[Intel-gfx] [RFC v3] drm/hdcp: drm enum property for CP State

[Sean Paul]

On Wed, Jul 26, 2017 at 12:51 PM, C, Ramalingam <ramalingam.c at intel.com> wrote:
>
>> -----Original Message-----
>> From: Sean Paul [mailto:seanpaul at chromium.org]
>> Sent: Wednesday, July 26, 2017 8:23 PM
>> To: C, Ramalingam <ramalingam.c at intel.com>
>> Cc: Sean Paul <seanpaul at chromium.org>; Vetter, Daniel
>> <daniel.vetter at intel.com>; Intel Graphics Development <intel-
>> gfx at lists.freedesktop.org>; dri-devel <dri-devel at lists.freedesktop.org>; Daniel
>> Vetter <daniel at ffwll.ch>
>> Subject: Re: [RFC v3] drm/hdcp: drm enum property for CP State
>>
>> On Wed, Jul 26, 2017 at 03:24:10PM +0530, Ramalingam C wrote:
>> >
>> >
>> > On Tuesday 25 July 2017 06:04 PM, Sean Paul wrote:
>> > > On Mon, Jul 24, 2017 at 2:12 PM, Ramalingam C <ramalingam.c at intel.com>
>> wrote:
>> > > > DRM connector property is created to represent the content
>> > > > protection state of the connector and to configure the same.
>> > > >
>> > > > Content protection states defined:
>> > > >          DRM_MODE_CONTENT_PROTECTION_UNSUPPORTED         -
>> Unsupported
>> > > >          DRM_MODE_CONTENT_PROTECTION_DISABLE             - Disabled
>> > > >          DRM_MODE_CONTENT_PROTECTION_ENABLE              - Enabled
>> > > >
>> > > > v2: Redesigned the property to match with CP needs of CrOS [Sean].
>> > > >
>> > > > v3: Renamed the state names. Header is removed [sean].
>> > > >
>> > > > Signed-off-by: Ramalingam C <ramalingam.c at intel.com>
>> > > > ---
>> > > >   drivers/gpu/drm/drm_connector.c | 14 ++++++++++++++
>> > > >   include/drm/drm_mode_config.h   |  5 +++++
>> > > >   include/uapi/drm/drm_mode.h     |  5 +++++
>> > > >   3 files changed, 24 insertions(+)
>> > > >
>> > > > diff --git a/drivers/gpu/drm/drm_connector.c
>> > > > b/drivers/gpu/drm/drm_connector.c index 5cd61af..d6aaa08 100644
>> > > > --- a/drivers/gpu/drm/drm_connector.c
>> > > > +++ b/drivers/gpu/drm/drm_connector.c
>> > > > @@ -617,6 +617,13 @@ static const struct drm_prop_enum_list
>> drm_link_status_enum_list[] = {
>> > > >   };
>> > > >   DRM_ENUM_NAME_FN(drm_get_link_status_name,
>> > > > drm_link_status_enum_list)
>> > > >
>> > > > +static const struct drm_prop_enum_list drm_cp_enum_list[] = {
>> > > > +       { DRM_MODE_CONTENT_PROTECTION_UNSUPPORTED,
>> "Unsupported" },
>> > > You're still changing the enum names from the original patch/CrOS
>> > > implementation.
>> > >
>> > > https://lists.freedesktop.org/archives/dri-devel/2014-December/07333
>> > > 6.html
>> > >
>> > > https://cs.chromium.org/chromium/src/ui/ozone/platform/drm/gpu/drm_d
>> > > isplay.cc?l=27
>> >
>> > Sean,
>> >
>> > I think we have bit of confusion here.
>>
>> Agreed :)
>>
>> >
>> > And in previous discussion you were fine with new state of property
>> > that is "unsupported" - indicates no common HDCP version is supported
>> > on HDCP src and Sink combo.
>> >
>> > when CP is not possible, if property exist, userspace will interpret
>> > it as CP is supported and attempt for enabling.
>> > I prefer indicating Unsupported state than failing such requests blindly.
>> >
>> > In that case to interpret the new state, we need to change CrOS User
>> > space code.
>> >
>> > In the RFC you mentioned above, two version of uapi interfaces are
>> > discussed.
>> >         V1 uses single property to configure the CP and also for
>> > status indication.
>> >         V2 uses two properties one for configuring and another one for
>> > status of Content protection.
>> >
>> > But CrOS user space is currently using the V1 interface.
>> >
>> > which will be preferred approach right now (V1/V2)?
>> > In either way we need to change the CrOS :(
>>
>> We can't upstream code without a userspace implementation, so V1. Without
>> any modifications.
>
> Based on uAPI rules mentioned by Daniel, what I infer is that once a uAPI is in place,
> usecase of uAPI cant be changed. It can be only be extended for future needs.
>
