Re: [PATCH 0/3] Add Amlogic stateless H.264 video decoder for S4

From: Zhentao Guo

Date: Tue Oct 28 2025 - 07:08:58 EST



在 2025/10/27 20:57, Christian Hewitt 写道:
[christianshewitt@xxxxxxxxx appears similar to someone who previously sent you email, but may not be that person. Learn why this could be a risk at https://aka.ms/LearnAboutSenderIdentification ]

[ EXTERNAL EMAIL ]

On 27 Oct 2025, at 9:42 am, Zhentao Guo via B4 Relay <devnull+zhentao.guo.amlogic.com@xxxxxxxxxx> wrote:

Introduce initial driver support for Amlogic's new video acceleration
hardware architecture, designed for video stream decoding.

Compared to the current Amlogic video decoder hardware architecture,
this new implementation eliminates the Esparser hardware component,
enabling direct vb2 buffer input. The driver is designed to support
the V4L2 M2M stateless decoder API. The initial phase includes support
for H.264 decoding on Amlogic S805X2 platform.
It would be nice if you can outline the additional hardware the driver
plans to support so reviewers have perspective on where we are now, and
what more be coming? - the community would also like to understand if
this driver will supersede the stalled driver attempt in staging.
Yes, we will follow Amlogic's overall upstream progress to support more hardware
platforms in the future. The next step will be adding support for A311D2(T7) platform.
We also plan to replace the stalled driver with the new stateless decoder driver.
The new stateless driver is fully compatible with the old G12/SM1 platforms, making
this transition feasible.
NB: I can only find two S805X2 devices to purchase for testing. The
first is a Mecool KD5 HDMI stick, but this has Widevine L1 certs so
will be unusable due to verified boot. The second is a no-name TV box
on Aliexpress where the advert claims S805X2 inside but also shows a
prominent Rockchip RK3228 logo :) .. please advise on some accessible
hardware that can be used for validating the driver.
I'm not familiar with the two S805X2 devices you mentioned. Maybe we can
post you a S805X2 reference board for testing. I will talk to my colleagues
about it and follow up.
The driver is capable of:
- Supporting stateless H.264 decoding up to a resolution 1920x1088(on the S805X2 platform).
It would be good to provide guidance to reviewers about other codecs
the driver plans to support and how (architecturally) the driver will
be expanded. I’m aware due to off-list discussions, but others are
not and will want to know.
Thanks for the suggestion. Our roadmap is to implement support for five
codecs―H.264, H.265, VP9, AV1, and MPEG-2―in phases. This patch is the
initial phase of development, including the V4L2 interface layer and
the H.264 driver code. The H.265 decoder is also in development.
I'll add our plan in the cover-letter of the next version (v2) patch
series to make it clear for the kernel reviewers.
- Supporting I/P/B frame handling.
- Supporting vb2 mmap and dma-buf modes.
- Supporting frame-based decode mode. (Note that some H.264 bitstreams require
DPB reordering to generate reference lists, the stateless decoder driver
cannot access reordered reference lists in this mode, requiring the driver
to perform reference list reordering itself)
- Supporting NV12/NV21 output.
- Supporting Annex B start codes.

This driver is tested with Gstreamer.
Example:
gst-launch-1.0 filesrc location=/tmp/video_640x360_mp4_hevc_450kbps_no_b.mp4 !
parsebin ! v4l2slh264dec ! filesink location=/tmp/output.yuv
Is this upstream gstreamer? if yes which version? If no, where can
we obtain sources for testing. Has ffmpeg been tested? - the same
questions apply.
The version is Gstreamer 1.24, I downloaded the source code from the official
website: https://gstreamer.freedesktop.org/.
FFmpeg hasn't been tested yet. We will also try to test the driver with ffmpeg.
The driver passes v4l2 compliance test, test reports :
v4l2-compliance 1.30.1, 64 bits, 64-bit time_t

