[PATCH v7 2/6] drm/mediatek: dsi: Improves the DSI lane setup robustness

[AngeloGioacchino Del Regno]

Il 17/02/25 16:03, Alexandre Mergnat ha scritto:
> Hi CK.
> 
> On 17/02/2025 08:56, CK Hu (胡俊光) wrote:
>> On Fri, 2025-01-10 at 14:31 +0100, Alexandre Mergnat wrote:
>>> External email : Please do not click links or open attachments until you have 
>>> verified the sender or the content.
>>>
>>>
>>> Currently, mtk_dsi_lane_ready (which setup the DSI lane) is triggered
>>> before mtk_dsi_poweron. lanes_ready flag toggle to true during
>>> mtk_dsi_lane_ready function, and the DSI module is set up during
>>> mtk_dsi_poweron.
>>>
>>> Later, during panel driver init, mtk_dsi_lane_ready is triggered but does
>>> nothing because lanes are considered ready. Unfortunately, when the panel
>>> driver try to communicate, the DSI returns a timeout.
>>>
>>> The solution found here is to put lanes_ready flag to false after the DSI
>>> module setup into mtk_dsi_poweron to init the DSI lanes after the power /
>>> setup of the DSI module.
>>
>> I'm not clear about what happen.
>> I think this DSI flow has worked for a long time.
>> So only some panel has problem?
> 
> I don't know if it's related to a specific panel or not.
> 
>>
>> And another question.
>> Do you mean mtk_dsi_lane_ready() do some setting to hardware, but lane is not 
>> actually ready?
> 
> The workflow should be:
> ... | dsi->lanes_ready = false | Power-on | setup dsi lanes | dsi->lanes_ready = 
> true (to avoid re-do dsi lanes setup) | ...
> 
> I observe (print function name called + dsi->lanes_ready value):

Alex, the first poweron is called by mtk_dsi_ddp_start() - and the start callback
is internal to the mediatek-drm driver.

That callback is called by mtk_crtc during setup and during bridge enable(), and
there we go with suboptimal code design backfiring - instead of using what the
DRM APIs provide, this driver uses something custom *and* the DRM APIs, giving
this issue.

Part of what mtk_crtc does is duplicated with what the DRM APIs want to do, so
there you go, that's your problem here :-)

Should I go on with describing the next step(s), or is that obvious for everyone?

:-)

Cheers,
Angelo

> 
> [    9.086030] mtk_dsi_probe 0
> [    9.662319] mtk_dsi_host_attach 0
> [    9.662941] mtk_dsi_encoder_init
> [    9.984685] mtk_dsi_poweron 0
> [   10.043755] mtk_dsi_host_transfer 0
> [   10.043769] mtk_dsi_lane_ready 0
> [   10.055837] mtk_dsi_host_transfer 1
> [   10.055853] mtk_dsi_lane_ready 1
> [   10.179788] mtk_dsi_host_transfer 1
> [   10.179803] mtk_dsi_lane_ready 1
> [   10.179880] mtk_dsi_host_transfer 1
> [   10.179885] mtk_dsi_lane_ready 1
> [   10.179920] mtk_dsi_host_transfer 1
> [   10.179923] mtk_dsi_lane_ready 1
> [   10.179986] mtk_dsi_host_transfer 1
> [   10.179993] mtk_dsi_lane_ready 1
> [   10.180134] mtk_dsi_host_transfer 1
> [   10.180143] mtk_dsi_lane_ready 1
> [   10.180175] mtk_dsi_host_transfer 1
> [   10.180178] mtk_dsi_lane_ready 1
> [   10.180223] mtk_dsi_host_transfer 1
> [   10.180226] mtk_dsi_lane_ready 1
> [   10.180245] mtk_dsi_host_transfer 1
> [   10.180248] mtk_dsi_lane_ready 1
> [   10.180278] mtk_dsi_host_transfer 1
> [   10.180280] mtk_dsi_lane_ready 1
> [   10.180312] mtk_dsi_host_transfer 1
> [   10.180314] mtk_dsi_lane_ready 1
> [   10.203774] mtk_dsi_bridge_atomic_pre_enable
> [   10.203787] mtk_dsi_poweron 1
> [   10.203793] mtk_output_dsi_enable
> [   10.203795] mtk_dsi_lane_ready 1
> [   10.471517] mtk_dsi_host_transfer 1
> [   10.486962] mtk_dsi_lane_ready 1
> [   10.487244] mtk_dsi_host_transfer 1
> [   10.503733] mtk_dsi_lane_ready 1
> 
> Here the mtk_dsi_lane_ready function:
> 
>      static void mtk_dsi_lane_ready(struct mtk_dsi *dsi)
>      {
>          if (!dsi->lanes_ready) {
>              dsi->lanes_ready = true;
>              mtk_dsi_rxtx_control(dsi);
>              usleep_range(30, 100);
>              mtk_dsi_reset_dphy(dsi);
>              mtk_dsi_clk_ulp_mode_leave(dsi);
>              mtk_dsi_lane0_ulp_mode_leave(dsi);
>              mtk_dsi_clk_hs_mode(dsi, 0);
>              usleep_range(1000, 3000);
>              /* The reaction time after pulling up the mipi signal for dsi_rx */
>          }
>      }
> 
> 
> As you can see, something call "mtk_dsi_bridge_atomic_pre_enable" then 
> mtk_dsi_poweron is called a second time. This issue is probably due to the probe 
> order (race condition).
> After all, IMHO, after a poweron, the registers status should be consider as 
> unknown (or at least HW default value), so, lanes setup has to be done. This 
> solution improve the driver's robustness.
> 
> 
>> Or mtk_dsi_lane_ready() configure the hardware and lane is is actually ready,
>> but something make it not ready again, what's the thing which break lane ready?
>>
>> If this is a bug fix, add Fixes tag.
>>
>> Regards,
>> CK
>>
>>>
>>> Signed-off-by: Alexandre Mergnat <amergnat at baylibre.com>
>>> ---
>>>  drivers/gpu/drm/mediatek/mtk_dsi.c | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/drivers/gpu/drm/mediatek/mtk_dsi.c b/drivers/gpu/drm/mediatek/ 
>>> mtk_dsi.c
>>> index e61b9bc68e9a..dcf0d93881b5 100644
>>> --- a/drivers/gpu/drm/mediatek/mtk_dsi.c
>>> +++ b/drivers/gpu/drm/mediatek/mtk_dsi.c
>>> @@ -724,6 +724,8 @@ static int mtk_dsi_poweron(struct mtk_dsi *dsi)
>>>         mtk_dsi_config_vdo_timing(dsi);
>>>         mtk_dsi_set_interrupt_enable(dsi);
>>>
>>> +       dsi->lanes_ready = false;
>>> +
>>>         return 0;
>>>  err_disable_engine_clk:
>>>         clk_disable_unprepare(dsi->engine_clk);
>>>
>>> -- 
>>> 2.25.1
>>>
>>