Re: [PATCH 4/4] drm/amdgpu: VCN avoid memory allocation during IB test

[Christian König <christian.koenig@xxxxxxx>]

Am 10.09.21 um 02:38 schrieb xinhui pan:
> alloc extra msg from direct IB pool.
>
> Signed-off-by: xinhui pan <xinhui.pan@xxxxxxx>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c | 99 +++++++++++--------------
>   1 file changed, 45 insertions(+), 54 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
> index 561296a85b43..b60d5f01fdae 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
> @@ -541,15 +541,14 @@ int amdgpu_vcn_dec_sw_ring_test_ring(struct amdgpu_ring *ring)
>   }
>   
>   static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
> -				   struct amdgpu_bo *bo,
> -				   struct dma_fence **fence)
> +		struct amdgpu_ib *ib_msg,
> +		struct dma_fence **fence)

The parameter indentation here and at a few other places doesn't look 
correct to me, what editor are you using BTW?

Apart from that the patch is Reviewed-by: Christian König 
<christian.koenig@xxxxxxx>.

Regards,
Christian.

>   {
>   	struct amdgpu_device *adev = ring->adev;
>   	struct dma_fence *f = NULL;
>   	struct amdgpu_job *job;
>   	struct amdgpu_ib *ib;
> -	uint64_t addr;
> -	void *msg = NULL;
> +	uint64_t addr = ib_msg->gpu_addr;
>   	int i, r;
>   
>   	r = amdgpu_job_alloc_with_ib(adev, 64,
> @@ -558,8 +557,6 @@ static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
>   		goto err;
>   
>   	ib = &job->ibs[0];
> -	addr = amdgpu_bo_gpu_offset(bo);
> -	msg = amdgpu_bo_kptr(bo);
>   	ib->ptr[0] = PACKET0(adev->vcn.internal.data0, 0);
>   	ib->ptr[1] = addr;
>   	ib->ptr[2] = PACKET0(adev->vcn.internal.data1, 0);
> @@ -576,9 +573,7 @@ static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
>   	if (r)
>   		goto err_free;
>   
> -	amdgpu_bo_fence(bo, f, false);
> -	amdgpu_bo_unreserve(bo);
> -	amdgpu_bo_free_kernel(&bo, NULL, (void **)&msg);
> +	amdgpu_ib_free(adev, ib_msg, f);
>   
>   	if (fence)
>   		*fence = dma_fence_get(f);
> @@ -588,27 +583,26 @@ static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
>   
>   err_free:
>   	amdgpu_job_free(job);
> -
>   err:
> -	amdgpu_bo_unreserve(bo);
> -	amdgpu_bo_free_kernel(&bo, NULL, (void **)&msg);
> +	amdgpu_ib_free(adev, ib_msg, f);
>   	return r;
>   }
>   
>   static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
> -					 struct amdgpu_bo **bo)
> +		struct amdgpu_ib *ib)
>   {
>   	struct amdgpu_device *adev = ring->adev;
>   	uint32_t *msg;
>   	int r, i;
>   
> -	*bo = NULL;
> -	r = amdgpu_bo_create_reserved(adev, 1024, PAGE_SIZE,
> -				      AMDGPU_GEM_DOMAIN_VRAM,
> -				      bo, NULL, (void **)&msg);
> +	memset(ib, 0, sizeof(*ib));
> +	r = amdgpu_ib_get(adev, NULL, PAGE_SIZE,
> +			AMDGPU_IB_POOL_DIRECT,
> +			ib);
>   	if (r)
>   		return r;
>   
> +	msg = ib->ptr;
>   	msg[0] = cpu_to_le32(0x00000028);
>   	msg[1] = cpu_to_le32(0x00000038);
>   	msg[2] = cpu_to_le32(0x00000001);
> @@ -630,19 +624,20 @@ static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
>   }
>   
