linux-kernel - Re: [PATCH REPOST] TDA1997x: replace video detection routine

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]

Message-ID: <798b2128-24a6-1a62-f98c-f228134e8d5c@xs4all.nl>
Date:   Fri, 23 Jul 2021 10:49:01 +0200
From:   Hans Verkuil <hverkuil@...all.nl>
To:     Krzysztof Hałasa <khalasa@...p.pl>
Cc:     Tim Harvey <tharvey@...eworks.com>,
        Mauro Carvalho Chehab <mchehab@...nel.org>,
        linux-media@...r.kernel.org, lkml <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH REPOST] TDA1997x: replace video detection routine

On 22/07/2021 08:27, Krzysztof Hałasa wrote:
> Hans,
> 
> I have a problem with matching H/V sync polarities. It appears the
> standard modes use different polarity settings, and HDMI doesn't have
> separate sync lines, it uses in-band signaling. I don't know what
> polarity should I set for the v4l2_find_dv_timings_cap() ->
> v4l2_match_dv_timings() to succeed.

The tda1997x doesn't have any method of obtaining the sync polarities?

HSync and VSync are, as you say, using in-band signaling, but the 'polarity'
still matters when it comes to detecting timings. All HDMI receivers I have
ever seen have a register that reports the polarities. I find it *very* hard
to imagine that this device doesn't support that.

Do you have a datasheet for this device?

> 
> 1. I can try all 4 combinations (+/+, +/-, -/+, -/-) and see if there is
> a match. Seems weird to me.
> 
> 2. I can add a flag to v4l2_find_dv_timings_cap() ->
> v4l2_match_dv_timings() which asks for the polarities to be ignored.
> 
> 3. I could add
>  #define V4L2_DV_VSYNC_POS_POL   0x00000001
>  #define V4L2_DV_HSYNC_POS_POL   0x00000002
> +#define V4L2_DV_VSYNC_ANY_POL   0x00000004
> +#define V4L2_DV_HSYNC_ANY_POL   0x00000008
> 
> and v4l2_match_dv_timings() would skip comparing them. A bit hacky.
> 
> 4. I could also add
>  #define V4L2_DV_VSYNC_POS_POL   0x00000001
>  #define V4L2_DV_HSYNC_POS_POL   0x00000002
> +#define V4L2_DV_VSYNC_NEG_POL   0x00000004
> +#define V4L2_DV_HSYNC_NEG_POL   0x00000008
> 
> and, likewise, v4l2_match_dv_timings() would know what to do. This is
> IMHO cleaner but requires much more changes in other code (like
> drivers).
> 
> Any other idea?
> Perhaps I should do "2"?

If there really is no register that reports the polarities, then option 2 is
the best approach. There are other reasons why you might want to ignore the
polarities: not all video sources use the correct polarities.

What I think would be best is to make a flags argument that allows the
match code to skip certain tests:

#define V4L2_DV_TIM_IGNORE_POLARITIES	BIT(0)
#define V4L2_DV_TIM_IGNORE_HPORCHES	BIT(1)
#define V4L2_DV_TIM_IGNORE_VPORCHES	BIT(2)
#define V4L2_DV_TIM_IGNORE_HSYNC	BIT(3)
#define V4L2_DV_TIM_IGNORE_VSYNC	BIT(4)

IGNORE_HPORCHES would just check hsync and total horizontal blanking,
IGNORE_HSYNC would just check total horizontal blanking.
IGNORE_HSYNC implies IGNORE_HPORCHES.

Regards,

	Hans