drm/amdgpu: use max_dw in ring_init

Instead of specifying the total ring size calculate that from the maximum
number of dw a submission can have and the number of concurrent submissions.

This fixes UVD with 8 concurrent submissions or more.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -215,18 +215,17 @@ int amdgpu_ring_restore(struct amdgpu_ring *ring,
  *
  * @adev: amdgpu_device pointer
  * @ring: amdgpu_ring structure holding ring information
- * @ring_size: size of the ring
+ * @max_ndw: maximum number of dw for ring alloc
  * @nop: nop packet for this ring
  *
  * Initialize the driver information for the selected ring (all asics).
  * Returns 0 on success, error on failure.
  */
 int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
-		     unsigned ring_size, u32 nop, u32 align_mask,
+		     unsigned max_dw, u32 nop, u32 align_mask,
 		     struct amdgpu_irq_src *irq_src, unsigned irq_type,
 		     enum amdgpu_ring_type ring_type)
 {
-	u32 rb_bufsz;
 	int r;
 
 	if (ring->adev == NULL) {
@@ -283,10 +282,8 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		return r;
 	}
 
-	/* Align ring size */
-	rb_bufsz = order_base_2(ring_size / 8);
-	ring_size = (1 << (rb_bufsz + 1)) * 4;
-	ring->ring_size = ring_size;
+	ring->ring_size = roundup_pow_of_two(max_dw * 4 *
+					     amdgpu_sched_hw_submission);
 	ring->align_mask = align_mask;
 	ring->nop = nop;
 	ring->type = ring_type;
@@ -319,8 +316,7 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		}
 	}
 	ring->ptr_mask = (ring->ring_size / 4) - 1;
-	ring->max_dw = DIV_ROUND_UP(ring->ring_size / 4,
-				    amdgpu_sched_hw_submission);
+	ring->max_dw = max_dw;
 
 	if (amdgpu_debugfs_ring_init(adev, ring)) {
 		DRM_ERROR("Failed to register debugfs file for rings !\n");

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

@@ -976,7 +976,7 @@ static int cik_sdma_sw_init(void *handle)
 		ring = &adev->sdma.instance[i].ring;
 		ring->ring_obj = NULL;
 		sprintf(ring->name, "sdma%d", i);
-		r = amdgpu_ring_init(adev, ring, 256 * 1024,
+		r = amdgpu_ring_init(adev, ring, 32 * 1024,
 				     SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0), 0xf,
 				     &adev->sdma.trap_irq,
 				     (i == 0) ?

drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c

@@ -4414,7 +4414,7 @@ static int gfx_v7_0_sw_init(void *handle)
 		ring = &adev->gfx.gfx_ring[i];
 		ring->ring_obj = NULL;
 		sprintf(ring->name, "gfx");
-		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+		r = amdgpu_ring_init(adev, ring, 128 * 1024,
 				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
 				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP,
 				     AMDGPU_RING_TYPE_GFX);
@@ -4441,7 +4441,7 @@ static int gfx_v7_0_sw_init(void *handle)
 		sprintf(ring->name, "comp %d.%d.%d", ring->me, ring->pipe, ring->queue);
 		irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe;
 		/* type-2 packets are deprecated on MEC, use type-3 instead */
-		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+		r = amdgpu_ring_init(adev, ring, 128 * 1024,
 				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
 				     &adev->gfx.eop_irq, irq_type,
 				     AMDGPU_RING_TYPE_COMPUTE);

drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c

@@ -1570,7 +1570,7 @@ static int gfx_v8_0_sw_init(void *handle)
 			ring->doorbell_index = AMDGPU_DOORBELL_GFX_RING0;
 		}
 
-		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+		r = amdgpu_ring_init(adev, ring, 128 * 1024,
 				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
 				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP,
 				     AMDGPU_RING_TYPE_GFX);
@@ -1597,7 +1597,7 @@ static int gfx_v8_0_sw_init(void *handle)
 		sprintf(ring->name, "comp %d.%d.%d", ring->me, ring->pipe, ring->queue);
 		irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe;
 		/* type-2 packets are deprecated on MEC, use type-3 instead */
-		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+		r = amdgpu_ring_init(adev, ring, 128 * 1024,
 				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
 				     &adev->gfx.eop_irq, irq_type,
 				     AMDGPU_RING_TYPE_COMPUTE);

drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c

@@ -990,7 +990,7 @@ static int sdma_v2_4_sw_init(void *handle)
 		ring->ring_obj = NULL;
 		ring->use_doorbell = false;
 		sprintf(ring->name, "sdma%d", i);
-		r = amdgpu_ring_init(adev, ring, 256 * 1024,
+		r = amdgpu_ring_init(adev, ring, 32 * 1024,
 				     SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), 0xf,
 				     &adev->sdma.trap_irq,
 				     (i == 0) ?

drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c

@@ -114,7 +114,7 @@ static int uvd_v4_2_sw_init(void *handle)
 
 	ring = &adev->uvd.ring;
 	sprintf(ring->name, "uvd");
-	r = amdgpu_ring_init(adev, ring, 4096, CP_PACKET2, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, CP_PACKET2, 0xf,
 			     &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD);
 
 	return r;

drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c

@@ -111,7 +111,7 @@ static int uvd_v5_0_sw_init(void *handle)
 
 	ring = &adev->uvd.ring;
 	sprintf(ring->name, "uvd");
-	r = amdgpu_ring_init(adev, ring, 4096, CP_PACKET2, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, CP_PACKET2, 0xf,
 			     &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD);
 
 	return r;

drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c

@@ -112,7 +112,7 @@ static int uvd_v6_0_sw_init(void *handle)
 
 	ring = &adev->uvd.ring;
 	sprintf(ring->name, "uvd");
-	r = amdgpu_ring_init(adev, ring, 4096, CP_PACKET2, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, CP_PACKET2, 0xf,
 			     &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD);
 
 	return r;

drivers/gpu/drm/amd/amdgpu/vce_v2_0.c

@@ -201,14 +201,14 @@ static int vce_v2_0_sw_init(void *handle)
 
 	ring = &adev->vce.ring[0];
 	sprintf(ring->name, "vce0");
-	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, VCE_CMD_NO_OP, 0xf,
 			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
 	if (r)
 		return r;
 
 	ring = &adev->vce.ring[1];
 	sprintf(ring->name, "vce1");
-	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, VCE_CMD_NO_OP, 0xf,
 			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
 	if (r)
 		return r;

drivers/gpu/drm/amd/amdgpu/vce_v3_0.c

@@ -381,14 +381,14 @@ static int vce_v3_0_sw_init(void *handle)
 
 	ring = &adev->vce.ring[0];
 	sprintf(ring->name, "vce0");
-	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, VCE_CMD_NO_OP, 0xf,
 			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
 	if (r)
 		return r;
 
 	ring = &adev->vce.ring[1];
 	sprintf(ring->name, "vce1");
-	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+	r = amdgpu_ring_init(adev, ring, 512, VCE_CMD_NO_OP, 0xf,
 			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
 	if (r)
 		return r;