ROUTING call would have done.

Quoted 29 lines
---
  drivers/media/v4l2-core/v4l2-subdev.c | 14 ++++++++++++++
  1 file changed, 14 insertions(+)
drivers/media/v4l2-core/v4l2-subdev.c +14 -0
index a6107e440ef0..c8c435df92c8 100644
@@ -930,6 +930,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
  		memset(routing->reserved, 0, sizeof(routing->reserved));
 
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+
  		for (i = 0; i < routing->num_routes; ++i) {
  			const struct v4l2_subdev_route *route = &routes[i];
  			const struct media_pad *pads = sd->entity.pads;
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer Reviewed-by on v8 17/38
#75 lore raw

Re: [PATCH v8 17/38] media: v4l: subdev: Add trivial set_routing support

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 5 lines · Sakari Ailus

Add trivial S_ROUTING IOCTL support for drivers where routing is static. Essentially this means returning the same information G_ROUTING call would have done.

Quoted 29 lines
---
 drivers/media/v4l2-core/v4l2-subdev.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/media/v4l2-core/v4l2-subdev.c +14 -0
index a6107e440ef0..c8c435df92c8 100644
@@ -930,6 +930,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
Signature · 3 lines
Regards,

Laurent Pinchart
maintainer
#76 lore raw

Re: [PATCH v8 17/38] media: v4l: subdev: Add trivial set_routing support

Quoted 4 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 8 lines
Quoted 5 lines · Sakari Ailus

Add trivial S_ROUTING IOCTL support for drivers where routing is static. Essentially this means returning the same information G_ROUTING call would have done.

Actually...

Quoted 12 lines
Quoted 12 lines
---
 drivers/media/v4l2-core/v4l2-subdev.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index a6107e440ef0..c8c435df92c8 100644
@@ -930,6 +930,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
Quoted 14 lines
Quoted 4 lines
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 946

Quoted 10 lines
drivers/media/v4l2-core/v4l2-subdev.c +10 -0
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+

I think this should be done after the next code block that validates the arguments. Otherwise the S_ROUTING ioctl will behave differently with regard to blatantly invalid arguments, depending on whether or not the subdev implements the .set_routing() operation. This can confuse userspace, and does confuse v4l2-compliance.

I have the following change in my tree that fixes the problem, feel free to squash it with this patch in v9.

commit 1e1f03eb8bc118c53a9deab05063d71a2fe7aa5f

Author: Laurent Pinchart <[email protected]>
Date:   Tue Apr 2 02:06:01 2024 +0300

fixup! media: v4l: subdev: Add trivial set_routing support

Validate arguments before handling the trivial set_routing support to expose a consistent behaviour to userspace. This fixes an issue with v4l2-compliance.

drivers/media/v4l2-core/v4l2-subdev.c +19 -12
index def91ab32c07..129867ed2bad 100644
@@ -933,19 +933,12 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
 		/*
-		 * If the driver doesn't support setting routing, just return
-		 * the routing table here.
+		 * Perform argument validation first, or subdevs that don't
+		 * support setting routing will not return an error when
+		 * arguments are blatantly wrong. The difference in behaviour
+		 * could be confusing for userspace, and in particular for API
+		 * compliance checkers.
 		 */
-		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
-			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
-			       state->routing.routes,
-			       min(state->routing.num_routes, routing->len_routes) *
-			       sizeof(*state->routing.routes));
-			routing->num_routes = state->routing.num_routes;
-
-			return 0;
-		}
-
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
@@ -969,6 +962,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 				return -EINVAL;
 		}
 
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+
 		krouting.num_routes = routing->num_routes;
 		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
Quoted 3 lines
Quoted 3 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#77 lore raw

Re: [PATCH v8 17/38] media: v4l: subdev: Add trivial set_routing support

Hi Laurent,

Quoted 56 lines · Laurent Pinchart
Quoted 4 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 8 lines
Quoted 5 lines · Sakari Ailus

Add trivial S_ROUTING IOCTL support for drivers where routing is static. Essentially this means returning the same information G_ROUTING call would have done.

Actually...

Quoted 12 lines
Quoted 12 lines
---
 drivers/media/v4l2-core/v4l2-subdev.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index a6107e440ef0..c8c435df92c8 100644
@@ -930,6 +930,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
Quoted 14 lines
Quoted 4 lines
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 978

Quoted 10 lines
drivers/media/v4l2-core/v4l2-subdev.c +10 -0
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+

I think this should be done after the next code block that validates the arguments. Otherwise the S_ROUTING ioctl will behave differently with regard to blatantly invalid arguments, depending on whether or not the subdev implements the .set_routing() operation. This can confuse userspace, and does confuse v4l2-compliance.

I have the following change in my tree that fixes the problem, feel free to squash it with this patch in v9.

commit 1e1f03eb8bc118c53a9deab05063d71a2fe7aa5f

Author: Laurent Pinchart <[email protected]>
Date:   Tue Apr 2 02:06:01 2024 +0300
Quoted 7 lines

fixup! media: v4l: subdev: Add trivial set_routing support

Validate arguments before handling the trivial set_routing support to expose a consistent behaviour to userspace. This fixes an issue with v4l2-compliance.

Thanks, I'll squash this into the patch.

Quoted 6 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index def91ab32c07..129867ed2bad 100644

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 940

Quoted 10 lines
drivers/media/v4l2-core/v4l2-subdev.c +5 -2
@@ -933,19 +933,12 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
 		/*
-		 * If the driver doesn't support setting routing, just return
-		 * the routing table here.
+		 * Perform argument validation first, or subdevs that don't
+		 * support setting routing will not return an error when
+		 * arguments are blatantly wrong. The difference in behaviour
+		 * could be confusing for userspace, and in particular for API
+		 * compliance checkers.

This is more fit for development time discussion, not something that should be left in the code IMO.

Quoted 41 lines
drivers/media/v4l2-core/v4l2-subdev.c +14 -10
 		 */
-		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
-			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
-			       state->routing.routes,
-			       min(state->routing.num_routes, routing->len_routes) *
-			       sizeof(*state->routing.routes));
-			routing->num_routes = state->routing.num_routes;
-
-			return 0;
-		}
-
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
@@ -969,6 +962,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 				return -EINVAL;
 		}
 
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+
 		krouting.num_routes = routing->num_routes;
 		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
Quoted 5 lines
Quoted 5 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
Signature · 3 lines
Regards,

Sakari Ailus
maintainer
#78 lore raw

Re: [PATCH v8 17/38] media: v4l: subdev: Add trivial set_routing support

Quoted 77 lines · Sakari Ailus

Hi Laurent,

Quoted 56 lines · Laurent Pinchart
Quoted 4 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 8 lines
Quoted 5 lines · Sakari Ailus

Add trivial S_ROUTING IOCTL support for drivers where routing is static. Essentially this means returning the same information G_ROUTING call would have done.

Actually...

Quoted 12 lines
Quoted 12 lines
---
 drivers/media/v4l2-core/v4l2-subdev.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index a6107e440ef0..c8c435df92c8 100644
@@ -930,6 +930,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
Quoted 14 lines
Quoted 4 lines
drivers/media/v4l2-core/v4l2-subdev.c +4 -0
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 978

Quoted 10 lines
drivers/media/v4l2-core/v4l2-subdev.c +10 -0
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+

I think this should be done after the next code block that validates the arguments. Otherwise the S_ROUTING ioctl will behave differently with regard to blatantly invalid arguments, depending on whether or not the subdev implements the .set_routing() operation. This can confuse userspace, and does confuse v4l2-compliance.

I have the following change in my tree that fixes the problem, feel free to squash it with this patch in v9.

commit 1e1f03eb8bc118c53a9deab05063d71a2fe7aa5f

Author: Laurent Pinchart <[email protected]>
Date:   Tue Apr 2 02:06:01 2024 +0300
Quoted 7 lines

fixup! media: v4l: subdev: Add trivial set_routing support

Validate arguments before handling the trivial set_routing support to expose a consistent behaviour to userspace. This fixes an issue with v4l2-compliance.

Thanks, I'll squash this into the patch.

Quoted 8 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index def91ab32c07..129867ed2bad 100644
@@ -933,19 +933,12 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
Quoted 11 lines

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 940

Quoted 8 lines
drivers/media/v4l2-core/v4l2-subdev.c +5 -2
 		/*
-		 * If the driver doesn't support setting routing, just return
-		 * the routing table here.
+		 * Perform argument validation first, or subdevs that don't
+		 * support setting routing will not return an error when
+		 * arguments are blatantly wrong. The difference in behaviour
+		 * could be confusing for userspace, and in particular for API
+		 * compliance checkers.

This is more fit for development time discussion, not something that should be left in the code IMO.

I added that comment to make sure that the next developer who refactors the code will not end up changing the order and introducing subtle breakages without realizing it. There's a high chance we wouldn't catch the problem during review if this happens after our brain caches get flushed.

Quoted 39 lines
Quoted 39 lines
drivers/media/v4l2-core/v4l2-subdev.c +14 -10
-		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
-			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
-			       state->routing.routes,
-			       min(state->routing.num_routes, routing->len_routes) *
-			       sizeof(*state->routing.routes));
-			routing->num_routes = state->routing.num_routes;
-
-			return 0;
-		}
-
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
@@ -969,6 +962,20 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 				return -EINVAL;
 		}
 
+		/*
+		 * If the driver doesn't support setting routing, just return
+		 * the routing table here.
+		 */
+		if (!v4l2_subdev_has_op(sd, pad, set_routing)) {
+			memcpy((struct v4l2_subdev_route *)(uintptr_t)routing->routes,
+			       state->routing.routes,
+			       min(state->routing.num_routes, routing->len_routes) *
+			       sizeof(*state->routing.routes));
+			routing->num_routes = state->routing.num_routes;
+
+			return 0;
+		}
+
 		krouting.num_routes = routing->num_routes;
 		krouting.len_routes = routing->len_routes;
 		krouting.routes = routes;
Quoted 5 lines
Quoted 5 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
 		for (i = 0; i < routing->num_routes; ++i) {
 			const struct v4l2_subdev_route *route = &routes[i];
 			const struct media_pad *pads = sd->entity.pads;
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#79 lore raw

Re: [PATCH v8 17/38] media: v4l: subdev: Add trivial set_routing support

Hi Laurent,

Quoted 43 lines · Laurent Pinchart
Quoted 31 lines
Quoted 13 lines

I think this should be done after the next code block that validates the arguments. Otherwise the S_ROUTING ioctl will behave differently with regard to blatantly invalid arguments, depending on whether or not the subdev implements the .set_routing() operation. This can confuse userspace, and does confuse v4l2-compliance.

I have the following change in my tree that fixes the problem, feel free to squash it with this patch in v9.

commit 1e1f03eb8bc118c53a9deab05063d71a2fe7aa5f

Author: Laurent Pinchart <[email protected]>
Date:   Tue Apr 2 02:06:01 2024 +0300
Quoted 7 lines

fixup! media: v4l: subdev: Add trivial set_routing support

Validate arguments before handling the trivial set_routing support to expose a consistent behaviour to userspace. This fixes an issue with v4l2-compliance.

Thanks, I'll squash this into the patch.

Quoted 8 lines
drivers/media/v4l2-core/v4l2-subdev.c +0 -0
index def91ab32c07..129867ed2bad 100644
@@ -933,19 +933,12 @@ static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg,
 		memset(routing->reserved, 0, sizeof(routing->reserved));
 
Quoted 11 lines

drivers/media/v4l2-core/v4l2-subdev.c static long subdev_do_ioctl(struct file *file, unsigned int cmd, void *arg, · line 940

Quoted 8 lines
drivers/media/v4l2-core/v4l2-subdev.c +5 -2
 		/*
-		 * If the driver doesn't support setting routing, just return
-		 * the routing table here.
+		 * Perform argument validation first, or subdevs that don't
+		 * support setting routing will not return an error when
+		 * arguments are blatantly wrong. The difference in behaviour
+		 * could be confusing for userspace, and in particular for API
+		 * compliance checkers.

This is more fit for development time discussion, not something that should be left in the code IMO.

Quoted 5 lines

I added that comment to make sure that the next developer who refactors the code will not end up changing the order and introducing subtle breakages without realizing it. There's a high chance we wouldn't catch the problem during review if this happens after our brain caches get flushed.

We don't have other comments like that in the code either. Input validation is generally needed, in this case it wasn't needed to carry out the task but to align the API behaviour with the drivers that do and don't support setting routing. It's not worth a comment here: it's the same for other IOCTLs as well. So if a comment were to be added, I'd add it before the switch.

Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#80 lore raw

PATCH v8 18/38 media: ccs: No need to set streaming to false in power off

In reply to Sakari Ailus

Streaming will have been stopped by the sensor is powered off, and so sensor->streaming is also false already. Do not set it as part of the runtime suspend callback.

---
 drivers/media/i2c/ccs/ccs-core.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/i2c/ccs/ccs-core.c +0 -1
index e21287d50c15..671540a8ab6a 100644
@@ -1710,7 +1710,6 @@ static int ccs_power_off(struct device *dev)
 	usleep_range(5000, 5000);
 	regulator_bulk_disable(ARRAY_SIZE(ccs_regulators),
 			       sensor->regulators);
-	sensor->streaming = false;
 
 	return 0;
 }
Signature · 1 line
2.39.2
Reviewed-by on v8 18/38
#81 lore raw

Re: [PATCH v8 18/38] media: ccs: No need to set streaming to false in power off

Quoting Sakari Ailus (2024-03-13 07:24:56)

Quoted 1 line

Streaming will have been stopped by the sensor is powered off, and so

by the time the sensor

or

before the sensor

Quoted 3 lines

sensor->streaming is also false already. Do not set it as part of the runtime suspend callback.

Quoted 20 lines
---
 drivers/media/i2c/ccs/ccs-core.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/i2c/ccs/ccs-core.c +0 -1
index e21287d50c15..671540a8ab6a 100644
@@ -1710,7 +1710,6 @@ static int ccs_power_off(struct device *dev)
        usleep_range(5000, 5000);
        regulator_bulk_disable(ARRAY_SIZE(ccs_regulators),
                               sensor->regulators);
-       sensor->streaming = false;
 
        return 0;
 }
Signature · 1 line
2.39.2
maintainer Reviewed-by on v8 18/38
#82 lore raw

Re: [PATCH v8 18/38] media: ccs: No need to set streaming to false in power off

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 5 lines · Sakari Ailus

Streaming will have been stopped by the sensor is powered off, and so sensor->streaming is also false already. Do not set it as part of the runtime suspend callback.

Quoted 16 lines
---
 drivers/media/i2c/ccs/ccs-core.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/i2c/ccs/ccs-core.c +0 -1
index e21287d50c15..671540a8ab6a 100644
@@ -1710,7 +1710,6 @@ static int ccs_power_off(struct device *dev)
 	usleep_range(5000, 5000);
 	regulator_bulk_disable(ARRAY_SIZE(ccs_regulators),
 			       sensor->regulators);
-	sensor->streaming = false;
 
 	return 0;
 }
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#83 lore raw

PATCH v8 19/38 media: ccs: Use {enable,disable}_streams operations

In reply to Sakari Ailus

Switch from s_stream() video op to enable_streams() and disable_streams() pad operations. They are preferred and required for streams support.

---
 drivers/media/i2c/ccs/ccs-core.c | 187 +++++++++++++++----------------
 1 file changed, 88 insertions(+), 99 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +88 -99
index 671540a8ab6a..9e70946653e9 100644
@@ -1715,22 +1715,66 @@ static int ccs_power_off(struct device *dev)
 }
 
 /* -----------------------------------------------------------------------------
- * Video stream management
+ * V4L2 subdev video operations
  */
 
-static int ccs_start_streaming(struct ccs_sensor *sensor)
+static int ccs_pm_get_init(struct ccs_sensor *sensor)
+{
+	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
+	int rval;
+
+	/*
+	 * It can't use pm_runtime_resume_and_get() here, as the driver
+	 * relies at the returned value to detect if the device was already
+	 * active or not.
+	 */
+	rval = pm_runtime_get_sync(&client->dev);
+	if (rval < 0)
+		goto error;
+
+	/* Device was already active, so don't set controls */
+	if (rval == 1 && !sensor->handler_setup_needed)
+		return 0;
+
+	sensor->handler_setup_needed = false;
+
+	/* Restore V4L2 controls to the previously suspended device */
+	rval = __v4l2_ctrl_handler_setup(&sensor->pixel_array->ctrl_handler);
+	if (rval)
+		goto error;
+
+	rval = __v4l2_ctrl_handler_setup(&sensor->src->ctrl_handler);
+	if (rval)
+		goto error;
+
+	/* Keep PM runtime usage_count incremented on success */
+	return 0;
+error:
+	pm_runtime_put(&client->dev);
+	return rval;
+}
+
+static int ccs_enable_streams(struct v4l2_subdev *subdev,
+			      struct v4l2_subdev_state *state, u32 pad,
+			      u64 streams_mask)
 {
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
 	unsigned int binning_mode;
 	int rval;
 
-	mutex_lock(&sensor->mutex);
+	if (pad != CCS_PAD_SRC)
+		return -EINVAL;
+
+	rval = ccs_pm_get_init(sensor);
+	if (rval)
+		return rval;
 
 	rval = ccs_write(sensor, CSI_DATA_FORMAT,
 			 (sensor->csi_format->width << 8) |
 			 sensor->csi_format->compressed);
 	if (rval)
-		goto out;
+		goto err_pm_put;
 
 	/* Binning configuration */
 	if (sensor->binning_horizontal == 1 &&
@@ -1743,38 +1787,38 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 
 		rval = ccs_write(sensor, BINNING_TYPE, binning_type);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		binning_mode = 1;
 	}
 	rval = ccs_write(sensor, BINNING_MODE, binning_mode);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	/* Set up PLL */
 	rval = ccs_pll_configure(sensor);
 	if (rval)
-		goto out;
+		goto err_pm_put;
 
 	/* Analog crop start coordinates */
 	rval = ccs_write(sensor, X_ADDR_START, sensor->pa_src.left);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	rval = ccs_write(sensor, Y_ADDR_START, sensor->pa_src.top);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	/* Analog crop end coordinates */
 	rval = ccs_write(sensor, X_ADDR_END,
 			 sensor->pa_src.left + sensor->pa_src.width - 1);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	rval = ccs_write(sensor, Y_ADDR_END,
 			 sensor->pa_src.top + sensor->pa_src.height - 1);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	/*
 	 * Output from pixel array, including blanking, is set using
@@ -1787,22 +1831,22 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 		rval = ccs_write(sensor, DIGITAL_CROP_X_OFFSET,
 				 sensor->scaler_sink.left);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_Y_OFFSET,
 				 sensor->scaler_sink.top);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_IMAGE_WIDTH,
 				 sensor->scaler_sink.width);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_IMAGE_HEIGHT,
 				 sensor->scaler_sink.height);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 	}
 
 	/* Scaling */
@@ -1810,20 +1854,20 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 	    != CCS_SCALING_CAPABILITY_NONE) {
 		rval = ccs_write(sensor, SCALING_MODE, sensor->scaling_mode);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, SCALE_M, sensor->scale_m);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 	}
 
 	/* Output size from sensor */
 	rval = ccs_write(sensor, X_OUTPUT_SIZE, sensor->src_src.width);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 	rval = ccs_write(sensor, Y_OUTPUT_SIZE, sensor->src_src.height);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	if (CCS_LIM(sensor, FLASH_MODE_CAPABILITY) &
 	    (CCS_FLASH_MODE_CAPABILITY_SINGLE_STROBE |
@@ -1832,111 +1876,52 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 	    sensor->hwcfg.strobe_setup->trigger != 0) {
 		rval = ccs_setup_flash_strobe(sensor);
 		if (rval)
-			goto out;
+			goto err_pm_put;
 	}
 
 	rval = ccs_call_quirk(sensor, pre_streamon);
 	if (rval) {
 		dev_err(&client->dev, "pre_streamon quirks failed\n");
-		goto out;
+		goto err_pm_put;
 	}
 
 	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_STREAMING);
 
-out:
-	mutex_unlock(&sensor->mutex);
-
-	return rval;
-}
-
-static int ccs_stop_streaming(struct ccs_sensor *sensor)
-{
-	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
-	int rval;
-
-	mutex_lock(&sensor->mutex);
-	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
-	if (rval)
-		goto out;
-
-	rval = ccs_call_quirk(sensor, post_streamoff);
-	if (rval)
-		dev_err(&client->dev, "post_streamoff quirks failed\n");
-
-out:
-	mutex_unlock(&sensor->mutex);
-	return rval;
-}
-
-/* -----------------------------------------------------------------------------
- * V4L2 subdev video operations
- */
-
-static int ccs_pm_get_init(struct ccs_sensor *sensor)
-{
-	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
-	int rval;
-
-	/*
-	 * It can't use pm_runtime_resume_and_get() here, as the driver
-	 * relies at the returned value to detect if the device was already
-	 * active or not.
-	 */
-	rval = pm_runtime_get_sync(&client->dev);
-	if (rval < 0)
-		goto error;
-
-	/* Device was already active, so don't set controls */
-	if (rval == 1 && !sensor->handler_setup_needed)
-		return 0;
-
-	sensor->handler_setup_needed = false;
+	sensor->streaming = true;
 
-	/* Restore V4L2 controls to the previously suspended device */
-	rval = v4l2_ctrl_handler_setup(&sensor->pixel_array->ctrl_handler);
-	if (rval)
-		goto error;
+	return 0;
 
-	rval = v4l2_ctrl_handler_setup(&sensor->src->ctrl_handler);
-	if (rval)
-		goto error;
+err_pm_put:
+	pm_runtime_mark_last_busy(&client->dev);
+	pm_runtime_put_autosuspend(&client->dev);
 
-	/* Keep PM runtime usage_count incremented on success */
-	return 0;
-error:
-	pm_runtime_put(&client->dev);
 	return rval;
 }
 
-static int ccs_set_stream(struct v4l2_subdev *subdev, int enable)
+static int ccs_disable_streams(struct v4l2_subdev *subdev,
+			       struct v4l2_subdev_state *state, u32 pad,
+			       u64 streams_mask)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
 	int rval;
 
-	if (!enable) {
-		ccs_stop_streaming(sensor);
-		sensor->streaming = false;
-		pm_runtime_mark_last_busy(&client->dev);
-		pm_runtime_put_autosuspend(&client->dev);
-
-		return 0;
-	}
+	if (pad != CCS_PAD_SRC)
+		return -EINVAL;
 
-	rval = ccs_pm_get_init(sensor);
+	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
 	if (rval)
 		return rval;
 
-	sensor->streaming = true;
+	rval = ccs_call_quirk(sensor, post_streamoff);
+	if (rval)
+		dev_err(&client->dev, "post_streamoff quirks failed\n");
 
-	rval = ccs_start_streaming(sensor);
-	if (rval < 0) {
-		sensor->streaming = false;
-		pm_runtime_mark_last_busy(&client->dev);
-		pm_runtime_put_autosuspend(&client->dev);
-	}
+	sensor->streaming = false;
+	pm_runtime_mark_last_busy(&client->dev);
+	pm_runtime_put_autosuspend(&client->dev);
 
-	return rval;
+	return 0;
 }
 
 static int ccs_pre_streamon(struct v4l2_subdev *subdev, u32 flags)
@@ -1962,7 +1947,9 @@ static int ccs_pre_streamon(struct v4l2_subdev *subdev, u32 flags)
 		}
 	}
 
+	mutex_lock(&sensor->mutex);
 	rval = ccs_pm_get_init(sensor);
+	mutex_unlock(&sensor->mutex);
 	if (rval)
 		return rval;
 
@@ -3046,7 +3033,7 @@ static int ccs_init_state(struct v4l2_subdev *sd,
 }
 
 static const struct v4l2_subdev_video_ops ccs_video_ops = {
-	.s_stream = ccs_set_stream,
+	.s_stream = v4l2_subdev_s_stream_helper,
 	.pre_streamon = ccs_pre_streamon,
 	.post_streamoff = ccs_post_streamoff,
 };
@@ -3057,6 +3044,8 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
 	.set_fmt = ccs_set_format,
 	.get_selection = ccs_get_selection,
 	.set_selection = ccs_set_selection,
+	.enable_streams = ccs_enable_streams,
+	.disable_streams = ccs_disable_streams,
 };
 
 static const struct v4l2_subdev_sensor_ops ccs_sensor_ops = {
Signature · 1 line
2.39.2
maintainer Reviewed-by on v8 19/38
#84 lore raw

Re: [PATCH v8 19/38] media: ccs: Use {enable,disable}_streams operations

Hi Sakari,

Thank you for the patch.

