Re: [PATCH v7 00/11] VP9 codec V4L2 control interface

From: Andrzej Pietrasiewicz
Date: Mon Nov 15 2021 - 07:57:36 EST

Next message: Nikolay Aleksandrov: "Re: [PATCH] net: bridge: Slightly optimize 'find_portno()'"
Previous message: Jason Gunthorpe: "Re: [PATCH net-next] devlink: Require devlink lock during device reload"
In reply to: Nicolas Dufresne: "Re: [PATCH v7 00/11] VP9 codec V4L2 control interface"
Next in thread: Andrzej Pietrasiewicz: "Re: [PATCH v7 00/11] VP9 codec V4L2 control interface"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi Hans,

W dniu 12.11.2021 o 16:27, Nicolas Dufresne pisze:

Hi Hans,

Le jeudi 11 novembre 2021 à 15:44 +0100, Hans Verkuil a écrit :

Hi all,

Andrzej, Jernej, Nicolas, if none of you (or anyone else for that matter)
objects, then I'll make a PR for this early next week.

I have no objection. I've myself delayed replying as we have been digging a lot
into our compliance failures, but I believe we have explained most of them by
now and nothing seems to be related to the API.

regards,
Nicolas

I'm fine with making a PR, too.

Andrzej

Regards,

Hans

On 29/09/2021 18:04, Andrzej Pietrasiewicz wrote:

Dear all,

This patch series adds VP9 codec V4L2 control interface and two drivers
using the new controls. It is a follow-up of previous v6 series [1].

In this iteration, we've implemented VP9 hardware decoding on two devices:
Rockchip VDEC and Hantro G2, and tested on RK3399, i.MX8MQ and i.MX8MP.
The i.MX8M driver needs proper power domains support, though, which is a
subject of a different effort, but in all 3 cases we were able to run the
drivers.

GStreamer support is also available, the needed changes have been submitted
by Daniel Almeida [2]. This MR is ready to be merged, and just needs the
VP9 V4L2 controls to be merged and released.

Both rkvdec and hantro drivers are passing a significant number of VP9 tests
using Fluster[3]. There are still a few tests that are not passing, due to
dynamic frame resize (not yet supported by V4L2) and small size videos
(due to IP block limitations).

The series adds the VP9 codec V4L2 control API as uAPI, so it aims at being
merged without passing through staging, as agreed[4]. The ABI has been checked
for padding and verified to contain no holes.

[1] https://patchwork.linuxtv.org/project/linux-media/list/?series=6377
[2] https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/2144
[3] https://github.com/fluendo/fluster
[4] https://lore.kernel.org/linux-media/b8f83c93-67fd-09f5-9314-15746cbfdc61@xxxxxxxxx/

The series depends on the YUV tiled format support prepared by Ezequiel:
https://www.spinics.net/lists/linux-media/msg197047.html

Rebased onto latest media_tree.

Changes related to v6:
- moved setting tile filter and tile bsd auxiliary buffer addresses so
that they are always set, even if no tiles are used (thanks, Jernej)
- added a comment near the place where the 32-bit DMA mask is applied
(thanks, Nicolas)
- improved consistency in register names (thanks, Nicolas)

Changes related to v5:
- improved the doc comments as per Ezequiel's review (thanks, Ezequiel)
- improved pdf output of documentation
- added Benjamin's Reviewed-by (thanks, Benjamin)

Changes related to v4:
- removed unused enum v4l2_vp9_intra_prediction_mode
- converted remaining enums to defines to follow the convention
- improved the documentation, in particular better documented how to use segmentation
features

Changes related to v3:

Apply suggestions from Jernej's review (thanks, Jernej):
- renamed a control and two structs:
V4L2_CTRL_TYPE_VP9_COMPRESSED_HDR_PROBS =>
V4L2_CTRL_TYPE_VP9_COMPRESSED_HDR
v4l2_ctrl_vp9_compressed_hdr_probs =>
v4l2_ctrl_vp9_compressed_hdr
v4l2_vp9_mv_compressed_hdr_probs => v4l2_vp9_mv_probs
- moved tx_mode to v4l2_ctrl_vp9_compressed_hdr
- fixed enum v4l2_vp9_ref_frame_sign_bias values (which are used to test a bitfield)
- explicitly assigned values to all other vp9 enums

Apply suggestion from Nicolas's review (thanks, Nicolas):
- explicitly stated that the v4l2_ctrl_vp9_compressed_hdr control is optional
and implemented only by drivers which need it

Changes related to the RFC v2:

