6 files changed, 84 insertions, 53 deletions

diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index 0e53416..6948eb8 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -2687,6 +2687,9 @@ void r600_uvd_rbc_stop(struct radeon_device *rdev)
 int r600_uvd_init(struct radeon_device *rdev)
 {
 	int i, j, r;
+	/* disable byte swapping */
+	u32 lmi_swap_cntl = 0;
+	u32 mp_swap_cntl = 0;
 
 	/* raise clocks while booting up the VCPU */
 	radeon_set_uvd_clocks(rdev, 53300, 40000);
@@ -2711,9 +2714,13 @@ int r600_uvd_init(struct radeon_device *rdev)
 	WREG32(UVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
 			     (1 << 21) | (1 << 9) | (1 << 20));
 
-	/* disable byte swapping */
-	WREG32(UVD_LMI_SWAP_CNTL, 0);
-	WREG32(UVD_MP_SWAP_CNTL, 0);
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(UVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(UVD_MP_SWAP_CNTL, mp_swap_cntl);
 
 	WREG32(UVD_MPC_SET_MUXA0, 0x40c2040);
 	WREG32(UVD_MPC_SET_MUXA1, 0x0);
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 1899738..b0dc0b6 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -244,16 +244,6 @@ void radeon_scratch_free(struct radeon_device *rdev, uint32_t reg)
  */
 void radeon_wb_disable(struct radeon_device *rdev)
 {
-	int r;
-
-	if (rdev->wb.wb_obj) {
-		r = radeon_bo_reserve(rdev->wb.wb_obj, false);
-		if (unlikely(r != 0))
-			return;
-		radeon_bo_kunmap(rdev->wb.wb_obj);
-		radeon_bo_unpin(rdev->wb.wb_obj);
-		radeon_bo_unreserve(rdev->wb.wb_obj);
-	}
 	rdev->wb.enabled = false;
 }
 
@@ -269,6 +259,11 @@ void radeon_wb_fini(struct radeon_device *rdev)
 {
 	radeon_wb_disable(rdev);
 	if (rdev->wb.wb_obj) {
+		if (!radeon_bo_reserve(rdev->wb.wb_obj, false)) {
+			radeon_bo_kunmap(rdev->wb.wb_obj);
+			radeon_bo_unpin(rdev->wb.wb_obj);
+			radeon_bo_unreserve(rdev->wb.wb_obj);
+		}
 		radeon_bo_unref(&rdev->wb.wb_obj);
 		rdev->wb.wb = NULL;
 		rdev->wb.wb_obj = NULL;
@@ -295,26 +290,26 @@ int radeon_wb_init(struct radeon_device *rdev)
 			dev_warn(rdev->dev, "(%d) create WB bo failed\n", r);
 			return r;
 		}
-	}
-	r = radeon_bo_reserve(rdev->wb.wb_obj, false);
-	if (unlikely(r != 0)) {
-		radeon_wb_fini(rdev);
-		return r;
-	}
-	r = radeon_bo_pin(rdev->wb.wb_obj, RADEON_GEM_DOMAIN_GTT,
-			  &rdev->wb.gpu_addr);
-	if (r) {
+		r = radeon_bo_reserve(rdev->wb.wb_obj, false);
+		if (unlikely(r != 0)) {
+			radeon_wb_fini(rdev);
+			return r;
+		}
+		r = radeon_bo_pin(rdev->wb.wb_obj, RADEON_GEM_DOMAIN_GTT,
+				&rdev->wb.gpu_addr);
+		if (r) {
+			radeon_bo_unreserve(rdev->wb.wb_obj);
+			dev_warn(rdev->dev, "(%d) pin WB bo failed\n", r);
+			radeon_wb_fini(rdev);
+			return r;
+		}
+		r = radeon_bo_kmap(rdev->wb.wb_obj, (void **)&rdev->wb.wb);
 		radeon_bo_unreserve(rdev->wb.wb_obj);
-		dev_warn(rdev->dev, "(%d) pin WB bo failed\n", r);
-		radeon_wb_fini(rdev);
-		return r;
-	}
-	r = radeon_bo_kmap(rdev->wb.wb_obj, (void **)&rdev->wb.wb);
-	radeon_bo_unreserve(rdev->wb.wb_obj);
-	if (r) {
-		dev_warn(rdev->dev, "(%d) map WB bo failed\n", r);
-		radeon_wb_fini(rdev);
-		return r;
+		if (r) {
+			dev_warn(rdev->dev, "(%d) map WB bo failed\n", r);
+			radeon_wb_fini(rdev);
+			return r;
+		}
 	}
 