> So IMHO it is better to shape the userspace as per the uAPI
> interfaces that community want to have for longer run. And mainly we should adapt to a
> interface which can be conveniently extended for HDCP2.2 and future specs versions.
>
> So IMO it is worth seeing what it takes for CrOS Userspace changes for HDCP2.2 interfaces.
> And In the interest of having HDCP2.2 in their stack sooner,  if the CrOS community or Google
>  is interested in providing that userspace support, it will be smoother to put the
> complete and preferred uAPI for HDCP2.2 and 1.4 in place.
>
> Sean, I think you will be better person to comment on this from both sides, CrOS and DRM.
>
>
>>
>> As has been mentioned before in this thread, it's a lot easier to upstream code
>> that is proven to work, than it is to merge speculative API. Further, if this patch
>> were merged, it would be CrOS which is currently the only consumer... not a
>> good look :)
>
> Not exactly. CrOS will be the first consumer, Android will follow soon after this.
> We need to enable this feature on HwC for that purpose. Yup might take required time :)
>
>>
>>
>> >
>> > Lets say we need to modify CrOS user space, will there be any help for that?
>> > I am not aware what it takes though.
>>
>> You could probably reach out to chromium-dev at chromium.org or the author of
>> the Chrome hdcp implementation for help.
>
> Thank you for the contact.
> And before reaching out, it is better to conclude the HDCP1.4/2.2 uAPI interfaces that is agreeable for this community.
>
> In that perspective, can we consider and review the uAPI interface with two properties discussed for V2 of your earlier patches?
> Complete use case is discussed in my previous mail.
>
> Based on userspace commitment for HDCP2.2, we can add blob property for SRM too.
>
>>
>> Sean
>>
>> >
>> > I am trying to discuss both uapi versions of V1 and V2. I prefer V2 though.
>> >
>> > If V1 is preferred we need a single property as below "content
>> > protection" property  with {"Unsupported", "Undesired", "Desired",
>> > "Enabled"}
>> >                         "Type1_desired" and "Type1_Enabled" are needed
>> > for
>> > HDCP2.2
>> >
>> >
>> > If V2 is preferred we need two properties as below
>> >
>> > "content protection" property with {"Unsupported", "Undesired", "Desired"}
>> >             ("Type1_desired" - needed For HDCP2.2) "content protection
>> > status" property with {"Disabled", "Enabled"}
>> >             ("Type1_enabled" - needed for HDCP2.2)
>
> This uAPI interface looks fit to serve the all needs of HDCP1.4 and 2.2 as explained below.
>
> Sean and Daniel,
>         I request you to share your views. Thanks.

Just going to summarize what I mentioned on IRC for the benefit of the list.

Any change to the original proposal will require a userspace
implementation to go along with it. We currently only have one
userspace implementation, which is Chrome. Chrome implements the
original RFC (linked above). If you change anything from the original
RFC, you no longer have a userspace implementation.

In order to have a userspace implementation, you can choose to alter
Chrome or another compositor.

In order to change Chrome, I suspect you'll need to do one of the
following to avoid breaking devices running old kernels with the RFC
uapi:
1- Make the proposed uapi changes compatible with the original RFC
2- Change all CrOS kernels to the new proposal at the same time as
userspace changes (this is actually possible with CrOS since it's all
packaged/updated up together)

In my opinion, option 1 is going to be the easiest path forward.
Others have also suggested same. So before we get too far ahead of
ourselves with HDCP 2.2, let's figure out how to upstream *something*
with a userspace implementation. If you want to go for option 2,
you'll need to do some legwork in Chrome beforehand.

Sean


>
> --Ram


<snip>

>> > > > --
>> > > > 2.7.4
>> > > >
>> >
>>
>> --
>> Sean Paul, Software Engineer, Google / Chromium OS