linux-kernel - Re: [PATCH 8/8] drm/tidss: Enable Dual and Duplicate Modes for OLDI

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 for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <0d9f4301-8af5-4497-8e59-affdc0541285@ti.com>
Date:   Thu, 28 Jul 2022 14:19:24 +0530
From:   Aradhya Bhatia <a-bhatia1@...com>
To:     Tomi Valkeinen <tomi.valkeinen@...asonboard.com>
CC:     Darren Etheridge <detheridge@...com>, Nishanth Menon <nm@...com>,
        Vignesh Raghavendra <vigneshr@...com>,
        Rahul T R <r-ravikumar@...com>,
        Krunal Bhargav <k-bhargav@...com>,
        Devarsh Thakkar <devarsht@...com>,
        DRI Development List <dri-devel@...ts.freedesktop.org>,
        Devicetree List <devicetree@...r.kernel.org>,
        Linux Kernel List <linux-kernel@...r.kernel.org>,
        Rob Herring <robh+dt@...nel.org>,
        Daniel Vetter <daniel@...ll.ch>,
        David Airlie <airlied@...ux.ie>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Jyri Sarha <jyri.sarha@....fi>
Subject: Re: [PATCH 8/8] drm/tidss: Enable Dual and Duplicate Modes for OLDI

Hi Tomi,

On 28-Jul-22 12:16, Tomi Valkeinen wrote:
> On 27/07/2022 16:22, Tomi Valkeinen wrote:
>> Hi,
>>
>> On 19/07/2022 11:08, Aradhya Bhatia wrote:
>>> The AM625 DSS peripheral supports 2 OLDI TXes which can work to enable 2
>>> duplicated displays of smaller resolutions or enable a single Dual-Link
>>> display with a higher resolution (1920x1200).
>>>
>>> Configure the necessary register to enable the different modes.
>>>
>>> Signed-off-by: Aradhya Bhatia <a-bhatia1@...com>
>>> ---
>>>   drivers/gpu/drm/tidss/tidss_dispc.c | 44 +++++++++++++++++++++++++++--
>>>   1 file changed, 41 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/tidss/tidss_dispc.c 
>>> b/drivers/gpu/drm/tidss/tidss_dispc.c
>>> index 0b9689453ee8..28cb61259471 100644
>>> --- a/drivers/gpu/drm/tidss/tidss_dispc.c
>>> +++ b/drivers/gpu/drm/tidss/tidss_dispc.c
>>> @@ -1021,8 +1021,8 @@ static void dispc_enable_oldi(struct 
>>> dispc_device *dispc, u32 hw_videoport,
>>>       int count = 0;
>>>       /*
>>> -     * For the moment DUALMODESYNC, MASTERSLAVE, MODE, and SRC
>>> -     * bits of DISPC_VP_DSS_OLDI_CFG are set statically to 0.
>>> +     * For the moment MASTERSLAVE, and SRC bits of 
>>> DISPC_VP_DSS_OLDI_CFG are
>>> +     * set statically to 0.
>>>        */
>>>       if (fmt->data_width == 24)
>>> @@ -1039,7 +1039,45 @@ static void dispc_enable_oldi(struct 
>>> dispc_device *dispc, u32 hw_videoport,
>>>       oldi_cfg |= BIT(0); /* ENABLE */
>>> -    dispc_vp_write(dispc, hw_videoport, DISPC_VP_DSS_OLDI_CFG, 
>>> oldi_cfg);
>>> +    /*
>>> +     * As per all the current implementations of DSS, the OLDI TXes 
>>> are present only on
>>> +     * hw_videoport = 0 (OLDI TX 0). However, the config register 
>>> for 2nd OLDI TX (OLDI TX 1)
>>> +     * is present in the address space of hw_videoport = 1. Hence, 
>>> using "hw_videoport + 1" to
>>> +     * configure OLDI TX 1.
>>> +     */
>>> +
>>> +    switch (dispc->oldi_mode) {
>>> +    case OLDI_MODE_OFF:
>>> +        oldi_cfg &= ~BIT(0); /* DISABLE */
>>> +        dispc_vp_write(dispc, hw_videoport, DISPC_VP_DSS_OLDI_CFG, 
