[PATCH] drm/i915/display/dp: 128/132b DP-capable with SST

[Jani Nikula]

On Tue, 09 Jan 2024, Jani Nikula <jani.nikula at intel.com> wrote:
> On Tue, 09 Jan 2024, "Murthy, Arun R" <arun.r.murthy at intel.com> wrote:
>>> -----Original Message-----
>>> From: Nikula, Jani <jani.nikula at intel.com>
>>> Sent: Monday, January 8, 2024 7:01 PM
>>> To: Murthy, Arun R <arun.r.murthy at intel.com>; intel-gfx at lists.freedesktop.org;
>>> Deak, Imre <imre.deak at intel.com>
>>> Cc: Murthy, Arun R <arun.r.murthy at intel.com>
>>> Subject: Re: [PATCH] drm/i915/display/dp: 128/132b DP-capable with SST
>>>
>>> On Wed, 03 Jan 2024, Arun R Murthy <arun.r.murthy at intel.com> wrote:
>>> > With a value of '0' read from MSTM_CAP register MST to be enabled.
>>> > DP2.1 SCR updates the spec for 128/132b DP capable supporting only one
>>> > stream and not supporting single stream sideband MSG.
>>> > The underlying protocol will be MST to enable use of MTP.
>>> >
>>> > Signed-off-by: Arun R Murthy <arun.r.murthy at intel.com>
>>> > ---
>>> >  drivers/gpu/drm/i915/display/intel_dp.c | 9 +++++++--
>>> >  1 file changed, 7 insertions(+), 2 deletions(-)
>>> >
>>> > diff --git a/drivers/gpu/drm/i915/display/intel_dp.c
>>> > b/drivers/gpu/drm/i915/display/intel_dp.c
>>> > index 9ff0cbd9c0df..40d3280f8d98 100644
>>> > --- a/drivers/gpu/drm/i915/display/intel_dp.c
>>> > +++ b/drivers/gpu/drm/i915/display/intel_dp.c
>>> > @@ -4038,8 +4038,13 @@ intel_dp_configure_mst(struct intel_dp *intel_dp)
>>> >     if (!intel_dp_mst_source_support(intel_dp))
>>> >             return;
>>> >
>>> > -   intel_dp->is_mst = sink_can_mst &&
>>> > -           i915->display.params.enable_dp_mst;
>>> > +   /*
>>> > +    * Even if dpcd reg MSTM_CAP is 0, if the sink supports UHBR rates
>>> then
>>> > +    * DP2.1 can be enabled with underlying protocol using MST for MTP
>>> > +    */
>>> > +   intel_dp->is_mst = (sink_can_mst ||
>>> > +
>>> drm_dp_is_uhbr_rate(intel_dp_max_common_rate(intel_dp)))
>>> > +                       && i915->display.params.enable_dp_mst;
>>>
>>> We use drm_dp_is_uhbr_rate() in intel_dp_is_uhbr() to determine whether the
>>> link rate in the *crtc state* is uhbr, and by proxy whether the link in the *crtc
>>> state* is 128b/132b.
>>>
>> Yes! If the link rate in crtc_state is not uhbr then we dont enable 128/132b. Also the return from drm_dp_read_mst_cap() return 0 or 1 and if not mst then 128/132b is not enabled. We need to deviate here and a value of bit[0]=0, bit[1]=0 in MSTM_CAP register is 128/132b with SST. Hence this hack is added to see if the return from MSTM_CAP is 0x00 and if uhbr rates are supported then enable 128/132b.
>
> Per spec it depends on intel_dp->dpcd[DP_MAIN_LINK_CHANNEL_CODING] &
> DP_CAP_ANSI_128B132B, why not use that here too?

Also, shouldn't we ensure we don't try to do more than one stream?

>
>>
>>> There, we've already decided to use uhbr and 128b/132b.
>>>
>>> This one here is different, and I think it's taking us to the wrong direction. For
>>> example, it should be possible to downgrade the link from uhbr to non-uhbr on
>>> link failures. We don't have that yet. But this makes untangling that even
>>> harder.
>>
>> Yes upon having the fallback, I think we will have to handle fallback in this case separately. Change might be required in drm core, drm_dp_read_mst_cap() should consider the DP2.1 changes to accommodate this 0x00 value.
>>
>> Will be floating the fallback patches very soon.
>>
>> Thanks and Regards,
>> Arun R Murthy
>> --------------------
>>>
>>>
>>> BR,
>>> Jani.
>>>
>>>
>>> >
>>> >     drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr,
>>> >                                     intel_dp->is_mst);
>>>
>>> --
>>> Jani Nikula, Intel

-- 
Jani Nikula, Intel