To properly define the USB-C DP altmode connectors, add the USB
subconnector type.
Suggested-by: Simon Ser <[email protected]>
Signed-off-by: Dmitry Baryshkov <[email protected]>
---
drivers/gpu/drm/drm_connector.c | 1 +
include/uapi/drm/drm_mode.h | 1 +
2 files changed, 2 insertions(+)
diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
index a6066e4a5e9a..9e96b038f5d0 100644
--- a/drivers/gpu/drm/drm_connector.c
+++ b/drivers/gpu/drm/drm_connector.c
@@ -1050,6 +1050,7 @@ static const struct drm_prop_enum_list drm_dp_subconnector_enum_list[] = {
{ DRM_MODE_SUBCONNECTOR_DisplayPort, "DP" }, /* DP */
{ DRM_MODE_SUBCONNECTOR_Wireless, "Wireless" }, /* DP */
{ DRM_MODE_SUBCONNECTOR_Native, "Native" }, /* DP */
+ { DRM_MODE_SUBCONNECTOR_USB, "USB" }, /* DP */
};
DRM_ENUM_NAME_FN(drm_get_dp_subconnector_name,
diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
index 92d96a2b6763..0f74918b011c 100644
--- a/include/uapi/drm/drm_mode.h
+++ b/include/uapi/drm/drm_mode.h
@@ -398,6 +398,7 @@ enum drm_mode_subconnector {
DRM_MODE_SUBCONNECTOR_HDMIA = 11, /* DP */
DRM_MODE_SUBCONNECTOR_Native = 15, /* DP */
DRM_MODE_SUBCONNECTOR_Wireless = 18, /* DP */
+ DRM_MODE_SUBCONNECTOR_USB = 20, /* DP */
};
#define DRM_MODE_CONNECTOR_Unknown 0
--
2.39.2
2 августа 2023 г. 22:13:51 GMT+03:00, Laurent Pinchart <[email protected]> пишет:
>On Wed, Aug 02, 2023 at 10:01:19PM +0300, Dmitry Baryshkov wrote:
>> On 02/08/2023 21:55, Laurent Pinchart wrote:
>> > Hi Dmitry,
>> >
>> > Thank you for the patch.
>> >
>> > On Sat, Jul 29, 2023 at 03:49:12AM +0300, Dmitry Baryshkov wrote:
>> >> To properly define the USB-C DP altmode connectors, add the USB
>> >> subconnector type.
>> >>
>> >> Suggested-by: Simon Ser <[email protected]>
>> >> Signed-off-by: Dmitry Baryshkov <[email protected]>
>> >> ---
>> >> drivers/gpu/drm/drm_connector.c | 1 +
>> >> include/uapi/drm/drm_mode.h | 1 +
>> >> 2 files changed, 2 insertions(+)
>> >>
>> >> diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
>> >> index a6066e4a5e9a..9e96b038f5d0 100644
>> >> --- a/drivers/gpu/drm/drm_connector.c
>> >> +++ b/drivers/gpu/drm/drm_connector.c
>> >> @@ -1050,6 +1050,7 @@ static const struct drm_prop_enum_list drm_dp_subconnector_enum_list[] = {
>> >> { DRM_MODE_SUBCONNECTOR_DisplayPort, "DP" }, /* DP */
>> >> { DRM_MODE_SUBCONNECTOR_Wireless, "Wireless" }, /* DP */
>> >> { DRM_MODE_SUBCONNECTOR_Native, "Native" }, /* DP */
>> >> + { DRM_MODE_SUBCONNECTOR_USB, "USB" }, /* DP */
>> >
>> > Should this be DRM_MODE_SUBCONNECTOR_USB_C and "USB-C", in case we get
>> > another USB type later ?
>>
>> Hmm, which id should I use for micro-USB then? (consider anx7808,
>> SlimPort). I thought about using DRM_MODE_SUBCONNECTOR_USB for both of
>> them. But maybe I should add another subtype for SlimPort.
>
>I suppose it depends on whether userspace needs a way to differentiate
>those. Do you have a good visibility on the userspace use cases ?
No. I'm not even sure, which userspace handles subtypes properly.