Quoted 13 lines · Sakari Ailus

Switch from s_stream() video op to enable_streams() and disable_streams() pad operations. They are preferred and required for streams support.

---
 drivers/media/i2c/ccs/ccs-core.c | 187 +++++++++++++++----------------
 1 file changed, 88 insertions(+), 99 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +0 -0
index 671540a8ab6a..9e70946653e9 100644

drivers/media/i2c/ccs/ccs-core.c static int ccs_power_off(struct device *dev) · line 1721

Quoted 9 lines
drivers/media/i2c/ccs/ccs-core.c +2 -2
@@ -1715,22 +1715,66 @@ static int ccs_power_off(struct device *dev)
 }
 
 /* -----------------------------------------------------------------------------
- * Video stream management
+ * V4L2 subdev video operations
  */
 
-static int ccs_start_streaming(struct ccs_sensor *sensor)
+static int ccs_pm_get_init(struct ccs_sensor *sensor)

This patch would have been easier to review if you had moved the function in a separate patch.

Quoted 20 lines
drivers/media/i2c/ccs/ccs-core.c +20 -0
+{
+	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
+	int rval;
+
+	/*
+	 * It can't use pm_runtime_resume_and_get() here, as the driver
+	 * relies at the returned value to detect if the device was already
+	 * active or not.
+	 */
+	rval = pm_runtime_get_sync(&client->dev);
+	if (rval < 0)
+		goto error;
+
+	/* Device was already active, so don't set controls */
+	if (rval == 1 && !sensor->handler_setup_needed)
+		return 0;
+
+	sensor->handler_setup_needed = false;
+
+	/* Restore V4L2 controls to the previously suspended device */

drivers/media/i2c/ccs/ccs-core.c static int ccs_power_off(struct device *dev) · line 1751

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +10 -0
+	rval = __v4l2_ctrl_handler_setup(&sensor->pixel_array->ctrl_handler);
+	if (rval)
+		goto error;
+
+	rval = __v4l2_ctrl_handler_setup(&sensor->src->ctrl_handler);
+	if (rval)
+		goto error;
+
+	/* Keep PM runtime usage_count incremented on success */
+	return 0;

Missing blank line.

Quoted 18 lines
drivers/media/i2c/ccs/ccs-core.c +12 -1
+error:
+	pm_runtime_put(&client->dev);
+	return rval;
+}
+
+static int ccs_enable_streams(struct v4l2_subdev *subdev,
+			      struct v4l2_subdev_state *state, u32 pad,
+			      u64 streams_mask)
 {
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
 	unsigned int binning_mode;
 	int rval;
 
-	mutex_lock(&sensor->mutex);
+	if (pad != CCS_PAD_SRC)
+		return -EINVAL;
+

drivers/media/i2c/ccs/ccs-core.c static int ccs_power_off(struct device *dev) · line 1777

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +4 -1
+	rval = ccs_pm_get_init(sensor);
+	if (rval)
+		return rval;
 
 	rval = ccs_write(sensor, CSI_DATA_FORMAT,
 			 (sensor->csi_format->width << 8) |
 			 sensor->csi_format->compressed);
 	if (rval)
-		goto out;
+		goto err_pm_put;

You could also have renamed the error label separately :-)

Quoted 170 lines
drivers/media/i2c/ccs/ccs-core.c +18 -67
 	/* Binning configuration */
 	if (sensor->binning_horizontal == 1 &&
@@ -1743,38 +1787,38 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 
 		rval = ccs_write(sensor, BINNING_TYPE, binning_type);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		binning_mode = 1;
 	}
 	rval = ccs_write(sensor, BINNING_MODE, binning_mode);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	/* Set up PLL */
 	rval = ccs_pll_configure(sensor);
 	if (rval)
-		goto out;
+		goto err_pm_put;
 
 	/* Analog crop start coordinates */
 	rval = ccs_write(sensor, X_ADDR_START, sensor->pa_src.left);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	rval = ccs_write(sensor, Y_ADDR_START, sensor->pa_src.top);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	/* Analog crop end coordinates */
 	rval = ccs_write(sensor, X_ADDR_END,
 			 sensor->pa_src.left + sensor->pa_src.width - 1);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	rval = ccs_write(sensor, Y_ADDR_END,
 			 sensor->pa_src.top + sensor->pa_src.height - 1);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	/*
 	 * Output from pixel array, including blanking, is set using
@@ -1787,22 +1831,22 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 		rval = ccs_write(sensor, DIGITAL_CROP_X_OFFSET,
 				 sensor->scaler_sink.left);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_Y_OFFSET,
 				 sensor->scaler_sink.top);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_IMAGE_WIDTH,
 				 sensor->scaler_sink.width);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_IMAGE_HEIGHT,
 				 sensor->scaler_sink.height);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 	}
 
 	/* Scaling */
@@ -1810,20 +1854,20 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 	    != CCS_SCALING_CAPABILITY_NONE) {
 		rval = ccs_write(sensor, SCALING_MODE, sensor->scaling_mode);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 
 		rval = ccs_write(sensor, SCALE_M, sensor->scale_m);
 		if (rval < 0)
-			goto out;
+			goto err_pm_put;
 	}
 
 	/* Output size from sensor */
 	rval = ccs_write(sensor, X_OUTPUT_SIZE, sensor->src_src.width);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 	rval = ccs_write(sensor, Y_OUTPUT_SIZE, sensor->src_src.height);
 	if (rval < 0)
-		goto out;
+		goto err_pm_put;
 
 	if (CCS_LIM(sensor, FLASH_MODE_CAPABILITY) &
 	    (CCS_FLASH_MODE_CAPABILITY_SINGLE_STROBE |
@@ -1832,111 +1876,52 @@ static int ccs_start_streaming(struct ccs_sensor *sensor)
 	    sensor->hwcfg.strobe_setup->trigger != 0) {
 		rval = ccs_setup_flash_strobe(sensor);
 		if (rval)
-			goto out;
+			goto err_pm_put;
 	}
 
 	rval = ccs_call_quirk(sensor, pre_streamon);
 	if (rval) {
 		dev_err(&client->dev, "pre_streamon quirks failed\n");
-		goto out;
+		goto err_pm_put;
 	}
 
 	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_STREAMING);
 
-out:
-	mutex_unlock(&sensor->mutex);
-
-	return rval;
-}
-
-static int ccs_stop_streaming(struct ccs_sensor *sensor)
-{
-	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
-	int rval;
-
-	mutex_lock(&sensor->mutex);
-	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
-	if (rval)
-		goto out;
-
-	rval = ccs_call_quirk(sensor, post_streamoff);
-	if (rval)
-		dev_err(&client->dev, "post_streamoff quirks failed\n");
-
-out:
-	mutex_unlock(&sensor->mutex);
-	return rval;
-}
-
-/* -----------------------------------------------------------------------------
- * V4L2 subdev video operations
- */
-
-static int ccs_pm_get_init(struct ccs_sensor *sensor)
-{
-	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
-	int rval;
-
-	/*
-	 * It can't use pm_runtime_resume_and_get() here, as the driver
-	 * relies at the returned value to detect if the device was already
-	 * active or not.
-	 */
-	rval = pm_runtime_get_sync(&client->dev);
-	if (rval < 0)
-		goto error;
-
-	/* Device was already active, so don't set controls */
-	if (rval == 1 && !sensor->handler_setup_needed)
-		return 0;
-
-	sensor->handler_setup_needed = false;
+	sensor->streaming = true;
 
-	/* Restore V4L2 controls to the previously suspended device */
-	rval = v4l2_ctrl_handler_setup(&sensor->pixel_array->ctrl_handler);

drivers/media/i2c/ccs/ccs-core.c static int ccs_start_streaming(struct ccs_sensor *sensor) · line 1896

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +4 -5
-	if (rval)
-		goto error;
+	return 0;
 
-	rval = v4l2_ctrl_handler_setup(&sensor->src->ctrl_handler);
-	if (rval)
-		goto error;
+err_pm_put:
+	pm_runtime_mark_last_busy(&client->dev);
+	pm_runtime_put_autosuspend(&client->dev);

On failures I tend not to use autosuspend, based on the idea that it may be easier to recover from a failure if the sensor got powered off. I'm not sure that's really relevant though.

Quoted 81 lines
drivers/media/i2c/ccs/ccs-core.c +18 -23
-	/* Keep PM runtime usage_count incremented on success */
-	return 0;
-error:
-	pm_runtime_put(&client->dev);
 	return rval;
 }
 
-static int ccs_set_stream(struct v4l2_subdev *subdev, int enable)
+static int ccs_disable_streams(struct v4l2_subdev *subdev,
+			       struct v4l2_subdev_state *state, u32 pad,
+			       u64 streams_mask)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
 	int rval;
 
-	if (!enable) {
-		ccs_stop_streaming(sensor);
-		sensor->streaming = false;
-		pm_runtime_mark_last_busy(&client->dev);
-		pm_runtime_put_autosuspend(&client->dev);
-
-		return 0;
-	}
+	if (pad != CCS_PAD_SRC)
+		return -EINVAL;
 
-	rval = ccs_pm_get_init(sensor);
+	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
 	if (rval)
 		return rval;
 
-	sensor->streaming = true;
+	rval = ccs_call_quirk(sensor, post_streamoff);
+	if (rval)
+		dev_err(&client->dev, "post_streamoff quirks failed\n");
 
-	rval = ccs_start_streaming(sensor);
-	if (rval < 0) {
-		sensor->streaming = false;
-		pm_runtime_mark_last_busy(&client->dev);
-		pm_runtime_put_autosuspend(&client->dev);
-	}
+	sensor->streaming = false;
+	pm_runtime_mark_last_busy(&client->dev);
+	pm_runtime_put_autosuspend(&client->dev);
 
-	return rval;
+	return 0;
 }
 
 static int ccs_pre_streamon(struct v4l2_subdev *subdev, u32 flags)
@@ -1962,7 +1947,9 @@ static int ccs_pre_streamon(struct v4l2_subdev *subdev, u32 flags)
 		}
 	}
 
+	mutex_lock(&sensor->mutex);
 	rval = ccs_pm_get_init(sensor);
+	mutex_unlock(&sensor->mutex);
 	if (rval)
 		return rval;
 
@@ -3046,7 +3033,7 @@ static int ccs_init_state(struct v4l2_subdev *sd,
 }
 
 static const struct v4l2_subdev_video_ops ccs_video_ops = {
-	.s_stream = ccs_set_stream,
+	.s_stream = v4l2_subdev_s_stream_helper,
 	.pre_streamon = ccs_pre_streamon,
 	.post_streamoff = ccs_post_streamoff,
 };
@@ -3057,6 +3044,8 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
 	.set_fmt = ccs_set_format,
 	.get_selection = ccs_get_selection,
 	.set_selection = ccs_set_selection,
+	.enable_streams = ccs_enable_streams,
+	.disable_streams = ccs_disable_streams,
 };
 
 static const struct v4l2_subdev_sensor_ops ccs_sensor_ops = {
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#85 lore raw

PATCH v8 20/38 media: ccs: Track streaming state

In reply to Sakari Ailus

With enable_streams and disable_streams, the driver for a device where streams are not independently started and stopped needs to maintain state information on streams that have been requested to be started. Do that now.

In the future, a helper function in the framework is a desirable way to do this instead.

---
 drivers/media/i2c/ccs/ccs-core.c | 13 ++++++++++---
 drivers/media/i2c/ccs/ccs.h      |  2 +-
 2 files changed, 11 insertions(+), 4 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +10 -3
index 9e70946653e9..0cd8ee957655 100644
@@ -1766,6 +1766,11 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 	if (pad != CCS_PAD_SRC)
 		return -EINVAL;
 
+	if (sensor->streaming) {
+		sensor->streaming |= streams_mask;
+		return 0;
+	}
+
 	rval = ccs_pm_get_init(sensor);
 	if (rval)
 		return rval;
@@ -1887,7 +1892,7 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 
 	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_STREAMING);
 
-	sensor->streaming = true;
+	sensor->streaming |= streams_mask;
 
 	return 0;
 
@@ -1909,6 +1914,10 @@ static int ccs_disable_streams(struct v4l2_subdev *subdev,
 	if (pad != CCS_PAD_SRC)
 		return -EINVAL;
 
+	sensor->streaming &= ~streams_mask;
+	if (sensor->streaming)
+		return 0;
+
 	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
 	if (rval)
 		return rval;
@@ -1917,7 +1926,6 @@ static int ccs_disable_streams(struct v4l2_subdev *subdev,
 	if (rval)
 		dev_err(&client->dev, "post_streamoff quirks failed\n");
 
-	sensor->streaming = false;
 	pm_runtime_mark_last_busy(&client->dev);
 	pm_runtime_put_autosuspend(&client->dev);
 
@@ -3525,7 +3533,6 @@ static int ccs_probe(struct i2c_client *client)
 		goto out_cleanup;
 	}
 
-	sensor->streaming = false;
 	sensor->dev_init_done = true;
 	sensor->handler_setup_needed = true;
 
drivers/media/i2c/ccs/ccs.h +1 -1
index 096573845a10..4725e6eca8d0 100644
@@ -236,7 +236,7 @@ struct ccs_sensor {
 	u16 image_start; /* image data start line */
 	u16 visible_pixel_start; /* start pixel of the visible image */
 
-	bool streaming;
+	u8 streaming;
 	bool dev_init_done;
 	bool handler_setup_needed;
 	u8 compressed_min_bpp;
Signature · 1 line
2.39.2
Reviewed-by on v8 20/38
#86 lore raw

Re: [PATCH v8 20/38] media: ccs: Track streaming state

Quoted 9 lines · Sakari Ailus

With enable_streams and disable_streams, the driver for a device where streams are not independently started and stopped needs to maintain state information on streams that have been requested to be started. Do that now.

In the future, a helper function in the framework is a desirable way to do this instead.

Quoted 70 lines
---
  drivers/media/i2c/ccs/ccs-core.c | 13 ++++++++++---
  drivers/media/i2c/ccs/ccs.h      |  2 +-
  2 files changed, 11 insertions(+), 4 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +10 -3
index 9e70946653e9..0cd8ee957655 100644
@@ -1766,6 +1766,11 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
  	if (pad != CCS_PAD_SRC)
  		return -EINVAL;
 
+	if (sensor->streaming) {
+		sensor->streaming |= streams_mask;
+		return 0;
+	}
+
  	rval = ccs_pm_get_init(sensor);
  	if (rval)
  		return rval;
@@ -1887,7 +1892,7 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 
  	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_STREAMING);
 
-	sensor->streaming = true;
+	sensor->streaming |= streams_mask;
 
  	return 0;
 
@@ -1909,6 +1914,10 @@ static int ccs_disable_streams(struct v4l2_subdev *subdev,
  	if (pad != CCS_PAD_SRC)
  		return -EINVAL;
 
+	sensor->streaming &= ~streams_mask;
+	if (sensor->streaming)
+		return 0;
+
  	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
  	if (rval)
  		return rval;
@@ -1917,7 +1926,6 @@ static int ccs_disable_streams(struct v4l2_subdev *subdev,
  	if (rval)
  		dev_err(&client->dev, "post_streamoff quirks failed\n");
 
-	sensor->streaming = false;
  	pm_runtime_mark_last_busy(&client->dev);
  	pm_runtime_put_autosuspend(&client->dev);
 
@@ -3525,7 +3533,6 @@ static int ccs_probe(struct i2c_client *client)
  		goto out_cleanup;
  	}
 
-	sensor->streaming = false;
  	sensor->dev_init_done = true;
  	sensor->handler_setup_needed = true;
 
drivers/media/i2c/ccs/ccs.h +1 -1
index 096573845a10..4725e6eca8d0 100644
@@ -236,7 +236,7 @@ struct ccs_sensor {
  	u16 image_start; /* image data start line */
  	u16 visible_pixel_start; /* start pixel of the visible image */
 
-	bool streaming;
+	u8 streaming;
  	bool dev_init_done;
  	bool handler_setup_needed;
  	u8 compressed_min_bpp;
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer Reviewed-by on v8 20/38
#87 lore raw

Re: [PATCH v8 20/38] media: ccs: Track streaming state

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 9 lines · Sakari Ailus

With enable_streams and disable_streams, the driver for a device where streams are not independently started and stopped needs to maintain state information on streams that have been requested to be started. Do that now.

In the future, a helper function in the framework is a desirable way to do this instead.

Quoted 70 lines
---
 drivers/media/i2c/ccs/ccs-core.c | 13 ++++++++++---
 drivers/media/i2c/ccs/ccs.h      |  2 +-
 2 files changed, 11 insertions(+), 4 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +10 -3
index 9e70946653e9..0cd8ee957655 100644
@@ -1766,6 +1766,11 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 	if (pad != CCS_PAD_SRC)
 		return -EINVAL;
 
+	if (sensor->streaming) {
+		sensor->streaming |= streams_mask;
+		return 0;
+	}
+
 	rval = ccs_pm_get_init(sensor);
 	if (rval)
 		return rval;
@@ -1887,7 +1892,7 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 
 	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_STREAMING);
 
-	sensor->streaming = true;
+	sensor->streaming |= streams_mask;
 
 	return 0;
 
@@ -1909,6 +1914,10 @@ static int ccs_disable_streams(struct v4l2_subdev *subdev,
 	if (pad != CCS_PAD_SRC)
 		return -EINVAL;
 
+	sensor->streaming &= ~streams_mask;
+	if (sensor->streaming)
+		return 0;
+
 	rval = ccs_write(sensor, MODE_SELECT, CCS_MODE_SELECT_SOFTWARE_STANDBY);
 	if (rval)
 		return rval;
@@ -1917,7 +1926,6 @@ static int ccs_disable_streams(struct v4l2_subdev *subdev,
 	if (rval)
 		dev_err(&client->dev, "post_streamoff quirks failed\n");
 
-	sensor->streaming = false;
 	pm_runtime_mark_last_busy(&client->dev);
 	pm_runtime_put_autosuspend(&client->dev);
 
@@ -3525,7 +3533,6 @@ static int ccs_probe(struct i2c_client *client)
 		goto out_cleanup;
 	}
 
-	sensor->streaming = false;
 	sensor->dev_init_done = true;
 	sensor->handler_setup_needed = true;
 
drivers/media/i2c/ccs/ccs.h +1 -1
index 096573845a10..4725e6eca8d0 100644
@@ -236,7 +236,7 @@ struct ccs_sensor {
 	u16 image_start; /* image data start line */
 	u16 visible_pixel_start; /* start pixel of the visible image */
 
-	bool streaming;
+	u8 streaming;
 	bool dev_init_done;
 	bool handler_setup_needed;
 	u8 compressed_min_bpp;
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#88 lore raw

PATCH v8 21/38 media: ccs: Move ccs_validate_csi_data_format up

In reply to Sakari Ailus

ccs_validate_csi_data_format() will soon be needed elsewhere, above its current location. Move it up.

---
 drivers/media/i2c/ccs/ccs-core.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +14 -14
index 0cd8ee957655..0efbc63534bc 100644
@@ -1979,6 +1979,20 @@ static int ccs_post_streamoff(struct v4l2_subdev *subdev)
 	return pm_runtime_put(&client->dev);
 }
 
+static const struct ccs_csi_data_format
+*ccs_validate_csi_data_format(struct ccs_sensor *sensor, u32 code)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
+		if (sensor->mbus_frame_fmts & (1 << i) &&
+		    ccs_csi_data_formats[i].code == code)
+			return &ccs_csi_data_formats[i];
+	}
+
+	return sensor->csi_format;
+}
+
 static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 			      struct v4l2_subdev_state *sd_state,
 			      struct v4l2_subdev_mbus_code_enum *code)
@@ -2114,20 +2128,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 	}
 }
 
-static const struct ccs_csi_data_format
-*ccs_validate_csi_data_format(struct ccs_sensor *sensor, u32 code)
-{
-	unsigned int i;
-
-	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
-		if (sensor->mbus_frame_fmts & (1 << i) &&
-		    ccs_csi_data_formats[i].code == code)
-			return &ccs_csi_data_formats[i];
-	}
-
-	return sensor->csi_format;
-}
-
 static int ccs_set_format_source(struct v4l2_subdev *subdev,
 				 struct v4l2_subdev_state *sd_state,
 				 struct v4l2_subdev_format *fmt)
Signature · 1 line
2.39.2
Reviewed-by on v8 21/38
#89 lore raw

Re: [PATCH v8 21/38] media: ccs: Move ccs_validate_csi_data_format up

Quoted 4 lines · Sakari Ailus

ccs_validate_csi_data_format() will soon be needed elsewhere, above its current location. Move it up.

