Re: [PATCH v10 07/69] drm/connector: Add HDMI 2.0 scrambler infrastructure

[Cristian Ciocaltea]

On 8/20/26 7:56 PM, Maxime Ripard wrote:
> On Wed, Aug 19, 2026 at 10:33:04PM +0300, Cristian Ciocaltea wrote:
>> On 8/19/26 1:12 PM, Maxime Ripard wrote:
>>> On Fri, Jul 31, 2026 at 07:19:14PM +0300, Cristian Ciocaltea wrote:
>>>> Add the connector-level infrastructure to support HDMI 2.0 scrambling:
>>>>
>>>> - A drm_connector_hdmi_scrambler_supported() helper to report whether
>>>>   the source supports the scrambling capability, based on the presence
>>>>   of the newly introduced .scrambler_{enable|disable}() callbacks in
>>>>   drm_connector_hdmi_funcs are mandatory
>>>> - A scrambler_needed flag to be managed by the hdmi state helpers based
>>>>   on the negotiated TMDS character rate and the source/sink scrambling
>>>>   capabilities
>>>> - A scrambler_enabled flag to track whether scrambling is currently
>>>>   active
>>>> - A delayed work item (scdc_work) to monitor sink-side scrambling status
>>>>   and retry the setup if the sink resets it
>>>> - A scdc_work_initialized flag to support lazy initialization of the
>>>>   work item on the first scrambling enable and guard the teardown paths
>>>>
>>>> These are intended to be used by SCDC scrambling helpers to coordinate
>>>> scrambling setup and teardown between the source driver and the DRM
>>>> core.
>>>>
>>>> Tested-by: Maud Spierings <maud_spierings@xxxxxxxxxxx>
>>>> Tested-by: Diederik de Haas <diederik@xxxxxxxxxxxxxx>  # NanoPC-T6 LTS, Rock 5B
>>>> Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@xxxxxxxxxxxxx>
>>>> ---
>>>>  drivers/gpu/drm/drm_connector.c | 31 ++++++++++++---
>>>>  include/drm/drm_connector.h     | 83 +++++++++++++++++++++++++++++++++++++++++
>>>>  2 files changed, 109 insertions(+), 5 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
>>>> index 4721cdeafc84..a18410faf040 100644
>>>> --- a/drivers/gpu/drm/drm_connector.c
>>>> +++ b/drivers/gpu/drm/drm_connector.c
>>>> @@ -622,12 +622,29 @@ int drmm_connector_hdmi_init(struct drm_device *dev,
>>>>  	 * default with the actual controller capability. A value of zero keeps
>>>>  	 * the limit inferred from supported_hdmi_ver.
>>>>  	 */
>>>> -	if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_2_0)
>>>> +	if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_2_0) {
>>>> +		if (!hdmi_funcs->scrambler_enable || !hdmi_funcs->scrambler_disable) {
>>>> +			drm_err(dev, "Scrambler callbacks missing for HDMI 2.x\n");
>>>> +			return -EINVAL;
>>>> +		}
>>>> +
>>>>  		connector->hdmi.max_tmds_char_rate = HDMI_2_0_TMDS_CHAR_RATE_MAX_HZ;
>>>> -	else if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_1_3)
>>>> -		connector->hdmi.max_tmds_char_rate = HDMI_1_3_TMDS_CHAR_RATE_MAX_HZ;
>>>> -	else if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_1_0)
>>>> -		connector->hdmi.max_tmds_char_rate = HDMI_1_0_TMDS_CHAR_RATE_MAX_HZ;
>>>> +	} else {
>>>> +		/*
>>>> +		 * Scrambler callbacks are only valid for connectors advertising
>>>> +		 * HDMI 2.0 capability. drm_connector_hdmi_scrambler_supported()
>>>> +		 * relies on their presence to report scrambling support.
>>>> +		 */
>>>> +		if (hdmi_funcs->scrambler_enable || hdmi_funcs->scrambler_disable) {
>>>> +			drm_err(dev, "Scrambler callbacks unexpected for HDMI 1.x\n");
>>>> +			return -EINVAL;
>>>> +		}
>>>> +
>>>> +		if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_1_3)
>>>> +			connector->hdmi.max_tmds_char_rate = HDMI_1_3_TMDS_CHAR_RATE_MAX_HZ;
>>>> +		else if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_1_0)
>>>> +			connector->hdmi.max_tmds_char_rate = HDMI_1_0_TMDS_CHAR_RATE_MAX_HZ;
>>>> +	}
>>>
>>> I'd put it into a separate test (possibly earlier). Merging both the
>>> tmds rate default and the scrambler callbacks check makes it messier
>>> than it would be if we had two separate tests.
>>
>> Ack. How about the following?
>>
>> 	if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_2_0)
>> 		connector->hdmi.max_tmds_char_rate = HDMI_2_0_TMDS_CHAR_RATE_MAX_HZ;
>> 	else if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_1_3)
>> 		connector->hdmi.max_tmds_char_rate = HDMI_1_3_TMDS_CHAR_RATE_MAX_HZ;
>> 	else if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_1_0)
>> 		connector->hdmi.max_tmds_char_rate = HDMI_1_0_TMDS_CHAR_RATE_MAX_HZ;
>>
>> 	if (hdmi_funcs->supported_tmds_char_rate) {
>> 		if (hdmi_funcs->supported_tmds_char_rate > connector->hdmi.max_tmds_char_rate) {
>> 			drm_err(dev, "Enforced max_tmds_char_rate exceeds %llu spec limit\n",
>> 				connector->hdmi.max_tmds_char_rate);
>> 			return -EINVAL;
>> 		}
>>
>> 		connector->hdmi.max_tmds_char_rate = hdmi_funcs->supported_tmds_char_rate;
>> 	}
>>
>> 	if (hdmi_funcs->supported_hdmi_ver >= HDMI_VERSION_2_0) {
>> 		if (!hdmi_funcs->scrambler_enable || !hdmi_funcs->scrambler_disable) {
>> 			drm_err(dev, "Scrambler callbacks missing for HDMI 2.x\n");
>> 			return -EINVAL;
>> 		}
>> 	} else {
>> 		/*
>> 		 * Scrambler callbacks are only valid for connectors advertising
>> 		 * HDMI 2.0 capability. drm_connector_hdmi_scrambler_supported()
>> 		 * relies on their presence to report scrambling support.
>> 		 */
>> 		if (hdmi_funcs->scrambler_enable || hdmi_funcs->scrambler_disable) {
>> 			drm_err(dev, "Scrambler callbacks unexpected for HDMI 1.x\n");
>> 			return -EINVAL;
>> 		}
>> 	}
> 
> I don't think we need the else clause at all. It's not valid, but it's
> also not creating any issue.

