WebSVN – Kolibri OS – Diff – /drivers/video/drm/radeon/radeon_vce.c

 /* 1 second timeout */
 #define VCE_IDLE_TIMEOUT_MS	1000
 /* Firmware Names */
+#define FIRMWARE_TAHITI	"radeon/TAHITI_vce.bin"
-/* Firmware Names */
+#define FIRMWARE_BONAIRE	"radeon/BONAIRE_vce.bin"
 #define FIRMWARE_BONAIRE	"radeon/BONAIRE_vce.bin"
 MODULE_FIRMWARE(FIRMWARE_TAHITI);
 	int i, r;
 	INIT_DELAYED_WORK(&rdev->vce.idle_work, radeon_vce_idle_work_handler);
+	switch (rdev->family) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+	case CHIP_ARUBA:
+		fw_name = FIRMWARE_TAHITI;
 		break;
 	switch (rdev->family) {
 	case CHIP_BONAIRE:
 	case CHIP_KAVERI:
 	case CHIP_KABINI:
 		 start, mid, end, rdev->vce.fb_version);
 	rdev->vce.fw_version = (start << 24) | (mid << 16) | (end << 8);
+	/* we can only work with this fw version for now */
+	if ((rdev->vce.fw_version != ((40 << 24) | (2 << 16) | (2 << 8))) &&
-	/* we can only work with this fw version for now */
+	    (rdev->vce.fw_version != ((50 << 24) | (0 << 16) | (1 << 8))) &&
-	if (rdev->vce.fw_version != ((40 << 24) | (2 << 16) | (2 << 8)))
+	    (rdev->vce.fw_version != ((50 << 24) | (1 << 16) | (2 << 8))))
+		return -EINVAL;
-		return -EINVAL;
+	/* allocate firmware, stack and heap BO */
-	/* allocate firmware, stack and heap BO */
+	if (rdev->family < CHIP_BONAIRE)
 		size = vce_v1_0_bo_size(rdev);
-	size = RADEON_GPU_PAGE_ALIGN(rdev->vce_fw->size) +
+	else
-	       RADEON_VCE_STACK_SIZE + RADEON_VCE_HEAP_SIZE;
+		size = vce_v2_0_bo_size(rdev);
 	r = radeon_bo_create(rdev, size, PAGE_SIZE, true,
 		radeon_bo_unreserve(rdev->vce.vcpu_bo);
 		dev_err(rdev->dev, "(%d) VCE map failed\n", r);
 		return r;
+	}
+	memset(cpu_addr, 0, radeon_bo_size(rdev->vce.vcpu_bo));
+	if (rdev->family < CHIP_BONAIRE)
+		r = vce_v1_0_load_fw(rdev, cpu_addr);
 	else
-	memcpy(cpu_addr, rdev->vce_fw->data, rdev->vce_fw->size);
+		memcpy(cpu_addr, rdev->vce_fw->data, rdev->vce_fw->size);
 	radeon_bo_kunmap(rdev->vce.vcpu_bo);
 	radeon_bo_unreserve(rdev->vce.vcpu_bo);
-	return 0;
+	return r;
 	dummy = ib.gpu_addr + 1024;
 	/* stitch together an VCE create msg */
 	ib.length_dw = 0;
-	ib.ptr[ib.length_dw++] = 0x0000000c; /* len */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x0000000c); /* len */
-	ib.ptr[ib.length_dw++] = 0x00000001; /* session cmd */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000001); /* session cmd */
-	ib.ptr[ib.length_dw++] = handle;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(handle);
-	ib.ptr[ib.length_dw++] = 0x00000030; /* len */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000030); /* len */
-	ib.ptr[ib.length_dw++] = 0x01000001; /* create cmd */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x01000001); /* create cmd */
-	ib.ptr[ib.length_dw++] = 0x00000000;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000000);
-	ib.ptr[ib.length_dw++] = 0x00000042;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000042);
-	ib.ptr[ib.length_dw++] = 0x0000000a;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x0000000a);