Quoted 50 lines
---
  drivers/media/i2c/ccs/ccs-core.c | 28 ++++++++++++++--------------
  1 file changed, 14 insertions(+), 14 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +14 -14
index 0cd8ee957655..0efbc63534bc 100644
@@ -1979,6 +1979,20 @@ static int ccs_post_streamoff(struct v4l2_subdev *subdev)
  	return pm_runtime_put(&client->dev);
  }
 
+static const struct ccs_csi_data_format
+*ccs_validate_csi_data_format(struct ccs_sensor *sensor, u32 code)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
+		if (sensor->mbus_frame_fmts & (1 << i) &&
+		    ccs_csi_data_formats[i].code == code)
+			return &ccs_csi_data_formats[i];
+	}
+
+	return sensor->csi_format;
+}
+
  static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
  			      struct v4l2_subdev_state *sd_state,
  			      struct v4l2_subdev_mbus_code_enum *code)
@@ -2114,20 +2128,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
  	}
  }
 
-static const struct ccs_csi_data_format
-*ccs_validate_csi_data_format(struct ccs_sensor *sensor, u32 code)
-{
-	unsigned int i;
-
-	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
-		if (sensor->mbus_frame_fmts & (1 << i) &&
-		    ccs_csi_data_formats[i].code == code)
-			return &ccs_csi_data_formats[i];
-	}
-
-	return sensor->csi_format;
-}
-
  static int ccs_set_format_source(struct v4l2_subdev *subdev,
  				 struct v4l2_subdev_state *sd_state,
  				 struct v4l2_subdev_format *fmt)
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer Reviewed-by on v8 21/38
#90 lore raw

Re: [PATCH v8 21/38] media: ccs: Move ccs_validate_csi_data_format up

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 4 lines · Sakari Ailus

ccs_validate_csi_data_format() will soon be needed elsewhere, above its current location. Move it up.

Quoted 50 lines
---
 drivers/media/i2c/ccs/ccs-core.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +14 -14
index 0cd8ee957655..0efbc63534bc 100644
@@ -1979,6 +1979,20 @@ static int ccs_post_streamoff(struct v4l2_subdev *subdev)
 	return pm_runtime_put(&client->dev);
 }
 
+static const struct ccs_csi_data_format
+*ccs_validate_csi_data_format(struct ccs_sensor *sensor, u32 code)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
+		if (sensor->mbus_frame_fmts & (1 << i) &&
+		    ccs_csi_data_formats[i].code == code)
+			return &ccs_csi_data_formats[i];
+	}
+
+	return sensor->csi_format;
+}
+
 static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 			      struct v4l2_subdev_state *sd_state,
 			      struct v4l2_subdev_mbus_code_enum *code)
@@ -2114,20 +2128,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 	}
 }
 
-static const struct ccs_csi_data_format
-*ccs_validate_csi_data_format(struct ccs_sensor *sensor, u32 code)
-{
-	unsigned int i;
-
-	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
-		if (sensor->mbus_frame_fmts & (1 << i) &&
-		    ccs_csi_data_formats[i].code == code)
-			return &ccs_csi_data_formats[i];
-	}
-
-	return sensor->csi_format;
-}
-
 static int ccs_set_format_source(struct v4l2_subdev *subdev,
 				 struct v4l2_subdev_state *sd_state,
 				 struct v4l2_subdev_format *fmt)
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#91 lore raw

PATCH v8 22/38 media: ccs: Support frame descriptors

In reply to Sakari Ailus

Provide information on the frame layout using frame descriptors.

---
 drivers/media/i2c/ccs/ccs-core.c  | 60 +++++++++++++++++++++++++++++++
 drivers/media/i2c/ccs/ccs-quirk.h |  7 ++++
 drivers/media/i2c/ccs/ccs.h       |  4 +++
 3 files changed, 71 insertions(+)
drivers/media/i2c/ccs/ccs-core.c +60 -0
index 0efbc63534bc..9cc2080b73ec 100644
@@ -25,6 +25,7 @@
 #include <linux/slab.h>
 #include <linux/smiapp.h>
 #include <linux/v4l2-mediabus.h>
+#include <media/mipi-csi2.h>
 #include <media/v4l2-cci.h>
 #include <media/v4l2-device.h>
 #include <media/v4l2-fwnode.h>
@@ -245,6 +246,33 @@ static int ccs_read_all_limits(struct ccs_sensor *sensor)
 	return ret;
 }
 
+static u8 ccs_mipi_csi2_data_type(unsigned int bpp)
+{
+	switch (bpp) {
+	case 6:
+		return MIPI_CSI2_DT_RAW6;
+	case 7:
+		return MIPI_CSI2_DT_RAW7;
+	case 8:
+		return MIPI_CSI2_DT_RAW8;
+	case 10:
+		return MIPI_CSI2_DT_RAW10;
+	case 12:
+		return MIPI_CSI2_DT_RAW12;
+	case 14:
+		return MIPI_CSI2_DT_RAW14;
+	case 16:
+		return MIPI_CSI2_DT_RAW16;
+	case 20:
+		return MIPI_CSI2_DT_RAW20;
+	case 24:
+		return MIPI_CSI2_DT_RAW24;
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+}
+
 static int ccs_read_frame_fmt(struct ccs_sensor *sensor)
 {
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
@@ -2632,6 +2660,37 @@ static int ccs_set_selection(struct v4l2_subdev *subdev,
 	return ret;
 }
 
+static int ccs_get_frame_desc(struct v4l2_subdev *subdev, unsigned int pad,
+				 struct v4l2_mbus_frame_desc *desc)
+{
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
+	struct v4l2_mbus_frame_desc_entry *entry = desc->entry;
+
+	if (ccs_has_quirk(sensor, frame_desc))
+		return ccs_call_quirk(sensor, frame_desc, desc);
+
+	switch (sensor->hwcfg.csi_signalling_mode) {
+	case CCS_CSI_SIGNALING_MODE_CSI_2_DPHY:
+	case CCS_CSI_SIGNALING_MODE_CSI_2_CPHY:
+		desc->type = V4L2_MBUS_FRAME_DESC_TYPE_CSI2;
+		break;
+	default:
+		/* FIXME: CCP2 support */
+		return -EINVAL;
+	}
+
+	entry->pixelcode = sensor->csi_format->code;
+	entry->stream = CCS_STREAM_PIXEL;
+	entry->bus.csi2.dt =
+		sensor->csi_format->width == sensor->csi_format->compressed ?
+		ccs_mipi_csi2_data_type(sensor->csi_format->compressed) :
+		CCS_DEFAULT_COMPRESSED_DT;
+	entry++;
+	desc->num_entries++;
+
+	return 0;
+}
+
 static int ccs_get_skip_frames(struct v4l2_subdev *subdev, u32 *frames)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
@@ -3054,6 +3113,7 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
 	.set_selection = ccs_set_selection,
 	.enable_streams = ccs_enable_streams,
 	.disable_streams = ccs_disable_streams,
+	.get_frame_desc = ccs_get_frame_desc,
 };
 
 static const struct v4l2_subdev_sensor_ops ccs_sensor_ops = {
drivers/media/i2c/ccs/ccs-quirk.h +7 -0
index 392c97109617..3e1d9eaa33fa 100644
@@ -36,6 +36,7 @@ struct ccs_sensor;
  *			 access may be done by the caller (default read
  *			 value is zero), else negative error code on error
  * @flags: Quirk flags
+ * @frame_desc: Obtain the frame descriptor
  */
 struct ccs_quirk {
 	int (*limits)(struct ccs_sensor *sensor);
@@ -46,6 +47,8 @@ struct ccs_quirk {
 	int (*init)(struct ccs_sensor *sensor);
 	int (*reg_access)(struct ccs_sensor *sensor, bool write, u32 *reg,
 			  u32 *val);
+	int (*frame_desc)(struct ccs_sensor *sensor,
+			  struct v4l2_mbus_frame_desc *desc);
 	unsigned long flags;
 };
 
@@ -62,6 +65,10 @@ struct ccs_reg_8 {
 		.val = _val,		\
 	}
 
+#define ccs_has_quirk(sensor, _quirk)					\
+	((sensor)->minfo.quirk &&					\
+	 (sensor)->minfo.quirk->_quirk)
+
 #define ccs_call_quirk(sensor, _quirk, ...)				\
 	((sensor)->minfo.quirk &&					\
 	 (sensor)->minfo.quirk->_quirk ?				\
drivers/media/i2c/ccs/ccs.h +4 -0
index 4725e6eca8d0..adb152366ea2 100644
@@ -46,6 +46,8 @@
 
 #define CCS_COLOUR_COMPONENTS		4
 
+#define CCS_DEFAULT_COMPRESSED_DT	0x30
+
 #define SMIAPP_NAME			"smiapp"
 #define CCS_NAME			"ccs"
 
@@ -175,6 +177,8 @@ struct ccs_csi_data_format {
 #define CCS_PAD_SRC			1
 #define CCS_PADS			2
 
+#define CCS_STREAM_PIXEL		0
+
 struct ccs_binning_subtype {
 	u8 horizontal:4;
 	u8 vertical:4;
Signature · 1 line
2.39.2
Reviewed-by on v8 22/38
#92 lore raw

Re: [PATCH v8 22/38] media: ccs: Support frame descriptors

Quoted 3 lines · Sakari Ailus

Provide information on the frame layout using frame descriptors.

Quoted 152 lines
---
  drivers/media/i2c/ccs/ccs-core.c  | 60 +++++++++++++++++++++++++++++++
  drivers/media/i2c/ccs/ccs-quirk.h |  7 ++++
  drivers/media/i2c/ccs/ccs.h       |  4 +++
  3 files changed, 71 insertions(+)
drivers/media/i2c/ccs/ccs-core.c +60 -0
index 0efbc63534bc..9cc2080b73ec 100644
@@ -25,6 +25,7 @@
  #include <linux/slab.h>
  #include <linux/smiapp.h>
  #include <linux/v4l2-mediabus.h>
+#include <media/mipi-csi2.h>
  #include <media/v4l2-cci.h>
  #include <media/v4l2-device.h>
  #include <media/v4l2-fwnode.h>
@@ -245,6 +246,33 @@ static int ccs_read_all_limits(struct ccs_sensor *sensor)
  	return ret;
  }
 
+static u8 ccs_mipi_csi2_data_type(unsigned int bpp)
+{
+	switch (bpp) {
+	case 6:
+		return MIPI_CSI2_DT_RAW6;
+	case 7:
+		return MIPI_CSI2_DT_RAW7;
+	case 8:
+		return MIPI_CSI2_DT_RAW8;
+	case 10:
+		return MIPI_CSI2_DT_RAW10;
+	case 12:
+		return MIPI_CSI2_DT_RAW12;
+	case 14:
+		return MIPI_CSI2_DT_RAW14;
+	case 16:
+		return MIPI_CSI2_DT_RAW16;
+	case 20:
+		return MIPI_CSI2_DT_RAW20;
+	case 24:
+		return MIPI_CSI2_DT_RAW24;
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+}
+
  static int ccs_read_frame_fmt(struct ccs_sensor *sensor)
  {
  	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
@@ -2632,6 +2660,37 @@ static int ccs_set_selection(struct v4l2_subdev *subdev,
  	return ret;
  }
 
+static int ccs_get_frame_desc(struct v4l2_subdev *subdev, unsigned int pad,
+				 struct v4l2_mbus_frame_desc *desc)
+{
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
+	struct v4l2_mbus_frame_desc_entry *entry = desc->entry;
+
+	if (ccs_has_quirk(sensor, frame_desc))
+		return ccs_call_quirk(sensor, frame_desc, desc);
+
+	switch (sensor->hwcfg.csi_signalling_mode) {
+	case CCS_CSI_SIGNALING_MODE_CSI_2_DPHY:
+	case CCS_CSI_SIGNALING_MODE_CSI_2_CPHY:
+		desc->type = V4L2_MBUS_FRAME_DESC_TYPE_CSI2;
+		break;
+	default:
+		/* FIXME: CCP2 support */
+		return -EINVAL;
+	}
+
+	entry->pixelcode = sensor->csi_format->code;
+	entry->stream = CCS_STREAM_PIXEL;
+	entry->bus.csi2.dt =
+		sensor->csi_format->width == sensor->csi_format->compressed ?
+		ccs_mipi_csi2_data_type(sensor->csi_format->compressed) :
+		CCS_DEFAULT_COMPRESSED_DT;
+	entry++;
+	desc->num_entries++;
+
+	return 0;
+}
+
  static int ccs_get_skip_frames(struct v4l2_subdev *subdev, u32 *frames)
  {
  	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
@@ -3054,6 +3113,7 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
  	.set_selection = ccs_set_selection,
  	.enable_streams = ccs_enable_streams,
  	.disable_streams = ccs_disable_streams,
+	.get_frame_desc = ccs_get_frame_desc,
  };
 
  static const struct v4l2_subdev_sensor_ops ccs_sensor_ops = {
drivers/media/i2c/ccs/ccs-quirk.h +7 -0
index 392c97109617..3e1d9eaa33fa 100644
@@ -36,6 +36,7 @@ struct ccs_sensor;
   *			 access may be done by the caller (default read
   *			 value is zero), else negative error code on error
   * @flags: Quirk flags
+ * @frame_desc: Obtain the frame descriptor
   */
  struct ccs_quirk {
  	int (*limits)(struct ccs_sensor *sensor);
@@ -46,6 +47,8 @@ struct ccs_quirk {
  	int (*init)(struct ccs_sensor *sensor);
  	int (*reg_access)(struct ccs_sensor *sensor, bool write, u32 *reg,
  			  u32 *val);
+	int (*frame_desc)(struct ccs_sensor *sensor,
+			  struct v4l2_mbus_frame_desc *desc);
  	unsigned long flags;
  };
 
@@ -62,6 +65,10 @@ struct ccs_reg_8 {
  		.val = _val,		\
  	}
 
+#define ccs_has_quirk(sensor, _quirk)					\
+	((sensor)->minfo.quirk &&					\
+	 (sensor)->minfo.quirk->_quirk)
+
  #define ccs_call_quirk(sensor, _quirk, ...)				\
  	((sensor)->minfo.quirk &&					\
  	 (sensor)->minfo.quirk->_quirk ?				\
drivers/media/i2c/ccs/ccs.h +4 -0
index 4725e6eca8d0..adb152366ea2 100644
@@ -46,6 +46,8 @@
 
  #define CCS_COLOUR_COMPONENTS		4
 
+#define CCS_DEFAULT_COMPRESSED_DT	0x30
+
  #define SMIAPP_NAME			"smiapp"
  #define CCS_NAME			"ccs"
 
@@ -175,6 +177,8 @@ struct ccs_csi_data_format {
  #define CCS_PAD_SRC			1
  #define CCS_PADS			2
 
+#define CCS_STREAM_PIXEL		0
+
  struct ccs_binning_subtype {
  	u8 horizontal:4;
  	u8 vertical:4;
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer
#93 lore raw

Re: [PATCH v8 22/38] media: ccs: Support frame descriptors

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 57 lines · Sakari Ailus

Provide information on the frame layout using frame descriptors.

---
 drivers/media/i2c/ccs/ccs-core.c  | 60 +++++++++++++++++++++++++++++++
 drivers/media/i2c/ccs/ccs-quirk.h |  7 ++++
 drivers/media/i2c/ccs/ccs.h       |  4 +++
 3 files changed, 71 insertions(+)
drivers/media/i2c/ccs/ccs-core.c +28 -0
index 0efbc63534bc..9cc2080b73ec 100644
@@ -25,6 +25,7 @@
 #include <linux/slab.h>
 #include <linux/smiapp.h>
 #include <linux/v4l2-mediabus.h>
+#include <media/mipi-csi2.h>
 #include <media/v4l2-cci.h>
 #include <media/v4l2-device.h>
 #include <media/v4l2-fwnode.h>
@@ -245,6 +246,33 @@ static int ccs_read_all_limits(struct ccs_sensor *sensor)
 	return ret;
 }
 
+static u8 ccs_mipi_csi2_data_type(unsigned int bpp)
+{
+	switch (bpp) {
+	case 6:
+		return MIPI_CSI2_DT_RAW6;
+	case 7:
+		return MIPI_CSI2_DT_RAW7;
+	case 8:
+		return MIPI_CSI2_DT_RAW8;
+	case 10:
+		return MIPI_CSI2_DT_RAW10;
+	case 12:
+		return MIPI_CSI2_DT_RAW12;
+	case 14:
+		return MIPI_CSI2_DT_RAW14;
+	case 16:
+		return MIPI_CSI2_DT_RAW16;
+	case 20:
+		return MIPI_CSI2_DT_RAW20;
+	case 24:
+		return MIPI_CSI2_DT_RAW24;
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+}
+
 static int ccs_read_frame_fmt(struct ccs_sensor *sensor)
 {
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
@@ -2632,6 +2660,37 @@ static int ccs_set_selection(struct v4l2_subdev *subdev,
 	return ret;

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_selection(struct v4l2_subdev *subdev, · line 2670

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +8 -0
 }
 
+static int ccs_get_frame_desc(struct v4l2_subdev *subdev, unsigned int pad,
+				 struct v4l2_mbus_frame_desc *desc)
+{
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
+	struct v4l2_mbus_frame_desc_entry *entry = desc->entry;
+
+	if (ccs_has_quirk(sensor, frame_desc))
+		return ccs_call_quirk(sensor, frame_desc, desc);

I would introduce the quirk later, along with the patch that will use it.

Quoted 6 lines
drivers/media/i2c/ccs/ccs-core.c +6 -0
+
+	switch (sensor->hwcfg.csi_signalling_mode) {
+	case CCS_CSI_SIGNALING_MODE_CSI_2_DPHY:
+	case CCS_CSI_SIGNALING_MODE_CSI_2_CPHY:
+		desc->type = V4L2_MBUS_FRAME_DESC_TYPE_CSI2;
+		break;

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_selection(struct v4l2_subdev *subdev, · line 2686

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +10 -0
+	default:
+		/* FIXME: CCP2 support */
+		return -EINVAL;
+	}
+
+	entry->pixelcode = sensor->csi_format->code;
+	entry->stream = CCS_STREAM_PIXEL;
+	entry->bus.csi2.dt =
+		sensor->csi_format->width == sensor->csi_format->compressed ?
+		ccs_mipi_csi2_data_type(sensor->csi_format->compressed) :

Functionally equivalent,

		ccs_mipi_csi2_data_type(sensor->csi_format->width) :

would be clearer I think. The way it's written today made me wonder why you want the DT for the compressed format, which is not what you're doing.

Quoted 55 lines
drivers/media/i2c/ccs/ccs-core.c +8 -0
+		CCS_DEFAULT_COMPRESSED_DT;
+	entry++;
+	desc->num_entries++;
+
+	return 0;
+}
+
 static int ccs_get_skip_frames(struct v4l2_subdev *subdev, u32 *frames)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
@@ -3054,6 +3113,7 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
 	.set_selection = ccs_set_selection,
 	.enable_streams = ccs_enable_streams,
 	.disable_streams = ccs_disable_streams,
+	.get_frame_desc = ccs_get_frame_desc,
 };
 
 static const struct v4l2_subdev_sensor_ops ccs_sensor_ops = {
drivers/media/i2c/ccs/ccs-quirk.h +7 -0
index 392c97109617..3e1d9eaa33fa 100644
@@ -36,6 +36,7 @@ struct ccs_sensor;
  *			 access may be done by the caller (default read
  *			 value is zero), else negative error code on error
  * @flags: Quirk flags
+ * @frame_desc: Obtain the frame descriptor
  */
 struct ccs_quirk {
 	int (*limits)(struct ccs_sensor *sensor);
@@ -46,6 +47,8 @@ struct ccs_quirk {
 	int (*init)(struct ccs_sensor *sensor);
 	int (*reg_access)(struct ccs_sensor *sensor, bool write, u32 *reg,
 			  u32 *val);
+	int (*frame_desc)(struct ccs_sensor *sensor,
+			  struct v4l2_mbus_frame_desc *desc);
 	unsigned long flags;
 };
 
@@ -62,6 +65,10 @@ struct ccs_reg_8 {
 		.val = _val,		\
 	}
 
+#define ccs_has_quirk(sensor, _quirk)					\
+	((sensor)->minfo.quirk &&					\
+	 (sensor)->minfo.quirk->_quirk)
+
 #define ccs_call_quirk(sensor, _quirk, ...)				\
 	((sensor)->minfo.quirk &&					\
 	 (sensor)->minfo.quirk->_quirk ?				\
drivers/media/i2c/ccs/ccs.h +0 -0
index 4725e6eca8d0..adb152366ea2 100644

drivers/media/i2c/ccs/ccs.h line 49

Quoted 4 lines
drivers/media/i2c/ccs/ccs.h +1 -0
@@ -46,6 +46,8 @@
 
 #define CCS_COLOUR_COMPONENTS		4
 
+#define CCS_DEFAULT_COMPRESSED_DT	0x30

I'd write

#define CCS_DEFAULT_COMPRESSED_DT MIPI_CSI2_DT_USER_DEFINED(0)

Quoted 13 lines
drivers/media/i2c/ccs/ccs.h +3 -0
+
 #define SMIAPP_NAME			"smiapp"
 #define CCS_NAME			"ccs"
 
@@ -175,6 +177,8 @@ struct ccs_csi_data_format {
 #define CCS_PAD_SRC			1
 #define CCS_PADS			2
 
+#define CCS_STREAM_PIXEL		0
+
 struct ccs_binning_subtype {
 	u8 horizontal:4;
 	u8 vertical:4;
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#94 lore raw

Re: [PATCH v8 22/38] media: ccs: Support frame descriptors

Hi Laurent,

Quoted 3 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Thanks for the review!

Quoted 61 lines · Sakari Ailus
Quoted 59 lines

Provide information on the frame layout using frame descriptors.

