[v3 1/2] drm/amd/pm: Add VCN reset support detection for SMU v13.0.6

Lazar, Lijo lijo.lazar at amd.com
Tue Aug 12 06:02:21 UTC 2025



On 8/12/2025 11:26 AM, Zhang, Jesse(Jie) wrote:
> [AMD Official Use Only - AMD Internal Distribution Only]
> 
> -----Original Message-----
> From: Lazar, Lijo <Lijo.Lazar at amd.com>
> Sent: Tuesday, August 12, 2025 1:38 PM
> To: Zhang, Jesse(Jie) <Jesse.Zhang at amd.com>; amd-gfx at lists.freedesktop.org
> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig, Christian <Christian.Koenig at amd.com>; Ji, Ruili <Ruili.Ji at amd.com>
> Subject: Re: [v3 1/2] drm/amd/pm: Add VCN reset support detection for SMU v13.0.6
> 
> 
> 
> On 8/12/2025 6:33 AM, Jesse.Zhang wrote:
>> This commit introduces support for detecting VCN reset capability
>> through the SMU interface. Key changes include:
>>
>> 1. Added amdgpu_dpm_reset_vcn_is_supported() interface to check VCN
>> reset support 2. Implemented SMU backend functions for VCN reset
>> capability detection 3. Added SMU_CAP(VCN_RESET) capability flag for
>> SMU v13.0.6 4. Updated PPSMC message definitions to accommodate VCN
>> reset functionality 5. Added version checks for VCN reset support
>> (fw_ver >= 0x04557100)
>>
>> The changes maintain backward compatibility while enabling proper
>> detection of VCN reset capabilities when supported by the firmware.
>>
>> v2: clean up debug info and adjust this message to be more meaningful
>> (Alex)
>>
>> Suggested-by: Alex Deucher <alexander.deucher at amd.com>
>> Signed-off-by: Ruili Ji <ruiliji2 at amd.com>
>> Signed-off-by: Jesse Zhang <Jesse.Zhang at amd.com>
>> ---
>>  drivers/gpu/drm/amd/pm/amdgpu_dpm.c             | 15 +++++++++++++++
>>  drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h         |  1 +
>>  drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c       | 10 ++++++++++
>>  drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h   |  5 +++++
>>  .../pm/swsmu/inc/pmfw_if/smu_v13_0_6_ppsmc.h    |  4 ++--
>>  .../drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c    | 17 +++++++++++++++++
>>  .../drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h    |  1 +
>>  7 files changed, 51 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
>> b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
>> index 6e0d711820ea..518d07afc7df 100644
>> --- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
>> +++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
>> @@ -820,6 +820,21 @@ int amdgpu_dpm_reset_vcn(struct amdgpu_device *adev, uint32_t inst_mask)
>>       return ret;
>>  }
>>
>> +bool amdgpu_dpm_reset_vcn_is_supported(struct amdgpu_device *adev) {
>> +     struct smu_context *smu = adev->powerplay.pp_handle;
>> +     bool ret;
>> +
>> +     if (!is_support_sw_smu(adev))
>> +             return false;
>> +
>> +     mutex_lock(&adev->pm.mutex);
>> +     ret = smu_reset_vcn_is_supported(smu);
>> +     mutex_unlock(&adev->pm.mutex);
>> +
>> +     return ret;
>> +}
>> +
>>  int amdgpu_dpm_get_dpm_freq_range(struct amdgpu_device *adev,
>>                                 enum pp_clock_type type,
>>                                 uint32_t *min,
>> diff --git a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
>> b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
>> index 09962db988d6..9748744133d9 100644
>> --- a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
>> +++ b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
>> @@ -615,6 +615,7 @@ ssize_t amdgpu_dpm_get_pm_policy_info(struct
>> amdgpu_device *adev,  int amdgpu_dpm_reset_sdma(struct amdgpu_device
>> *adev, uint32_t inst_mask);  bool
>> amdgpu_dpm_reset_sdma_is_supported(struct amdgpu_device *adev);  int
>> amdgpu_dpm_reset_vcn(struct amdgpu_device *adev, uint32_t inst_mask);
>> +bool amdgpu_dpm_reset_vcn_is_supported(struct amdgpu_device *adev);
>>  bool amdgpu_dpm_is_temp_metrics_supported(struct amdgpu_device *adev,
>>                                         enum smu_temp_metric_type type);
>>
>> diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
>> b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
>> index dc48a1dd8be4..f9a350a82764 100644
>> --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
>> +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
>> @@ -4106,6 +4106,16 @@ int smu_reset_sdma(struct smu_context *smu, uint32_t inst_mask)
>>       return ret;
>>  }
>>
>> +bool smu_reset_vcn_is_supported(struct smu_context *smu) {
>> +     bool ret = false;
>> +
>> +     if (smu->ppt_funcs && smu->ppt_funcs->reset_vcn_is_supported)
>> +             ret = smu->ppt_funcs->reset_vcn_is_supported(smu);
>> +
>> +     return ret;
>> +}
>> +
>>  int smu_reset_vcn(struct smu_context *smu, uint32_t inst_mask)  {
>>       if (smu->ppt_funcs && smu->ppt_funcs->dpm_reset_vcn) diff --git
>> a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
>> b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
>> index 611b381b9147..7990771151be 100644
>> --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
>> +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
>> @@ -1426,6 +1426,10 @@ struct pptable_funcs {
>>        * @reset_vcn: message SMU to soft reset vcn instance.
>>        */
>>       int (*dpm_reset_vcn)(struct smu_context *smu, uint32_t inst_mask);
>> +     /**
>> +      * @reset_vcn_is_supported: Check if support resets vcn.
>> +      */
>> +     bool (*reset_vcn_is_supported)(struct smu_context *smu);
>>
>>       /**
>>        * @get_ecc_table:  message SMU to get ECC INFO table.
>> @@ -1702,6 +1706,7 @@ int smu_send_rma_reason(struct smu_context
>> *smu);  int smu_reset_sdma(struct smu_context *smu, uint32_t
>> inst_mask);  bool smu_reset_sdma_is_supported(struct smu_context
>> *smu);  int smu_reset_vcn(struct smu_context *smu, uint32_t
>> inst_mask);
>> +bool smu_reset_vcn_is_supported(struct smu_context *smu);
>>  int smu_set_pm_policy(struct smu_context *smu, enum pp_pm_policy p_type,
>>                     int level);
>>  ssize_t smu_get_pm_policy_info(struct smu_context *smu,
> 
> It's better to split smu v13.0.6 changes to a separate patch.
> 
>> diff --git
>> a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_ppsmc.h
>> b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_ppsmc.h
>> index 41f268313613..63a088ef7169 100644
>> --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_ppsmc.h
>> +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_ppsmc.h
>> @@ -94,9 +94,9 @@
>>  #define PPSMC_MSG_RmaDueToBadPageThreshold          0x43
>>  #define PPSMC_MSG_SetThrottlingPolicy               0x44
>>  #define PPSMC_MSG_ResetSDMA                         0x4D
>> -#define PPSMC_MSG_ResetVCN                          0x4E
>>  #define PPSMC_MSG_GetStaticMetricsTable             0x59
>> -#define PPSMC_Message_Count                         0x5A
>> +#define PPSMC_MSG_ResetVCN                          0x5B
>> +#define PPSMC_Message_Count                         0x5C
>>
>>  //PPSMC Reset Types for driver msg argument
>>  #define PPSMC_RESET_TYPE_DRIVER_MODE_1_RESET        0x1
>> diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
>> b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
>> index 90e66c8f2f82..60aaf0e2ce8f 100644
>> --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
>> +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
>> @@ -434,6 +434,9 @@ static void smu_v13_0_6_init_caps(struct smu_context *smu)
>>           ((pgm == 0) && (fw_ver >= 0x00557900)) ||
>>           ((pgm == 4) && (fw_ver >= 0x4557000)))
>>               smu_v13_0_6_cap_set(smu, SMU_CAP(SDMA_RESET));
>> +
>> +     if ((pgm == 4) && (fw_ver >= 0x04557100))
>> +             smu_v13_0_6_cap_set(smu, SMU_CAP(VCN_RESET));
>>  }
>>
>>  static void smu_v13_0_x_init_caps(struct smu_context *smu) @@ -3171,6
>> +3174,19 @@ static int smu_v13_0_6_reset_sdma(struct smu_context *smu, uint32_t inst_mask)
>>       return ret;
>>  }
>>
>> +static bool smu_v13_0_6_reset_vcn_is_supported(struct smu_context
>> +*smu) {
>> +     bool ret = true;
>> +
>> +     if (!smu_v13_0_6_cap_supported(smu, SMU_CAP(VCN_RESET))) {
>> +             dev_info(smu->adev->dev,
>> +                     "SMU VCN reset not supported.  Please update SMU firmware.\n");
> 
> This will come for every reset attempt for programs other than 4. Better restrict this to once, or not to keep it as it requires IFWI update.
> 
> Hi Lijo, Once we got their confirm from PMFW, we will add more programs for review.
> Some programs are still under testing.
> 

That is fine, but any user on an older IFWI will be affected by these
messages. Since IFWI upgrade is a major change, I don't know whether we
should keep this or at minimum restrict the messaging to only once.

Thanks,
Lijo

> Thanks
> Jesse
> 
> Thanks,
> Lijo
> 
>> +             ret = false;
>> +     }
>> +
>> +     return ret;
>> +}
>> +
>>  static int smu_v13_0_6_reset_vcn(struct smu_context *smu, uint32_t
>> inst_mask)  {
>>       int ret = 0;
>> @@ -3859,6 +3875,7 @@ static const struct pptable_funcs smu_v13_0_6_ppt_funcs = {
>>       .reset_sdma = smu_v13_0_6_reset_sdma,
>>       .reset_sdma_is_supported = smu_v13_0_6_reset_sdma_is_supported,
>>       .dpm_reset_vcn = smu_v13_0_6_reset_vcn,
>> +     .reset_vcn_is_supported = smu_v13_0_6_reset_vcn_is_supported,
>>  };
>>
>>  void smu_v13_0_6_set_ppt_funcs(struct smu_context *smu) diff --git
>> a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h
>> b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h
>> index ece04ad724fb..7c98f77c429d 100644
>> --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h
>> +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h
>> @@ -64,6 +64,7 @@ enum smu_v13_0_6_caps {
>>       SMU_CAP(RMA_MSG),
>>       SMU_CAP(ACA_SYND),
>>       SMU_CAP(SDMA_RESET),
>> +     SMU_CAP(VCN_RESET),
>>       SMU_CAP(STATIC_METRICS),
>>       SMU_CAP(HST_LIMIT_METRICS),
>>       SMU_CAP(BOARD_VOLTAGE),
> 



More information about the amd-gfx mailing list