Re: [PATCH] drm: Update MST First Link Slot Information Based on Encoding Format

[Bhawanpreet Lakha <Bhawanpreet.lakha@amd.com>]

Adding Mikita aswell

On 2021-10-14 4:21 p.m., Bhawanpreet Lakha wrote:
>
> On 2021-10-13 6:25 p.m., Lyude Paul wrote:
>> Some comments below (also, sorry again for the mixup on the last 
>> review!)
>>
>> On Tue, 2021-10-12 at 17:58 -0400, Bhawanpreet Lakha wrote:
>>> 8b/10b encoding format requires to reserve the first slot for
>>> recording metadata. Real data transmission starts from the second slot,
>>> with a total of available 63 slots available.
>>>
>>> In 128b/132b encoding format, metadata is transmitted separately
>>> in LLCP packet before MTP. Real data transmission starts from
>>> the first slot, with a total of 64 slots available.
>>>
>>> v2:
>>> * Remove get_mst_link_encoding_cap
>>> * Move total/start slots to mst_state, and copy it to mst_mgr in
>>> atomic_check
>>>
>>> Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
>>> Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
>>> ---
>>>   .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 28 +++++++++++++++
>>>   drivers/gpu/drm/drm_dp_mst_topology.c         | 35 
>>> +++++++++++++++----
>>>   include/drm/drm_dp_mst_helper.h               | 13 +++++++
>>>   3 files changed, 69 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>>> b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>>> index 5020f2d36fe1..4ad50eb0091a 100644
>>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>>> @@ -10612,6 +10612,8 @@ static int amdgpu_dm_atomic_check(struct 
>>> drm_device
>>> *dev,
>>>   #if defined(CONFIG_DRM_AMD_DC_DCN)
>>>          struct dsc_mst_fairness_vars vars[MAX_PIPES];
>>>   #endif
>>> +       struct drm_dp_mst_topology_state *mst_state;
>>> +       struct drm_dp_mst_topology_mgr *mgr;
>>>            trace_amdgpu_dm_atomic_check_begin(state);
>>>   @@ -10819,6 +10821,32 @@ static int amdgpu_dm_atomic_check(struct 
>>> drm_device
>>> *dev,
>>>                  lock_and_validation_needed = true;
>>>          }
>>>   +#if defined(CONFIG_DRM_AMD_DC_DCN)
>>> +       for_each_new_mst_mgr_in_state(state, mgr, mst_state, i) {
>>> +               struct amdgpu_dm_connector *aconnector;
>>> +               struct drm_connector *connector;
>>> +               struct drm_connector_list_iter iter;
>>> +               u8 link_coding_cap;
>>> +
>>> +               if (!mgr->mst_state )
>>> +                       continue;
>> extraneous space
>>
>>> +
>>> +               drm_connector_list_iter_begin(dev, &iter);
>>> +               drm_for_each_connector_iter(connector, &iter) {
>>> +                       int id = connector->index;
>>> +
>>> +                       if (id == mst_state->mgr->conn_base_id) {
>>> +                               aconnector =
>>> to_amdgpu_dm_connector(connector);
>>> +                               link_coding_cap =
>>> dc_link_dp_mst_decide_link_encoding_format(aconnector->dc_link);
>>> +                               drm_dp_mst_update_coding_cap(mst_state,
>>> link_coding_cap);
>>> +
>>> +                               break;
>>> +                       }
>>> +               }
>>> +               drm_connector_list_iter_end(&iter);
>>> +
>>> +       }
>>> +#endif
>>>          /**
>>>           * Streams and planes are reset when there are changes that 
>>> affect
>>>           * bandwidth. Anything that affects bandwidth needs to go 
>>> through
>>> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
>>> b/drivers/gpu/drm/drm_dp_mst_topology.c
>>> index ad0795afc21c..fb5c47c4cb2e 100644
>>> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
>>> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
>>> @@ -3368,7 +3368,7 @@ int drm_dp_update_payload_part1(struct
>>> drm_dp_mst_topology_mgr *mgr)
>>>          struct drm_dp_payload req_payload;
>>>          struct drm_dp_mst_port *port;
>>>          int i, j;
>>> -       int cur_slots = 1;
>>> +       int cur_slots = mgr->start_slot;
>>>          bool skip;
>>>            mutex_lock(&mgr->payload_lock);
>>> @@ -4321,7 +4321,7 @@ int drm_dp_find_vcpi_slots(struct
>>> drm_dp_mst_topology_mgr *mgr,
>>>          num_slots = DIV_ROUND_UP(pbn, mgr->pbn_div);
>>>            /* max. time slots - one slot for MTP header */
>>> -       if (num_slots > 63)
>>> +       if (num_slots > mgr->total_avail_slots)
>>>                  return -ENOSPC;
>> For reasons I will explain a little further in this email, we might 
>> want to
>> drop this…
>>
>>>          return num_slots;
>>>   }
>>> @@ -4333,7 +4333,7 @@ static int drm_dp_init_vcpi(struct
>>> drm_dp_mst_topology_mgr *mgr,
>>>          int ret;
>>>            /* max. time slots - one slot for MTP header */
>>> -       if (slots > 63)
>>> +       if (slots > mgr->total_avail_slots)
>> …and this
>>
>>>                  return -ENOSPC;
>>>            vcpi->pbn = pbn;
>>> @@ -4507,6 +4507,18 @@ int drm_dp_atomic_release_vcpi_slots(struct
>>> drm_atomic_state *state,
>>>   }
>>>   EXPORT_SYMBOL(drm_dp_atomic_release_vcpi_slots);
>>>   +void drm_dp_mst_update_coding_cap(struct drm_dp_mst_topology_state
>>> *mst_state, uint8_t link_coding_cap)
>> Need some kdocs here
>>
>>> +{
>>> +       if (link_coding_cap == DP_CAP_ANSI_128B132B) {
>>> +               mst_state->total_avail_slots = 64;
>>> +               mst_state->start_slot = 0;
>>> +       }
>>> +
>>> +       DRM_DEBUG_KMS("%s coding format on mgr 0x%p\n",
>>> +                       (link_coding_cap == DP_CAP_ANSI_128B132B) ?