---
 drivers/media/i2c/ccs/ccs-core.c  | 60 +++++++++++++++++++++++++++++++
 drivers/media/i2c/ccs/ccs-quirk.h |  7 ++++
 drivers/media/i2c/ccs/ccs.h       |  4 +++
 3 files changed, 71 insertions(+)
drivers/media/i2c/ccs/ccs-core.c +28 -0
index 0efbc63534bc..9cc2080b73ec 100644
@@ -25,6 +25,7 @@
 #include <linux/slab.h>
 #include <linux/smiapp.h>
 #include <linux/v4l2-mediabus.h>
+#include <media/mipi-csi2.h>
 #include <media/v4l2-cci.h>
 #include <media/v4l2-device.h>
 #include <media/v4l2-fwnode.h>
@@ -245,6 +246,33 @@ static int ccs_read_all_limits(struct ccs_sensor *sensor)
 	return ret;
 }
 
+static u8 ccs_mipi_csi2_data_type(unsigned int bpp)
+{
+	switch (bpp) {
+	case 6:
+		return MIPI_CSI2_DT_RAW6;
+	case 7:
+		return MIPI_CSI2_DT_RAW7;
+	case 8:
+		return MIPI_CSI2_DT_RAW8;
+	case 10:
+		return MIPI_CSI2_DT_RAW10;
+	case 12:
+		return MIPI_CSI2_DT_RAW12;
+	case 14:
+		return MIPI_CSI2_DT_RAW14;
+	case 16:
+		return MIPI_CSI2_DT_RAW16;
+	case 20:
+		return MIPI_CSI2_DT_RAW20;
+	case 24:
+		return MIPI_CSI2_DT_RAW24;
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+}
+
 static int ccs_read_frame_fmt(struct ccs_sensor *sensor)
 {
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
@@ -2632,6 +2660,37 @@ static int ccs_set_selection(struct v4l2_subdev *subdev,
 	return ret;
 }
 
Quoted 11 lines

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_selection(struct v4l2_subdev *subdev, · line 2670

Quoted 8 lines
drivers/media/i2c/ccs/ccs-core.c +8 -0
+static int ccs_get_frame_desc(struct v4l2_subdev *subdev, unsigned int pad,
+				 struct v4l2_mbus_frame_desc *desc)
+{
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
+	struct v4l2_mbus_frame_desc_entry *entry = desc->entry;
+
+	if (ccs_has_quirk(sensor, frame_desc))
+		return ccs_call_quirk(sensor, frame_desc, desc);

I would introduce the quirk later, along with the patch that will use it.

There is no user for this quirk in the series. It's been added for quirk infrastructure completeness, I guess I can move it in a separate patch and postpone that.

Quoted 20 lines
Quoted 7 lines
drivers/media/i2c/ccs/ccs-core.c +6 -0
+
+	switch (sensor->hwcfg.csi_signalling_mode) {
+	case CCS_CSI_SIGNALING_MODE_CSI_2_DPHY:
+	case CCS_CSI_SIGNALING_MODE_CSI_2_CPHY:
+		desc->type = V4L2_MBUS_FRAME_DESC_TYPE_CSI2;
+		break;

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_selection(struct v4l2_subdev *subdev, · line 2686

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +10 -0
+	default:
+		/* FIXME: CCP2 support */
+		return -EINVAL;
+	}
+
+	entry->pixelcode = sensor->csi_format->code;
+	entry->stream = CCS_STREAM_PIXEL;
+	entry->bus.csi2.dt =
+		sensor->csi_format->width == sensor->csi_format->compressed ?
+		ccs_mipi_csi2_data_type(sensor->csi_format->compressed) :

Functionally equivalent,

Quoted 5 lines
		ccs_mipi_csi2_data_type(sensor->csi_format->width) :

would be clearer I think. The way it's written today made me wonder why you want the DT for the compressed format, which is not what you're doing.

Sounds good.

Quoted 57 lines
Quoted 57 lines
drivers/media/i2c/ccs/ccs-core.c +8 -0
+		CCS_DEFAULT_COMPRESSED_DT;
+	entry++;
+	desc->num_entries++;
+
+	return 0;
+}
+
 static int ccs_get_skip_frames(struct v4l2_subdev *subdev, u32 *frames)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
@@ -3054,6 +3113,7 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
 	.set_selection = ccs_set_selection,
 	.enable_streams = ccs_enable_streams,
 	.disable_streams = ccs_disable_streams,
+	.get_frame_desc = ccs_get_frame_desc,
 };
 
 static const struct v4l2_subdev_sensor_ops ccs_sensor_ops = {
drivers/media/i2c/ccs/ccs-quirk.h +7 -0
index 392c97109617..3e1d9eaa33fa 100644
@@ -36,6 +36,7 @@ struct ccs_sensor;
  *			 access may be done by the caller (default read
  *			 value is zero), else negative error code on error
  * @flags: Quirk flags
+ * @frame_desc: Obtain the frame descriptor
  */
 struct ccs_quirk {
 	int (*limits)(struct ccs_sensor *sensor);
@@ -46,6 +47,8 @@ struct ccs_quirk {
 	int (*init)(struct ccs_sensor *sensor);
 	int (*reg_access)(struct ccs_sensor *sensor, bool write, u32 *reg,
 			  u32 *val);
+	int (*frame_desc)(struct ccs_sensor *sensor,
+			  struct v4l2_mbus_frame_desc *desc);
 	unsigned long flags;
 };
 
@@ -62,6 +65,10 @@ struct ccs_reg_8 {
 		.val = _val,		\
 	}
 
+#define ccs_has_quirk(sensor, _quirk)					\
+	((sensor)->minfo.quirk &&					\
+	 (sensor)->minfo.quirk->_quirk)
+
 #define ccs_call_quirk(sensor, _quirk, ...)				\
 	((sensor)->minfo.quirk &&					\
 	 (sensor)->minfo.quirk->_quirk ?				\
drivers/media/i2c/ccs/ccs.h +0 -0
index 4725e6eca8d0..adb152366ea2 100644
@@ -46,6 +46,8 @@
 
Quoted 7 lines

drivers/media/i2c/ccs/ccs.h line 49

Quoted 3 lines
drivers/media/i2c/ccs/ccs.h +1 -0
 #define CCS_COLOUR_COMPONENTS		4
 
+#define CCS_DEFAULT_COMPRESSED_DT	0x30

I'd write

#define CCS_DEFAULT_COMPRESSED_DT MIPI_CSI2_DT_USER_DEFINED(0)

Yes.

Quoted 15 lines
Quoted 14 lines
drivers/media/i2c/ccs/ccs.h +3 -0
+
 #define SMIAPP_NAME			"smiapp"
 #define CCS_NAME			"ccs"
 
@@ -175,6 +177,8 @@ struct ccs_csi_data_format {
 #define CCS_PAD_SRC			1
 #define CCS_PADS			2
 
+#define CCS_STREAM_PIXEL		0
+
 struct ccs_binning_subtype {
 	u8 horizontal:4;
 	u8 vertical:4;
Signature · 1 line
Sakari Ailus
author maintainer
#95 lore raw

PATCH v8 23/38 media: uapi: ccs: Add media bus code for MIPI CCS embedded data

In reply to Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

---
 .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |  3 ++
 2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +28 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
       - X
       - X
       - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
+
+The format definition shall be used only by devices that fulfill all three
+levels above.
+
+This mbus code are only used for "2-byte simplified tagged data format" (code
+0xa) but their use may be extended further in the future, to cover other CCS
+embedded data format codes.
include/uapi/linux/media-bus-format.h +3 -0
index d4c1d991014b..03f7e9ab517b 100644
@@ -183,4 +183,7 @@
 #define MEDIA_BUS_FMT_META_20			0x8006
 #define MEDIA_BUS_FMT_META_24			0x8007
 
+/* Specific metadata formats. Next is 0x9002. */
+#define MEDIA_BUS_FMT_CCS_EMBEDDED		0x9001
+
 #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 1 line
2.39.2
Reviewed-by on v8 23/38
#96 lore raw

Re: [PATCH v8 23/38] media: uapi: ccs: Add media bus code for MIPI CCS embedded data

Quoted 4 lines · Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

Quoted 53 lines
---
  .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
  include/uapi/linux/media-bus-format.h         |  3 ++
  2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +28 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
        - X
        - X
        - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
+
+The format definition shall be used only by devices that fulfill all three
+levels above.
+
+This mbus code are only used for "2-byte simplified tagged data format" (code
+0xa) but their use may be extended further in the future, to cover other CCS
+embedded data format codes.
include/uapi/linux/media-bus-format.h +3 -0
index d4c1d991014b..03f7e9ab517b 100644
@@ -183,4 +183,7 @@
  #define MEDIA_BUS_FMT_META_20			0x8006
  #define MEDIA_BUS_FMT_META_24			0x8007
 
+/* Specific metadata formats. Next is 0x9002. */
+#define MEDIA_BUS_FMT_CCS_EMBEDDED		0x9001
+
  #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 5 lines
Julien Massot
Senior Software Engineer
Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, UK
Registered in England & Wales, no. 5513718
maintainer
#97 lore raw

Re: [PATCH v8 23/38] media: uapi: ccs: Add media bus code for MIPI CCS embedded data

In reply to Sakari Ailus

Hi Sakari,

Thank you for the patch.

Quoted 22 lines · Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

---
 .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |  3 ++
 2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +5 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
       - X
       - X
       - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8581

Quoted 10 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +10 -0
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as

s/Data unit/Data Unit/

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8590

Quoted 9 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
+
+The format definition shall be used only by devices that fulfill all three
+levels above.

Hmmmm... Do we need to mandate level 3 ? There are lots of sensors that comply with the first two levels but have their own register set. Would you like a sensor-specific embedded data format for each of them ? If so, how would we document it without essentially copying the datasheet ?

Quoted 16 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +4 -0
+
+This mbus code are only used for "2-byte simplified tagged data format" (code
+0xa) but their use may be extended further in the future, to cover other CCS
+embedded data format codes.
include/uapi/linux/media-bus-format.h +3 -0
index d4c1d991014b..03f7e9ab517b 100644
@@ -183,4 +183,7 @@
 #define MEDIA_BUS_FMT_META_20			0x8006
 #define MEDIA_BUS_FMT_META_24			0x8007
 
+/* Specific metadata formats. Next is 0x9002. */
+#define MEDIA_BUS_FMT_CCS_EMBEDDED		0x9001
+
 #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#98 lore raw

Re: [PATCH v8 23/38] media: uapi: ccs: Add media bus code for MIPI CCS embedded data

Hi Laurent,

Quoted 31 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 26 lines · Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

---
 .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |  3 ++
 2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
       - X
       - X
       - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
Quoted 8 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8581

Quoted 6 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +6 -0
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as

s/Data unit/Data Unit/

Yes.

Quoted 15 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8590

Quoted 10 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
+
+The format definition shall be used only by devices that fulfill all three
+levels above.

Hmmmm... Do we need to mandate level 3 ? There are lots of sensors that comply with the first two levels but have their own register set. Would you like a sensor-specific embedded data format for each of them ? If so, how would we document it without essentially copying the datasheet ?

You could refer to the datasheet if it's publicly available.

The CCS static data format is meaningful only for accessing (level 1) and parsing (level 2) the data without knowing its meaning, which is the most important aspect. The user space will need to know the meaning of the data and can additionally select the same data unpacker than for CCS and possibly also the CCS data parser if applicable.

The user space should especially not have to have a list of CCS-compliant sensors which would become necessary, should we skip on the 3rd level requirement here.

Quoted 18 lines
Quoted 17 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +4 -0
+
+This mbus code are only used for "2-byte simplified tagged data format" (code
+0xa) but their use may be extended further in the future, to cover other CCS
+embedded data format codes.
include/uapi/linux/media-bus-format.h +3 -0
index d4c1d991014b..03f7e9ab517b 100644
@@ -183,4 +183,7 @@
 #define MEDIA_BUS_FMT_META_20			0x8006
 #define MEDIA_BUS_FMT_META_24			0x8007
 
+/* Specific metadata formats. Next is 0x9002. */
+#define MEDIA_BUS_FMT_CCS_EMBEDDED		0x9001
+
 #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 3 lines
Regards,

Sakari Ailus
maintainer
#99 lore raw

Re: [PATCH v8 23/38] media: uapi: ccs: Add media bus code for MIPI CCS embedded data

Hi Sakari,

Quoted 54 lines · Sakari Ailus
Quoted 27 lines · Laurent Pinchart
Quoted 26 lines · Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

---
 .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |  3 ++
 2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
       - X
       - X
       - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
Quoted 8 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8581

Quoted 6 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +6 -0
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as

s/Data unit/Data Unit/

Yes.

Quoted 6 lines
Quoted 6 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +6 -0
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
Quoted 8 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8590

Quoted 3 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +3 -0
+
+The format definition shall be used only by devices that fulfill all three
+levels above.

Hmmmm... Do we need to mandate level 3 ? There are lots of sensors that comply with the first two levels but have their own register set. Would you like a sensor-specific embedded data format for each of them ? If so, how would we document it without essentially copying the datasheet ?

Quoted 1 line

You could refer to the datasheet if it's publicly available.

What if it's not ? How would you expect documentation to be organized ? Will we then end up adding one media bus code per sensor ?

Quoted 26 lines

The CCS static data format is meaningful only for accessing (level 1) and parsing (level 2) the data without knowing its meaning, which is the most important aspect. The user space will need to know the meaning of the data and can additionally select the same data unpacker than for CCS and possibly also the CCS data parser if applicable.

The user space should especially not have to have a list of CCS-compliant sensors which would become necessary, should we skip on the 3rd level requirement here.

Quoted 16 lines
Quoted 16 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +4 -0
+
+This mbus code are only used for "2-byte simplified tagged data format" (code
+0xa) but their use may be extended further in the future, to cover other CCS
+embedded data format codes.
include/uapi/linux/media-bus-format.h +3 -0
index d4c1d991014b..03f7e9ab517b 100644
@@ -183,4 +183,7 @@
 #define MEDIA_BUS_FMT_META_20			0x8006
 #define MEDIA_BUS_FMT_META_24			0x8007
 
+/* Specific metadata formats. Next is 0x9002. */
+#define MEDIA_BUS_FMT_CCS_EMBEDDED		0x9001
+
 #endif /* __LINUX_MEDIA_BUS_FORMAT_H */
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#100 lore raw

Re: [PATCH v8 23/38] media: uapi: ccs: Add media bus code for MIPI CCS embedded data

Hi Laurent,

Quoted 57 lines · Laurent Pinchart

Hi Sakari,

Quoted 54 lines · Sakari Ailus
Quoted 27 lines · Laurent Pinchart
Quoted 26 lines · Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

---
 .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |  3 ++
 2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
       - X
       - X
       - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
Quoted 8 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8581

Quoted 6 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +6 -0
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as

s/Data unit/Data Unit/

Yes.

Quoted 14 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8590

Quoted 9 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
+
+The format definition shall be used only by devices that fulfill all three
+levels above.

Hmmmm... Do we need to mandate level 3 ? There are lots of sensors that comply with the first two levels but have their own register set. Would you like a sensor-specific embedded data format for each of them ? If so, how would we document it without essentially copying the datasheet ?

Quoted 4 lines
Quoted 1 line

You could refer to the datasheet if it's publicly available.

What if it's not ? How would you expect documentation to be organized ? Will we then end up adding one media bus code per sensor ?

Probably. The user space needs a way to find out what to expect. If a sensor produces its own custom metadata format, it needs to be documented.

There's also documentation for OV2740 embedded data in the set.

Signature · 1 line
Sakari Ailus
maintainer
#101 lore raw

Re: [PATCH v8 23/38] media: uapi: ccs: Add media bus code for MIPI CCS embedded data

Hi Sakari,

Quoted 58 lines · Sakari Ailus
Quoted 57 lines · Laurent Pinchart
Quoted 55 lines · Sakari Ailus
Quoted 27 lines · Laurent Pinchart
Quoted 26 lines · Sakari Ailus

Add new MIPI CCS embedded data media bus code
(MEDIA_BUS_FMT_CCS_EMBEDDED).

---
 .../media/v4l/subdev-formats.rst              | 28 +++++++++++++++++++
 include/uapi/linux/media-bus-format.h         |  3 ++
 2 files changed, 31 insertions(+)
Documentation/userspace-api/media/v4l/subdev-formats.rst +9 -0
index cbd475f7cae9..c8f982411e70 100644
@@ -8564,3 +8564,31 @@ and finally the bit number in subscript. "X" indicates a padding bit.
       - X
       - X
       - X
+
+.. _MEDIA-BUS-FMT-CCS-EMBEDDED:
+
+MIPI CCS Embedded Data Formats
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+`MIPI CCS <https://www.mipi.org/specifications/camera-command-set>`_ defines a
+metadata format for sensor embedded data, which is used to store the register
+configuration used for capturing a given frame. The format is defined in the CCS
Quoted 8 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8581

Quoted 6 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +6 -0
+specification. The media bus code for this format is
+``MEDIA_BUS_FMT_CCS_EMBEDDED``.
+
+The CCS embedded data format definition includes three levels:
+
+1. Padding within CSI-2 bus :ref:`Data unit <media-glossary-data-unit>` as

s/Data unit/Data Unit/

Yes.

Quoted 6 lines
Quoted 6 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +6 -0
+   documented in the MIPI CCS specification.
+
+2. The tagged data format as documented in the MIPI CCS specification.
+
+3. Register addresses and register documentation as documented in the MIPI CCS
+   specification.
Quoted 8 lines

Documentation/userspace-api/media/v4l/subdev-formats.rst and finally the bit number in subscript. "X" indicates a padding bit. · line 8590

Quoted 3 lines
Documentation/userspace-api/media/v4l/subdev-formats.rst +3 -0
+
+The format definition shall be used only by devices that fulfill all three
+levels above.

Hmmmm... Do we need to mandate level 3 ? There are lots of sensors that comply with the first two levels but have their own register set. Would you like a sensor-specific embedded data format for each of them ? If so, how would we document it without essentially copying the datasheet ?

You could refer to the datasheet if it's publicly available.

Quoted 7 lines
Quoted 2 lines

What if it's not ? How would you expect documentation to be organized ? Will we then end up adding one media bus code per sensor ?

Probably. The user space needs a way to find out what to expect. If a sensor produces its own custom metadata format, it needs to be documented.

There's also documentation for OV2740 embedded data in the set.

The OV2740 is a different case than what I had in mind. It uses level 1 of the CCS embedded data only, without using the CCS tagged format. The documentation you've included in this patch series looks good to me.

For sensors that are compatible with CCS embedded data level 2, I'm not sure how to proceed. Those sensors send (register, value) pairs in their embedded data, but the register set is not CCS-compatible. Even when documentation is available, it's hard to verify that the sensor will never output the value of undocumented registers, as the registers included in the embedded data may vary depending on the sensor configuration.

How would you proceed in such cases ? I can try to write a sample documentation patch, but I don't know what you expect. I'm thinking about the IMX219 sensor for instance, but any sensor compatible with CCS embedded data level 2 without a public datasheet is a good example.

Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#102 lore raw

PATCH v8 24/38 media: ccs: Add support for embedded data stream

In reply to Sakari Ailus

Add support for embedded data stream, in UAPI and frame descriptor.

This patch adds also a new embedded data pad (2) to the source sub-device.

---
 drivers/media/i2c/ccs/ccs-core.c | 370 +++++++++++++++++++++++++++++--
 drivers/media/i2c/ccs/ccs.h      |  18 +-
 2 files changed, 360 insertions(+), 28 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +346 -24
index 9cc2080b73ec..1f91af25eac6 100644
@@ -1902,6 +1902,13 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 	if (rval < 0)
 		goto err_pm_put;
 
+	/* Configure embedded data */
+	if (sensor->csi_format->compressed >= 16) {
+		rval = ccs_write(sensor, EMB_DATA_CTRL, sensor->emb_data_ctrl);
+		if (rval < 0)
+			goto err_pm_put;
+	}
+
 	if (CCS_LIM(sensor, FLASH_MODE_CAPABILITY) &
 	    (CCS_FLASH_MODE_CAPABILITY_SINGLE_STROBE |
 	     SMIAPP_FLASH_MODE_CAPABILITY_MULTIPLE_STROBE) &&
@@ -2021,6 +2028,57 @@ static const struct ccs_csi_data_format
 	return sensor->csi_format;
 }
 
+#define CCS_EMBEDDED_CODE_DEPTH(depth, half_depth)			\
+	depth,								\
+	CCS_EMB_DATA_CAPABILITY_TWO_BYTES_PER_RAW##depth,		\
+	CCS_EMB_DATA_CAPABILITY_NO_ONE_BYTE_PER_RAW##depth,		\
+	CCS_EMB_DATA_CTRL_RAW##half_depth##_PACKING_FOR_RAW##depth,	\
+	MEDIA_BUS_FMT_META_##half_depth,				\
+	MEDIA_BUS_FMT_META_##depth,					\
+
+static const struct ccs_embedded_code {
+	u8 depth;
+	u8 cap_two_bytes_per_sample;
+	u8 cap_no_legacy;
+	u8 ctrl;
+	u32 code_two_bytes;
+	u32 code_legacy;
+} ccs_embedded_codes[] = {
+	{ CCS_EMBEDDED_CODE_DEPTH(16, 8) },
+	{ CCS_EMBEDDED_CODE_DEPTH(20, 10) },
+	{ CCS_EMBEDDED_CODE_DEPTH(24, 12) },
+};
+
+static const struct ccs_embedded_code *ccs_embedded_code(unsigned int bpp)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(ccs_embedded_codes); i++)
+		if (ccs_embedded_codes[i].depth == bpp)
+			return ccs_embedded_codes + i;
+
+	WARN_ON(1);
+
+	return ccs_embedded_codes;
+}
+
+static u32
+ccs_default_embedded_code(struct ccs_sensor *sensor,
+			  const struct ccs_embedded_code *embedded_code)
+{
+	if (CCS_LIM(sensor, EMB_DATA_CAPABILITY) &
+	    BIT(embedded_code->cap_two_bytes_per_sample))
+		return embedded_code->code_two_bytes;
+
+	if (!(CCS_LIM(sensor, EMB_DATA_CAPABILITY) &
+	      BIT(embedded_code->cap_no_legacy)))
+		return embedded_code->code_legacy;
+
+	WARN_ON(1);
+
+	return embedded_code->code_legacy;
+}
+
 static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 			      struct v4l2_subdev_state *sd_state,
 			      struct v4l2_subdev_mbus_code_enum *code)
@@ -2036,6 +2094,69 @@ static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 	dev_err(&client->dev, "subdev %s, pad %u, index %u\n",
 		subdev->name, code->pad, code->index);
 
