linux-kernel - Re: [PATCH v2 1/2] drm: bridge: dw-hdmi: Take input format from plat

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 3 Mar 2017 17:42:08 +0100
From:   Neil Armstrong <narmstrong@...libre.com>
To:     Jose Abreu <Jose.Abreu@...opsys.com>,
        Laurent Pinchart <laurent.pinchart@...asonboard.com>
Cc:     dri-devel@...ts.freedesktop.org,
        laurent.pinchart+renesas@...asonboard.com,
        kieran.bingham@...asonboard.com, linux-amlogic@...ts.infradead.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 1/2] drm: bridge: dw-hdmi: Take input format from
 plat_data

On 03/03/2017 05:39 PM, Jose Abreu wrote:
> Hi Neil,
> 
> 
> On 03-03-2017 11:31, Neil Armstrong wrote:
>>
>> Sure, but I was struggling about finding an equivalence.
>>
>> How should I replace these ?
>>
>> DW_HDMI_ENC_FMT_RGB
>> DW_HDMI_ENC_FMT_YCBCR444
>> DW_HDMI_ENC_FMT_YCBCR422_16BITS
>> DW_HDMI_ENC_FMT_YCBCR422_8BITS
>> DW_HDMI_ENC_FMT_XVYCC444
>>
>> I do not have the strict information about the bus format, what is RGB in 8bit per pixel ?
>> Are the 3 samples transmitted separately ?
>> What should be the RGB colorspace ?
>> Should we use the "Packed" formats, LVDS or a new buf format ?
>>
>> Jose, what are the *exact* bus formats supported ?
> 
> Well, the controller supports everything that is in the HDMI spec
> (1.4 and 2.0), the implementation is the one that limits the
> supported formats. There is also a CSC but it does not convert
> from anything to everything :)

Sure, I was meaning the *input* format the controller receives from the
pixel encoder, I'm quite sure the format is strict.

> 
> I think you can safely add every MBUS code that is compatible
> with HDMI. Namely:
>     -    24/30/36/48-bit RGB 4:4:4
>     -    24/30/36/48-bit YCbCr 4:4:4
>     -    16/20/24-bit YCbCr 4:2:2
>     -    24/30/36/48-bit YCbCr 4:2:0
> 
> And then let the glue drivers decide which they support in input
> and what they want in output (the output is a little tricky
> because it will depend in EDID, this should be handled by
> userspace + drm core and not the drivers so let it fixed to RGB,
> just in case).
> 
>>
>> Same for DW_HDMI_ENC_FMT_YCBCR* stuff, are they packed ?
> 
> Hmm, this depends on the implementation. The controller always
> receives 48bits of data per pixel, its the implementation that is
> responsible to correctly align that data.
> 
>>
>> I understand the YCBCR444/XVYCC444 needs a color space information instead.
>>
>> Could we use the following list ?
>>
>> Bus Format			| Colorspace 		| Encoding
>> -------------------------------------------------------------------------------
>> MEDIA_BUS_FMT_RGB888_1X24	| V4L2_COLORSPACE_SRGB	| V4L2_XFER_FUNC_SRGB
>> MEDIA_BUS_FMT_RGB101010_1X30	| V4L2_COLORSPACE_SRGB	| V4L2_XFER_FUNC_SRGB
>> MEDIA_BUS_FMT_RGB121212_1X36	| V4L2_COLORSPACE_SRGB	| V4L2_XFER_FUNC_SRGB
>> MEDIA_BUS_FMT_VUY8_1X24		| V4L2_XFER_FUNC_709	| V4L2_YCBCR_ENC_709
>> MEDIA_BUS_FMT_YUV10_1X30	| V4L2_XFER_FUNC_709	| V4L2_YCBCR_ENC_709
>> MEDIA_BUS_FMT_YUV10_1X32	| V4L2_XFER_FUNC_709	| V4L2_YCBCR_ENC_709
>> MEDIA_BUS_FMT_YUV10_1X32	| V4L2_XFER_FUNC_709	| V4L2_YCBCR_ENC_709
>> MEDIA_BUS_FMT_YUV10_1X32	| V4L2_XFER_FUNC_709	| V4L2_YCBCR_ENC_XV709
> 
> I think the HDMI spec dictates the default colorspace+encoding
> for each bus format, not sure though, Laurent?
> 
> Best regards,
> Jose Miguel Abreu
> 
>>
>> But all of these is complex, and I'm not sure how we should handle SDTV modes here,
>> and without knowing the exact inputs types supported by the IP, this needs to be
>> confirmed.
>>
>> Meanwhile, all this can be fixed later, at least this patch moves the
>> definition out of the C file and uses better names for these custom enums.
>>
>