[PATCH 04/17] drm/amdgpu: Make enforce_isolation setting per GPU

Alex Deucher alexander.deucher at amd.com
Thu Aug 15 00:04:48 UTC 2024


From: Srinivasan Shanmugam <srinivasan.shanmugam at amd.com>

This commit makes enforce_isolation setting to be per GPU and per
partition by adding the enforce_isolation array to the adev structure.
The adev variable is set based on the global enforce_isolation module
parameter during device initialization.

In amdgpu_ids.c, the adev->enforce_isolation value for the current GPU
is used to determine whether to enforce isolation between graphics and
compute processes on that GPU.

In amdgpu_ids.c, the adev->enforce_isolation value for the current GPU
and partition is used to determine whether to enforce isolation between
graphics and compute processes on that GPU and partition.

This allows the enforce_isolation setting to be controlled individually
for each GPU and each partition, which is useful in a system with
multiple GPUs and partitions where different isolation settings might be
desired for different GPUs and partitions.

v2: fix loop in amdgpu_vmid_mgr_init() (Alex)

Cc: Christian König <christian.koenig at amd.com>
Cc: Alex Deucher <alexander.deucher at amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam at amd.com>
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
Suggested-by: Christian König <christian.koenig at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  2 ++
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c     |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c |  5 +++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c    | 17 +++++++++++------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h    |  3 ++-
 5 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index f3980b40f2ce..ae3e827da5ec 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1162,6 +1162,8 @@ struct amdgpu_device {
 	bool                            debug_disable_soft_recovery;
 	bool                            debug_use_vram_fw_buf;
 	bool                            debug_enable_ras_aca;
+
+	bool				enforce_isolation[MAX_XCP];
 };
 
 static inline uint32_t amdgpu_ip_version(const struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 78b3c067fea7..5d5ba1e3d90f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -1110,7 +1110,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 			struct drm_gpu_scheduler *sched = entity->rq->sched;
 			struct amdgpu_ring *ring = to_amdgpu_ring(sched);
 
-			if (amdgpu_vmid_uses_reserved(vm, ring->vm_hub))
+			if (amdgpu_vmid_uses_reserved(adev, vm, ring->vm_hub))
 				return -EINVAL;
 		}
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index a6b8d0ba4758..95953c028ca5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -1916,6 +1916,8 @@ static int amdgpu_device_init_apu_flags(struct amdgpu_device *adev)
  */
 static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 {
+	int i;
+
 	if (amdgpu_sched_jobs < 4) {
 		dev_warn(adev->dev, "sched jobs (%d) must be at least 4\n",
 			 amdgpu_sched_jobs);
@@ -1970,6 +1972,9 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 
 	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
 
+	for (i = 0; i < MAX_XCP; i++)
+		adev->enforce_isolation[i] = !!enforce_isolation;
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
index 6608eeb61e5a..92d27d32de41 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
@@ -424,7 +424,7 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 	if (r || !idle)
 		goto error;
 
-	if (amdgpu_vmid_uses_reserved(vm, vmhub)) {
+	if (amdgpu_vmid_uses_reserved(adev, vm, vmhub)) {
 		r = amdgpu_vmid_grab_reserved(vm, ring, job, &id, fence);
 		if (r || !id)
 			goto error;
@@ -476,15 +476,19 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 
 /*
  * amdgpu_vmid_uses_reserved - check if a VM will use a reserved VMID
+ * @adev: amdgpu_device pointer
  * @vm: the VM to check
  * @vmhub: the VMHUB which will be used
  *
  * Returns: True if the VM will use a reserved VMID.
  */
-bool amdgpu_vmid_uses_reserved(struct amdgpu_vm *vm, unsigned int vmhub)
+bool amdgpu_vmid_uses_reserved(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm, unsigned int vmhub)
 {
 	return vm->reserved_vmid[vmhub] ||
-		(enforce_isolation && AMDGPU_IS_GFXHUB(vmhub));
+		(adev->enforce_isolation[(vm->root.bo->xcp_id != AMDGPU_XCP_NO_PARTITION) ?
+					 vm->root.bo->xcp_id : 0] &&
+		 AMDGPU_IS_GFXHUB(vmhub));
 }
 
 int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev,
@@ -600,9 +604,10 @@ void amdgpu_vmid_mgr_init(struct amdgpu_device *adev)
 		}
 	}
 	/* alloc a default reserved vmid to enforce isolation */
-	if (enforce_isolation)
-		amdgpu_vmid_alloc_reserved(adev, AMDGPU_GFXHUB(0));
-
+	for (i = 0; i < (adev->xcp_mgr ? adev->xcp_mgr->num_xcps : 1); i++) {
+		if (adev->enforce_isolation[i])
+			amdgpu_vmid_alloc_reserved(adev, AMDGPU_GFXHUB(i));
+	}
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
index 240fa6751260..4012fb2dd08a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
@@ -78,7 +78,8 @@ void amdgpu_pasid_free_delayed(struct dma_resv *resv,
 
 bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev,
 			       struct amdgpu_vmid *id);
-bool amdgpu_vmid_uses_reserved(struct amdgpu_vm *vm, unsigned int vmhub);
+bool amdgpu_vmid_uses_reserved(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm, unsigned int vmhub);
 int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev,
 				unsigned vmhub);
 void amdgpu_vmid_free_reserved(struct amdgpu_device *adev,
-- 
2.46.0



More information about the amd-gfx mailing list