>>> "128b/132b":"8b/10b", mst_state->mgr);
>> need to fix indenting here, and wrap this to 100 chars
>>
>>> +}
>>> +EXPORT_SYMBOL(drm_dp_mst_update_coding_cap);
>>> +
>>>   /**
>>>    * drm_dp_mst_allocate_vcpi() - Allocate a virtual channel
>>>    * @mgr: manager for this port
>>> @@ -4538,8 +4550,8 @@ bool drm_dp_mst_allocate_vcpi(struct
>>> drm_dp_mst_topology_mgr *mgr,
>>>            ret = drm_dp_init_vcpi(mgr, &port->vcpi, pbn, slots);
>>>          if (ret) {
>>> -               drm_dbg_kms(mgr->dev, "failed to init vcpi slots=%d 
>>> max=63
>>> ret=%d\n",
>>> -                           DIV_ROUND_UP(pbn, mgr->pbn_div), ret);
>>> +               drm_dbg_kms(mgr->dev, "failed to init vcpi slots=%d 
>>> max=%d
>>> ret=%d\n",
>>> +                           DIV_ROUND_UP(pbn, mgr->pbn_div), mgr-
>>>> total_avail_slots, ret);
>>>                  drm_dp_mst_topology_put_port(port);
>>>                  goto out;
>>>          }
>>> @@ -5226,7 +5238,7 @@ drm_dp_mst_atomic_check_vcpi_alloc_limit(struct
>>> drm_dp_mst_topology_mgr *mgr,
>>>                                           struct 
>>> drm_dp_mst_topology_state
>>> *mst_state)
>>>   {
>>>          struct drm_dp_vcpi_allocation *vcpi;
>>> -       int avail_slots = 63, payload_count = 0;
>>> +       int avail_slots = mgr->total_avail_slots, payload_count = 0;
>>>            list_for_each_entry(vcpi, &mst_state->vcpis, next) {
>>>                  /* Releasing VCPI is always OK-even if the port is 
>>> gone */
>>> @@ -5255,7 +5267,7 @@ drm_dp_mst_atomic_check_vcpi_alloc_limit(struct
>>> drm_dp_mst_topology_mgr *mgr,
>>>                  }
>>>          }
>>>          drm_dbg_atomic(mgr->dev, "[MST MGR:%p] mst state %p VCPI 
>>> avail=%d
>>> used=%d\n",
>>> -                      mgr, mst_state, avail_slots, 63 - avail_slots);
>>> +                      mgr, mst_state, avail_slots, 
>>> mgr->total_avail_slots -
>>> avail_slots);
>>>            return 0;
>>>   }
>>> @@ -5421,6 +5433,10 @@ int drm_dp_mst_atomic_check(struct 
>>> drm_atomic_state
>>> *state)
>>>                          break;
>>>                    mutex_lock(&mgr->lock);
>>> +
>>> +               mgr->start_slot = mst_state->start_slot;
>>> +               mgr->total_avail_slots = mst_state->total_avail_slots;
>>> +
>> this isn't correct - atomic checks aren't allowed to mutate anything 
>> besides
>> the atomic state structure that we're checking since we don't know 
>> whether or
>> not the display state is going to be committed when checking it. If 
>> we're
>> storing state in mgr, that state needs to be written to mgr during 
>> the atomic
>> commit instead of the atomic check.