 	/* clear wb memory */
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 5b937df..ddb8f8e 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -63,7 +63,9 @@ static void radeon_fence_write(struct radeon_device *rdev, u32 seq, int ring)
 {
 	struct radeon_fence_driver *drv = &rdev->fence_drv[ring];
 	if (likely(rdev->wb.enabled || !drv->scratch_reg)) {
-		*drv->cpu_addr = cpu_to_le32(seq);
+		if (drv->cpu_addr) {
+			*drv->cpu_addr = cpu_to_le32(seq);
+		}
 	} else {
 		WREG32(drv->scratch_reg, seq);
 	}
@@ -84,7 +86,11 @@ static u32 radeon_fence_read(struct radeon_device *rdev, int ring)
 	u32 seq = 0;
 
 	if (likely(rdev->wb.enabled || !drv->scratch_reg)) {
-		seq = le32_to_cpu(*drv->cpu_addr);
+		if (drv->cpu_addr) {
+			seq = le32_to_cpu(*drv->cpu_addr);
+		} else {
+			seq = lower_32_bits(atomic64_read(&drv->last_seq));
+		}
 	} else {
 		seq = RREG32(drv->scratch_reg);
 	}
diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c
index 2c1341f..43ec4a4 100644
--- a/drivers/gpu/drm/radeon/radeon_gart.c
+++ b/drivers/gpu/drm/radeon/radeon_gart.c
@@ -1197,11 +1197,13 @@ int radeon_vm_bo_update_pte(struct radeon_device *rdev,
 int radeon_vm_bo_rmv(struct radeon_device *rdev,
 		     struct radeon_bo_va *bo_va)
 {
-	int r;
+	int r = 0;
 
 	mutex_lock(&rdev->vm_manager.lock);
 	mutex_lock(&bo_va->vm->mutex);
-	r = radeon_vm_bo_update_pte(rdev, bo_va->vm, bo_va->bo, NULL);
+	if (bo_va->soffset) {
+		r = radeon_vm_bo_update_pte(rdev, bo_va->vm, bo_va->bo, NULL);
+	}
 	mutex_unlock(&rdev->vm_manager.lock);
 	list_del(&bo_va->vm_list);
 	mutex_unlock(&bo_va->vm->mutex);
diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c
index e17faa7..82434018 100644
--- a/drivers/gpu/drm/radeon/radeon_ring.c
+++ b/drivers/gpu/drm/radeon/radeon_ring.c
@@ -402,6 +402,13 @@ int radeon_ring_alloc(struct radeon_device *rdev, struct radeon_ring *ring, unsi
 		return -ENOMEM;
 	/* Align requested size with padding so unlock_commit can
 	 * pad safely */
+	radeon_ring_free_size(rdev, ring);
+	if (ring->ring_free_dw == (ring->ring_size / 4)) {
+		/* This is an empty ring update lockup info to avoid
+		 * false positive.
+		 */
+		radeon_ring_lockup_update(ring);
+	}
 	ndw = (ndw + ring->align_mask) & ~ring->align_mask;
 	while (ndw > (ring->ring_free_dw - 1)) {
 		radeon_ring_free_size(rdev, ring);
diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c b/drivers/gpu/drm/radeon/radeon_uvd.c
index 906e5c0..cad735d 100644
--- a/drivers/gpu/drm/radeon/radeon_uvd.c
+++ b/drivers/gpu/drm/radeon/radeon_uvd.c
@@ -159,7 +159,17 @@ int radeon_uvd_suspend(struct radeon_device *rdev)
 	if (!r) {
 		radeon_bo_kunmap(rdev->uvd.vcpu_bo);
 		radeon_bo_unpin(rdev->uvd.vcpu_bo);
+		rdev->uvd.cpu_addr = NULL;
+		if (!radeon_bo_pin(rdev->uvd.vcpu_bo, RADEON_GEM_DOMAIN_CPU, NULL)) {
+			radeon_bo_kmap(rdev->uvd.vcpu_bo, &rdev->uvd.cpu_addr);
+		}
 		radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+
+		if (rdev->uvd.cpu_addr) {
+			radeon_fence_driver_start_ring(rdev, R600_RING_TYPE_UVD_INDEX);
+		} else {
+			rdev->fence_drv[R600_RING_TYPE_UVD_INDEX].cpu_addr = NULL;
+		}
 	}
 	return r;
 }
@@ -178,6 +188,10 @@ int radeon_uvd_resume(struct radeon_device *rdev)
 		return r;
 	}
 