For the reference, SlimPort is mostly legacy hardware, think about Nexus 4, 5, 6, 7 (2013)
>
>> >> };
>> >>
>> >> DRM_ENUM_NAME_FN(drm_get_dp_subconnector_name,
>> >> diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
>> >> index 92d96a2b6763..0f74918b011c 100644
>> >> --- a/include/uapi/drm/drm_mode.h
>> >> +++ b/include/uapi/drm/drm_mode.h
>> >> @@ -398,6 +398,7 @@ enum drm_mode_subconnector {
>> >> DRM_MODE_SUBCONNECTOR_HDMIA = 11, /* DP */
>> >> DRM_MODE_SUBCONNECTOR_Native = 15, /* DP */
>> >> DRM_MODE_SUBCONNECTOR_Wireless = 18, /* DP */
>> >> + DRM_MODE_SUBCONNECTOR_USB = 20, /* DP */
>> >> };
>> >>
>> >> #define DRM_MODE_CONNECTOR_Unknown 0
>
On 02/08/2023 21:55, Laurent Pinchart wrote:
> Hi Dmitry,
>
> Thank you for the patch.
>
> On Sat, Jul 29, 2023 at 03:49:12AM +0300, Dmitry Baryshkov wrote:
>> To properly define the USB-C DP altmode connectors, add the USB
>> subconnector type.
>>
>> Suggested-by: Simon Ser <[email protected]>
>> Signed-off-by: Dmitry Baryshkov <[email protected]>
>> ---
>> drivers/gpu/drm/drm_connector.c | 1 +
>> include/uapi/drm/drm_mode.h | 1 +
>> 2 files changed, 2 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
>> index a6066e4a5e9a..9e96b038f5d0 100644
>> --- a/drivers/gpu/drm/drm_connector.c
>> +++ b/drivers/gpu/drm/drm_connector.c
>> @@ -1050,6 +1050,7 @@ static const struct drm_prop_enum_list drm_dp_subconnector_enum_list[] = {
>> { DRM_MODE_SUBCONNECTOR_DisplayPort, "DP" }, /* DP */
>> { DRM_MODE_SUBCONNECTOR_Wireless, "Wireless" }, /* DP */
>> { DRM_MODE_SUBCONNECTOR_Native, "Native" }, /* DP */
>> + { DRM_MODE_SUBCONNECTOR_USB, "USB" }, /* DP */
>
> Should this be DRM_MODE_SUBCONNECTOR_USB_C and "USB-C", in case we get
> another USB type later ?
Hmm, which id should I use for micro-USB then? (consider anx7808,
SlimPort). I thought about using DRM_MODE_SUBCONNECTOR_USB for both of
them. But maybe I should add another subtype for SlimPort.
>
>> };
>>
>> DRM_ENUM_NAME_FN(drm_get_dp_subconnector_name,
>> diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
>> index 92d96a2b6763..0f74918b011c 100644
>> --- a/include/uapi/drm/drm_mode.h
>> +++ b/include/uapi/drm/drm_mode.h
>> @@ -398,6 +398,7 @@ enum drm_mode_subconnector {
>> DRM_MODE_SUBCONNECTOR_HDMIA = 11, /* DP */
>> DRM_MODE_SUBCONNECTOR_Native = 15, /* DP */
>> DRM_MODE_SUBCONNECTOR_Wireless = 18, /* DP */
>> + DRM_MODE_SUBCONNECTOR_USB = 20, /* DP */
>> };
>>
>> #define DRM_MODE_CONNECTOR_Unknown 0
>
--
With best wishes
Dmitry
On Wed, Aug 02, 2023 at 10:01:19PM +0300, Dmitry Baryshkov wrote:
> On 02/08/2023 21:55, Laurent Pinchart wrote:
> > Hi Dmitry,
> >
> > Thank you for the patch.
> >
> > On Sat, Jul 29, 2023 at 03:49:12AM +0300, Dmitry Baryshkov wrote:
> >> To properly define the USB-C DP altmode connectors, add the USB
> >> subconnector type.
> >>
> >> Suggested-by: Simon Ser <[email protected]>
> >> Signed-off-by: Dmitry Baryshkov <[email protected]>
> >> ---
> >> drivers/gpu/drm/drm_connector.c | 1 +
> >> include/uapi/drm/drm_mode.h | 1 +
> >> 2 files changed, 2 insertions(+)
> >>
> >> diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
> >> index a6066e4a5e9a..9e96b038f5d0 100644
> >> --- a/drivers/gpu/drm/drm_connector.c
> >> +++ b/drivers/gpu/drm/drm_connector.c
> >> @@ -1050,6 +1050,7 @@ static const struct drm_prop_enum_list drm_dp_subconnector_enum_list[] = {
> >> { DRM_MODE_SUBCONNECTOR_DisplayPort, "DP" }, /* DP */
> >> { DRM_MODE_SUBCONNECTOR_Wireless, "Wireless" }, /* DP */
> >> { DRM_MODE_SUBCONNECTOR_Native, "Native" }, /* DP */
> >> + { DRM_MODE_SUBCONNECTOR_USB, "USB" }, /* DP */
> >
> > Should this be DRM_MODE_SUBCONNECTOR_USB_C and "USB-C", in case we get
> > another USB type later ?