As discussed a while ago, we used to have a scrambler_supported flag, inferred
from supported_hdmi_ver, which allowed helpers to verify the capability when
needed.  That flag has now been removed and replaced by
drm_connector_hdmi_scrambler_supported(), which relies exclusively on the
presence of the scrambler callbacks to report whether the capability is
supported. 

If we don't ensure that these callbacks are *not* set for HDMI 1.x cases, one
could set supported_hdmi_ver to HDMI_VERSION_1_4, for example, while still
providing the scrambler_{enable,disable} funcs.  This would lead to an
inconsistency between the maximum TMDS character rate inferred from
supported_hdmi_ver and the capability reported by
drm_connector_hdmi_scrambler_supported().

> I'd move that second check earlier together with the infoframe callbacks
> checks and so on too.

Ack.

>>>>  	if (hdmi_funcs->supported_tmds_char_rate) {
>>>>  		if (hdmi_funcs->supported_tmds_char_rate > connector->hdmi.max_tmds_char_rate) {
>>>> @@ -635,6 +652,7 @@ int drmm_connector_hdmi_init(struct drm_device *dev,
>>>>  				connector->hdmi.max_tmds_char_rate);
>>>>  			return -EINVAL;
>>>>  		}
>>>> +
>>>>  		connector->hdmi.max_tmds_char_rate = hdmi_funcs->supported_tmds_char_rate;
>>>>  	}
>>
>> [...]
>>
>>>> +	/**
>>>> +	 * @scdc_work: Work item currently used to monitor sink-side scrambling
>>>> +	 * status and retry setup if the sink resets it.
>>>> +	 */
>>>> +	struct delayed_work scdc_work;
>>>> +
>>>> +	/**
>>>> +	 * @scdc_work_initialized: Tracks whether @scdc_work has been set up via
>>>> +	 * INIT_DELAYED_WORK(). The work item is initialized lazily on the first
>>>> +	 * scrambling enable, so this guards the teardown paths against touching
>>>> +	 * an uninitialized work item.
>>>> +	 */
>>>> +	bool scdc_work_initialized;
>>>> +
>>>
>>> Why should we track whether it's initialized or not? I'd always
>>> initialize it, but only ever schedule something if we're using the
>>> scrambler.
>>
>> Having this initialized in the connector would lead to a module dependency
>> cycle.
>>
>> Currently the work function lives in drm_hdmi_helper.c, which is built into
>> drm_display_helper module:
>>
>> static void drm_connector_hdmi_scdc_work(struct work_struct *work)
>> {
>> 	[...]
>> 	if (READ_ONCE(connector->hdmi.scrambler_enabled) &&
>> 	    !drm_scdc_get_scrambling_status(connector))
>> 		drm_connector_hdmi_try_scrambling_setup(connector);
>> 	[...]
>> }
>>
>> int drm_connector_hdmi_enable_scrambling(struct drm_connector *connector,
>> 					 const struct drm_connector_state *conn_state)
>> {
>>
>> 	[...]
>> 	if (!hdmi->scdc_work_initialized) {
>> 		INIT_DELAYED_WORK(&hdmi->scdc_work,
>> 				  drm_connector_hdmi_scdc_work);
>> 		hdmi->scdc_work_initialized = true;
>> 	}
>> 	[...]
>> }
>>
>> If we move INIT_DELAYED_WORK() into the connector (i.e. in drm.ko), the work
>> function has to be reachable from there.  The following attempts to accomplish
>> that would fail:
>>
>> - Keep the work function in drm_hdmi_helper.c and export it from
>>   drm_display_helper.
>>
>> - Move the work function into drm_connector.c and export 
>>   drm_connector_hdmi_try_scrambling_setup(), or a wrapper function, from 
>>   drm_display_helper.
> 
> An alternative could be to move drm_connector_hdmi_init to
> drm_hdmi_helper.c, no?

I haven't considered this option so far, as I believe it would also require some
refactoring to get right - for example, moving HDMI-related initialization from
the generic drm_connector_init_only() to drm_connector_hdmi_init(), and
splitting drm_connector_cleanup() into a dedicated drm_connector_hdmi_cleanup()
utility.

> But yeah, if we can't let's keep it like that

Should I proceed with this refactoring, or would it be better to postpone it
until I send out the HDMI 2.1 patches, to avoid expanding this series even
further?

Thanks,
Cristian