+	/* Have been pin in cpu unmap unpin */
+	radeon_bo_kunmap(rdev->uvd.vcpu_bo);
+	radeon_bo_unpin(rdev->uvd.vcpu_bo);
+
 	r = radeon_bo_pin(rdev->uvd.vcpu_bo, RADEON_GEM_DOMAIN_VRAM,
 			  &rdev->uvd.gpu_addr);
 	if (r) {
@@ -613,19 +627,19 @@ int radeon_uvd_get_create_msg(struct radeon_device *rdev, int ring,
 	}
 
 	/* stitch together an UVD create msg */
-	msg[0] = 0x00000de4;
-	msg[1] = 0x00000000;
-	msg[2] = handle;
-	msg[3] = 0x00000000;
-	msg[4] = 0x00000000;
-	msg[5] = 0x00000000;
-	msg[6] = 0x00000000;
-	msg[7] = 0x00000780;
-	msg[8] = 0x00000440;
-	msg[9] = 0x00000000;
-	msg[10] = 0x01b37000;
+	msg[0] = cpu_to_le32(0x00000de4);
+	msg[1] = cpu_to_le32(0x00000000);
+	msg[2] = cpu_to_le32(handle);
+	msg[3] = cpu_to_le32(0x00000000);
+	msg[4] = cpu_to_le32(0x00000000);
+	msg[5] = cpu_to_le32(0x00000000);
+	msg[6] = cpu_to_le32(0x00000000);
+	msg[7] = cpu_to_le32(0x00000780);
+	msg[8] = cpu_to_le32(0x00000440);
+	msg[9] = cpu_to_le32(0x00000000);
+	msg[10] = cpu_to_le32(0x01b37000);
 	for (i = 11; i < 1024; ++i)
-		msg[i] = 0x0;
+		msg[i] = cpu_to_le32(0x0);
 
 	radeon_bo_kunmap(bo);
 	radeon_bo_unreserve(bo);
@@ -659,12 +673,12 @@ int radeon_uvd_get_destroy_msg(struct radeon_device *rdev, int ring,
 	}
 
 	/* stitch together an UVD destroy msg */
-	msg[0] = 0x00000de4;
-	msg[1] = 0x00000002;
-	msg[2] = handle;
-	msg[3] = 0x00000000;
+	msg[0] = cpu_to_le32(0x00000de4);
+	msg[1] = cpu_to_le32(0x00000002);
+	msg[2] = cpu_to_le32(handle);
+	msg[3] = cpu_to_le32(0x00000000);
 	for (i = 4; i < 1024; ++i)
-		msg[i] = 0x0;
+		msg[i] = cpu_to_le32(0x0);
 
 	radeon_bo_kunmap(bo);
 	radeon_bo_unreserve(bo);