- added another driver including a postprocessor to de-tile
codec-specific tiling
- reworked uAPI structs layout to follow VP8 style
- changed validation of loop filter params
- changed validation of segmentation params
- changed validation of VP9 frame params
- removed level lookup array from loop filter struct
(can be computed by drivers)
- renamed some enum values to match the spec more closely
- V4L2 VP9 library changed the 'eob' member of
'struct v4l2_vp9_frame_symbol_counts' so that it is an array
of pointers instead of an array of pointers to arrays
(IPs such as g2 creatively pass parts of the 'eob' counts in
the 'coeff' counts)
- factored out several repeated portions of code
- minor nitpicks and cleanups

Andrzej Pietrasiewicz (6):
media: uapi: Add VP9 stateless decoder controls
media: Add VP9 v4l2 library
media: hantro: Rename registers
media: hantro: Prepare for other G2 codecs
media: hantro: Support VP9 on the G2 core
media: hantro: Support NV12 on the G2 core

Boris Brezillon (1):
media: rkvdec: Add the VP9 backend

Ezequiel Garcia (4):
hantro: postproc: Fix motion vector space size
hantro: postproc: Introduce struct hantro_postproc_ops
hantro: Simplify postprocessor
hantro: Add quirk for NV12/NV12_4L4 capture format

.../userspace-api/media/v4l/biblio.rst | 10 +
.../media/v4l/ext-ctrls-codec-stateless.rst | 573 +++++
.../media/v4l/pixfmt-compressed.rst | 15 +
.../media/v4l/vidioc-g-ext-ctrls.rst | 8 +
.../media/v4l/vidioc-queryctrl.rst | 12 +
.../media/videodev2.h.rst.exceptions | 2 +
drivers/media/v4l2-core/Kconfig | 4 +
drivers/media/v4l2-core/Makefile | 1 +
drivers/media/v4l2-core/v4l2-ctrls-core.c | 180 ++
drivers/media/v4l2-core/v4l2-ctrls-defs.c | 8 +
drivers/media/v4l2-core/v4l2-ioctl.c | 1 +
drivers/media/v4l2-core/v4l2-vp9.c | 1850 +++++++++++++++++
drivers/staging/media/hantro/Kconfig | 1 +
drivers/staging/media/hantro/Makefile | 7 +-
drivers/staging/media/hantro/hantro.h | 40 +-
drivers/staging/media/hantro/hantro_drv.c | 23 +-
drivers/staging/media/hantro/hantro_g2.c | 27 +
.../staging/media/hantro/hantro_g2_hevc_dec.c | 69 +-
drivers/staging/media/hantro/hantro_g2_regs.h | 132 +-
.../staging/media/hantro/hantro_g2_vp9_dec.c | 980 +++++++++
drivers/staging/media/hantro/hantro_hw.h | 83 +-
.../staging/media/hantro/hantro_postproc.c | 79 +-
drivers/staging/media/hantro/hantro_v4l2.c | 20 +
drivers/staging/media/hantro/hantro_vp9.c | 240 +++
drivers/staging/media/hantro/hantro_vp9.h | 103 +
drivers/staging/media/hantro/imx8m_vpu_hw.c | 38 +-
.../staging/media/hantro/rockchip_vpu_hw.c | 7 +-
.../staging/media/hantro/sama5d4_vdec_hw.c | 3 +-
drivers/staging/media/rkvdec/Kconfig | 1 +
drivers/staging/media/rkvdec/Makefile | 2 +-
drivers/staging/media/rkvdec/rkvdec-vp9.c | 1078 ++++++++++
drivers/staging/media/rkvdec/rkvdec.c | 52 +-
drivers/staging/media/rkvdec/rkvdec.h | 12 +-
include/media/v4l2-ctrls.h | 4 +
include/media/v4l2-vp9.h | 182 ++
include/uapi/linux/v4l2-controls.h | 284 +++
include/uapi/linux/videodev2.h | 6 +
37 files changed, 6033 insertions(+), 104 deletions(-)
create mode 100644 drivers/media/v4l2-core/v4l2-vp9.c
create mode 100644 drivers/staging/media/hantro/hantro_g2.c
create mode 100644 drivers/staging/media/hantro/hantro_g2_vp9_dec.c
create mode 100644 drivers/staging/media/hantro/hantro_vp9.c
create mode 100644 drivers/staging/media/hantro/hantro_vp9.h
create mode 100644 drivers/staging/media/rkvdec/rkvdec-vp9.c
create mode 100644 include/media/v4l2-vp9.h

base-commit: e4e737bb5c170df6135a127739a9e6148ee3da82

Next message: Nikolay Aleksandrov: "Re: [PATCH] net: bridge: Slightly optimize 'find_portno()'"
Previous message: Jason Gunthorpe: "Re: [PATCH net-next] devlink: Require devlink lock during device reload"
In reply to: Nicolas Dufresne: "Re: [PATCH v7 00/11] VP9 codec V4L2 control interface"
Next in thread: Andrzej Pietrasiewicz: "Re: [PATCH v7 00/11] VP9 codec V4L2 control interface"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]