>
> Hmm, which id should I use for micro-USB then? (consider anx7808,
> SlimPort). I thought about using DRM_MODE_SUBCONNECTOR_USB for both of
> them. But maybe I should add another subtype for SlimPort.
I suppose it depends on whether userspace needs a way to differentiate
those. Do you have a good visibility on the userspace use cases ?
> >> };
> >>
> >> DRM_ENUM_NAME_FN(drm_get_dp_subconnector_name,
> >> diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
> >> index 92d96a2b6763..0f74918b011c 100644
> >> --- a/include/uapi/drm/drm_mode.h
> >> +++ b/include/uapi/drm/drm_mode.h
> >> @@ -398,6 +398,7 @@ enum drm_mode_subconnector {
> >> DRM_MODE_SUBCONNECTOR_HDMIA = 11, /* DP */
> >> DRM_MODE_SUBCONNECTOR_Native = 15, /* DP */
> >> DRM_MODE_SUBCONNECTOR_Wireless = 18, /* DP */
> >> + DRM_MODE_SUBCONNECTOR_USB = 20, /* DP */
> >> };
> >>
> >> #define DRM_MODE_CONNECTOR_Unknown 0
--
Regards,
Laurent Pinchart
Hi Dmitry,
Thank you for the patch.
On Sat, Jul 29, 2023 at 03:49:12AM +0300, Dmitry Baryshkov wrote:
> To properly define the USB-C DP altmode connectors, add the USB
> subconnector type.
>
> Suggested-by: Simon Ser <[email protected]>
> Signed-off-by: Dmitry Baryshkov <[email protected]>
> ---
> drivers/gpu/drm/drm_connector.c | 1 +
> include/uapi/drm/drm_mode.h | 1 +
> 2 files changed, 2 insertions(+)
>
> diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
> index a6066e4a5e9a..9e96b038f5d0 100644
> --- a/drivers/gpu/drm/drm_connector.c
> +++ b/drivers/gpu/drm/drm_connector.c
> @@ -1050,6 +1050,7 @@ static const struct drm_prop_enum_list drm_dp_subconnector_enum_list[] = {
> { DRM_MODE_SUBCONNECTOR_DisplayPort, "DP" }, /* DP */
> { DRM_MODE_SUBCONNECTOR_Wireless, "Wireless" }, /* DP */
> { DRM_MODE_SUBCONNECTOR_Native, "Native" }, /* DP */
> + { DRM_MODE_SUBCONNECTOR_USB, "USB" }, /* DP */
Should this be DRM_MODE_SUBCONNECTOR_USB_C and "USB-C", in case we get
another USB type later ?
> };
>
> DRM_ENUM_NAME_FN(drm_get_dp_subconnector_name,
> diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
> index 92d96a2b6763..0f74918b011c 100644
> --- a/include/uapi/drm/drm_mode.h
> +++ b/include/uapi/drm/drm_mode.h
> @@ -398,6 +398,7 @@ enum drm_mode_subconnector {
> DRM_MODE_SUBCONNECTOR_HDMIA = 11, /* DP */
> DRM_MODE_SUBCONNECTOR_Native = 15, /* DP */
> DRM_MODE_SUBCONNECTOR_Wireless = 18, /* DP */
> + DRM_MODE_SUBCONNECTOR_USB = 20, /* DP */
> };
>
> #define DRM_MODE_CONNECTOR_Unknown 0
--
Regards,
Laurent Pinchart
On Thu, 3 Aug 2023 at 18:31, Simon Ser <[email protected]> wrote:
>
> On Thursday, August 3rd, 2023 at 17:22, Simon Ser <[email protected]> wrote:
>
> > The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> > Can USB-C (or USB) be seen as a DP downstream port?
>
> To expand on this a bit: I'm wondering if we're mixing apples and
> oranges here. The current values of "subconnector" typically describe
> the lower-level protocol tunneled inside DP. For instance, VGA can be
> tunneled inside the DP cable when using DP → VGA adapter.
My opinion hasn't changed: I think this should be the USB connector
with proper DP / DVI / HDMI / etc. subconnector type (or lack of it).
In the end, the physical connector on the side of laptop is USB-C.
If we want to make it different from GUD, we might want to define a
USB-DP connector type (which would also include SlimPort).
