Re: [PATCH v5 5/8] clk: qcom: gcc-msm8939: mark Venus core GDSCs as hardware controlled

[Erikas Bitovtas <xerikasxx@gmail.com>]

On 5/8/26 6:55 PM, Bryan O'Donoghue wrote:
> On 07/05/2026 16:51, Erikas Bitovtas wrote:
>>
>>
>> On 5/7/26 4:37 PM, Bryan O'Donoghue wrote:
>>> v4l2-ctl --verbose --set-fmt-video-
>>> out=width=1280,height=720,pixelformat=NV12 --set-selection-output
>>> target=crop,top=0,left=0,width=1280,height=720 --set-fmt-
>>> video=pixelformat=H264 --stream-mmap --stream-out-mmap --stream-
>>> from=cyclists_1280x720_92frames.yuv --stream-to=/tmp/
>>> cyclists_1280x720_92frames.h264 -d /dev/video1
>>
>> With the patch kept:
>> v4l2-ctl --verbose
>> --set-fmt-video-out=width=1280,height=720,pixelformat=NV12
>> --set-selection-output target=crop,top=0,left=0,width=1280,height=720
>> --set-fmt-video=pixelformat=H26
>> 4 --stream-mmap --stream-out-mmap
>> --stream-from=cyclists_1280x720_92frames.yuv
>> --stream-to=/tmp/cyclists_1280x720_92frames.h264 -d /dev/video1
>> VIDIOC_QUERYCAP: ok
>> VIDIOC_G_FMT: ok
>> VIDIOC_S_FMT: ok
>> Format Video Capture Multiplanar:
>>     Width/Height      : 1280/736
>>     Pixel Format      : 'H264' (H.264)
>>     Field             : None
>>     Number of planes  : 1
>>     Flags             :
>>     Colorspace        : Default
>>     Transfer Function : Default
>>     YCbCr/HSV Encoding: Default
>>     Quantization      : Default
>>     Plane 0           :
>>        Bytes per Line : 0
>>        Size Image     : 708608
>> VIDIOC_G_FMT: ok
>> VIDIOC_S_FMT: ok
>> Format Video Output Multiplanar:
>>     Width/Height      : 1280/736
>>     Pixel Format      : 'NV12' (Y/UV 4:2:0)
>>     Field             : None
>>     Number of planes  : 1
>>     Flags             :
>>     Colorspace        : Default
>>     Transfer Function : Default
>>     YCbCr/HSV Encoding: Default
>>     Quantization      : Default
>>     Plane 0           :
>>        Bytes per Line : 1280
>>        Size Image     : 1425408
>> VIDIOC_G_SELECTION: ok
>> VIDIOC_S_SELECTION: ok
>>         VIDIOC_G_FMT returned 0 (No error information)
>>         VIDIOC_G_FMT returned 0 (No error information)
>>         VIDIOC_G_FMT returned 0 (No error information)
>>         VIDIOC_REQBUFS returned -1 (Invalid argument)
>>
>> With the patch reverted:
>> v4l2-ctl --verbose
>> --set-fmt-video-out=width=1280,height=720,pixelformat=NV12
>> --set-selection-output target=crop,top=0,left=0,width=1280,height=720
>> --set-fmt-video=pixelformat=H26
>> 4 --stream-mmap --stream-out-mmap
>> --stream-from=cyclists_1280x720_92frames.yuv
>> --stream-to=/tmp/cyclists_1280x720_92frames.h264 -d /dev/video1
>> VIDIOC_QUERYCAP: ok
>> VIDIOC_G_FMT: ok
>> The pixelformat 'H264' is invalid
>> VIDIOC_G_FMT: ok
>> The pixelformat 'NV12' is invalid
>>
>> Decoding breaks if I revert the patch as well, it fails with a power
>> collapse fail.
> 
> If you declare both as decoders can you successfully decode @ video0 and
> video1 ?
> Apparently device nodes for encoder and decoder are not persistent
across reboots, so when I reverted the patch, I was actually testing
against decoder, not encoder.

So, the result is the same when the patch is applied, when it is
reverted, and when testing against v1 where the cores are enabled only
for decoding.