Compliance test for aml-vdec-drv device /dev/video0:

Driver Info:
Driver name : aml-vdec-drv
Card type : platform:aml-vdec-drv
Bus info : platform:fe320000.video-codec
Driver version : 6.16.0
It would be preferable to test the next iteration of patches on
the latest kernel release, e.g. 6.18-rc, or recent media tree or
perhaps drm-tip source.
Thanks for the suggestion, we will do the compliance test on kernel
6.18 and post the result of the next version(v2) patch series.
Capabilities : 0x84204000
Video Memory-to-Memory Multiplanar
Streaming
Extended Pix Format
Device Capabilities
Device Caps : 0x04204000
Video Memory-to-Memory Multiplanar
Streaming
Extended Pix Format
Detected Stateless Decoder
Media Driver Info:
Driver name : aml-vdec-drv
Model : aml-vdec-drv
Serial :
Bus info : platform:fe320000.video-codec
Media version : 6.16.0
Hardware revision: 0x00000000 (0)
Driver version : 6.16.0
Interface Info:
ID : 0x0300000c
Type : V4L Video
Entity Info:
ID : 0x00000001 (1)
Name : aml_dev_drv-source
Function : V4L2 I/O
Pad 0x01000002 : 0: Source
Link 0x02000008: to remote pad 0x1000004 of entity 'aml_dev_drv-proc' (Video Decoder): Data, Enabled, Immutable

Required ioctls:
test MC information (see 'Media Driver Info' above): OK
test VIDIOC_QUERYCAP: OK
test invalid ioctls: OK

Allow for multiple opens:
test second /dev/video0 open: OK
test VIDIOC_QUERYCAP: OK
test VIDIOC_G/S_PRIORITY: OK
test for unlimited opens: OK

Debug ioctls:
test VIDIOC_DBG_G/S_REGISTER: OK (Not Supported)
test VIDIOC_LOG_STATUS: OK (Not Supported)

Input ioctls:
test VIDIOC_G/S_TUNER/ENUM_FREQ_BANDS: OK (Not Supported)
test VIDIOC_G/S_FREQUENCY: OK (Not Supported)
test VIDIOC_S_HW_FREQ_SEEK: OK (Not Supported)
test VIDIOC_ENUMAUDIO: OK (Not Supported)
test VIDIOC_G/S/ENUMINPUT: OK (Not Supported)
test VIDIOC_G/S_AUDIO: OK (Not Supported)
Inputs: 0 Audio Inputs: 0 Tuners: 0

Output ioctls:
test VIDIOC_G/S_MODULATOR: OK (Not Supported)
test VIDIOC_G/S_FREQUENCY: OK (Not Supported)
test VIDIOC_ENUMAUDOUT: OK (Not Supported)
test VIDIOC_G/S/ENUMOUTPUT: OK (Not Supported)
test VIDIOC_G/S_AUDOUT: OK (Not Supported)
Outputs: 0 Audio Outputs: 0 Modulators: 0

Input/Output configuration ioctls:
test VIDIOC_ENUM/G/S/QUERY_STD: OK (Not Supported)
test VIDIOC_ENUM/G/S/QUERY_DV_TIMINGS: OK (Not Supported)
test VIDIOC_DV_TIMINGS_CAP: OK (Not Supported)
test VIDIOC_G/S_EDID: OK (Not Supported)

Control ioctls:
test VIDIOC_QUERY_EXT_CTRL/QUERYMENU: OK
test VIDIOC_QUERYCTRL: OK
test VIDIOC_G/S_CTRL: OK
test VIDIOC_G/S/TRY_EXT_CTRLS: OK
test VIDIOC_(UN)SUBSCRIBE_EVENT/DQEVENT: OK
test VIDIOC_G/S_JPEGCOMP: OK (Not Supported)
Standard Controls: 6 Private Controls: 0
Standard Compound Controls: 4 Private Compound Controls: 0