>>> oldi_cfg);
>>> +        dispc_vp_write(dispc, hw_videoport + 1, 
>>> DISPC_VP_DSS_OLDI_CFG, oldi_cfg);
>>> +        break;
>>> +
>>> +    case OLDI_SINGLE_LINK_SINGLE_MODE_0:
>>> +        dispc_vp_write(dispc, hw_videoport, DISPC_VP_DSS_OLDI_CFG, 
>>> oldi_cfg);
>>> +        break;
>>> +
>>> +    case OLDI_SINGLE_LINK_SINGLE_MODE_1:
>>> +        dispc_vp_write(dispc, hw_videoport + 1, 
>>> DISPC_VP_DSS_OLDI_CFG, oldi_cfg);
>>> +        break;
>>> +
>>> +    case OLDI_SINGLE_LINK_DUPLICATE_MODE:
>>> +        oldi_cfg |= BIT(5); /* DUPLICATE MODE */
>>> +        dispc_vp_write(dispc, hw_videoport, DISPC_VP_DSS_OLDI_CFG, 
>>> oldi_cfg);
>>> +        dispc_vp_write(dispc, hw_videoport + 1, 
>>> DISPC_VP_DSS_OLDI_CFG, oldi_cfg);
>>> +        break;
>>> +
>>> +    case OLDI_DUAL_LINK:
>>> +        oldi_cfg |= BIT(11); /* DUALMODESYNC */
>>> +        dispc_vp_write(dispc, hw_videoport, DISPC_VP_DSS_OLDI_CFG, 
>>> oldi_cfg);
>>> +        dispc_vp_write(dispc, hw_videoport + 1, 
>>> DISPC_VP_DSS_OLDI_CFG, oldi_cfg);
>>> +        break;
>>> +
>>> +    default:
>>> +        dev_warn(dispc->dev, "%s: Incorrect oldi mode. Returning.\n",
>>> +             __func__);
>>> +        return;
>>> +    }
>>>       while (!(oldi_reset_bit & dispc_read(dispc, DSS_SYSSTATUS)) &&
>>>              count < 10000)
>>
>> This feels a bit hacky:
>>
>> - The function is dispc_enable_oldi, but the above code also disables 
>> oldi. We have code in dispc_vp_unprepare() which disables OLDI at the 
>> moment.
>>
>> - The function takes hw_videoport as a parameter, and is designed to 
>> work on that videoport. The above operates on two videoports. Isn't 
>> the function also called for hw_videoport +1, which would result in 
>> reg writes to hw_videoport + 2?
>>
>> - No matching code in dispc_vp_unprepare
>>
>> Obviously the duplicate mode (I presume that's "cloning") and the dual 
>> link complicate things here, and I have to say I haven't worked with 
>> such setups. But I think somehow this should be restructured so that 
>> common configuration (common to the OLDIs) is done somewhere else.
>>
>> I would guess that there are other drivers that support cloning and 
>> dual mode. Did you have a look how they handle things?
> 
> Oh, I see now... There's just one dss video port for OLDI, the same as 
> in am65x, but that single video port is now connected to two OLDI TXes. 
> And thus this function will only be called for the single video port.
> > But... The registers for the second OLDI are part of the second video
> port (DPI) register block?

Yes! The config register for the second OLDI TX has been (incorrectly)
added in the register space for the DPI video port. 'dispc_vp_prepare'
is the only function calling 'dispc_enable_oldi', and
'dispc_enable_oldi' would not be called for hw_videoports = 1 (DPI)
because of the conditional check.

Hence, to activate both the OLDI-TXes connected to the OLDI video port,
I had to use the (hw_videoport + 1) way.

However, I will remove the disable part from the 'dispc_enable_oldi' and
I will implement the disabling properly under 'dispc_vp_unprepare', in
the next version.

Regards
Aradhya