>
> However, in the USB-C case, DP itself is tunneled inside USB-C. And you
> might use a USB-C → DP adapter. So it's not really *sub*connector, it's
> more of a *super*connector, right?
>
> I think [1] is somewhat related, since it also allows user-space to
> discover whether a connector uses USB-C. But relying on sysfs to figure
> this out isn't super optimal perhaps.
>
> [1]: https://lore.kernel.org/dri-devel/[email protected]/
--
With best wishes
Dmitry
On Thursday, August 3rd, 2023 at 17:22, Simon Ser <[email protected]> wrote:
> The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> Can USB-C (or USB) be seen as a DP downstream port?
To expand on this a bit: I'm wondering if we're mixing apples and
oranges here. The current values of "subconnector" typically describe
the lower-level protocol tunneled inside DP. For instance, VGA can be
tunneled inside the DP cable when using DP → VGA adapter.
However, in the USB-C case, DP itself is tunneled inside USB-C. And you
might use a USB-C → DP adapter. So it's not really *sub*connector, it's
more of a *super*connector, right?
I think [1] is somewhat related, since it also allows user-space to
discover whether a connector uses USB-C. But relying on sysfs to figure
this out isn't super optimal perhaps.
[1]: https://lore.kernel.org/dri-devel/[email protected]/
On Wednesday, August 2nd, 2023 at 21:23, Dmitry Baryshkov <[email protected]> wrote:
> >> >> + { DRM_MODE_SUBCONNECTOR_USB, "USB" }, /* DP */
> >> >
> >> > Should this be DRM_MODE_SUBCONNECTOR_USB_C and "USB-C", in case we get
> >> > another USB type later ?
> >>
> >> Hmm, which id should I use for micro-USB then? (consider anx7808,
> >> SlimPort). I thought about using DRM_MODE_SUBCONNECTOR_USB for both of
> >> them. But maybe I should add another subtype for SlimPort.
> >
> >I suppose it depends on whether userspace needs a way to differentiate
> >those. Do you have a good visibility on the userspace use cases ?
>
> No. I'm not even sure, which userspace handles subtypes properly.
wlroots uses it for human-readable output descriptions, e.g.
> wayland-info
interface: 'wl_output', version: 4, name: 49
name: DP-3
description: Samsung Electric Company SyncMaster HS3P505873 (DP-3 via DVI-D)
The "via DVI-D" bit comes from subconnector.
The description is displayed to the user when picking an output to screen
capture, among other things. It is helpful to users because they can better
understand why their output connected via DVI shows up as "DP".
The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
Can USB-C (or USB) be seen as a DP downstream port?
On Thu, 3 Aug 2023 at 18:43, Simon Ser <[email protected]> wrote:
>
> On Thursday, August 3rd, 2023 at 17:36, Dmitry Baryshkov <[email protected]> wrote:
>
> > On Thu, 3 Aug 2023 at 18:31, Simon Ser [email protected] wrote:
> >
> > > On Thursday, August 3rd, 2023 at 17:22, Simon Ser [email protected] wrote:
> > >
> > > > The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> > > > Can USB-C (or USB) be seen as a DP downstream port?
> > >
> > > To expand on this a bit: I'm wondering if we're mixing apples and
> > > oranges here. The current values of "subconnector" typically describe
> > > the lower-level protocol tunneled inside DP. For instance, VGA can be
> > > tunneled inside the DP cable when using DP → VGA adapter.
> >
> > My opinion hasn't changed: I think this should be the USB connector
> > with proper DP / DVI / HDMI / etc. subconnector type (or lack of it).
> > In the end, the physical connector on the side of laptop is USB-C.
>
> - Even if the connector is USB-C, the protocol used for display is
> still DP. There's also the case of Thunderbolt.
Yes. But the connector type is not about the protocol.
> - This is inconsistent with existing drivers. i915 and amdgpu expose
> DP ports for their USB-C ports. Changing that isn't possible without
> causing user-space regressions (compositor config files use the
> connector type).
Yes, I know. Consider my phrase as a personal opinion or minority report.
I think that using DisplayPort for USB-C connectors was a mistake,
which we now have to cope with somehow.
--
With best wishes
Dmitry
On Thursday, August 3rd, 2023 at 17:36, Dmitry Baryshkov <[email protected]> wrote:
> On Thu, 3 Aug 2023 at 18:31, Simon Ser [email protected] wrote:
>
> > On Thursday, August 3rd, 2023 at 17:22, Simon Ser [email protected] wrote:
> >
> > > The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> > > Can USB-C (or USB) be seen as a DP downstream port?