Format ioctls:
test VIDIOC_ENUM_FMT/FRAMESIZES/FRAMEINTERVALS: OK
test VIDIOC_G/S_PARM: OK (Not Supported)
test VIDIOC_G_FBUF: OK (Not Supported)
test VIDIOC_G_FMT: OK
test VIDIOC_TRY_FMT: OK
test VIDIOC_S_FMT: OK
test VIDIOC_G_SLICED_VBI_CAP: OK (Not Supported)
test Cropping: OK (Not Supported)
test Composing: OK (Not Supported)
test Scaling: OK (Not Supported)

Codec ioctls:
test VIDIOC_(TRY_)ENCODER_CMD: OK (Not Supported)
test VIDIOC_G_ENC_INDEX: OK (Not Supported)
test VIDIOC_(TRY_)DECODER_CMD: OK

Buffer ioctls:
test VIDIOC_REQBUFS/CREATE_BUFS/QUERYBUF: OK
test CREATE_BUFS maximum buffers: OK
test VIDIOC_REMOVE_BUFS: OK
test VIDIOC_EXPBUF: OK
test Requests: OK
test blocking wait: OK

Total for aml-vdec-drv device /dev/video0: 49, Succeeded: 49, Failed: 0, Warnings: 0

Some Fluster test cases are still failing. We will publish the final results
once all these Fluster test failures have been resolved.
IMHO good drivers that merge quickly have more merit than perfect
drivers that take forever. No driver passes all fluster tests. If
you share the results, people can advise whether the failures are
normal, something that must be improved, or something that can be
improved with time.
Thanks. We have now tested the JVT-AVC_V1.
Result:
Ran 54/135 tests successfully