-	ib.ptr[ib.length_dw++] = 0x00000001;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000001);
-	ib.ptr[ib.length_dw++] = 0x00000080;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000080);
-	ib.ptr[ib.length_dw++] = 0x00000060;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000060);
-	ib.ptr[ib.length_dw++] = 0x00000100;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000100);
-	ib.ptr[ib.length_dw++] = 0x00000100;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000100);
-	ib.ptr[ib.length_dw++] = 0x0000000c;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x0000000c);
-	ib.ptr[ib.length_dw++] = 0x00000000;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000000);
-	ib.ptr[ib.length_dw++] = 0x00000014; /* len */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000014); /* len */
-	ib.ptr[ib.length_dw++] = 0x05000005; /* feedback buffer */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x05000005); /* feedback buffer */
-	ib.ptr[ib.length_dw++] = upper_32_bits(dummy);
+	ib.ptr[ib.length_dw++] = cpu_to_le32(upper_32_bits(dummy));
-	ib.ptr[ib.length_dw++] = dummy;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(dummy);
-	ib.ptr[ib.length_dw++] = 0x00000001;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000001);
 	for (i = ib.length_dw; i < ib_size_dw; ++i)
-		ib.ptr[i] = 0x0;
+		ib.ptr[i] = cpu_to_le32(0x0);
 	dummy = ib.gpu_addr + 1024;
 	/* stitch together an VCE destroy msg */
 	ib.length_dw = 0;
-	ib.ptr[ib.length_dw++] = 0x0000000c; /* len */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x0000000c); /* len */
-	ib.ptr[ib.length_dw++] = 0x00000001; /* session cmd */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000001); /* session cmd */
-	ib.ptr[ib.length_dw++] = handle;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(handle);
-	ib.ptr[ib.length_dw++] = 0x00000014; /* len */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000014); /* len */
-	ib.ptr[ib.length_dw++] = 0x05000005; /* feedback buffer */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x05000005); /* feedback buffer */
-	ib.ptr[ib.length_dw++] = upper_32_bits(dummy);
+	ib.ptr[ib.length_dw++] = cpu_to_le32(upper_32_bits(dummy));
-	ib.ptr[ib.length_dw++] = dummy;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(dummy);
-	ib.ptr[ib.length_dw++] = 0x00000001;
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000001);
-	ib.ptr[ib.length_dw++] = 0x00000008; /* len */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x00000008); /* len */
-	ib.ptr[ib.length_dw++] = 0x02000001; /* destroy cmd */
+	ib.ptr[ib.length_dw++] = cpu_to_le32(0x02000001); /* destroy cmd */
 	for (i = ib.length_dw; i < ib_size_dw; ++i)
-		ib.ptr[i] = 0x0;
+		ib.ptr[i] = cpu_to_le32(0x0);
 /**
  * radeon_vce_validate_handle - validate stream handle
+ *
  * @p: parser context
  * @handle: handle to validate
+ * @allocated: allocated a new handle?
+ *
  * Validates the handle and return the found session index or -EINVAL
  * we we don't have another free session index.