>   static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
> -					  struct amdgpu_bo **bo)
> +					  struct amdgpu_ib *ib)
>   {
>   	struct amdgpu_device *adev = ring->adev;
>   	uint32_t *msg;
>   	int r, i;
>   
> -	*bo = NULL;
> -	r = amdgpu_bo_create_reserved(adev, 1024, PAGE_SIZE,
> -				      AMDGPU_GEM_DOMAIN_VRAM,
> -				      bo, NULL, (void **)&msg);
> +	memset(ib, 0, sizeof(*ib));
> +	r = amdgpu_ib_get(adev, NULL, PAGE_SIZE,
> +			AMDGPU_IB_POOL_DIRECT,
> +			ib);
>   	if (r)
>   		return r;
>   
> +	msg = ib->ptr;
>   	msg[0] = cpu_to_le32(0x00000028);
>   	msg[1] = cpu_to_le32(0x00000018);
>   	msg[2] = cpu_to_le32(0x00000000);
> @@ -658,21 +653,21 @@ static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
>   int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
>   {
>   	struct dma_fence *fence = NULL;
> -	struct amdgpu_bo *bo;
> +	struct amdgpu_ib ib;
>   	long r;
>   
> -	r = amdgpu_vcn_dec_get_create_msg(ring, 1, &bo);
> +	r = amdgpu_vcn_dec_get_create_msg(ring, 1, &ib);
>   	if (r)
>   		goto error;
>   
> -	r = amdgpu_vcn_dec_send_msg(ring, bo, NULL);
> +	r = amdgpu_vcn_dec_send_msg(ring, &ib, NULL);
>   	if (r)
>   		goto error;
> -	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &bo);
> +	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &ib);
>   	if (r)
>   		goto error;
>   
> -	r = amdgpu_vcn_dec_send_msg(ring, bo, &fence);
> +	r = amdgpu_vcn_dec_send_msg(ring, &ib, &fence);
>   	if (r)
>   		goto error;
>   
> @@ -688,8 +683,8 @@ int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
>   }
>   
>   static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
> -				   struct amdgpu_bo *bo,
> -				   struct dma_fence **fence)
> +		struct amdgpu_ib *ib_msg,
> +		struct dma_fence **fence)
>   {
>   	struct amdgpu_vcn_decode_buffer *decode_buffer = NULL;
>   	const unsigned int ib_size_dw = 64;
> @@ -697,7 +692,7 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
>   	struct dma_fence *f = NULL;
>   	struct amdgpu_job *job;
>   	struct amdgpu_ib *ib;
> -	uint64_t addr;
> +	uint64_t addr = ib_msg->gpu_addr;
>   	int i, r;
>   
>   	r = amdgpu_job_alloc_with_ib(adev, ib_size_dw * 4,
> @@ -706,7 +701,6 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
>   		goto err;
>   
>   	ib = &job->ibs[0];
> -	addr = amdgpu_bo_gpu_offset(bo);
>   	ib->length_dw = 0;
>   
>   	ib->ptr[ib->length_dw++] = sizeof(struct amdgpu_vcn_decode_buffer) + 8;
> @@ -726,9 +720,7 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
>   	if (r)
>   		goto err_free;
>   
> -	amdgpu_bo_fence(bo, f, false);
> -	amdgpu_bo_unreserve(bo);
> -	amdgpu_bo_unref(&bo);
> +	amdgpu_ib_free(adev, ib_msg, f);
>   
>   	if (fence)
>   		*fence = dma_fence_get(f);
> @@ -738,31 +730,29 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
>   
>   err_free:
>   	amdgpu_job_free(job);
> -
>   err:
> -	amdgpu_bo_unreserve(bo);
> -	amdgpu_bo_unref(&bo);
> +	amdgpu_ib_free(adev, ib_msg, f);
>   	return r;
>   }
>   
>   int amdgpu_vcn_dec_sw_ring_test_ib(struct amdgpu_ring *ring, long timeout)