>>
>> ...but, coming back to this MST code after being gone for a while, I 
>> think it
>> might be time for us to stop worrying about the non-atomic state. 
>> Especially
>> since there's only one driver using it; the legacy radeon.ko; and 
>> right now
>> the plan is either to just drop MST support on radeon.ko or move the 
>> MST code
>> it uses into radeon.ko.Which brings me to say - I think we can drop 
>> some of
>> the hunks I mentioned above (e.g. the changes to drm_dp_init_vcpi and
>> drm_dp_find_vcpi_slots I mentioned above). We can then just update 
>> the kdocs
>> for these functions in a separate patch to clarify that now in 
>> addition to
>> being deprecated, these functions are just broken and will eventually be
>> removed.
>>
>> So - doing that allows us to get rid of mgr->total_avail_slots and mgr-
>>> start_slot entirely, just set the slot info in the atomic state 
>>> during atomic
>> check, and then just rely on the atomic state for
>> drm_dp_atomic_find_vcpi_slots() and friends. Which seems much simpler 
>> to me,
>> does this sound alrght with you?
>
> Thanks for the response,
>
> That function is per port (drm_dp_atomic_find_vcpi_slots) so not sure 
> how that will work, maybe I don't understand what you mean?
>
> Also we only need to check this inside 
> drm_dp_mst_atomic_check_vcpi_alloc_limit(), which doesn't have a 
> state, so we still need to have this on the mgr somehow.
>
> I was thinking we could add the slots(or some DP version indicator) 
> inside the drm_connector, and add a parameter to 
> drm_dp_mst_atomic_check_vcpi_alloc_limit(int slots) and call it with 
> this info via drm_dp_mst_atomic_check().
>
>
> Bhawan
>
>>
>>>                  ret = drm_dp_mst_atomic_check_mstb_bw_limit(mgr-
>>>> mst_primary,
>>> mst_state);
>>>                  mutex_unlock(&mgr->lock);
>>> @@ -5527,11 +5543,16 @@ int drm_dp_mst_topology_mgr_init(struct
>>> drm_dp_mst_topology_mgr *mgr,
>>>          if (!mgr->proposed_vcpis)
>>>                  return -ENOMEM;
>>>          set_bit(0, &mgr->payload_mask);
>>> +       mgr->total_avail_slots = 63;
>>> +       mgr->start_slot = 1;
>>>            mst_state = kzalloc(sizeof(*mst_state), GFP_KERNEL);
>>>          if (mst_state == NULL)
>>>                  return -ENOMEM;
>>>   +       mst_state->total_avail_slots = 63;
>>> +       mst_state->start_slot = 1;
>>> +
>>>          mst_state->mgr = mgr;
>>>          INIT_LIST_HEAD(&mst_state->vcpis);
>>>   diff --git a/include/drm/drm_dp_mst_helper.h
>>> b/include/drm/drm_dp_mst_helper.h
>>> index ddb9231d0309..f8152dfb34ed 100644
>>> --- a/include/drm/drm_dp_mst_helper.h
>>> +++ b/include/drm/drm_dp_mst_helper.h
>>> @@ -554,6 +554,8 @@ struct drm_dp_mst_topology_state {
>>>          struct drm_private_state base;
>>>          struct list_head vcpis;
>>>          struct drm_dp_mst_topology_mgr *mgr;
>>> +       u8 total_avail_slots;
>>> +       u8 start_slot;
>>>   };
>>>     #define to_dp_mst_topology_mgr(x) container_of(x, struct
>>> drm_dp_mst_topology_mgr, base)
>>> @@ -661,6 +663,16 @@ struct drm_dp_mst_topology_mgr {
>>>           */
>>>          int pbn_div;
>>>   +       /**
>>> +        * @total_avail_slots: 63 for 8b/10b, 64 for 128/132b
>>> +        */
>>> +       u8 total_avail_slots;
>>> +
>>> +       /**
>>> +        * @start_slot: 1 for 8b/10b, 0 for 128/132b
>>> +        */
>>> +       u8 start_slot;
>>> +
>>>          /**
>>>           * @funcs: Atomic helper callbacks
>>>           */
>>> @@ -806,6 +818,7 @@ int drm_dp_mst_get_vcpi_slots(struct
>>> drm_dp_mst_topology_mgr *mgr, struct drm_dp
>>>     void drm_dp_mst_reset_vcpi_slots(struct drm_dp_mst_topology_mgr 
>>> *mgr,
>>> struct drm_dp_mst_port *port);
>>>   +void drm_dp_mst_update_coding_cap(struct drm_dp_mst_topology_state
>>> *mst_state, uint8_t link_coding_cap);
>>>     void drm_dp_mst_deallocate_vcpi(struct drm_dp_mst_topology_mgr 
>>> *mgr,
>>>                                  struct drm_dp_mst_port *port);