Re: [PATCH v4 15/25] drm/msm/dpu: Add CWB to msm_display_topology

[Jessica Zhang <quic_jesszhan@quicinc.com>]

On 1/9/2025 4:00 PM, Dmitry Baryshkov wrote:
> On Thu, Jan 09, 2025 at 02:34:44PM -0800, Jessica Zhang wrote:
>>
>>
>> On 1/3/2025 10:16 AM, Dmitry Baryshkov wrote:
>>> On Fri, Jan 03, 2025 at 10:03:35AM -0800, Jessica Zhang wrote:
>>>>
>>>>
>>>> On 12/19/2024 9:03 PM, Dmitry Baryshkov wrote:
>>>>> On Mon, Dec 16, 2024 at 04:43:26PM -0800, Jessica Zhang wrote:
>>>>>> Add the cwb_enabled flag to msm_display topology and adjust the toplogy
>>>>>> to account for concurrent writeback
>>>>>
>>>>> Why?
>>>>
>>>> Hi Dmitry,
>>>>
>>>> This flag is necessary to specify that CWB mux(es) need to be assigned for
>>>> the given reqeusted topology.
>>>
>>> Why is necessary? Please rephrase your statement (we need foo bar, so do
>>> baz).
>>
>> Ack, what do you think of rephrasing the commit msg to this:
>>
>> ```
>> Add support for adjusting topology based on if concurrent writeback is
>> enabled.
>>
>> Currently, the topology is calculated based on the assumption that the user
>> cannot request real-time and writeback simultaneously. For example, the
>> number of LMs and CTLs are currently based off the number of phys encoders
>> under the assumption there will be at least 1 LM/CTL per phys encoder.
>>
>> This will not hold true for concurrent writeback as 2 phys encoders (1
>> real-time and 1 writeback) can be driven by 1 LM/CTL when concurrent
>> writeback is enabled.
>>
>> To account for this, add a cwb_enabled flag and only adjust the number of
>> CTL/LMs needed by a given topology based on the number of phys encoders only
>> if CWB is not enabled.
>>
>> ```
>>
>>>
>>>>
>>>>>
>>>>>>
>>>>>> Signed-off-by: Jessica Zhang <quic_jesszhan@quicinc.com>
>>>>>> ---
>>>>>>     drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c | 11 ++++++++++-
>>>>>>     drivers/gpu/drm/msm/disp/dpu1/dpu_rm.c   | 10 ++++++++--
>>>>>>     drivers/gpu/drm/msm/disp/dpu1/dpu_rm.h   |  2 ++
>>>>>>     3 files changed, 20 insertions(+), 3 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>>>>>> index b4bfded3d53025853cee112ca598533ece290318..b063c8fe4c0594772d84401fa56c9c21afc0ad18 100644
>>>>>> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>>>>>> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>>>>>> @@ -1198,6 +1198,8 @@ static struct msm_display_topology dpu_crtc_get_topology(
>>>>>>     		dpu_encoder_update_topology(drm_enc, &topology, crtc_state->state,
>>>>>>     					    &crtc_state->adjusted_mode);
>>>>>> +	topology.cwb_enabled = drm_crtc_in_clone_mode(crtc_state);
>>>>>> +
>>>>>>     	/*
>>>>>>     	 * Datapath topology selection
>>>>>>     	 *
>>>>>> @@ -1209,9 +1211,16 @@ static struct msm_display_topology dpu_crtc_get_topology(
>>>>>>     	 * 2 LM, 1 INTF (stream merge to support high resolution interfaces)
>>>>>>     	 *
>>>>>>     	 * Add dspps to the reservation requirements if ctm is requested
>>>>>> +	 *
>>>>>> +	 * Only hardcode num_lm to 2 for cases where num_intf == 2 and CWB is not
>>>>>> +	 * enabled. This is because in cases where CWB is enabled, num_intf will
>>>>>> +	 * count both the WB and real-time phys encoders.
>>>>>> +	 *
>>>>>> +	 * For non-DSC CWB usecases, have the num_lm be decided by the
>>>>>> +	 * (mode->hdisplay > MAX_HDISPLAY_SPLIT) check.
>>>>>>     	 */
>>>>>> -	if (topology.num_intf == 2)
>>>>>> +	if (topology.num_intf == 2 && !topology.cwb_enabled)
>>>>>>     		topology.num_lm = 2;
>>>>>>     	else if (topology.num_dsc == 2)
>>>>>>     		topology.num_lm = 2;
>>>>>> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.c
>>>>>> index b763ef19f4c60ae8a35df6a6ffb19e8411bc63f8..85adaf256b2c705d2d7df378b6ffc0e578f52bc3 100644
>>>>>> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.c
>>>>>> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.c
>>>>>> @@ -382,8 +382,14 @@ static int _dpu_rm_reserve_ctls(
>>>>>>     	int i = 0, j, num_ctls;
>>>>>>     	bool needs_split_display;
>>>>>> -	/* each hw_intf needs its own hw_ctrl to program its control path */
>>>>>> -	num_ctls = top->num_intf;
>>>>>> +	/*
>>>>>> +	 * For non-CWB mode, each hw_intf needs its own hw_ctl to program its
>>>>>> +	 * control path. Hardcode num_ctls to 1 if CWB is enabled
>>>>>> +	 */
>>>>>
>>>>> Why?
>>>>
>>>> This is because num_intf is based on the number of phys_encs. Since in the
>>>> CWB case, the WB and real-time encoders will be driven by the same CTL. I
>>>> can add this to the comment doc.
>>>
>>> Why are they driven by the same CTL? Is it also the case for platforms
>>> before DPU 5.x?
>>
>> This is because the WB and real-time path for a given topology would be
>> driven by the same data path so the same CTL should enable the real-time and
>> WB active bits.
>>
>> This is the same for pre-DPU 5.x.
> 
> But pre-5.x platforms didn't have ACTIVE_CTL, so they should be using
> separte CTL for each of the physical encoders.