> >
> > To expand on this a bit: I'm wondering if we're mixing apples and
> > oranges here. The current values of "subconnector" typically describe
> > the lower-level protocol tunneled inside DP. For instance, VGA can be
> > tunneled inside the DP cable when using DP → VGA adapter.
>
> My opinion hasn't changed: I think this should be the USB connector
> with proper DP / DVI / HDMI / etc. subconnector type (or lack of it).
> In the end, the physical connector on the side of laptop is USB-C.
- Even if the connector is USB-C, the protocol used for display is
still DP. There's also the case of Thunderbolt.
- This is inconsistent with existing drivers. i915 and amdgpu expose
DP ports for their USB-C ports. Changing that isn't possible without
causing user-space regressions (compositor config files use the
connector type).
On Thu, Aug 03, 2023 at 03:31:16PM +0000, Simon Ser wrote:
> On Thursday, August 3rd, 2023 at 17:22, Simon Ser <[email protected]> wrote:
>
> > The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> > Can USB-C (or USB) be seen as a DP downstream port?
>
> To expand on this a bit: I'm wondering if we're mixing apples and
> oranges here. The current values of "subconnector" typically describe
> the lower-level protocol tunneled inside DP. For instance, VGA can be
> tunneled inside the DP cable when using DP → VGA adapter.
Doesn't this contradict the example use case you gave in your previous
e-mail, with wlroots stating "DP-3 via DVI-D" ? I understand that as DP
carried over a DVI-D physical connector, did I get it wrong ?
> However, in the USB-C case, DP itself is tunneled inside USB-C. And you
> might use a USB-C → DP adapter. So it's not really *sub*connector, it's
> more of a *super*connector, right?
>
> I think [1] is somewhat related, since it also allows user-space to
> discover whether a connector uses USB-C. But relying on sysfs to figure
> this out isn't super optimal perhaps.
>
> [1]: https://lore.kernel.org/dri-devel/[email protected]/
--
Regards,
Laurent Pinchart
On Thu, 3 Aug 2023 at 23:47, Simon Ser <[email protected]> wrote:
>
> On Thursday, August 3rd, 2023 at 22:44, Laurent Pinchart <[email protected]> wrote:
>
> > On Thu, Aug 03, 2023 at 03:31:16PM +0000, Simon Ser wrote:
> >
> > > On Thursday, August 3rd, 2023 at 17:22, Simon Ser [email protected] wrote:
> > >
> > > > The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> > > > Can USB-C (or USB) be seen as a DP downstream port?
> > >
> > > To expand on this a bit: I'm wondering if we're mixing apples and
> > > oranges here. The current values of "subconnector" typically describe
> > > the lower-level protocol tunneled inside DP. For instance, VGA can be
> > > tunneled inside the DP cable when using DP → VGA adapter.
> >
> > Doesn't this contradict the example use case you gave in your previous
> > e-mail, with wlroots stating "DP-3 via DVI-D" ? I understand that as DP
> > carried over a DVI-D physical connector, did I get it wrong ?
>
> No, this is DVI carried over DP. DP cannot be carried over VGA/DVI/HDMI,
> but VGA/DVI/HDMI can be carried over DP.
Well, not quite. It means that the sink (display) connected to this
port identifies itself as an VGA / DVI / HDMI monitor.
E.g. on if connect HDMI/DVI monitor through the DP-DVI dongle (native
DP connector), AMD driver still identifies it as 'subconnector HDMI',
despite dongle a DVI connector on the other side.
--
With best wishes
Dmitry
On Thursday, August 3rd, 2023 at 22:44, Laurent Pinchart <[email protected]> wrote:
> On Thu, Aug 03, 2023 at 03:31:16PM +0000, Simon Ser wrote:
>
> > On Thursday, August 3rd, 2023 at 17:22, Simon Ser [email protected] wrote:
> >
> > > The KMS docs describe "subconnector" to be defined as "downstream port" for DP.
> > > Can USB-C (or USB) be seen as a DP downstream port?
> >
> > To expand on this a bit: I'm wondering if we're mixing apples and
> > oranges here. The current values of "subconnector" typically describe
> > the lower-level protocol tunneled inside DP. For instance, VGA can be
> > tunneled inside the DP cable when using DP → VGA adapter.
>
> Doesn't this contradict the example use case you gave in your previous
> e-mail, with wlroots stating "DP-3 via DVI-D" ? I understand that as DP
> carried over a DVI-D physical connector, did I get it wrong ?
No, this is DVI carried over DP. DP cannot be carried over VGA/DVI/HDMI,
but VGA/DVI/HDMI can be carried over DP.