  */
-int radeon_vce_validate_handle(struct radeon_cs_parser *p, uint32_t handle)
+static int radeon_vce_validate_handle(struct radeon_cs_parser *p,
+				      uint32_t handle, bool *allocated)
+{
 	unsigned i;
+	*allocated = false;
 	/* validate the handle */
 	for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) {
+		if (atomic_read(&p->rdev->vce.handles[i]) == handle) {
+			if (p->rdev->vce.filp[i] != p->filp) {
+				DRM_ERROR("VCE handle collision detected!\n");
+				return -EINVAL;
 		if (atomic_read(&p->rdev->vce.handles[i]) == handle)
 			return i;
+		}
+	}
 	/* handle not found try to alloc a new one */
 	for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) {
 		if (!atomic_cmpxchg(&p->rdev->vce.handles[i], 0, handle)) {
+			p->rdev->vce.filp[i] = p->filp;
-			p->rdev->vce.filp[i] = p->filp;
+			p->rdev->vce.img_size[i] = 0;
-			p->rdev->vce.img_size[i] = 0;
+			*allocated = true;
 			return i;
+		}
+ *
  */
 int radeon_vce_cs_parse(struct radeon_cs_parser *p)
+{
 	int session_idx = -1;
-	bool destroyed = false;
+	bool destroyed = false, created = false, allocated = false;
 	uint32_t tmp, handle = 0;
 	uint32_t *size = &tmp;
-	int i, r;
+	int i, r = 0;
 	while (p->idx < p->chunk_ib->length_dw) {
 		uint32_t len = radeon_get_ib_value(p, p->idx);
 		uint32_t cmd = radeon_get_ib_value(p, p->idx + 1);
 		if ((len < 8) || (len & 3)) {
+			DRM_ERROR("invalid VCE command length (%d)!\n", len);
-			DRM_ERROR("invalid VCE command length (%d)!\n", len);
+			r = -EINVAL;
-                	return -EINVAL;
+			goto out;
+		}
+		if (destroyed) {
-		if (destroyed) {
+			DRM_ERROR("No other command allowed after destroy!\n");
-			DRM_ERROR("No other command allowed after destroy!\n");
+			r = -EINVAL;
-			return -EINVAL;
+			goto out;
+		}
+		switch (cmd) {
-		switch (cmd) {
+		case 0x00000001: // session
-		case 0x00000001: // session
+			handle = radeon_get_ib_value(p, p->idx + 2);
-			handle = radeon_get_ib_value(p, p->idx + 2);
+			session_idx = radeon_vce_validate_handle(p, handle,
-			session_idx = radeon_vce_validate_handle(p, handle);
+								 &allocated);
 			if (session_idx < 0)
 				return session_idx;
 			size = &p->rdev->vce.img_size[session_idx];
+			break;
+		case 0x00000002: // task info
+			break;
+		case 0x01000001: // create
+			created = true;
-			break;
+			if (!allocated) {
 				DRM_ERROR("Handle already in use!\n");
-		case 0x00000002: // task info
+				r = -EINVAL;
-			break;
+				goto out;
+			}
 		case 0x01000001: // create
 			*size = radeon_get_ib_value(p, p->idx + 8) *
 				radeon_get_ib_value(p, p->idx + 10) *
 * 3 / 2;
+			break;
 			break;
 		case 0x04000001: // config extension
-		case 0x04000001: // config extension
+		case 0x04000002: // pic control
-		case 0x04000002: // pic control
+		case 0x04000005: // rate control
-		case 0x04000005: // rate control
+		case 0x04000007: // motion estimation
-		case 0x04000007: // motion estimation
+		case 0x04000008: // rdo
-		case 0x04000008: // rdo
+		case 0x04000009: // vui
 			break;
 		case 0x03000001: // encode
 			r = radeon_vce_cs_reloc(p, p->idx + 10, p->idx + 9,
 						*size);
 			if (r)
-				return r;
+				goto out;
 			r = radeon_vce_cs_reloc(p, p->idx + 12, p->idx + 11,
 						*size / 3);
 			if (r)
-				return r;
+				goto out;
 			break;
 		case 0x02000001: // destroy
 			destroyed = true;
 			break;
 		case 0x05000001: // context buffer
 			r = radeon_vce_cs_reloc(p, p->idx + 3, p->idx + 2,
 						*size * 2);
 			if (r)
-				return r;