>   {
>   	struct dma_fence *fence = NULL;
> -	struct amdgpu_bo *bo;
> +	struct amdgpu_ib ib;
>   	long r;
>   
> -	r = amdgpu_vcn_dec_get_create_msg(ring, 1, &bo);
> +	r = amdgpu_vcn_dec_get_create_msg(ring, 1, &ib);
>   	if (r)
>   		goto error;
>   
> -	r = amdgpu_vcn_dec_sw_send_msg(ring, bo, NULL);
> +	r = amdgpu_vcn_dec_sw_send_msg(ring, &ib, NULL);
>   	if (r)
>   		goto error;
> -	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &bo);
> +	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &ib);
>   	if (r)
>   		goto error;
>   
> -	r = amdgpu_vcn_dec_sw_send_msg(ring, bo, &fence);
> +	r = amdgpu_vcn_dec_sw_send_msg(ring, &ib, &fence);
>   	if (r)
>   		goto error;
>   
> @@ -809,7 +799,7 @@ int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring)
>   }
>   
>   static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
> -					 struct amdgpu_bo *bo,
> +					 struct amdgpu_ib *ib_msg,
>   					 struct dma_fence **fence)
>   {
>   	const unsigned ib_size_dw = 16;
> @@ -825,7 +815,7 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
>   		return r;
>   
>   	ib = &job->ibs[0];
> -	addr = amdgpu_bo_gpu_offset(bo);
> +	addr = ib_msg->gpu_addr;
>   
>   	ib->length_dw = 0;
>   	ib->ptr[ib->length_dw++] = 0x00000018;
> @@ -863,7 +853,7 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
>   }
>   
>   static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
> -					  struct amdgpu_bo *bo,
> +					  struct amdgpu_ib *ib_msg,
>   					  struct dma_fence **fence)
>   {
>   	const unsigned ib_size_dw = 16;
> @@ -879,7 +869,7 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
>   		return r;
>   
>   	ib = &job->ibs[0];
> -	addr = amdgpu_bo_gpu_offset(bo);
> +	addr = ib_msg->gpu_addr;
>   
>   	ib->length_dw = 0;
>   	ib->ptr[ib->length_dw++] = 0x00000018;
> @@ -918,21 +908,23 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
>   
>   int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
>   {
> +	struct amdgpu_device *adev = ring->adev;
>   	struct dma_fence *fence = NULL;
> -	struct amdgpu_bo *bo = NULL;
> +	struct amdgpu_ib ib;
>   	long r;
>   
> -	r = amdgpu_bo_create_reserved(ring->adev, 128 * 1024, PAGE_SIZE,
> -				      AMDGPU_GEM_DOMAIN_VRAM,
> -				      &bo, NULL, NULL);
> +	memset(&ib, 0, sizeof(ib));
> +	r = amdgpu_ib_get(adev, NULL, 128 << 10,
> +			AMDGPU_IB_POOL_DIRECT,
> +			&ib);
>   	if (r)
>   		return r;
>   
> -	r = amdgpu_vcn_enc_get_create_msg(ring, 1, bo, NULL);
> +	r = amdgpu_vcn_enc_get_create_msg(ring, 1, &ib, NULL);
>   	if (r)
>   		goto error;
>   
> -	r = amdgpu_vcn_enc_get_destroy_msg(ring, 1, bo, &fence);
> +	r = amdgpu_vcn_enc_get_destroy_msg(ring, 1, &ib, &fence);
>   	if (r)
>   		goto error;
>   
> @@ -943,9 +935,8 @@ int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
>   		r = 0;
>   
>   error:
> +	amdgpu_ib_free(adev, &ib, fence);
>   	dma_fence_put(fence);
> -	amdgpu_bo_unreserve(bo);
> -	amdgpu_bo_free_kernel(&bo, NULL, NULL);
>   
>   	return r;
>   }