[libcamera-devel] [PATCH v2] v4l2: v4l2_camera_proxy: Support MJPEG

[Kieran Bingham]

Hi Paul / Laurent,

On 08/06/2020 15:12, Laurent Pinchart wrote:
> Hi Paul,
> 
> Thank you for the patch.
> 
> On Mon, Jun 08, 2020 at 05:05:29PM +0900, Paul Elder wrote:

<snip>

>>  namespace {
>>  
>> -static const std::array<PixelFormatInfo, 13> pixelFormatInfo = {{
>> +static const std::array<PixelFormatInfo, 14> pixelFormatInfo = {{
>>  	/* RGB formats. */
>>  	{ PixelFormat(DRM_FORMAT_RGB888),	V4L2_PIX_FMT_BGR24,	1, {{ { 24, 1, 1 }, {  0, 0, 0 }, {  0, 0, 0 } }} },
>>  	{ PixelFormat(DRM_FORMAT_BGR888),	V4L2_PIX_FMT_RGB24,	1, {{ { 24, 1, 1 }, {  0, 0, 0 }, {  0, 0, 0 } }} },
>> @@ -573,6 +574,9 @@ static const std::array<PixelFormatInfo, 13> pixelFormatInfo = {{
>>  	{ PixelFormat(DRM_FORMAT_NV61),		V4L2_PIX_FMT_NV61,	2, {{ {  8, 1, 1 }, { 16, 2, 1 }, {  0, 0, 0 } }} },


>>  	{ PixelFormat(DRM_FORMAT_NV24),		V4L2_PIX_FMT_NV24,	2, {{ {  8, 1, 1 }, { 16, 2, 1 }, {  0, 0, 0 } }} },
>>  	{ PixelFormat(DRM_FORMAT_NV42),		V4L2_PIX_FMT_NV42,	2, {{ {  8, 1, 1 }, { 16, 1, 1 }, {  0, 0, 0 } }} },

Not related to this patch, but I only commenting here because I can see it:

Above I see NV24 and NV42 which I expect are nvSwap versions of each other.

The second plane for NV24 shows h/v subsampling of 2,1 where NV42 shows 1,1.

That looks suspicious to me.

Looking at the rest of the table, perhaps NV24 should be 16,1,1 too ?

--
Kieran