- 52 test vectors failed due to interlaced or mbaff clips: The Amlogic stateless
  decoder driver only support bitstreams with frame_mbs_only_flags == 1.
  Test Vectors:
        cabac_mot_fld0_full
        cabac_mot_mbaff0_full
        cabac_mot_picaff0_full
        CABREF3_Sand_D
        CAFI1_SVA_C
        CAMA1_Sony_C
        CAMA1_TOSHIBA_B
        cama1_vtc_c
        cama2_vtc_b
        CAMA3_Sand_E
        cama3_vtc_b
        CAMACI3_Sony_C
        CAMANL1_TOSHIBA_B
        CAMANL2_TOSHIBA_B
        CAMANL3_Sand_E
        CAMASL3_Sony_B
        CAMP_MOT_MBAFF_L30
        CAMP_MOT_MBAFF_L31
        CANLMA2_Sony_C
        CANLMA3_Sony_C
        CAPA1_TOSHIBA_B
        CAPAMA3_Sand_F
        cavlc_mot_fld0_full_B
        cavlc_mot_mbaff0_full_B
        cavlc_mot_picaff0_full_B
        CVCANLMA2_Sony_C
        CVFI1_Sony_D
        CVFI1_SVA_C
        CVFI2_Sony_H
        CVFI2_SVA_C
        CVMA1_Sony_D
        CVMA1_TOSHIBA_B
        CVMANL1_TOSHIBA_B
        CVMANL2_TOSHIBA_B
        CVMAPAQP3_Sony_E
        CVMAQP2_Sony_G
        CVMAQP3_Sony_D
        CVMP_MOT_FLD_L30_B
        CVNLFI1_Sony_C
        CVNLFI2_Sony_H
        CVPA1_TOSHIBA_B
        FI1_Sony_E
        MR6_BT_B
        MR7_BT_B
        MR8_BT_B
        MR9_BT_B
        Sharp_MP_Field_1_B
        Sharp_MP_Field_2_B
        Sharp_MP_Field_3_B
        Sharp_MP_PAFF_1r2
        Sharp_MP_PAFF_2r
        CVMP_MOT_FRM_L31_B
- 3 test vectors failed due to unsupported bitstream.
  num_slice_group_minus1 greater than zero is not supported by the
  hardware.
  Test Vectors:
        FM1_BT_B
        FM1_FT_E
        FM2_SVA_C
- 2 test vectors failed because SP_SLICE type is not supported by the
  hardware.
  Test Vectors:
        SP1_BT_A
        sp2_bt_b
We are working with the remain failures, we think that these fail cases
must be resolved.
Signed-off-by: Zhentao Guo <zhentao.guo@xxxxxxxxxxx>
---
Zhentao Guo (3):
dt-bindings: vdec: Add binding document of Amlogic decoder accelerator
dts: decoder: Support V4L2 stateless decoder dt node for S4
decoder: Add V4L2 stateless H.264 decoder driver
None of the patch subject lines are following the naming formats
used for the subsystems. Patch order is also important. Bindings
go first, device-tree changes to expose support normally go last.
Expect maintainer complaints about not reading kernel etiquette
rules before submitting.
Thanks for the reminder. I will rebase the patches in correct order
and revise the subject according to the specifications.
.../bindings/media/amlogic,vcodec-dec.yaml | 96 ++
MAINTAINERS | 7 +
arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 24 +
drivers/media/platform/amlogic/Kconfig | 2 +
drivers/media/platform/amlogic/Makefile | 1 +
drivers/media/platform/amlogic/vdec/Kconfig | 15 +
drivers/media/platform/amlogic/vdec/Makefile | 4 +
drivers/media/platform/amlogic/vdec/aml_vdec.c | 759 +++++++++
drivers/media/platform/amlogic/vdec/aml_vdec.h | 31 +
.../platform/amlogic/vdec/aml_vdec_canvas_utils.c | 154 ++
.../platform/amlogic/vdec/aml_vdec_canvas_utils.h | 22 +
drivers/media/platform/amlogic/vdec/aml_vdec_drv.c | 263 +++
drivers/media/platform/amlogic/vdec/aml_vdec_drv.h | 194 +++
drivers/media/platform/amlogic/vdec/aml_vdec_hw.c | 652 +++++++
drivers/media/platform/amlogic/vdec/aml_vdec_hw.h | 182 ++
.../platform/amlogic/vdec/aml_vdec_platform.c | 37 +
.../platform/amlogic/vdec/aml_vdec_platform.h | 62 +
drivers/media/platform/amlogic/vdec/h264.c | 1790 ++++++++++++++++++++
drivers/media/platform/amlogic/vdec/h264.h | 300 ++++
drivers/media/platform/amlogic/vdec/reg_defines.h | 175 ++
20 files changed, 4770 insertions(+)
The driver requires closed-source firmware blobs (sadly) for each
codec and hardware target. Will that change in future iterations?
If not, the firmware blobs are not provided here so it would be
useful if you could provide a public source for them. There are
known FOSS tools for extracting files from the Android ucode blob
but testing presumably needs a specific blob version?
Yes, we plan to address this in future iterations. We intend  to pack
the firmware blobs for all codecs together but they will still remain
categorized by different hardware platforms sine we want to avoid making
the firmware package excessively large. The closed-source firmware will be
uploaded to the firmware repo:
https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git.
I will share the commit link once the upload is completed.
Most maintainers also prefer a complex driver to be submitted in a
series of patches that break the driver up into logical and easier
to-review pieces. Large uber-patches are harder to digest and are
likely to take much longer to review, and more effort for everyone
to track feedback on over time.
Thanks for reminder. I admittedly overlooked these considerations.
The next version of the patch series will be improved accordingly.
***

NB: The fact this driver is being submitted, is still a positive
and welcomed step towards better upstream Amlogic support and it’s
great to see! I would encourage you and colleagues to establish a
public working space to share code and tools so the community can
directly engage and support driver testing and development.
Yes, we will take this into consideration. Thanks again for all your
suggestions above, we look forward to cooperate with community.

Christian
Best Regrads
Zhentao