+	if (subdev == &sensor->src->sd) {
+		if (code->pad == CCS_PAD_META) {
+			if (code->index)
+				goto out;
+
+			code->code = MEDIA_BUS_FMT_CCS_EMBEDDED;
+
+			rval = 0;
+			goto out;
+		}
+		if (code->stream == CCS_STREAM_META) {
+			struct v4l2_mbus_framefmt *pix_fmt =
+				v4l2_subdev_state_get_format(sd_state,
+							     CCS_PAD_SRC,
+							     CCS_STREAM_PIXEL);
+			const struct ccs_csi_data_format *csi_format =
+				ccs_validate_csi_data_format(sensor,
+							     pix_fmt->code);
+			unsigned int i = 0;
+			u32 codes[2];
+
+			switch (csi_format->compressed) {
+			case 8:
+				codes[i++] = MEDIA_BUS_FMT_META_8;
+				break;
+			case 10:
+				codes[i++] = MEDIA_BUS_FMT_META_10;
+				break;
+			case 12:
+				codes[i++] = MEDIA_BUS_FMT_META_12;
+				break;
+			case 14:
+				codes[i++] = MEDIA_BUS_FMT_META_14;
+				break;
+			case 16:
+			case 20:
+			case 24: {
+				const struct ccs_embedded_code *embedded_code =
+					ccs_embedded_code(csi_format->compressed);
+
+				if (CCS_LIM(sensor, EMB_DATA_CAPABILITY) &
+				    BIT(embedded_code->cap_two_bytes_per_sample))
+					codes[i++] =
+						embedded_code->code_two_bytes;
+
+				if (!(CCS_LIM(sensor, EMB_DATA_CAPABILITY) &
+				      BIT(embedded_code->cap_no_legacy)))
+					codes[i++] = embedded_code->code_legacy;
+				break;
+			}
+			default:
+				WARN_ON(1);
+			}
+
+			if (WARN_ON(i > ARRAY_SIZE(codes)) || code->index >= i)
+				goto out;
+
+			code->code = codes[code->index];
+			rval = 0;
+			goto out;
+		}
+	}
+
 	if (subdev != &sensor->src->sd || code->pad != CCS_PAD_SRC) {
 		if (code->index)
 			goto out;
@@ -2078,8 +2199,11 @@ static int __ccs_get_format(struct v4l2_subdev *subdev,
 			    struct v4l2_subdev_state *sd_state,
 			    struct v4l2_subdev_format *fmt)
 {
-	fmt->format = *v4l2_subdev_state_get_format(sd_state, fmt->pad);
-	fmt->format.code = __ccs_get_mbus_code(subdev, fmt->pad);
+	fmt->format = *v4l2_subdev_state_get_format(sd_state, fmt->pad,
+						    fmt->stream);
+
+	if (fmt->pad != CCS_PAD_META && fmt->stream != CCS_STREAM_META)
+		fmt->format.code = __ccs_get_mbus_code(subdev, fmt->pad);
 
 	return 0;
 }
@@ -2109,10 +2233,11 @@ static void ccs_get_crop_compose(struct v4l2_subdev *subdev,
 	if (crops)
 		for (i = 0; i < subdev->entity.num_pads; i++)
 			crops[i] =
-				v4l2_subdev_state_get_crop(sd_state, i);
+				v4l2_subdev_state_get_crop(sd_state, i,
+							   CCS_STREAM_PIXEL);
 	if (comps)
-		*comps = v4l2_subdev_state_get_compose(sd_state,
-						       ssd->sink_pad);
+		*comps = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+						       CCS_STREAM_PIXEL);
 }
 
 /* Changes require propagation only on sink pad. */
@@ -2145,7 +2270,8 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 		fallthrough;
 	case V4L2_SEL_TGT_COMPOSE:
 		*crops[CCS_PAD_SRC] = *comp;
-		fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC);
+		fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
+						   CCS_STREAM_PIXEL);
 		fmt->width = comp->width;
 		fmt->height = comp->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE && ssd == sensor->src)
@@ -2209,6 +2335,81 @@ static int ccs_set_format_source(struct v4l2_subdev *subdev,
 	return ccs_pll_update(sensor);
 }
 
+static int ccs_set_format_meta(struct v4l2_subdev *subdev,
+			       struct v4l2_subdev_state *sd_state,
+			       struct v4l2_mbus_framefmt *fmt)
+{
+	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
+	const struct ccs_csi_data_format *csi_format;
+	struct v4l2_mbus_framefmt *pix_fmt;
+	struct v4l2_mbus_framefmt *meta_fmt;
+	struct v4l2_mbus_framefmt *meta_out_fmt;
+	u32 code;
+
+	pix_fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
+					       CCS_STREAM_PIXEL);
+	meta_fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_META, 0);
+	meta_out_fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
+						    CCS_STREAM_META);
+
+	code = fmt ? fmt->code : meta_out_fmt->code;
+
+	meta_out_fmt->width = meta_fmt->width = pix_fmt->width;
+	meta_out_fmt->height = meta_fmt->height =
+		sensor->embedded_end - sensor->embedded_start;
+	meta_fmt->code = MEDIA_BUS_FMT_CCS_EMBEDDED;
+
+	csi_format = ccs_validate_csi_data_format(sensor, pix_fmt->code);
+
+	switch (csi_format->compressed) {
+	case 8:
+		meta_out_fmt->code = MEDIA_BUS_FMT_META_8;
+		break;
+	case 10:
+		meta_out_fmt->code = MEDIA_BUS_FMT_META_10;
+		break;
+	case 12:
+		meta_out_fmt->code = MEDIA_BUS_FMT_META_12;
+		break;
+	case 14:
+		meta_out_fmt->code = MEDIA_BUS_FMT_META_14;
+		break;
+	case 16:
+	case 20:
+	case 24: {
+		const struct ccs_embedded_code *embedded_code;
+
+		embedded_code = ccs_embedded_code(csi_format->compressed);
+		meta_out_fmt->code =
+			ccs_default_embedded_code(sensor, embedded_code);
+
+		if (!(CCS_LIM(sensor, EMB_DATA_CAPABILITY) &
+		      BIT(embedded_code->cap_no_legacy)) &&
+		    code == embedded_code->code_legacy) {
+			meta_out_fmt->code = embedded_code->code_legacy;
+			sensor->emb_data_ctrl = 0;
+		}
+
+		if (CCS_LIM(sensor, EMB_DATA_CAPABILITY) &
+		    BIT(embedded_code->cap_two_bytes_per_sample) &&
+		    code == embedded_code->code_two_bytes) {
+			meta_out_fmt->code = embedded_code->code_two_bytes;
+			sensor->emb_data_ctrl = embedded_code->ctrl;
+		}
+
+		break;
+	}
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+
+	if (fmt)
+		*fmt = *meta_out_fmt;
+
+	return 0;
+}
+
 static int ccs_set_format(struct v4l2_subdev *subdev,
 			  struct v4l2_subdev_state *sd_state,
 			  struct v4l2_subdev_format *fmt)
@@ -2217,12 +2418,25 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
 	struct v4l2_rect *crops[CCS_PADS];
 
+	if (subdev == &sensor->src->sd && fmt->pad == CCS_PAD_META)
+		return ccs_get_format(subdev, sd_state, fmt);
+
 	mutex_lock(&sensor->mutex);
 
+	if (subdev == &sensor->src->sd && fmt->stream == CCS_STREAM_META) {
+		ccs_set_format_meta(subdev, sd_state, &fmt->format);
+
+		mutex_unlock(&sensor->mutex);
+
+		return 0;
+	}
+
 	if (fmt->pad == ssd->source_pad) {
 		int rval;
 
 		rval = ccs_set_format_source(subdev, sd_state, fmt);
+		if (sensor->embedded_start != sensor->embedded_end)
+			ccs_set_format_meta(subdev, sd_state, NULL);
 
 		mutex_unlock(&sensor->mutex);
 
@@ -2497,6 +2711,12 @@ static int ccs_sel_supported(struct v4l2_subdev *subdev,
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
 
+	if (sel->stream != CCS_STREAM_PIXEL)
+		return -EINVAL;
+
+	if (sel->pad == CCS_PAD_META)
+		return -EINVAL;
+
 	/* We only implement crop in three places. */
 	switch (sel->target) {
 	case V4L2_SEL_TGT_CROP:
@@ -2541,7 +2761,8 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 
 	if (sel->pad == ssd->sink_pad) {
 		struct v4l2_mbus_framefmt *mfmt =
-			v4l2_subdev_state_get_format(sd_state, sel->pad);
+			v4l2_subdev_state_get_format(sd_state, sel->pad,
+						     CCS_STREAM_PIXEL);
 
 		src_size.width = mfmt->width;
 		src_size.height = mfmt->height;
@@ -2602,7 +2823,9 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 		} else if (sel->pad == ssd->sink_pad) {
 			struct v4l2_mbus_framefmt *sink_fmt =
 				v4l2_subdev_state_get_format(sd_state,
-							     ssd->sink_pad);
+							     ssd->sink_pad,
+							     CCS_STREAM_PIXEL);
+
 			sel->r.top = sel->r.left = 0;
 			sel->r.width = sink_fmt->width;
 			sel->r.height = sink_fmt->height;
@@ -2688,6 +2911,14 @@ static int ccs_get_frame_desc(struct v4l2_subdev *subdev, unsigned int pad,
 	entry++;
 	desc->num_entries++;
 
+	if (sensor->embedded_start != sensor->embedded_end) {
+		entry->pixelcode = MEDIA_BUS_FMT_CCS_EMBEDDED;
+		entry->stream = CCS_STREAM_META;
+		entry->bus.csi2.dt = MIPI_CSI2_DT_EMBEDDED_8B;
+		entry++;
+		desc->num_entries++;
+	}
+
 	return 0;
 }
 
@@ -3006,6 +3237,8 @@ static void ccs_cleanup(struct ccs_sensor *sensor)
 	ccs_free_controls(sensor);
 }
 
+static const struct v4l2_subdev_internal_ops ccs_internal_ops;
+
 static int ccs_init_subdev(struct ccs_sensor *sensor,
 			   struct ccs_subdev *ssd, const char *name,
 			   unsigned short num_pads, u32 function,
@@ -3018,15 +3251,18 @@ static int ccs_init_subdev(struct ccs_sensor *sensor,
 	if (!ssd)
 		return 0;
 
-	if (ssd != sensor->src)
+	if (ssd != sensor->src) {
 		v4l2_subdev_init(&ssd->sd, &ccs_ops);
+		ssd->sd.internal_ops = &ccs_internal_ops;
+	}
 
 	ssd->sd.flags |= V4L2_SUBDEV_FL_HAS_DEVNODE;
 	ssd->sd.entity.function = function;
 	ssd->sensor = sensor;
 
 	ssd->npads = num_pads;
-	ssd->source_pad = num_pads - 1;
+	ssd->source_pad =
+		ssd == sensor->pixel_array ? CCS_PA_PAD_SRC : CCS_PAD_SRC;
 
 	v4l2_i2c_subdev_set_name(&ssd->sd, client, sensor->minfo.name, name);
 
@@ -3040,6 +3276,10 @@ static int ccs_init_subdev(struct ccs_sensor *sensor,
 		ssd->sd.owner = THIS_MODULE;
 		ssd->sd.dev = &client->dev;
 		v4l2_set_subdevdata(&ssd->sd, client);
+	} else {
+		ssd->sd.flags |= V4L2_SUBDEV_FL_STREAMS;
+		ssd->pads[CCS_PAD_META].flags =
+			MEDIA_PAD_FL_SINK | MEDIA_PAD_FL_INTERNAL;
 	}
 
 	rval = media_entity_pads_init(&ssd->sd.entity, ssd->npads, ssd->pads);
@@ -3057,21 +3297,19 @@ static int ccs_init_subdev(struct ccs_sensor *sensor,
 	return 0;
 }
 
-static int ccs_init_state(struct v4l2_subdev *sd,
-			  struct v4l2_subdev_state *sd_state)
+static int __ccs_init_state(struct v4l2_subdev *sd,
+			    struct v4l2_subdev_state *sd_state)
 {
 	struct ccs_subdev *ssd = to_ccs_subdev(sd);
 	struct ccs_sensor *sensor = ssd->sensor;
 	unsigned int pad = ssd == sensor->pixel_array ?
 		CCS_PA_PAD_SRC : CCS_PAD_SINK;
 	struct v4l2_mbus_framefmt *fmt =
-		v4l2_subdev_state_get_format(sd_state, pad);
+		v4l2_subdev_state_get_format(sd_state, pad, CCS_STREAM_PIXEL);
 	struct v4l2_rect *crop =
-		v4l2_subdev_state_get_crop(sd_state, pad);
+		v4l2_subdev_state_get_crop(sd_state, pad, CCS_STREAM_PIXEL);
 	bool is_active = !sd->active_state || sd->active_state == sd_state;
 
-	mutex_lock(&sensor->mutex);
-
 	ccs_get_native_size(ssd, crop);
 
 	fmt->width = crop->width;
@@ -3083,20 +3321,78 @@ static int ccs_init_state(struct v4l2_subdev *sd,
 		if (is_active)
 			sensor->pa_src = *crop;
 
-		mutex_unlock(&sensor->mutex);
 		return 0;
 	}
 
-	fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC);
+	fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
+					   CCS_STREAM_PIXEL);
 	fmt->code = ssd == sensor->src ?
 		sensor->csi_format->code : sensor->internal_csi_format->code;
 	fmt->field = V4L2_FIELD_NONE;
 
 	ccs_propagate(sd, sd_state, is_active, V4L2_SEL_TGT_CROP);
 
+	return 0;
+}
+
+static int ccs_init_state(struct v4l2_subdev *sd,
+			  struct v4l2_subdev_state *sd_state)
+{
+	struct ccs_subdev *ssd = to_ccs_subdev(sd);
+	struct ccs_sensor *sensor = ssd->sensor;
+	int rval;
+
+	mutex_lock(&sensor->mutex);
+	rval = __ccs_init_state(sd, sd_state);
 	mutex_unlock(&sensor->mutex);
 
-	return 0;
+	return rval;
+}
+
+static int ccs_src_init_state(struct v4l2_subdev *sd,
+			      struct v4l2_subdev_state *sd_state)
+{
+	struct v4l2_subdev_route routes[] = {
+		{
+			.sink_pad = CCS_PAD_SINK,
+			.source_pad = CCS_PAD_SRC,
+			.source_stream = CCS_STREAM_PIXEL,
+			.flags = V4L2_SUBDEV_ROUTE_FL_ACTIVE,
+		}, {
+			.sink_pad = CCS_PAD_META,
+			.source_pad = CCS_PAD_SRC,
+			.source_stream = CCS_STREAM_META,
+			.flags = V4L2_SUBDEV_ROUTE_FL_ACTIVE,
+		}
+	};
+	struct v4l2_subdev_krouting routing = {
+		.routes = routes,
+		.num_routes = 1,
+	};
+	struct ccs_subdev *ssd = to_ccs_subdev(sd);
+	struct ccs_sensor *sensor = ssd->sensor;
+	int rval;
+
+	mutex_lock(&sensor->mutex);
+
+	if (sensor->embedded_start != sensor->embedded_end)
+		routing.num_routes++;
+
+	rval = v4l2_subdev_set_routing(sd, sd_state, &routing);
+	if (rval)
+		goto out;
+
+	rval = __ccs_init_state(sd, sd_state);
+	if (rval)
+		goto out;
+
+	if (sensor->embedded_start != sensor->embedded_end)
+		ccs_set_format_meta(sd, sd_state, NULL);
+
+out:
+	mutex_unlock(&sensor->mutex);
+
+	return rval;
 }
 
 static const struct v4l2_subdev_video_ops ccs_video_ops = {
@@ -3111,6 +3407,14 @@ static const struct v4l2_subdev_pad_ops ccs_pad_ops = {
 	.set_fmt = ccs_set_format,
 	.get_selection = ccs_get_selection,
 	.set_selection = ccs_set_selection,
+};
+
+static const struct v4l2_subdev_pad_ops ccs_src_pad_ops = {
+	.enum_mbus_code = ccs_enum_mbus_code,
+	.get_fmt = ccs_get_format,
+	.set_fmt = ccs_set_format,
+	.get_selection = ccs_get_selection,
+	.set_selection = ccs_set_selection,
 	.enable_streams = ccs_enable_streams,
 	.disable_streams = ccs_disable_streams,
 	.get_frame_desc = ccs_get_frame_desc,
@@ -3127,12 +3431,22 @@ static const struct v4l2_subdev_ops ccs_ops = {
 	.sensor = &ccs_sensor_ops,
 };
 
+static const struct v4l2_subdev_ops ccs_src_ops = {
+	.video = &ccs_video_ops,
+	.pad = &ccs_src_pad_ops,
+	.sensor = &ccs_sensor_ops,
+};
+
 static const struct media_entity_operations ccs_entity_ops = {
 	.link_validate = v4l2_subdev_link_validate,
 };
 
-static const struct v4l2_subdev_internal_ops ccs_internal_src_ops = {
+static const struct v4l2_subdev_internal_ops ccs_internal_ops = {
 	.init_state = ccs_init_state,
+};
+
+static const struct v4l2_subdev_internal_ops ccs_src_internal_ops = {
+	.init_state = ccs_src_init_state,
 	.registered = ccs_registered,
 	.unregistered = ccs_unregistered,
 };
@@ -3283,8 +3597,8 @@ static int ccs_probe(struct i2c_client *client)
 
 	sensor->src = &sensor->ssds[sensor->ssds_used];
 
-	v4l2_i2c_subdev_init(&sensor->src->sd, client, &ccs_ops);
-	sensor->src->sd.internal_ops = &ccs_internal_src_ops;
+	v4l2_i2c_subdev_init(&sensor->src->sd, client, &ccs_src_ops);
+	sensor->src->sd.internal_ops = &ccs_src_internal_ops;
 
 	sensor->regulators = devm_kcalloc(&client->dev,
 					  ARRAY_SIZE(ccs_regulators),
@@ -3555,12 +3869,20 @@ static int ccs_probe(struct i2c_client *client)
 		goto out_cleanup;
 	}
 
-	rval = ccs_init_subdev(sensor, sensor->scaler, " scaler", 2,
+	rval = ccs_init_subdev(sensor, sensor->scaler, " scaler",
+			       sensor->ssds_used != CCS_SUBDEVS ?
+			       CCS_PADS_NOMETA :
+			       sensor->embedded_start == sensor->embedded_end ?
+			       CCS_PADS_NOMETA : CCS_PADS,
 			       MEDIA_ENT_F_PROC_VIDEO_SCALER,
 			       "ccs scaler mutex", &scaler_lock_key);
 	if (rval)
 		goto out_cleanup;
-	rval = ccs_init_subdev(sensor, sensor->binner, " binner", 2,
+	rval = ccs_init_subdev(sensor, sensor->binner, " binner",
+			       sensor->ssds_used == CCS_SUBDEVS ?
+			       CCS_PADS_NOMETA :
+			       sensor->embedded_start == sensor->embedded_end ?
+			       CCS_PADS_NOMETA : CCS_PADS,
 			       MEDIA_ENT_F_PROC_VIDEO_SCALER,
 			       "ccs binner mutex", &binner_lock_key);
 	if (rval)
drivers/media/i2c/ccs/ccs.h +14 -4
index adb152366ea2..207d33de3eee 100644
@@ -173,11 +173,18 @@ struct ccs_csi_data_format {
 #define CCS_SUBDEVS			3
 
 #define CCS_PA_PAD_SRC			0
-#define CCS_PAD_SINK			0
-#define CCS_PAD_SRC			1
-#define CCS_PADS			2
+enum {
+	CCS_PAD_SINK,
+	CCS_PAD_SRC,
+	CCS_PAD_META,
+	CCS_PADS_NOMETA = CCS_PAD_META,
+	CCS_PADS,
+};
 
-#define CCS_STREAM_PIXEL		0
+enum {
+	CCS_STREAM_PIXEL,
+	CCS_STREAM_META,
+};
 
 struct ccs_binning_subtype {
 	u8 horizontal:4;
@@ -228,6 +235,9 @@ struct ccs_sensor {
 	int default_pixel_order;
 	struct ccs_data_container sdata, mdata;
 
+	u32 embedded_mbus_code;
+	u8 emb_data_ctrl;
+
 	u8 binning_horizontal;
 	u8 binning_vertical;
 
Signature · 1 line
2.39.2
author maintainer
#103 lore raw

PATCH v8 25/38 media: ccs: Remove ccs_get_crop_compose helper

In reply to Sakari Ailus

As it's now easier to obtain the necessary information on crop and compose rectangles after moving to sub-device state, remove the ccs_get_crop_compose helper.

