Commit 96595204 authored by Srinivasan Shanmugam's avatar Srinivasan Shanmugam Committed by Alex Deucher
Browse files

drm/amdgpu: Make enforce_isolation setting per GPU 



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@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: default avatarSrinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
Suggested-by: default avatarChristian König <christian.koenig@amd.com>
parent ee7a846e

drivers/gpu/drm/amd/amdgpu/amdgpu.h

+2 −0
Original line number Diff line number Diff line
@@ -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,

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

+1 −1
Original line number Diff line number Diff line
@@ -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;

		}

	}

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

+5 −0
Original line number Diff line number Diff line
@@ -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;

}

drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c

+11 −6
Original line number Diff line number Diff line
@@ -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));

	}

}



/**

drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h

+2 −1
Original line number Diff line number Diff line
@@ -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,