For pre-DPU 5.x, enabling CWB would mean configuring the registers under 
both the WB and MODE_SEL_* cases here [1]

[1] 
https://elixir.bootlin.com/linux/v6.12.6/source/drivers/gpu/drm/msm/disp/dpu1/dpu_hw_ctl.c#L588

> 
>>
>>>
>>>>
>>>> Thanks,
>>>>
>>>> Jessica Zhang
>>>>
>>>>>
>>>>>> +	if (top->cwb_enabled)
>>>>>> +		num_ctls = 1;
>>>>>> +	else
>>>>>> +		num_ctls = top->num_intf;
>>>>>>     	needs_split_display = _dpu_rm_needs_split_display(top);
>>>>>> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.h b/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.h
>>>>>> index b061dfdab52e04ab7d777e912a30173273cb3db7..12db21a2403ec6930894c36a58e898c5d94c2568 100644
>>>>>> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.h
>>>>>> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_rm.h
>>>>>> @@ -46,6 +46,7 @@ struct dpu_rm {
>>>>>>      * @num_dspp:     number of dspp blocks used
>>>>>>      * @num_dsc:      number of Display Stream Compression (DSC) blocks used
>>>>>>      * @needs_cdm:    indicates whether cdm block is needed for this display topology
>>>>>> + * @cwb_enabled:  indicates whether CWB is enabled for this display topology
>>>>>>      */
>>>>>>     struct msm_display_topology {
>>>>>>     	u32 num_lm;
>>>>>> @@ -53,6 +54,7 @@ struct msm_display_topology {
>>>>>>     	u32 num_dspp;
>>>>>>     	u32 num_dsc;
>>>>>>     	bool needs_cdm;
>>>>>> +	bool cwb_enabled;
>>>>>>     };
>>>>>>     int dpu_rm_init(struct drm_device *dev,
>>>>>>
>>>>>> -- 
>>>>>> 2.34.1
>>>>>>
>>>>>
>>>>> -- 
>>>>> With best wishes
>>>>> Dmitry
>>>>
>>>
>>> -- 
>>> With best wishes
>>> Dmitry
>>
> 
> -- 
> With best wishes
> Dmitry