---
 drivers/media/i2c/ccs/ccs-core.c | 130 ++++++++++++++-----------------
 1 file changed, 58 insertions(+), 72 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +58 -72
index 1f91af25eac6..6c7eb4f04070 100644
@@ -2222,24 +2222,6 @@ static int ccs_get_format(struct v4l2_subdev *subdev,
 	return rval;
 }
 
-static void ccs_get_crop_compose(struct v4l2_subdev *subdev,
-				 struct v4l2_subdev_state *sd_state,
-				 struct v4l2_rect **crops,
-				 struct v4l2_rect **comps)
-{
-	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	unsigned int i;
-
-	if (crops)
-		for (i = 0; i < subdev->entity.num_pads; i++)
-			crops[i] =
-				v4l2_subdev_state_get_crop(sd_state, i,
-							   CCS_STREAM_PIXEL);
-	if (comps)
-		*comps = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
-						       CCS_STREAM_PIXEL);
-}
-
 /* Changes require propagation only on sink pad. */
 static void ccs_propagate(struct v4l2_subdev *subdev,
 			  struct v4l2_subdev_state *sd_state, int which,
@@ -2247,15 +2229,17 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *comp, *crop;
 	struct v4l2_mbus_framefmt *fmt;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
-
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 	switch (target) {
 	case V4L2_SEL_TGT_CROP:
-		comp->width = crops[CCS_PAD_SINK]->width;
-		comp->height = crops[CCS_PAD_SINK]->height;
+		crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SINK,
+						  CCS_STREAM_PIXEL);
+		comp->width = crop->width;
+		comp->height = crop->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE) {
 			if (ssd == sensor->scaler) {
 				sensor->scale_m = CCS_LIM(sensor, SCALER_N_MIN);
@@ -2269,13 +2253,15 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 		}
 		fallthrough;
 	case V4L2_SEL_TGT_COMPOSE:
-		*crops[CCS_PAD_SRC] = *comp;
+		crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SRC,
+						  CCS_STREAM_PIXEL);
+		*crop = *comp;
 		fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
 						   CCS_STREAM_PIXEL);
 		fmt->width = comp->width;
 		fmt->height = comp->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE && ssd == sensor->src)
-			sensor->src_src = *crops[CCS_PAD_SRC];
+			sensor->src_src = *crop;
 		break;
 	default:
 		WARN_ON_ONCE(1);
@@ -2416,7 +2402,7 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *crops[CCS_PADS];
+	struct v4l2_rect *crop;
 
 	if (subdev == &sensor->src->sd && fmt->pad == CCS_PAD_META)
 		return ccs_get_format(subdev, sd_state, fmt);
@@ -2458,12 +2444,13 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 		      CCS_LIM(sensor, MIN_Y_OUTPUT_SIZE),
 		      CCS_LIM(sensor, MAX_Y_OUTPUT_SIZE));
 
-	ccs_get_crop_compose(subdev, sd_state, crops, NULL);
+	crop = v4l2_subdev_state_get_crop(sd_state, ssd->sink_pad,
+					  CCS_STREAM_PIXEL);
 
-	crops[ssd->sink_pad]->left = 0;
-	crops[ssd->sink_pad]->top = 0;
-	crops[ssd->sink_pad]->width = fmt->format.width;
-	crops[ssd->sink_pad]->height = fmt->format.height;
+	crop->left = 0;
+	crop->top = 0;
+	crop->width = fmt->format.width;
+	crop->height = fmt->format.height;
 	ccs_propagate(subdev, sd_state, fmt->which, V4L2_SEL_TGT_CROP);
 
 	mutex_unlock(&sensor->mutex);
@@ -2518,24 +2505,23 @@ static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w,
 static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 				   struct v4l2_subdev_state *sd_state,
 				   struct v4l2_subdev_selection *sel,
-				   struct v4l2_rect **crops,
+				   struct v4l2_rect *sink_crop,
 				   struct v4l2_rect *comp)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	unsigned int i;
 	unsigned int binh = 1, binv = 1;
 	int best = scaling_goodness(
-		subdev,
-		crops[CCS_PAD_SINK]->width, sel->r.width,
-		crops[CCS_PAD_SINK]->height, sel->r.height, sel->flags);
+		subdev,	sink_crop->width, sel->r.width,
+		sink_crop->height, sel->r.height, sel->flags);
 
 	for (i = 0; i < sensor->nbinning_subtypes; i++) {
 		int this = scaling_goodness(
 			subdev,
-			crops[CCS_PAD_SINK]->width
+			sink_crop->width
 			/ sensor->binning_subtypes[i].horizontal,
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height
+			sink_crop->height
 			/ sensor->binning_subtypes[i].vertical,
 			sel->r.height, sel->flags);
 
@@ -2550,8 +2536,8 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 		sensor->binning_horizontal = binh;
 	}
 
-	sel->r.width = (crops[CCS_PAD_SINK]->width / binh) & ~1;
-	sel->r.height = (crops[CCS_PAD_SINK]->height / binv) & ~1;
+	sel->r.width = (sink_crop->width / binh) & ~1;
+	sel->r.height = (sink_crop->height / binv) & ~1;
 }
 
 /*
@@ -2566,7 +2552,7 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 				   struct v4l2_subdev_state *sd_state,
 				   struct v4l2_subdev_selection *sel,
-				   struct v4l2_rect **crops,
+				   struct v4l2_rect *sink_crop,
 				   struct v4l2_rect *comp)
 {
 	struct i2c_client *client = v4l2_get_subdevdata(subdev);
@@ -2579,17 +2565,12 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	unsigned int i;
 	int best = INT_MIN;
 
-	sel->r.width = min_t(unsigned int, sel->r.width,
-			     crops[CCS_PAD_SINK]->width);
-	sel->r.height = min_t(unsigned int, sel->r.height,
-			      crops[CCS_PAD_SINK]->height);
-
-	a = crops[CCS_PAD_SINK]->width
-		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.width;
-	b = crops[CCS_PAD_SINK]->height
-		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
-	max_m = crops[CCS_PAD_SINK]->width
-		* CCS_LIM(sensor, SCALER_N_MIN)
+	sel->r.width = min_t(unsigned int, sel->r.width, sink_crop->width);
+	sel->r.height = min_t(unsigned int, sel->r.height, sink_crop->height);
+
+	a = sink_crop->width * CCS_LIM(sensor, SCALER_N_MIN) / sel->r.width;
+	b = sink_crop->height * CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
+	max_m = sink_crop->width * CCS_LIM(sensor, SCALER_N_MIN)
 		/ CCS_LIM(sensor, MIN_X_OUTPUT_SIZE);
 
 	a = clamp(a, CCS_LIM(sensor, SCALER_M_MIN),
@@ -2622,10 +2603,10 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	for (i = 0; i < ntry; i++) {
 		int this = scaling_goodness(
 			subdev,
-			crops[CCS_PAD_SINK]->width
+			sink_crop->width
 			/ try[i] * CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height,
+			sink_crop->height,
 			sel->r.height,
 			sel->flags);
 
@@ -2642,12 +2623,10 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 			continue;
 
 		this = scaling_goodness(
-			subdev, crops[CCS_PAD_SINK]->width
-			/ try[i]
+			subdev, sink_crop->width / try[i]
 			* CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height
-			/ try[i]
+			sink_crop->height / try[i]
 			* CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.height,
 			sel->flags);
@@ -2660,17 +2639,15 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	}
 
 	sel->r.width =
-		(crops[CCS_PAD_SINK]->width
-		 / scale_m
+		(sink_crop->width / scale_m
 		 * CCS_LIM(sensor, SCALER_N_MIN)) & ~1;
 	if (mode == SMIAPP_SCALING_MODE_BOTH)
 		sel->r.height =
-			(crops[CCS_PAD_SINK]->height
-			 / scale_m
+			(sink_crop->height / scale_m
 			 * CCS_LIM(sensor, SCALER_N_MIN))
 			& ~1;
 	else
-		sel->r.height = crops[CCS_PAD_SINK]->height;
+		sel->r.height = sink_crop->height;
 
 	if (sel->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
 		sensor->scale_m = scale_m;
@@ -2684,17 +2661,20 @@ static int ccs_set_compose(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *comp, *sink_crop;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	sink_crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SINK,
+					       CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	sel->r.top = 0;
 	sel->r.left = 0;
 
 	if (ssd == sensor->binner)
-		ccs_set_compose_binner(subdev, sd_state, sel, crops, comp);
+		ccs_set_compose_binner(subdev, sd_state, sel, sink_crop, comp);
 	else
-		ccs_set_compose_scaler(subdev, sd_state, sel, crops, comp);
+		ccs_set_compose_scaler(subdev, sd_state, sel, sink_crop, comp);
 
 	*comp = sel->r;
 	ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_COMPOSE);
@@ -2755,9 +2735,12 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect src_size = { 0 }, *crops[CCS_PADS], *comp;
+	struct v4l2_rect src_size = { 0 }, *crop, *comp;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	crop = v4l2_subdev_state_get_crop(sd_state, sel->pad,
+					  CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	if (sel->pad == ssd->sink_pad) {
 		struct v4l2_mbus_framefmt *mfmt =
@@ -2781,7 +2764,7 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 	sel->r.left = min_t(int, sel->r.left, src_size.width - sel->r.width);
 	sel->r.top = min_t(int, sel->r.top, src_size.height - sel->r.height);
 
-	*crops[sel->pad] = sel->r;
+	*crop = sel->r;
 
 	if (ssd != sensor->pixel_array && sel->pad == CCS_PAD_SINK)
 		ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_CROP);
@@ -2806,14 +2789,17 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *crop, *comp;
 	int ret;
 
 	ret = ccs_sel_supported(subdev, sel);
 	if (ret)
 		return ret;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	crop = v4l2_subdev_state_get_crop(sd_state, sel->pad,
+					  CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	switch (sel->target) {
 	case V4L2_SEL_TGT_CROP_BOUNDS:
@@ -2835,7 +2821,7 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 		break;
 	case V4L2_SEL_TGT_CROP:
 	case V4L2_SEL_TGT_COMPOSE_BOUNDS:
-		sel->r = *crops[sel->pad];
+		sel->r = *crop;
 		break;
 	case V4L2_SEL_TGT_COMPOSE:
 		sel->r = *comp;
Signature · 1 line
2.39.2
maintainer
#104 lore raw

Re: [PATCH v8 25/38] media: ccs: Remove ccs_get_crop_compose helper

Hi Sakari,

Thank you for the patch.

Quoted 108 lines · Sakari Ailus

As it's now easier to obtain the necessary information on crop and compose rectangles after moving to sub-device state, remove the ccs_get_crop_compose helper.

---
 drivers/media/i2c/ccs/ccs-core.c | 130 ++++++++++++++-----------------
 1 file changed, 58 insertions(+), 72 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +18 -31
index 1f91af25eac6..6c7eb4f04070 100644
@@ -2222,24 +2222,6 @@ static int ccs_get_format(struct v4l2_subdev *subdev,
 	return rval;
 }
 
-static void ccs_get_crop_compose(struct v4l2_subdev *subdev,
-				 struct v4l2_subdev_state *sd_state,
-				 struct v4l2_rect **crops,
-				 struct v4l2_rect **comps)
-{
-	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	unsigned int i;
-
-	if (crops)
-		for (i = 0; i < subdev->entity.num_pads; i++)
-			crops[i] =
-				v4l2_subdev_state_get_crop(sd_state, i,
-							   CCS_STREAM_PIXEL);
-	if (comps)
-		*comps = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
-						       CCS_STREAM_PIXEL);
-}
-
 /* Changes require propagation only on sink pad. */
 static void ccs_propagate(struct v4l2_subdev *subdev,
 			  struct v4l2_subdev_state *sd_state, int which,
@@ -2247,15 +2229,17 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *comp, *crop;
 	struct v4l2_mbus_framefmt *fmt;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
-
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 	switch (target) {
 	case V4L2_SEL_TGT_CROP:
-		comp->width = crops[CCS_PAD_SINK]->width;
-		comp->height = crops[CCS_PAD_SINK]->height;
+		crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SINK,
+						  CCS_STREAM_PIXEL);
+		comp->width = crop->width;
+		comp->height = crop->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE) {
 			if (ssd == sensor->scaler) {
 				sensor->scale_m = CCS_LIM(sensor, SCALER_N_MIN);
@@ -2269,13 +2253,15 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 		}
 		fallthrough;
 	case V4L2_SEL_TGT_COMPOSE:
-		*crops[CCS_PAD_SRC] = *comp;
+		crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SRC,
+						  CCS_STREAM_PIXEL);
+		*crop = *comp;
 		fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
 						   CCS_STREAM_PIXEL);
 		fmt->width = comp->width;
 		fmt->height = comp->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE && ssd == sensor->src)
-			sensor->src_src = *crops[CCS_PAD_SRC];
+			sensor->src_src = *crop;
 		break;
 	default:
 		WARN_ON_ONCE(1);
@@ -2416,7 +2402,7 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *crops[CCS_PADS];
+	struct v4l2_rect *crop;
 
 	if (subdev == &sensor->src->sd && fmt->pad == CCS_PAD_META)
 		return ccs_get_format(subdev, sd_state, fmt);
@@ -2458,12 +2444,13 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 		      CCS_LIM(sensor, MIN_Y_OUTPUT_SIZE),
 		      CCS_LIM(sensor, MAX_Y_OUTPUT_SIZE));
 
-	ccs_get_crop_compose(subdev, sd_state, crops, NULL);
+	crop = v4l2_subdev_state_get_crop(sd_state, ssd->sink_pad,
+					  CCS_STREAM_PIXEL);
 
-	crops[ssd->sink_pad]->left = 0;
-	crops[ssd->sink_pad]->top = 0;
-	crops[ssd->sink_pad]->width = fmt->format.width;
-	crops[ssd->sink_pad]->height = fmt->format.height;
+	crop->left = 0;
+	crop->top = 0;
+	crop->width = fmt->format.width;
+	crop->height = fmt->format.height;
 	ccs_propagate(subdev, sd_state, fmt->which, V4L2_SEL_TGT_CROP);
 
 	mutex_unlock(&sensor->mutex);

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2508

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2518,24 +2505,23 @@ static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w,
 static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 				   struct v4l2_subdev_state *sd_state,
 				   struct v4l2_subdev_selection *sel,
-				   struct v4l2_rect **crops,
+				   struct v4l2_rect *sink_crop,

const

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2515

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +1 -3
 				   struct v4l2_rect *comp)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	unsigned int i;
 	unsigned int binh = 1, binv = 1;
 	int best = scaling_goodness(
-		subdev,
-		crops[CCS_PAD_SINK]->width, sel->r.width,
-		crops[CCS_PAD_SINK]->height, sel->r.height, sel->flags);
+		subdev,	sink_crop->width, sel->r.width,

s/\tsink_crop/ sink_crop/

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2516

Quoted 1 line
drivers/media/i2c/ccs/ccs-core.c +1 -0
+		sink_crop->height, sel->r.height, sel->flags);
	int best = scaling_goodness(subdev, sink_crop->width, sel->r.width,
				    sink_crop->height, sel->r.height,
				    sel->flags);

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2526

Quoted 12 lines
drivers/media/i2c/ccs/ccs-core.c +2 -2
 	for (i = 0; i < sensor->nbinning_subtypes; i++) {
 		int this = scaling_goodness(
 			subdev,
-			crops[CCS_PAD_SINK]->width
+			sink_crop->width
 			/ sensor->binning_subtypes[i].horizontal,
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height
+			sink_crop->height
 			/ sensor->binning_subtypes[i].vertical,
 			sel->r.height, sel->flags);

You know this is horrible to read, right ? :-)

Quoted 13 lines
drivers/media/i2c/ccs/ccs-core.c +2 -2
@@ -2550,8 +2536,8 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 		sensor->binning_horizontal = binh;
 	}
 
-	sel->r.width = (crops[CCS_PAD_SINK]->width / binh) & ~1;
-	sel->r.height = (crops[CCS_PAD_SINK]->height / binv) & ~1;
+	sel->r.width = (sink_crop->width / binh) & ~1;
+	sel->r.height = (sink_crop->height / binv) & ~1;
 }
 
 /*

drivers/media/i2c/ccs/ccs-core.c static void ccs_set_compose_binner(struct v4l2_subdev *subdev, · line 2555

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2566,7 +2552,7 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 				   struct v4l2_subdev_state *sd_state,
 				   struct v4l2_subdev_selection *sel,
-				   struct v4l2_rect **crops,
+				   struct v4l2_rect *sink_crop,

const

Quoted 56 lines
drivers/media/i2c/ccs/ccs-core.c +10 -17
 				   struct v4l2_rect *comp)
 {
 	struct i2c_client *client = v4l2_get_subdevdata(subdev);
@@ -2579,17 +2565,12 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	unsigned int i;
 	int best = INT_MIN;
 
-	sel->r.width = min_t(unsigned int, sel->r.width,
-			     crops[CCS_PAD_SINK]->width);
-	sel->r.height = min_t(unsigned int, sel->r.height,
-			      crops[CCS_PAD_SINK]->height);
-
-	a = crops[CCS_PAD_SINK]->width
-		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.width;
-	b = crops[CCS_PAD_SINK]->height
-		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
-	max_m = crops[CCS_PAD_SINK]->width
-		* CCS_LIM(sensor, SCALER_N_MIN)
+	sel->r.width = min_t(unsigned int, sel->r.width, sink_crop->width);
+	sel->r.height = min_t(unsigned int, sel->r.height, sink_crop->height);
+
+	a = sink_crop->width * CCS_LIM(sensor, SCALER_N_MIN) / sel->r.width;
+	b = sink_crop->height * CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
+	max_m = sink_crop->width * CCS_LIM(sensor, SCALER_N_MIN)
 		/ CCS_LIM(sensor, MIN_X_OUTPUT_SIZE);
 
 	a = clamp(a, CCS_LIM(sensor, SCALER_M_MIN),
@@ -2622,10 +2603,10 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	for (i = 0; i < ntry; i++) {
 		int this = scaling_goodness(
 			subdev,
-			crops[CCS_PAD_SINK]->width
+			sink_crop->width
 			/ try[i] * CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height,
+			sink_crop->height,
 			sel->r.height,
 			sel->flags);
 
@@ -2642,12 +2623,10 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 			continue;
 
 		this = scaling_goodness(
-			subdev, crops[CCS_PAD_SINK]->width
-			/ try[i]
+			subdev, sink_crop->width / try[i]
 			* CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height
-			/ try[i]
+			sink_crop->height / try[i]
 			* CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.height,
 			sel->flags);

drivers/media/i2c/ccs/ccs-core.c static void ccs_set_compose_scaler(struct v4l2_subdev *subdev, · line 2643

Quoted 7 lines
drivers/media/i2c/ccs/ccs-core.c +1 -2
@@ -2660,17 +2639,15 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	}
 
 	sel->r.width =
-		(crops[CCS_PAD_SINK]->width
-		 / scale_m
+		(sink_crop->width / scale_m
 		 * CCS_LIM(sensor, SCALER_N_MIN)) & ~1;
	sel->r.width = (sink_crop->width / scale_m
		* CCS_LIM(sensor, SCALER_N_MIN)) & ~1;
Quoted 14 lines
drivers/media/i2c/ccs/ccs-core.c +2 -3
 	if (mode == SMIAPP_SCALING_MODE_BOTH)
 		sel->r.height =
-			(crops[CCS_PAD_SINK]->height
-			 / scale_m
+			(sink_crop->height / scale_m
 			 * CCS_LIM(sensor, SCALER_N_MIN))
 			& ~1;
 	else
-		sel->r.height = crops[CCS_PAD_SINK]->height;
+		sel->r.height = sink_crop->height;
 
 	if (sel->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
 		sensor->scale_m = scale_m;

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_compose(struct v4l2_subdev *subdev, · line 2664

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2684,17 +2661,20 @@ static int ccs_set_compose(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *comp, *sink_crop;

sink_crop can be const.

Quoted 20 lines
drivers/media/i2c/ccs/ccs-core.c +6 -3
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	sink_crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SINK,
+					       CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	sel->r.top = 0;
 	sel->r.left = 0;
 
 	if (ssd == sensor->binner)
-		ccs_set_compose_binner(subdev, sd_state, sel, crops, comp);
+		ccs_set_compose_binner(subdev, sd_state, sel, sink_crop, comp);
 	else
-		ccs_set_compose_scaler(subdev, sd_state, sel, crops, comp);
+		ccs_set_compose_scaler(subdev, sd_state, sel, sink_crop, comp);
 
 	*comp = sel->r;
 	ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_COMPOSE);

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_crop(struct v4l2_subdev *subdev, · line 2738

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2755,9 +2735,12 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect src_size = { 0 }, *crops[CCS_PADS], *comp;
+	struct v4l2_rect src_size = { 0 }, *crop, *comp;

comp can be const.

Quoted 19 lines
drivers/media/i2c/ccs/ccs-core.c +5 -2
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	crop = v4l2_subdev_state_get_crop(sd_state, sel->pad,
+					  CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	if (sel->pad == ssd->sink_pad) {
 		struct v4l2_mbus_framefmt *mfmt =
@@ -2781,7 +2764,7 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 	sel->r.left = min_t(int, sel->r.left, src_size.width - sel->r.width);
 	sel->r.top = min_t(int, sel->r.top, src_size.height - sel->r.height);
 
-	*crops[sel->pad] = sel->r;
+	*crop = sel->r;
 
 	if (ssd != sensor->pixel_array && sel->pad == CCS_PAD_SINK)
 		ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_CROP);

drivers/media/i2c/ccs/ccs-core.c static int ccs_get_selection(struct v4l2_subdev *subdev, · line 2792

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2806,14 +2789,17 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *crop, *comp;

const

Quoted 23 lines
drivers/media/i2c/ccs/ccs-core.c +5 -2
 	int ret;
 
 	ret = ccs_sel_supported(subdev, sel);
 	if (ret)
 		return ret;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	crop = v4l2_subdev_state_get_crop(sd_state, sel->pad,
+					  CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	switch (sel->target) {
 	case V4L2_SEL_TGT_CROP_BOUNDS:
@@ -2835,7 +2821,7 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 		break;
 	case V4L2_SEL_TGT_CROP:
 	case V4L2_SEL_TGT_COMPOSE_BOUNDS:
-		sel->r = *crops[sel->pad];
+		sel->r = *crop;
 		break;
 	case V4L2_SEL_TGT_COMPOSE:
 		sel->r = *comp;
Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#105 lore raw

Re: [PATCH v8 25/38] media: ccs: Remove ccs_get_crop_compose helper

Hi Laurent,

Quoted 111 lines · Laurent Pinchart

Hi Sakari,

Thank you for the patch.

Quoted 106 lines · Sakari Ailus

As it's now easier to obtain the necessary information on crop and compose rectangles after moving to sub-device state, remove the ccs_get_crop_compose helper.

---
 drivers/media/i2c/ccs/ccs-core.c | 130 ++++++++++++++-----------------
 1 file changed, 58 insertions(+), 72 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +18 -31
index 1f91af25eac6..6c7eb4f04070 100644
@@ -2222,24 +2222,6 @@ static int ccs_get_format(struct v4l2_subdev *subdev,
 	return rval;
 }
 
-static void ccs_get_crop_compose(struct v4l2_subdev *subdev,
-				 struct v4l2_subdev_state *sd_state,
-				 struct v4l2_rect **crops,
-				 struct v4l2_rect **comps)
-{
-	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	unsigned int i;
-
-	if (crops)
-		for (i = 0; i < subdev->entity.num_pads; i++)
-			crops[i] =
-				v4l2_subdev_state_get_crop(sd_state, i,
-							   CCS_STREAM_PIXEL);
-	if (comps)
-		*comps = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
-						       CCS_STREAM_PIXEL);
-}
-
 /* Changes require propagation only on sink pad. */
 static void ccs_propagate(struct v4l2_subdev *subdev,
 			  struct v4l2_subdev_state *sd_state, int which,
@@ -2247,15 +2229,17 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *comp, *crop;
 	struct v4l2_mbus_framefmt *fmt;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
-
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 	switch (target) {
 	case V4L2_SEL_TGT_CROP:
-		comp->width = crops[CCS_PAD_SINK]->width;
-		comp->height = crops[CCS_PAD_SINK]->height;
+		crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SINK,
+						  CCS_STREAM_PIXEL);
+		comp->width = crop->width;
+		comp->height = crop->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE) {
 			if (ssd == sensor->scaler) {
 				sensor->scale_m = CCS_LIM(sensor, SCALER_N_MIN);
@@ -2269,13 +2253,15 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 		}
 		fallthrough;
 	case V4L2_SEL_TGT_COMPOSE:
-		*crops[CCS_PAD_SRC] = *comp;
+		crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SRC,
+						  CCS_STREAM_PIXEL);
+		*crop = *comp;
 		fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
 						   CCS_STREAM_PIXEL);
 		fmt->width = comp->width;
 		fmt->height = comp->height;
 		if (which == V4L2_SUBDEV_FORMAT_ACTIVE && ssd == sensor->src)
-			sensor->src_src = *crops[CCS_PAD_SRC];
+			sensor->src_src = *crop;
 		break;
 	default:
 		WARN_ON_ONCE(1);
@@ -2416,7 +2402,7 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *crops[CCS_PADS];
+	struct v4l2_rect *crop;
 
 	if (subdev == &sensor->src->sd && fmt->pad == CCS_PAD_META)
 		return ccs_get_format(subdev, sd_state, fmt);
@@ -2458,12 +2444,13 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 		      CCS_LIM(sensor, MIN_Y_OUTPUT_SIZE),
 		      CCS_LIM(sensor, MAX_Y_OUTPUT_SIZE));
 
-	ccs_get_crop_compose(subdev, sd_state, crops, NULL);
+	crop = v4l2_subdev_state_get_crop(sd_state, ssd->sink_pad,
+					  CCS_STREAM_PIXEL);
 
-	crops[ssd->sink_pad]->left = 0;
-	crops[ssd->sink_pad]->top = 0;
-	crops[ssd->sink_pad]->width = fmt->format.width;
-	crops[ssd->sink_pad]->height = fmt->format.height;
+	crop->left = 0;
+	crop->top = 0;
+	crop->width = fmt->format.width;
+	crop->height = fmt->format.height;
 	ccs_propagate(subdev, sd_state, fmt->which, V4L2_SEL_TGT_CROP);
 
Quoted 9 lines

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2508

Quoted 7 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
 	mutex_unlock(&sensor->mutex);
@@ -2518,24 +2505,23 @@ static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w,
 static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 				   struct v4l2_subdev_state *sd_state,
 				   struct v4l2_subdev_selection *sel,
-				   struct v4l2_rect **crops,
+				   struct v4l2_rect *sink_crop,

const

Yes. Also it seems the comp argument can be dropped.

