[PATCH 07/11] drm/vc4: hdmi: Disable connector status polling during suspend

[Stefan Wahren]

Am 03.07.24 um 12:28 schrieb Stefan Wahren:
> Hi Maxime,
>
> Am 02.07.24 um 15:48 schrieb Maxime Ripard:
>> Hi,
>>
>> On Sun, Jun 30, 2024 at 05:36:48PM GMT, Stefan Wahren wrote:
>>> Suspend of VC4 HDMI will likely triggers a warning from
>>> vc4_hdmi_connector_detect_ctx() during poll of connector status.
>>> The power management will prevent the resume and keep the relevant
>>> power domain disabled.
>>>
>>> Since there is no reason to poll the connector status during
>>> suspend, the polling should be disabled during this.
>>>
>>> It not possible to use drm_mode_config_helper_suspend() here,
>>> because the callbacks might be called during bind phase and not all
>>> components are fully initialized.
>>>
>>> Link:
>>> https://lore.kernel.org/dri-devel/7003512d-7303-4f41-b0d6-a8af5bf8e497@gmx.net/
>>> Signed-off-by: Stefan Wahren <wahrenst at gmx.net>
>>> ---
>>>   drivers/gpu/drm/vc4/vc4_hdmi.c | 11 +++++++++++
>>>   1 file changed, 11 insertions(+)
>>>
>>> diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c
>>> b/drivers/gpu/drm/vc4/vc4_hdmi.c
>>> index b3a42b709718..e80495cea6ac 100644
>>> --- a/drivers/gpu/drm/vc4/vc4_hdmi.c
>>> +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c
>>> @@ -3106,6 +3106,13 @@ static int vc5_hdmi_init_resources(struct
>>> drm_device *drm,
>>>   static int vc4_hdmi_runtime_suspend(struct device *dev)
>>>   {
>>>       struct vc4_hdmi *vc4_hdmi = dev_get_drvdata(dev);
>>> +    struct drm_device *drm = vc4_hdmi->connector.dev;
>>> +
>>> +    /*
>>> +     * Don't disable polling if it was never initialized
>>> +     */
>>> +    if (drm && drm->mode_config.poll_enabled)
>>> +        drm_kms_helper_poll_disable(drm);
>> Does it make sense to add it to runtime_suspend?
> i saw that other drm drivers used drm_mode_config_helper_suspend() in
> the RUNTIME_PM_OPS. But i agree, it should be better moved to
> SYSTEM_SLEEP_PM_OPS.
>> What if the board boots without a display connected, and only after a
>> while one is connected? Wouldn't that prevent the driver from detecting
>> it?
> tbh I noticed that HDMI re-detection didn't worked in my setup
> 6.10-rcX before this series. I need to investigate ...
Okay, this patch breaks definitely HDMI re-detection. So please ignore
this patch. Sorry, about this mess.

There is another minor issue which already exists before that cause the
following log entry on HDMI reconnect:

[   74.078745] vc4-drm soc:gpu: [drm] User-defined mode not supported:
"1920x1200": 60 154000 1920 1968 2000 2080 1200 1203 1209 1235 0x68 0x9

But in this case HDMI works.

Regards
>>
>> Maxime
>