+				goto out;
 			break;
 		case 0x05000004: // video bitstream buffer
 			tmp = radeon_get_ib_value(p, p->idx + 4);
 			r = radeon_vce_cs_reloc(p, p->idx + 3, p->idx + 2,
 						tmp);
+			if (r)
-			if (r)
+				goto out;
-				return r;
+			break;
 			break;
+		case 0x05000005: // feedback buffer
+			r = radeon_vce_cs_reloc(p, p->idx + 3, p->idx + 2,
-		case 0x05000005: // feedback buffer
+);
-			r = radeon_vce_cs_reloc(p, p->idx + 3, p->idx + 2,
+			if (r)
-);
+				goto out;
-			if (r)
+			break;
+		default:
+			DRM_ERROR("invalid VCE command (0x%x)!\n", cmd);
+			r = -EINVAL;
+			goto out;
+		}
-				return r;
+		if (session_idx == -1) {
+			DRM_ERROR("no session command at start of IB\n");
+			r = -EINVAL;
-			break;
+			goto out;
+		}
 		default:
-			DRM_ERROR("invalid VCE command (0x%x)!\n", cmd);
+		p->idx += len / 4;
 			return -EINVAL;
 	if (allocated && !created) {
-		if (session_idx == -1) {
+		DRM_ERROR("New session without create command!\n");
 			       struct radeon_semaphore *semaphore,
 			       bool emit_wait)
+{
 	uint64_t addr = semaphore->gpu_addr;
-	radeon_ring_write(ring, VCE_CMD_SEMAPHORE);
+	radeon_ring_write(ring, cpu_to_le32(VCE_CMD_SEMAPHORE));
-	radeon_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+	radeon_ring_write(ring, cpu_to_le32((addr >> 3) & 0x000FFFFF));
-	radeon_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+	radeon_ring_write(ring, cpu_to_le32((addr >> 23) & 0x000FFFFF));
-	radeon_ring_write(ring, 0x01003000 | (emit_wait ? 1 : 0));
+	radeon_ring_write(ring, cpu_to_le32(0x01003000 | (emit_wait ? 1 : 0)));
 	if (!emit_wait)
-		radeon_ring_write(ring, VCE_CMD_END);
+		radeon_ring_write(ring, cpu_to_le32(VCE_CMD_END));
 	return true;
+ *
  */
 void radeon_vce_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
+{
 	struct radeon_ring *ring = &rdev->ring[ib->ring];
-	radeon_ring_write(ring, VCE_CMD_IB);
+	radeon_ring_write(ring, cpu_to_le32(VCE_CMD_IB));
-	radeon_ring_write(ring, ib->gpu_addr);
+	radeon_ring_write(ring, cpu_to_le32(ib->gpu_addr));
-	radeon_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	radeon_ring_write(ring, cpu_to_le32(upper_32_bits(ib->gpu_addr)));
-	radeon_ring_write(ring, ib->length_dw);
+	radeon_ring_write(ring, cpu_to_le32(ib->length_dw));
+}
 /**
  * radeon_vce_fence_emit - add a fence command to the ring
 			   struct radeon_fence *fence)
+{
 	struct radeon_ring *ring = &rdev->ring[fence->ring];
 	uint64_t addr = rdev->fence_drv[fence->ring].gpu_addr;
-	radeon_ring_write(ring, VCE_CMD_FENCE);
+	radeon_ring_write(ring, cpu_to_le32(VCE_CMD_FENCE));
-	radeon_ring_write(ring, addr);
+	radeon_ring_write(ring, cpu_to_le32(addr));
-	radeon_ring_write(ring, upper_32_bits(addr));
+	radeon_ring_write(ring, cpu_to_le32(upper_32_bits(addr)));
-	radeon_ring_write(ring, fence->seq);
+	radeon_ring_write(ring, cpu_to_le32(fence->seq));
-	radeon_ring_write(ring, VCE_CMD_TRAP);
+	radeon_ring_write(ring, cpu_to_le32(VCE_CMD_TRAP));
-	radeon_ring_write(ring, VCE_CMD_END);
+	radeon_ring_write(ring, cpu_to_le32(VCE_CMD_END));
+}
 /**
 	if (r) {
 		DRM_ERROR("radeon: vce failed to lock ring %d (%d).\n",
 			  ring->idx, r);
 		return r;
+	}
-	radeon_ring_write(ring, VCE_CMD_END);
+	radeon_ring_write(ring, cpu_to_le32(VCE_CMD_END));
 	radeon_ring_unlock_commit(rdev, ring, false);
 	for (i = 0; i < rdev->usec_timeout; i++) {
 	        if (vce_v1_0_get_rptr(rdev, ring) != rptr)

Subversion Repositories Kolibri OS

(root)/drivers/video/drm/radeon/radeon_vce.c – Rev 5271 → 6104