Quoted 13 lines

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2515

Quoted 11 lines
drivers/media/i2c/ccs/ccs-core.c +1 -3
 				   struct v4l2_rect *comp)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	unsigned int i;
 	unsigned int binh = 1, binv = 1;
 	int best = scaling_goodness(
-		subdev,
-		crops[CCS_PAD_SINK]->width, sel->r.width,
-		crops[CCS_PAD_SINK]->height, sel->r.height, sel->flags);
+		subdev,	sink_crop->width, sel->r.width,

s/\tsink_crop/ sink_crop/

Yes.

Quoted 8 lines

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2516

Quoted 2 lines
drivers/media/i2c/ccs/ccs-core.c +1 -0
+		sink_crop->height, sel->r.height, sel->flags);
	int best = scaling_goodness(subdev, sink_crop->width, sel->r.width,
				    sink_crop->height, sel->r.height,
				    sel->flags);
Quoted 1 line
Quoted 13 lines

drivers/media/i2c/ccs/ccs-core.c static int scaling_goodness(struct v4l2_subdev *subdev, int w, int ask_w, · line 2526

Quoted 11 lines
drivers/media/i2c/ccs/ccs-core.c +2 -2
 	for (i = 0; i < sensor->nbinning_subtypes; i++) {
 		int this = scaling_goodness(
 			subdev,
-			crops[CCS_PAD_SINK]->width
+			sink_crop->width
 			/ sensor->binning_subtypes[i].horizontal,
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height
+			sink_crop->height
 			/ sensor->binning_subtypes[i].vertical,
 			sel->r.height, sel->flags);

You know this is horrible to read, right ? :-)

I'll rewrap the calls to start the lines on the opening parenthesis.

Quoted 12 lines
Quoted 12 lines
drivers/media/i2c/ccs/ccs-core.c +2 -2
@@ -2550,8 +2536,8 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 		sensor->binning_horizontal = binh;
 	}
 
-	sel->r.width = (crops[CCS_PAD_SINK]->width / binh) & ~1;
-	sel->r.height = (crops[CCS_PAD_SINK]->height / binv) & ~1;
+	sel->r.width = (sink_crop->width / binh) & ~1;
+	sel->r.height = (sink_crop->height / binv) & ~1;
 }
 
Quoted 9 lines

drivers/media/i2c/ccs/ccs-core.c static void ccs_set_compose_binner(struct v4l2_subdev *subdev, · line 2555

Quoted 7 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
 /*
@@ -2566,7 +2552,7 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 				   struct v4l2_subdev_state *sd_state,
 				   struct v4l2_subdev_selection *sel,
-				   struct v4l2_rect **crops,
+				   struct v4l2_rect *sink_crop,

const

Yes.

Quoted 169 lines
Quoted 57 lines
drivers/media/i2c/ccs/ccs-core.c +10 -17
 				   struct v4l2_rect *comp)
 {
 	struct i2c_client *client = v4l2_get_subdevdata(subdev);
@@ -2579,17 +2565,12 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	unsigned int i;
 	int best = INT_MIN;
 
-	sel->r.width = min_t(unsigned int, sel->r.width,
-			     crops[CCS_PAD_SINK]->width);
-	sel->r.height = min_t(unsigned int, sel->r.height,
-			      crops[CCS_PAD_SINK]->height);
-
-	a = crops[CCS_PAD_SINK]->width
-		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.width;
-	b = crops[CCS_PAD_SINK]->height
-		* CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
-	max_m = crops[CCS_PAD_SINK]->width
-		* CCS_LIM(sensor, SCALER_N_MIN)
+	sel->r.width = min_t(unsigned int, sel->r.width, sink_crop->width);
+	sel->r.height = min_t(unsigned int, sel->r.height, sink_crop->height);
+
+	a = sink_crop->width * CCS_LIM(sensor, SCALER_N_MIN) / sel->r.width;
+	b = sink_crop->height * CCS_LIM(sensor, SCALER_N_MIN) / sel->r.height;
+	max_m = sink_crop->width * CCS_LIM(sensor, SCALER_N_MIN)
 		/ CCS_LIM(sensor, MIN_X_OUTPUT_SIZE);
 
 	a = clamp(a, CCS_LIM(sensor, SCALER_M_MIN),
@@ -2622,10 +2603,10 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	for (i = 0; i < ntry; i++) {
 		int this = scaling_goodness(
 			subdev,
-			crops[CCS_PAD_SINK]->width
+			sink_crop->width
 			/ try[i] * CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height,
+			sink_crop->height,
 			sel->r.height,
 			sel->flags);
 
@@ -2642,12 +2623,10 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 			continue;
 
 		this = scaling_goodness(
-			subdev, crops[CCS_PAD_SINK]->width
-			/ try[i]
+			subdev, sink_crop->width / try[i]
 			* CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.width,
-			crops[CCS_PAD_SINK]->height
-			/ try[i]
+			sink_crop->height / try[i]
 			* CCS_LIM(sensor, SCALER_N_MIN),
 			sel->r.height,
 			sel->flags);

drivers/media/i2c/ccs/ccs-core.c static void ccs_set_compose_scaler(struct v4l2_subdev *subdev, · line 2643

Quoted 7 lines
drivers/media/i2c/ccs/ccs-core.c +1 -2
@@ -2660,17 +2639,15 @@ static void ccs_set_compose_scaler(struct v4l2_subdev *subdev,
 	}
 
 	sel->r.width =
-		(crops[CCS_PAD_SINK]->width
-		 / scale_m
+		(sink_crop->width / scale_m
 		 * CCS_LIM(sensor, SCALER_N_MIN)) & ~1;
	sel->r.width = (sink_crop->width / scale_m
		* CCS_LIM(sensor, SCALER_N_MIN)) & ~1;
Quoted 14 lines
drivers/media/i2c/ccs/ccs-core.c +2 -3
 	if (mode == SMIAPP_SCALING_MODE_BOTH)
 		sel->r.height =
-			(crops[CCS_PAD_SINK]->height
-			 / scale_m
+			(sink_crop->height / scale_m
 			 * CCS_LIM(sensor, SCALER_N_MIN))
 			& ~1;
 	else
-		sel->r.height = crops[CCS_PAD_SINK]->height;
+		sel->r.height = sink_crop->height;
 
 	if (sel->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
 		sensor->scale_m = scale_m;

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_compose(struct v4l2_subdev *subdev, · line 2664

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2684,17 +2661,20 @@ static int ccs_set_compose(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *comp, *sink_crop;

sink_crop can be const.

Quoted 20 lines
drivers/media/i2c/ccs/ccs-core.c +6 -3
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	sink_crop = v4l2_subdev_state_get_crop(sd_state, CCS_PAD_SINK,
+					       CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	sel->r.top = 0;
 	sel->r.left = 0;
 
 	if (ssd == sensor->binner)
-		ccs_set_compose_binner(subdev, sd_state, sel, crops, comp);
+		ccs_set_compose_binner(subdev, sd_state, sel, sink_crop, comp);
 	else
-		ccs_set_compose_scaler(subdev, sd_state, sel, crops, comp);
+		ccs_set_compose_scaler(subdev, sd_state, sel, sink_crop, comp);
 
 	*comp = sel->r;
 	ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_COMPOSE);

drivers/media/i2c/ccs/ccs-core.c static int ccs_set_crop(struct v4l2_subdev *subdev, · line 2738

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2755,9 +2735,12 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect src_size = { 0 }, *crops[CCS_PADS], *comp;
+	struct v4l2_rect src_size = { 0 }, *crop, *comp;

comp can be const.

Quoted 19 lines
drivers/media/i2c/ccs/ccs-core.c +5 -2
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	crop = v4l2_subdev_state_get_crop(sd_state, sel->pad,
+					  CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	if (sel->pad == ssd->sink_pad) {
 		struct v4l2_mbus_framefmt *mfmt =
@@ -2781,7 +2764,7 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 	sel->r.left = min_t(int, sel->r.left, src_size.width - sel->r.width);
 	sel->r.top = min_t(int, sel->r.top, src_size.height - sel->r.height);
 
-	*crops[sel->pad] = sel->r;
+	*crop = sel->r;
 
 	if (ssd != sensor->pixel_array && sel->pad == CCS_PAD_SINK)
 		ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_CROP);

drivers/media/i2c/ccs/ccs-core.c static int ccs_get_selection(struct v4l2_subdev *subdev, · line 2792

Quoted 5 lines
drivers/media/i2c/ccs/ccs-core.c +1 -1
@@ -2806,14 +2789,17 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	struct ccs_subdev *ssd = to_ccs_subdev(subdev);
-	struct v4l2_rect *comp, *crops[CCS_PADS];
+	struct v4l2_rect *crop, *comp;

const

Quoted 23 lines
drivers/media/i2c/ccs/ccs-core.c +5 -2
 	int ret;
 
 	ret = ccs_sel_supported(subdev, sel);
 	if (ret)
 		return ret;
 
-	ccs_get_crop_compose(subdev, sd_state, crops, &comp);
+	crop = v4l2_subdev_state_get_crop(sd_state, sel->pad,
+					  CCS_STREAM_PIXEL);
+	comp = v4l2_subdev_state_get_compose(sd_state, ssd->sink_pad,
+					     CCS_STREAM_PIXEL);
 
 	switch (sel->target) {
 	case V4L2_SEL_TGT_CROP_BOUNDS:
@@ -2835,7 +2821,7 @@ static int ccs_get_selection(struct v4l2_subdev *subdev,
 		break;
 	case V4L2_SEL_TGT_CROP:
 	case V4L2_SEL_TGT_COMPOSE_BOUNDS:
-		sel->r = *crops[sel->pad];
+		sel->r = *crop;
 		break;
 	case V4L2_SEL_TGT_COMPOSE:
 		sel->r = *comp;
Signature · 3 lines
Regards,

Sakari Ailus
author maintainer
#106 lore raw

PATCH v8 26/38 media: ccs: Rely on sub-device state locking

In reply to Sakari Ailus

Rely on sub-device state locking to serialise access to driver's data structures. The driver-provided mutex is used as the state lock for all driver sub-devices.

---
 drivers/media/i2c/ccs/ccs-core.c | 192 +++++++++++++------------------
 drivers/media/i2c/ccs/ccs.h      |   1 -
 2 files changed, 81 insertions(+), 112 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +81 -111
index 6c7eb4f04070..a8c48abd2e30 100644
@@ -541,12 +541,13 @@ static int ccs_pll_update(struct ccs_sensor *sensor)
  *
  */
 
-static void __ccs_update_exposure_limits(struct ccs_sensor *sensor)
+static void __ccs_update_exposure_limits(struct ccs_sensor *sensor,
+					 struct v4l2_rect *pa_src)
 {
 	struct v4l2_ctrl *ctrl = sensor->exposure;
 	int max;
 
-	max = sensor->pa_src.height + sensor->vblank->val -
+	max = pa_src->height + sensor->vblank->val -
 		CCS_LIM(sensor, COARSE_INTEGRATION_TIME_MAX_MARGIN);
 
 	__v4l2_ctrl_modify_range(ctrl, ctrl->minimum, max, ctrl->step, max);
@@ -649,12 +650,20 @@ static int ccs_set_ctrl(struct v4l2_ctrl *ctrl)
 		container_of(ctrl->handler, struct ccs_subdev, ctrl_handler)
 			->sensor;
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
+	struct v4l2_subdev_state *state;
+	struct v4l2_rect *pa_src;
 	int pm_status;
 	u32 orient = 0;
 	unsigned int i;
 	int exposure;
 	int rval;
 
+	if (ctrl->id == V4L2_CID_VBLANK || ctrl->id == V4L2_CID_HBLANK) {
+		state = v4l2_subdev_get_locked_active_state(&sensor->pixel_array->sd);
+		pa_src = v4l2_subdev_state_get_crop(state, CCS_PA_PAD_SRC,
+						    CCS_STREAM_PIXEL);
+	}
+
 	switch (ctrl->id) {
 	case V4L2_CID_HFLIP:
 	case V4L2_CID_VFLIP:
@@ -673,7 +682,7 @@ static int ccs_set_ctrl(struct v4l2_ctrl *ctrl)
 	case V4L2_CID_VBLANK:
 		exposure = sensor->exposure->val;
 
-		__ccs_update_exposure_limits(sensor);
+		__ccs_update_exposure_limits(sensor, pa_src);
 
 		if (exposure > sensor->exposure->maximum) {
 			sensor->exposure->val =	sensor->exposure->maximum;
@@ -765,12 +774,12 @@ static int ccs_set_ctrl(struct v4l2_ctrl *ctrl)
 		break;
 	case V4L2_CID_VBLANK:
 		rval = ccs_write(sensor, FRAME_LENGTH_LINES,
-				 sensor->pa_src.height + ctrl->val);
+				 pa_src->height + ctrl->val);
 
 		break;
 	case V4L2_CID_HBLANK:
 		rval = ccs_write(sensor, LINE_LENGTH_PCK,
-				 sensor->pa_src.width + ctrl->val);
+				 pa_src->width + ctrl->val);
 
 		break;
 	case V4L2_CID_TEST_PATTERN:
@@ -1225,7 +1234,8 @@ static int ccs_get_mbus_formats(struct ccs_sensor *sensor)
 	return 0;
 }
 
-static void ccs_update_blanking(struct ccs_sensor *sensor)
+static void ccs_update_blanking(struct ccs_sensor *sensor,
+				struct v4l2_rect *pa_src)
 {
 	struct v4l2_ctrl *vblank = sensor->vblank;
 	struct v4l2_ctrl *hblank = sensor->hblank;
@@ -1248,21 +1258,26 @@ static void ccs_update_blanking(struct ccs_sensor *sensor)
 
 	min = max_t(int,
 		    CCS_LIM(sensor, MIN_FRAME_BLANKING_LINES),
-		    min_fll - sensor->pa_src.height);
-	max = max_fll -	sensor->pa_src.height;
+		    min_fll - pa_src->height);
+	max = max_fll -	pa_src->height;
 
 	__v4l2_ctrl_modify_range(vblank, min, max, vblank->step, min);
 
-	min = max_t(int, min_llp - sensor->pa_src.width, min_lbp);
-	max = max_llp - sensor->pa_src.width;
+	min = max_t(int, min_llp - pa_src->width, min_lbp);
+	max = max_llp - pa_src->width;
 
 	__v4l2_ctrl_modify_range(hblank, min, max, hblank->step, min);
 
-	__ccs_update_exposure_limits(sensor);
+	__ccs_update_exposure_limits(sensor, pa_src);
 }
 
 static int ccs_pll_blanking_update(struct ccs_sensor *sensor)
 {
+	struct v4l2_subdev_state *state =
+		v4l2_subdev_get_locked_active_state(&sensor->pixel_array->sd);
+	struct v4l2_rect *pa_src =
+		v4l2_subdev_state_get_crop(state, CCS_PA_PAD_SRC,
+					   CCS_STREAM_PIXEL);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
 	int rval;
 
@@ -1271,15 +1286,15 @@ static int ccs_pll_blanking_update(struct ccs_sensor *sensor)
 		return rval;
 
 	/* Output from pixel array, including blanking */
-	ccs_update_blanking(sensor);
+	ccs_update_blanking(sensor, pa_src);
 
 	dev_dbg(&client->dev, "vblank\t\t%d\n", sensor->vblank->val);
 	dev_dbg(&client->dev, "hblank\t\t%d\n", sensor->hblank->val);
 
 	dev_dbg(&client->dev, "real timeperframe\t100/%d\n",
 		sensor->pll.pixel_rate_pixel_array /
-		((sensor->pa_src.width + sensor->hblank->val) *
-		 (sensor->pa_src.height + sensor->vblank->val) / 100));
+		((pa_src->width + sensor->hblank->val) *
+		 (pa_src->height + sensor->vblank->val) / 100));
 
 	return 0;
 }
@@ -1787,6 +1802,16 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 			      u64 streams_mask)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
+	struct v4l2_subdev_state *pa_state =
+		v4l2_subdev_get_locked_active_state(&sensor->pixel_array->sd);
+	struct v4l2_subdev_state *src_state =
+		v4l2_subdev_get_locked_active_state(&sensor->src->sd);
+	struct v4l2_rect *pa_src =
+		v4l2_subdev_state_get_crop(pa_state, CCS_PA_PAD_SRC,
+					   CCS_STREAM_PIXEL);
+	struct v4l2_rect *src_src =
+		v4l2_subdev_state_get_crop(src_state, CCS_PAD_SRC,
+					   CCS_STREAM_PIXEL);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
 	unsigned int binning_mode;
 	int rval;
@@ -1834,22 +1859,20 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 		goto err_pm_put;
 
 	/* Analog crop start coordinates */
-	rval = ccs_write(sensor, X_ADDR_START, sensor->pa_src.left);
+	rval = ccs_write(sensor, X_ADDR_START, pa_src->left);
 	if (rval < 0)
 		goto err_pm_put;
 
-	rval = ccs_write(sensor, Y_ADDR_START, sensor->pa_src.top);
+	rval = ccs_write(sensor, Y_ADDR_START, pa_src->top);
 	if (rval < 0)
 		goto err_pm_put;
 
 	/* Analog crop end coordinates */
-	rval = ccs_write(sensor, X_ADDR_END,
-			 sensor->pa_src.left + sensor->pa_src.width - 1);
+	rval = ccs_write(sensor, X_ADDR_END, pa_src->left + pa_src->width - 1);
 	if (rval < 0)
 		goto err_pm_put;
 
-	rval = ccs_write(sensor, Y_ADDR_END,
-			 sensor->pa_src.top + sensor->pa_src.height - 1);
+	rval = ccs_write(sensor, Y_ADDR_END, pa_src->top + pa_src->height - 1);
 	if (rval < 0)
 		goto err_pm_put;
 
@@ -1861,23 +1884,30 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 	/* Digital crop */
 	if (CCS_LIM(sensor, DIGITAL_CROP_CAPABILITY)
 	    == CCS_DIGITAL_CROP_CAPABILITY_INPUT_CROP) {
+		struct v4l2_subdev_state *scaler_state =
+			v4l2_subdev_get_locked_active_state(&sensor->scaler->sd);
+		struct v4l2_rect *scaler_sink =
+			v4l2_subdev_state_get_crop(scaler_state,
+						   sensor->scaler->sink_pad,
+						   CCS_STREAM_PIXEL);
+
 		rval = ccs_write(sensor, DIGITAL_CROP_X_OFFSET,
-				 sensor->scaler_sink.left);
+				 scaler_sink->left);
 		if (rval < 0)
 			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_Y_OFFSET,
-				 sensor->scaler_sink.top);
+				 scaler_sink->top);
 		if (rval < 0)
 			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_IMAGE_WIDTH,
-				 sensor->scaler_sink.width);
+				 scaler_sink->width);
 		if (rval < 0)
 			goto err_pm_put;
 
 		rval = ccs_write(sensor, DIGITAL_CROP_IMAGE_HEIGHT,
-				 sensor->scaler_sink.height);
+				 scaler_sink->height);
 		if (rval < 0)
 			goto err_pm_put;
 	}
@@ -1895,10 +1925,10 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 	}
 
 	/* Output size from sensor */
-	rval = ccs_write(sensor, X_OUTPUT_SIZE, sensor->src_src.width);
+	rval = ccs_write(sensor, X_OUTPUT_SIZE, src_src->width);
 	if (rval < 0)
 		goto err_pm_put;
-	rval = ccs_write(sensor, Y_OUTPUT_SIZE, sensor->src_src.height);
+	rval = ccs_write(sensor, Y_OUTPUT_SIZE, src_src->height);
 	if (rval < 0)
 		goto err_pm_put;
 
@@ -2087,9 +2117,6 @@ static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 	unsigned int i;
 	int idx = -1;
-	int rval = -EINVAL;
-
-	mutex_lock(&sensor->mutex);
 
 	dev_err(&client->dev, "subdev %s, pad %u, index %u\n",
 		subdev->name, code->pad, code->index);
@@ -2097,12 +2124,11 @@ static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 	if (subdev == &sensor->src->sd) {
 		if (code->pad == CCS_PAD_META) {
 			if (code->index)
-				goto out;
+				return -EINVAL;
 
 			code->code = MEDIA_BUS_FMT_CCS_EMBEDDED;
 
-			rval = 0;
-			goto out;
+			return 0;
 		}
 		if (code->stream == CCS_STREAM_META) {
 			struct v4l2_mbus_framefmt *pix_fmt =
@@ -2149,21 +2175,21 @@ static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 			}
 
 			if (WARN_ON(i > ARRAY_SIZE(codes)) || code->index >= i)
-				goto out;
+				return -EINVAL;
 
 			code->code = codes[code->index];
-			rval = 0;
-			goto out;
+
+			return 0;
 		}
 	}
 
 	if (subdev != &sensor->src->sd || code->pad != CCS_PAD_SRC) {
 		if (code->index)
-			goto out;
+			return -EINVAL;
 
 		code->code = sensor->internal_csi_format->code;
-		rval = 0;
-		goto out;
+
+		return 0;
 	}
 
 	for (i = 0; i < ARRAY_SIZE(ccs_csi_data_formats); i++) {
@@ -2174,18 +2200,14 @@ static int ccs_enum_mbus_code(struct v4l2_subdev *subdev,
 			code->code = ccs_csi_data_formats[i].code;
 			dev_err(&client->dev, "found index %u, i %u, code %x\n",
 				code->index, i, code->code);
-			rval = 0;
-			break;
+			return 0;
 		}
 	}
 
-out:
-	mutex_unlock(&sensor->mutex);
-
-	return rval;
+	return -EINVAL;
 }
 
-static u32 __ccs_get_mbus_code(struct v4l2_subdev *subdev, unsigned int pad)
+static u32 ccs_get_mbus_code(struct v4l2_subdev *subdev, unsigned int pad)
 {
 	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
 
