Re: [PATCH v2 1/7] media: Add 2-10-10-10 RGB formats

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 20/12/2022 16:24, Laurent Pinchart wrote:
Hi Tomi,

On Tue, Dec 20, 2022 at 04:12:29PM +0200, Tomi Valkeinen wrote:
On 19/12/2022 21:10, Laurent Pinchart wrote:
On Mon, Dec 19, 2022 at 04:01:33PM +0200, Tomi Valkeinen wrote:
Add XBGR2101010, ABGR2101010 and BGRA1010102 formats.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen+renesas@xxxxxxxxxxxxxxxx>
---
   .../userspace-api/media/v4l/pixfmt-rgb.rst    | 194 ++++++++++++++++++
   drivers/media/v4l2-core/v4l2-ioctl.c          |   3 +
   include/uapi/linux/videodev2.h                |   3 +
   3 files changed, 200 insertions(+)

diff --git a/Documentation/userspace-api/media/v4l/pixfmt-rgb.rst b/Documentation/userspace-api/media/v4l/pixfmt-rgb.rst
index 30f51cd33f99..de78cd2dcd73 100644
--- a/Documentation/userspace-api/media/v4l/pixfmt-rgb.rst
+++ b/Documentation/userspace-api/media/v4l/pixfmt-rgb.rst
@@ -763,6 +763,200 @@ nomenclature that instead use the order of components as seen in a 24- or
       \normalsize
+10 Bits Per Component
+=====================
+
+These formats store a 30-bit RGB triplet with an optional 2 bit alpha in four
+bytes. They are named based on the order of the RGB components as seen in a
+32-bit word, which is then stored in memory in little endian byte order
+(unless otherwise noted by the presence of bit 31 in the 4CC value), and on the
+number of bits for each component.
+
+.. raw:: latex
+
+    \begingroup
+    \tiny
+    \setlength{\tabcolsep}{2pt}
+
+.. tabularcolumns:: |p{2.8cm}|p{2.0cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|p{0.22cm}|
+
+
+.. flat-table:: RGB Formats 10 Bits Per Color Component
+    :header-rows:  2
+    :stub-columns: 0
+
+    * - Identifier
+      - Code
+      - :cspan:`7` Byte 0 in memory
+      - :cspan:`7` Byte 1
+      - :cspan:`7` Byte 2
+      - :cspan:`7` Byte 3
+    * -
+      -
+      - 7
+      - 6
+      - 5
+      - 4
+      - 3
+      - 2
+      - 1
+      - 0
+
+      - 7
+      - 6
+      - 5
+      - 4
+      - 3
+      - 2
+      - 1
+      - 0
+
+      - 7
+      - 6
+      - 5
+      - 4
+      - 3
+      - 2
+      - 1
+      - 0
+
+      - 7
+      - 6
+      - 5
+      - 4
+      - 3
+      - 2
+      - 1
+      - 0
+    * .. _V4L2-PIX-FMT-XBGR2101010:
+
+      - ``V4L2_PIX_FMT_XBGR2101010``
+      - 'RX30'
+
+      - b\ :sub:`5`
+      - b\ :sub:`4`
+      - b\ :sub:`3`
+      - b\ :sub:`2`
+      - b\ :sub:`1`
+      - b\ :sub:`0`
+      - x
+      - x
+
+      - g\ :sub:`3`
+      - g\ :sub:`2`
+      - g\ :sub:`1`
+      - g\ :sub:`0`
+      - b\ :sub:`9`
+      - b\ :sub:`8`
+      - b\ :sub:`7`
+      - b\ :sub:`6`
+
+      - r\ :sub:`1`
+      - r\ :sub:`0`
+      - g\ :sub:`9`
+      - g\ :sub:`8`
+      - g\ :sub:`7`
+      - g\ :sub:`6`
+      - g\ :sub:`5`
+      - g\ :sub:`4`
+
+      - r\ :sub:`9`
+      - r\ :sub:`8`
+      - r\ :sub:`7`
+      - r\ :sub:`6`
+      - r\ :sub:`5`
+      - r\ :sub:`4`
+      - r\ :sub:`3`
+      - r\ :sub:`2`
+      -

This doesn't match the text above. This would be RGBX2101010. I'm not
sure which format you want, so I don't know if it's the documentation or
the format name that is incorrect. The next two formats also seem
incorrect to me.

Right, the text should say big endian instead of little endian.

No, in big-endian format, you would have, for instance,
V4L2_PIX_FMT_XBGR2101010 defined as

	[x, x, B[9:4]], [B[3:0], G[9:6]], [G[5:0], R[1:0]], [R[7:0]]

in memory byte order, while the format you want to define is

	[B[5:0], x, x], [G[3:0], B[9:6]], [R[1:0], G[9:4]], [R[9:2]]

Yes, I see your point.

The issue here is that 10-bpp formats don't have an integer number of
bytes per component. They are thus more similar to the 16-bit RGB
formats, where the macro named defined the order in a 16-bit word, which
was then stored in little-endian format in memory. For 24-bit and 32-bit
formats, we departed from that rule by using the byte memory order in
the macro name. For 10-bpp RGB formats we can't do so anymore. The most
sensible option is thus, I think, to use the same naming scheme as the
16-bit RGB formats, which incidentaly matches the DRM naming scheme.

I agree. It wasn't quite clear to me if Hans agreed to that in the patch 7 discussions, but as you say, maybe that's the only option that makes sense.

 Tomi




[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux