[PATCH] drm/amdgpu: disable workload profile switching when OD is enabled

[Feng, Kenneth]

[AMD Official Use Only - AMD Internal Distribution Only]

Reviewed-by: Kenneth Feng <kenneth.feng at amd.com>


-----Original Message-----
From: amd-gfx <amd-gfx-bounces at lists.freedesktop.org> On Behalf Of Alex Deucher
Sent: Thursday, May 29, 2025 10:19 PM
To: amd-gfx at lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher at amd.com>
Subject: [PATCH] drm/amdgpu: disable workload profile switching when OD is enabled

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.


Users have reported that they have to reduce the level of undervolting to acheive stability when dynamic workload profiles are enabled on GC 10.3.x. Disable dynamic workload profiles if the user has enabled OD.

Fixes: b9467983b774 ("drm/amdgpu: add dynamic workload profile switching for gfx10")
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4262
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c |  8 ++++++++
 drivers/gpu/drm/amd/pm/amdgpu_dpm.c     | 22 ++++++++++++++++++++++
 drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h |  1 +
 3 files changed, 31 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 1db1e6ec0184f..c5646af055abc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -2228,6 +2228,9 @@ void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring)
        enum PP_SMC_POWER_PROFILE profile;
        int r;

+       if (amdgpu_dpm_is_overdrive_enabled(adev))
+               return;
+
        if (adev->gfx.num_gfx_rings)
                profile = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
        else
@@ -2258,6 +2261,11 @@ void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring)

 void amdgpu_gfx_profile_ring_end_use(struct amdgpu_ring *ring)  {
+       struct amdgpu_device *adev = ring->adev;
+
+       if (amdgpu_dpm_is_overdrive_enabled(adev))
+               return;
+
        atomic_dec(&ring->adev->gfx.total_submission_cnt);

        schedule_delayed_work(&ring->adev->gfx.idle_work, GFX_PROFILE_IDLE_TIMEOUT); diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
index b5477e02fe996..353f01c0d1000 100644
--- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
+++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
@@ -1701,6 +1701,28 @@ int amdgpu_dpm_is_overdrive_supported(struct amdgpu_device *adev)
        }
 }

+int amdgpu_dpm_is_overdrive_enabled(struct amdgpu_device *adev) {
+       if (is_support_sw_smu(adev)) {
+               struct smu_context *smu = adev->powerplay.pp_handle;
+
+               return smu->od_enabled;
+       } else {
+               struct pp_hwmgr *hwmgr;
+
+               /*
+                * dpm on some legacy asics don't carry od_enabled member
+                * as its pp_handle is casted directly from adev.
+                */
+               if (amdgpu_dpm_is_legacy_dpm(adev))
+                       return false;
+
+               hwmgr = (struct pp_hwmgr *)adev->powerplay.pp_handle;
+
+               return hwmgr->od_enabled;
+       }
+}
+
 int amdgpu_dpm_set_pp_table(struct amdgpu_device *adev,
                            const char *buf,
                            size_t size) diff --git a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
index b52162112fa68..9163962814ea8 100644
--- a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
+++ b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h
@@ -563,6 +563,7 @@ int amdgpu_dpm_get_smu_prv_buf_details(struct amdgpu_device *adev,
                                       void **addr,
                                       size_t *size);  int amdgpu_dpm_is_overdrive_supported(struct amdgpu_device *adev);
+int amdgpu_dpm_is_overdrive_enabled(struct amdgpu_device *adev);
 int amdgpu_dpm_set_pp_table(struct amdgpu_device *adev,
                            const char *buf,
                            size_t size);
--
2.49.0