@@ -2195,33 +2217,19 @@ static u32 __ccs_get_mbus_code(struct v4l2_subdev *subdev, unsigned int pad)
 		return sensor->internal_csi_format->code;
 }
 
-static int __ccs_get_format(struct v4l2_subdev *subdev,
-			    struct v4l2_subdev_state *sd_state,
-			    struct v4l2_subdev_format *fmt)
+static int ccs_get_format(struct v4l2_subdev *subdev,
+			  struct v4l2_subdev_state *sd_state,
+			  struct v4l2_subdev_format *fmt)
 {
 	fmt->format = *v4l2_subdev_state_get_format(sd_state, fmt->pad,
 						    fmt->stream);
 
 	if (fmt->pad != CCS_PAD_META && fmt->stream != CCS_STREAM_META)
-		fmt->format.code = __ccs_get_mbus_code(subdev, fmt->pad);
+		fmt->format.code = ccs_get_mbus_code(subdev, fmt->pad);
 
 	return 0;
 }
 
-static int ccs_get_format(struct v4l2_subdev *subdev,
-			  struct v4l2_subdev_state *sd_state,
-			  struct v4l2_subdev_format *fmt)
-{
-	struct ccs_sensor *sensor = to_ccs_sensor(subdev);
-	int rval;
-
-	mutex_lock(&sensor->mutex);
-	rval = __ccs_get_format(subdev, sd_state, fmt);
-	mutex_unlock(&sensor->mutex);
-
-	return rval;
-}
-
 /* Changes require propagation only on sink pad. */
 static void ccs_propagate(struct v4l2_subdev *subdev,
 			  struct v4l2_subdev_state *sd_state, int which,
@@ -2245,7 +2253,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 				sensor->scale_m = CCS_LIM(sensor, SCALER_N_MIN);
 				sensor->scaling_mode =
 					CCS_SCALING_MODE_NO_SCALING;
-				sensor->scaler_sink = *comp;
 			} else if (ssd == sensor->binner) {
 				sensor->binning_horizontal = 1;
 				sensor->binning_vertical = 1;
@@ -2260,8 +2267,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 						   CCS_STREAM_PIXEL);
 		fmt->width = comp->width;
 		fmt->height = comp->height;
-		if (which == V4L2_SUBDEV_FORMAT_ACTIVE && ssd == sensor->src)
-			sensor->src_src = *crop;
 		break;
 	default:
 		WARN_ON_ONCE(1);
@@ -2280,7 +2285,7 @@ static int ccs_set_format_source(struct v4l2_subdev *subdev,
 	unsigned int i;
 	int rval;
 
-	rval = __ccs_get_format(subdev, sd_state, fmt);
+	rval = ccs_get_format(subdev, sd_state, fmt);
 	if (rval)
 		return rval;
 
@@ -2407,13 +2412,9 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 	if (subdev == &sensor->src->sd && fmt->pad == CCS_PAD_META)
 		return ccs_get_format(subdev, sd_state, fmt);
 
-	mutex_lock(&sensor->mutex);
-
 	if (subdev == &sensor->src->sd && fmt->stream == CCS_STREAM_META) {
 		ccs_set_format_meta(subdev, sd_state, &fmt->format);
 
-		mutex_unlock(&sensor->mutex);
-
 		return 0;
 	}
 
@@ -2424,13 +2425,12 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 		if (sensor->embedded_start != sensor->embedded_end)
 			ccs_set_format_meta(subdev, sd_state, NULL);
 
-		mutex_unlock(&sensor->mutex);
-
 		return rval;
 	}
 
 	/* Sink pad. Width and height are changeable here. */
-	fmt->format.code = __ccs_get_mbus_code(subdev, fmt->pad);
+	fmt->format.code = ccs_get_mbus_code(subdev, fmt->pad);
+
 	fmt->format.width &= ~1;
 	fmt->format.height &= ~1;
 	fmt->format.field = V4L2_FIELD_NONE;
@@ -2453,8 +2453,6 @@ static int ccs_set_format(struct v4l2_subdev *subdev,
 	crop->height = fmt->format.height;
 	ccs_propagate(subdev, sd_state, fmt->which, V4L2_SEL_TGT_CROP);
 
-	mutex_unlock(&sensor->mutex);
-
 	return 0;
 }
 
@@ -2768,9 +2766,6 @@ static int ccs_set_crop(struct v4l2_subdev *subdev,
 
 	if (ssd != sensor->pixel_array && sel->pad == CCS_PAD_SINK)
 		ccs_propagate(subdev, sd_state, sel->which, V4L2_SEL_TGT_CROP);
-	else if (sel->which == V4L2_SUBDEV_FORMAT_ACTIVE &&
-		 ssd == sensor->pixel_array)
-		sensor->pa_src = sel->r;
 
 	return 0;
 }
@@ -2842,8 +2837,6 @@ static int ccs_set_selection(struct v4l2_subdev *subdev,
 	if (ret)
 		return ret;
 
-	mutex_lock(&sensor->mutex);
-
 	sel->r.left = max(0, sel->r.left & ~1);
 	sel->r.top = max(0, sel->r.top & ~1);
 	sel->r.width = CCS_ALIGN_DIM(sel->r.width, sel->flags);
@@ -2865,7 +2858,6 @@ static int ccs_set_selection(struct v4l2_subdev *subdev,
 		ret = -EINVAL;
 	}
 
-	mutex_unlock(&sensor->mutex);
 	return ret;
 }
 
@@ -3244,6 +3236,7 @@ static int ccs_init_subdev(struct ccs_sensor *sensor,
 
 	ssd->sd.flags |= V4L2_SUBDEV_FL_HAS_DEVNODE;
 	ssd->sd.entity.function = function;
+	ssd->sd.state_lock = &sensor->mutex;
 	ssd->sensor = sensor;
 
 	ssd->npads = num_pads;
@@ -3283,8 +3276,8 @@ static int ccs_init_subdev(struct ccs_sensor *sensor,
 	return 0;
 }
 
-static int __ccs_init_state(struct v4l2_subdev *sd,
-			    struct v4l2_subdev_state *sd_state)
+static int ccs_init_state(struct v4l2_subdev *sd,
+			  struct v4l2_subdev_state *sd_state)
 {
 	struct ccs_subdev *ssd = to_ccs_subdev(sd);
 	struct ccs_sensor *sensor = ssd->sensor;
@@ -3303,12 +3296,8 @@ static int __ccs_init_state(struct v4l2_subdev *sd,
 	fmt->code = sensor->internal_csi_format->code;
 	fmt->field = V4L2_FIELD_NONE;
 
-	if (ssd == sensor->pixel_array) {
-		if (is_active)
-			sensor->pa_src = *crop;
-
+	if (ssd == sensor->pixel_array)
 		return 0;
-	}
 
 	fmt = v4l2_subdev_state_get_format(sd_state, CCS_PAD_SRC,
 					   CCS_STREAM_PIXEL);
@@ -3321,20 +3310,6 @@ static int __ccs_init_state(struct v4l2_subdev *sd,
 	return 0;
 }
 
-static int ccs_init_state(struct v4l2_subdev *sd,
-			  struct v4l2_subdev_state *sd_state)
-{
-	struct ccs_subdev *ssd = to_ccs_subdev(sd);
-	struct ccs_sensor *sensor = ssd->sensor;
-	int rval;
-
-	mutex_lock(&sensor->mutex);
-	rval = __ccs_init_state(sd, sd_state);
-	mutex_unlock(&sensor->mutex);
-
-	return rval;
-}
-
 static int ccs_src_init_state(struct v4l2_subdev *sd,
 			      struct v4l2_subdev_state *sd_state)
 {
@@ -3359,25 +3334,20 @@ static int ccs_src_init_state(struct v4l2_subdev *sd,
 	struct ccs_sensor *sensor = ssd->sensor;
 	int rval;
 
-	mutex_lock(&sensor->mutex);
-
 	if (sensor->embedded_start != sensor->embedded_end)
 		routing.num_routes++;
 
 	rval = v4l2_subdev_set_routing(sd, sd_state, &routing);
 	if (rval)
-		goto out;
+		return 0;
 
-	rval = __ccs_init_state(sd, sd_state);
+	rval = ccs_init_state(sd, sd_state);
 	if (rval)
-		goto out;
+		return 0;
 
 	if (sensor->embedded_start != sensor->embedded_end)
 		ccs_set_format_meta(sd, sd_state, NULL);
 
-out:
-	mutex_unlock(&sensor->mutex);
-
 	return rval;
 }
 
drivers/media/i2c/ccs/ccs.h +0 -1
index 207d33de3eee..a2a461cda338 100644
@@ -230,7 +230,6 @@ struct ccs_sensor {
 	u32 mbus_frame_fmts;
 	const struct ccs_csi_data_format *csi_format;
 	const struct ccs_csi_data_format *internal_csi_format;
-	struct v4l2_rect pa_src, scaler_sink, src_src;
 	u32 default_mbus_frame_fmts;
 	int default_pixel_order;
 	struct ccs_data_container sdata, mdata;
Signature · 1 line
2.39.2
author maintainer
#107 lore raw

PATCH v8 27/38 media: ccs: Compute binning configuration from sub-device state

In reply to Sakari Ailus

Calculate binning configuration from sub-device state so the state related configuration can be removed from the driver's device context struct.

---
 drivers/media/i2c/ccs/ccs-core.c | 72 ++++++++++++++++++++++----------
 drivers/media/i2c/ccs/ccs.h      |  3 --
 2 files changed, 49 insertions(+), 26 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +49 -23
index a8c48abd2e30..3b80c54453cc 100644
@@ -511,13 +511,52 @@ static int ccs_pll_try(struct ccs_sensor *sensor, struct ccs_pll *pll)
 	return ccs_pll_calculate(&client->dev, &lim, pll);
 }
 
+static void
+ccs_get_binning(struct ccs_sensor *sensor, u8 *binning_mode, u8 *binh, u8 *binv)
+{
+	struct v4l2_subdev_state *binner_state =
+		v4l2_subdev_get_locked_active_state(&sensor->binner->sd);
+	struct v4l2_rect *binner_sink_crop =
+		v4l2_subdev_state_get_crop(binner_state, CCS_PAD_SINK,
+					   CCS_STREAM_PIXEL);
+	struct v4l2_rect *binner_sink_comp =
+		v4l2_subdev_state_get_compose(binner_state, CCS_PAD_SINK,
+					      CCS_STREAM_PIXEL);
+
+	if (binner_sink_crop->width == binner_sink_comp->width &&
+	    binner_sink_crop->height == binner_sink_comp->height) {
+		if (binning_mode)
+			*binning_mode = 0;
+
+		if (binh)
+			*binh = 1;
+
+		if (binv)
+			*binv = 1;
+
+		return;
+	}
+
+	if (binning_mode)
+		*binning_mode = 1;
+
+	if (binh)
+		*binh = binner_sink_crop->width / binner_sink_comp->width;
+
+	if (binv)
+		*binv = binner_sink_crop->height / binner_sink_comp->height;
+}
+
 static int ccs_pll_update(struct ccs_sensor *sensor)
 {
 	struct ccs_pll *pll = &sensor->pll;
+	u8 binh, binv;
 	int rval;
 
-	pll->binning_horizontal = sensor->binning_horizontal;
-	pll->binning_vertical = sensor->binning_vertical;
+	ccs_get_binning(sensor, NULL, &binh, &binv);
+
+	pll->binning_horizontal = binh;
+	pll->binning_vertical = binv;
 	pll->link_freq =
 		sensor->link_freq->qmenu_int[sensor->link_freq->val];
 	pll->scale_m = sensor->scale_m;
@@ -1241,8 +1280,11 @@ static void ccs_update_blanking(struct ccs_sensor *sensor,
 	struct v4l2_ctrl *hblank = sensor->hblank;
 	u16 min_fll, max_fll, min_llp, max_llp, min_lbp;
 	int min, max;
+	u8 binh, binv;
+
+	ccs_get_binning(sensor, NULL, &binh, &binv);
 
-	if (sensor->binning_vertical > 1 || sensor->binning_horizontal > 1) {
+	if (binv > 1 || binh > 1) {
 		min_fll = CCS_LIM(sensor, MIN_FRAME_LENGTH_LINES_BIN);
 		max_fll = CCS_LIM(sensor, MAX_FRAME_LENGTH_LINES_BIN);
 		min_llp = CCS_LIM(sensor, MIN_LINE_LENGTH_PCK_BIN);
@@ -1813,7 +1855,7 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 		v4l2_subdev_state_get_crop(src_state, CCS_PAD_SRC,
 					   CCS_STREAM_PIXEL);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
-	unsigned int binning_mode;
+	u8 binning_mode, binh, binv;
 	int rval;
 
 	if (pad != CCS_PAD_SRC)
@@ -1835,19 +1877,12 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 		goto err_pm_put;
 
 	/* Binning configuration */
-	if (sensor->binning_horizontal == 1 &&
-	    sensor->binning_vertical == 1) {
-		binning_mode = 0;
-	} else {
-		u8 binning_type =
-			(sensor->binning_horizontal << 4)
-			| sensor->binning_vertical;
+	ccs_get_binning(sensor,	&binning_mode, &binh, &binv);
 
-		rval = ccs_write(sensor, BINNING_TYPE, binning_type);
+	if (binning_mode) {
+		rval = ccs_write(sensor, BINNING_TYPE, (binh << 4) | binv);
 		if (rval < 0)
 			goto err_pm_put;
-
-		binning_mode = 1;
 	}
 	rval = ccs_write(sensor, BINNING_MODE, binning_mode);
 	if (rval < 0)
@@ -2253,9 +2288,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 				sensor->scale_m = CCS_LIM(sensor, SCALER_N_MIN);
 				sensor->scaling_mode =
 					CCS_SCALING_MODE_NO_SCALING;
-			} else if (ssd == sensor->binner) {
-				sensor->binning_horizontal = 1;
-				sensor->binning_vertical = 1;
 			}
 		}
 		fallthrough;
@@ -2529,10 +2561,6 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 			best = this;
 		}
 	}
-	if (sel->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
-		sensor->binning_vertical = binv;
-		sensor->binning_horizontal = binh;
-	}
 
 	sel->r.width = (sink_crop->width / binh) & ~1;
 	sel->r.height = (sink_crop->height / binv) & ~1;
@@ -3721,8 +3749,6 @@ static int ccs_probe(struct i2c_client *client)
 				sensor->binning_subtypes[i].vertical);
 		}
 	}
-	sensor->binning_horizontal = 1;
-	sensor->binning_vertical = 1;
 
 	if (device_create_file(&client->dev, &dev_attr_ident) != 0) {
 		dev_err(&client->dev, "sysfs ident entry creation failed\n");
drivers/media/i2c/ccs/ccs.h +0 -3
index a2a461cda338..e6fc00a9fa11 100644
@@ -237,9 +237,6 @@ struct ccs_sensor {
 	u32 embedded_mbus_code;
 	u8 emb_data_ctrl;
 
-	u8 binning_horizontal;
-	u8 binning_vertical;
-
 	u8 scale_m;
 	u8 scaling_mode;
 
Signature · 1 line
2.39.2
maintainer
#108 lore raw

Re: [PATCH v8 27/38] media: ccs: Compute binning configuration from sub-device state

Hi Sakari,

Thank you for the patch.

Quoted 18 lines · Sakari Ailus

Calculate binning configuration from sub-device state so the state related configuration can be removed from the driver's device context struct.

---
 drivers/media/i2c/ccs/ccs-core.c | 72 ++++++++++++++++++++++----------
 drivers/media/i2c/ccs/ccs.h      |  3 --
 2 files changed, 49 insertions(+), 26 deletions(-)
drivers/media/i2c/ccs/ccs-core.c +1 -0
index a8c48abd2e30..3b80c54453cc 100644
@@ -511,13 +511,52 @@ static int ccs_pll_try(struct ccs_sensor *sensor, struct ccs_pll *pll)
 	return ccs_pll_calculate(&client->dev, &lim, pll);
 }
 
+static void

drivers/media/i2c/ccs/ccs-core.c static int ccs_pll_try(struct ccs_sensor *sensor, struct ccs_pll *pll) · line 524

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +10 -0
+ccs_get_binning(struct ccs_sensor *sensor, u8 *binning_mode, u8 *binh, u8 *binv)
+{
+	struct v4l2_subdev_state *binner_state =
+		v4l2_subdev_get_locked_active_state(&sensor->binner->sd);
+	struct v4l2_rect *binner_sink_crop =
+		v4l2_subdev_state_get_crop(binner_state, CCS_PAD_SINK,
+					   CCS_STREAM_PIXEL);
+	struct v4l2_rect *binner_sink_comp =
+		v4l2_subdev_state_get_compose(binner_state, CCS_PAD_SINK,
+					      CCS_STREAM_PIXEL);

I like this patch very much :-) The comments I made to 28/28 apply there too.

Quoted 13 lines
drivers/media/i2c/ccs/ccs-core.c +13 -0
+
+	if (binner_sink_crop->width == binner_sink_comp->width &&
+	    binner_sink_crop->height == binner_sink_comp->height) {
+		if (binning_mode)
+			*binning_mode = 0;
+
+		if (binh)
+			*binh = 1;
+
+		if (binv)
+			*binv = 1;
+
+		return;

drivers/media/i2c/ccs/ccs-core.c static int ccs_pll_try(struct ccs_sensor *sensor, struct ccs_pll *pll) · line 547

Quoted 10 lines
drivers/media/i2c/ccs/ccs-core.c +10 -0
+	}
+
+	if (binning_mode)
+		*binning_mode = 1;
+
+	if (binh)
+		*binh = binner_sink_crop->width / binner_sink_comp->width;
+
+	if (binv)
+		*binv = binner_sink_crop->height / binner_sink_comp->height;

binh and binv are never NULL. How about the following ?

	if (binning_mode)
		*binning_mode = sink_crop->width != sink_comp->width ||
				sink_crop->height != sink_comp->height;

	*binh = sink_crop->width / sink_comp->width;
	*binv = sink_crop->height / sink_comp->height;
Quoted 106 lines
drivers/media/i2c/ccs/ccs-core.c +15 -23
+}
+
 static int ccs_pll_update(struct ccs_sensor *sensor)
 {
 	struct ccs_pll *pll = &sensor->pll;
+	u8 binh, binv;
 	int rval;
 
-	pll->binning_horizontal = sensor->binning_horizontal;
-	pll->binning_vertical = sensor->binning_vertical;
+	ccs_get_binning(sensor, NULL, &binh, &binv);
+
+	pll->binning_horizontal = binh;
+	pll->binning_vertical = binv;
 	pll->link_freq =
 		sensor->link_freq->qmenu_int[sensor->link_freq->val];
 	pll->scale_m = sensor->scale_m;
@@ -1241,8 +1280,11 @@ static void ccs_update_blanking(struct ccs_sensor *sensor,
 	struct v4l2_ctrl *hblank = sensor->hblank;
 	u16 min_fll, max_fll, min_llp, max_llp, min_lbp;
 	int min, max;
+	u8 binh, binv;
+
+	ccs_get_binning(sensor, NULL, &binh, &binv);
 
-	if (sensor->binning_vertical > 1 || sensor->binning_horizontal > 1) {
+	if (binv > 1 || binh > 1) {
 		min_fll = CCS_LIM(sensor, MIN_FRAME_LENGTH_LINES_BIN);
 		max_fll = CCS_LIM(sensor, MAX_FRAME_LENGTH_LINES_BIN);
 		min_llp = CCS_LIM(sensor, MIN_LINE_LENGTH_PCK_BIN);
@@ -1813,7 +1855,7 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 		v4l2_subdev_state_get_crop(src_state, CCS_PAD_SRC,
 					   CCS_STREAM_PIXEL);
 	struct i2c_client *client = v4l2_get_subdevdata(&sensor->src->sd);
-	unsigned int binning_mode;
+	u8 binning_mode, binh, binv;
 	int rval;
 
 	if (pad != CCS_PAD_SRC)
@@ -1835,19 +1877,12 @@ static int ccs_enable_streams(struct v4l2_subdev *subdev,
 		goto err_pm_put;
 
 	/* Binning configuration */
-	if (sensor->binning_horizontal == 1 &&
-	    sensor->binning_vertical == 1) {
-		binning_mode = 0;
-	} else {
-		u8 binning_type =
-			(sensor->binning_horizontal << 4)
-			| sensor->binning_vertical;
+	ccs_get_binning(sensor,	&binning_mode, &binh, &binv);
 
-		rval = ccs_write(sensor, BINNING_TYPE, binning_type);
+	if (binning_mode) {
+		rval = ccs_write(sensor, BINNING_TYPE, (binh << 4) | binv);
 		if (rval < 0)
 			goto err_pm_put;
-
-		binning_mode = 1;
 	}
 	rval = ccs_write(sensor, BINNING_MODE, binning_mode);
 	if (rval < 0)
@@ -2253,9 +2288,6 @@ static void ccs_propagate(struct v4l2_subdev *subdev,
 				sensor->scale_m = CCS_LIM(sensor, SCALER_N_MIN);
 				sensor->scaling_mode =
 					CCS_SCALING_MODE_NO_SCALING;
-			} else if (ssd == sensor->binner) {
-				sensor->binning_horizontal = 1;
-				sensor->binning_vertical = 1;
 			}
 		}
 		fallthrough;
@@ -2529,10 +2561,6 @@ static void ccs_set_compose_binner(struct v4l2_subdev *subdev,
 			best = this;
 		}
 	}
-	if (sel->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
-		sensor->binning_vertical = binv;
-		sensor->binning_horizontal = binh;
-	}
 
 	sel->r.width = (sink_crop->width / binh) & ~1;
 	sel->r.height = (sink_crop->height / binv) & ~1;
@@ -3721,8 +3749,6 @@ static int ccs_probe(struct i2c_client *client)
 				sensor->binning_subtypes[i].vertical);
 		}
 	}
-	sensor->binning_horizontal = 1;
-	sensor->binning_vertical = 1;
 
 	if (device_create_file(&client->dev, &dev_attr_ident) != 0) {
 		dev_err(&client->dev, "sysfs ident entry creation failed\n");
drivers/media/i2c/ccs/ccs.h +0 -3
index a2a461cda338..e6fc00a9fa11 100644
@@ -237,9 +237,6 @@ struct ccs_sensor {
 	u32 embedded_mbus_code;
 	u8 emb_data_ctrl;
 
-	u8 binning_horizontal;
-	u8 binning_vertical;
-
 	u8 scale_m;
 	u8 scaling_mode;
 
Signature · 3 lines
Regards,

Laurent Pinchart